Friction Stir Weld Integrity and the Rolling Stock Industry

Friction Stir Weld Integrity and its Importance to the Rolling Stock Industry

Blignault C¹, Kallee S W¹, Thomas W M¹, Russell M J¹

Author details

¹Friction and Forge Processes Department, TWI, Ltd., UK.

Paper presented at Southern African Institute of Welding (SAIW) conference, Integrity of Welded Structures in the Energy, Processing and Transport Industries in Southern Africa, Gold Reef City, 28-29 May 2008.

Abstract

Since the invention of friction stir welding (FSW) at TWI in 1991, companies from all parts of the world have implemented the process, predominantly in the fabrication of aluminium components and panels. Friction stir welded structures are now revolutionising the way in which trains and trams are built. Recent rail accidents such as the Ladbroke Grove rail collision in the UK have led to further investigations into the crashworthiness of aluminium rail cars. The results of these investigations have clearly demonstrated the importance of using appropriate joint designs and welding procedures.

This paper summarises the benefits of FSW for the fabrication of rolling stock and also discusses the key factors which influence the quality and mechanical properties of a friction stir weld. With the increasing demand of quality standards during production of rail cars, it is important to have a clear understanding of the typical imperfections that are associated with the selected joining method. Examples of flaws and defects are defined and summarised for some typical FSW joint designs. Finally, this paper introduces the potential of a new 'through-hole' impact testing technique by which surface imperfections, localised metallurgical notches, and root defects of friction stir welded joints can be evaluated. This technique is particularly useful over Charpy, Izod or the more sophisticated Crack Tip Opening Displacement (CTOD) test methods, because the latter cannot accurately test surface or root imperfections in a friction stir welded sample. Comparative data obtained from small scale impact performance tests between metal inert gas (MIG) and FSW joints are also presented in context with crashworthiness of rail vehicles.

1 Introduction

European rail accidents such as the high-speed train disaster in Eschede, Germany, in June 1998 and the Ladbroke Grove accident in Britain in October 1999 are two examples of accidents which have highlighted the need to further improve crashworthiness of rail vehicles. In the event of impact the carriage crumple zone absorbs a significant amount of the crash energy, it is important however that the part of the rail vehicle containing passengers remains substantially intact. Premature failure of welds in the rail vehicle would not be desirable. It was for this reason that industry wished to understand the root causes of certain aluminium welds 'unzipping', a phenomenon reported in a number of recent rail accidents. This called for further investigation into alternatives to fusion welding, the use of grades of aluminium that are less susceptible to fusion weld weakening, and further developments in structural joint design.

Further exploration to identify alternative welding techniques recognised that an inherent feature of all fusion welded joints in aged hardened aluminium alloys is the considerable strength reduction in the weld region compared with parent material. Unless adequate consideration is given to this point at the design and build stage, fracture may occur during accidents in the vicinity of the weld, which could compromise the crashworthiness of the welded structure. Friction stir welding (FSW) was adopted by rolling stock manufacturers and suppliers (such as Hitachi,^[1] Nippon Sharyo^[2] and Bombardier^[3]) as an alternative welding technique for rail carriage structures. If used correctly, the FSW process has significant potential to reduce the width of heat affected zone (HAZ) and the degree of thermal softening experienced by the weld region. Another significant driver for the uptake of this patented process is its combination of cost effectiveness and good weld performance. A further benefit is that the heat input during the FSW process is relatively low compared to MIG welding, therefore reducing the overall level of component distortion. This stems from the fact that the FSW process operates below the melting point of the material to be joined.

Different aluminium grades and joint designs are also being considered in addition to alternative joining technologies. In principle, there are many aluminium alloys to choose from; however, modern aluminium railcars are commonly made from complex double-skinned extrusions so good formability is required. The need for high strength and excellent extrudability limits the choice to those alloys which do soften on heating (welding). This prompted work to explore the extent of softening, depending on the alloy used and manufacturing process adopted. This paper discusses friction stir weld properties, weld integrity, their relevance to the manufacturing of rolling stock, and provides typical examples of the use of FSW in industry.

2 Applications of fsw in the rolling stock industry

2.1 Factors to consider during fabrication of rail vehicles

Many joining techniques are commonly used for rail vehicles, including metal inert gas (MIG), friction stir welding (FSW), resistance spot welding, and bolting. Increasing interest is also being shown in laser-arc hybrid welding. Whatever the joining method, many factors are taken into account in the fabrication of rail vehicles, including:

2.1.1 Cost

The initial cost factor remains a major consideration but, increasingly, ways of reducing through life costs are being adopted. Reducing the anticipated through life maintenance bill is crucial, whether the vehicles are being purchased outright or leased. The adoption of new manufacturing methods, such as advanced welding technologies, can have a significant impact on both initial investment and recurring production costs.

2.1.2 Robust design

Manufacturers will always fabricate designs which meet the standards of the times. In recent years more attention has been given to crashworthiness. Modern designs must meet existing crashworthiness requirements and, as understanding of the key issues grows, designs will increasingly reflect the next generation of standards. Other factors that influence structural integrity and component assembly procedures are the method of joining, itsreliability/repeatability, and the distortion characteristics associated with that method.

2.1.3 End of life

The rail sector, like all other industry sectors, is increasingly turning its attention to end of life dismantling and recycling issues. For example, in vehicles designed from stainless steel there is great benefit in avoiding a mixture of materials, so as to simplify the end of life dismantling and maximise the value of the scrap.

2.1.4 European standards

Manufacturers must be alert to the fact that regulations covering the fabrication of rail vehicles are changing. The process of welding has always required a high level of skill and control to ensure that welds are fit for purpose.This recognises the fact that it can be difficult or expensive to check all welds once they are made, while at the same time a poor quality weld could have disastrous implications in terms of safety. European standards such as prEN15085-2 on 'Railway applications - welding of railway vehicles and components' are therefore adopting an approach which is based on two standards already used in the German rail sector, DIN 6700 and DIN 18800. The new approach will require fabricators to demonstrate compliance with EN 719 and EN 729. Together, these standards cover requirements for the welders and the manufacturing facility. Compliance with these standards must be demonstrated by independent third party assessment. These changes are familiar to manufacturers who are already supplying into the German market. All suppliers to the European market will, in future, be required to engage with an assessor organisation, with necessary skills and accreditation to do this work.

2.1.5 Selection of material and joint designs

Fully heat treated 6xxx series aluminium alloys (typically with 12-14% elongation properties) are commonly recommended for use in rail vehicle structures. These alloys have excellent buckling properties and often 180° folds around very small radii can be achieved in aluminium sheet, with no signs of cracking. A fine grain structure is an advantage, however the thermal cycles experienced during welding will lead to a reduction in mechanical properties, and a change in grain structure. Weld seams and heat affected zones are often the weakest areas, however good design can overcome potential problems. The use of longitudinal aluminium extrusions with integral stiffeners is attractive for rail car design as the wavelength of buckling can be dictated by the spacing of stiffeners; hence energy absorption can be very high in such aluminium structures. Use of light weight aluminium alloys is also advantageous in terms of crash performance, as heavy vehicles have more kinetic energy, which will result in higher impact forces and a requirement to absorb more energy on impact. Light-weighing is also important to reduce fuel usage/environmental impact, and to protect the track at high speeds.

2.2 European railway industry

The Scandinavian aluminium extruders Sapa and Hydro Marine Aluminium were the first in Europe to commercially apply the friction stir welding process for the manufacture of single-wall aluminium roof panels for rolling stock applications. Since 1997, Alstom LHB in Germany has purchased such FSW prefabricated panels for Copenhagen suburban trains (Figure 1 and 2). Since early 2001 they have also used friction stir welded aluminium side walls and since 2002 FSW floor panels for Munich suburban trains. These panels are made by Sapa in Sweden.

Fig.1. Alstom LHB trains for DSB Danish State Railways during production. FSW roof panels for these trains are made at Hydro Marine Aluminium under a contract with Sapa

Fig.2. Friction stir welded roof panel produced at Hydro Marine Aluminium for Sapa for delivery to Alstom LHB (Germany)

In March 1999, Alstom LHB engineers considered the friction stir welding of hollow aluminium profiles for making floor and side panels. Successful FSW experiments were conducted in up to 23mm thick aluminium plates, to demonstrate how MIG welds could be replaced in the underframe area of rolling stock. They are now assembling friction stir welded roof panels for train carriages as shown in Figure 3. These panels are made by Hydro Marine Aluminium and are typically 3m x 14.5m long.

Fig.3. Alstom LHB assembling friction stir welded roof panels on rail carriages

Bombardier in Derby (United Kingdom) has also carried out FSW experiments for butt and lap welds and has worked closely with TWI to assess these welded joints. Friction stir welding is now being used to build replacement stock for the London Underground system. No fewer than 376 friction stir welded vehicles have been ordered from the TWI Industrial Member Bombardier for the latest Victoria Line upgrade.

Up to 16m long FSW machines have been designed, built, and commissioned by a number of international machine manufacturers. One of them has been installed at Sapa and is used for the production of large panels and heavy profiles with a welding length of up to 14.5m and a maximum width of 3m (see Figure 4). This machine has three welding heads, which means that it is possible to weld from two sides of the panel at the same time, or to use two welding heads (positioned on the same side of the panel) starting at the centre of the workpiece and welding in opposite directions.

Fig.4. Sapa in Finspång (Sweden) installed an ESAB SuperStir^TM machine with three heads, which can be used simultaneously to increase the throughput

A number of European companies have also requested the provision of job shop services and low cost feasibility studies, to reduce the risks of capital investment and R&Dmp;D efforts. Some of them proposed teaming-up in a collaborative project, which has the overall objective to accelerate the use of friction stir welding in Europe. A EuroStir^® project was launched in December 2000 and it was part-funded by EUREKA, which is apan-European initiative for promoting collaborative research in advanced technology. Part of this programme was to perform small-scale impact tests on both FSW and MIG welded specimens. ^[3]

Further work was also conducted between Bombardier Transportation UK and TWI Ltd during 2005 to test the performance of full-scale aluminium components under high-speed impact. General results obtained from this work are discussed in Section 4.4.

2.3 Rolling stock in Japan

Hitachi was amongst the first train manufacturers to recognise the technical and economic benefits of FSW. They use a double-skinned design for their vehicles, which are constructed from hollow aluminium extrusions joined along their length by FSW. Hitachi comment positively on the low distortion and excellent mechanical properties of friction stir welds compared to MIG welds. Hitachi has delivered friction stir welded vehicles for both commuter and express trains as illustrated in Figures 5 and 6.

Fig.5. Hitachi uses FSW for assembling the new Channel Tunnel Rail Link trains

Fig.6. Commuter train built by Hitachi with full length friction stir welds of double-skin side and roof panels (welded from one side)

Hitachi use a number of different FSW joint designs, some of these are illustrated in Figures 7 and 8. Hitachi claims that by using FSW they typically experience only one twelfth of the distortion when compared to MIG welding.

Fig.7. Hitachi extrusion designs used in manufacture of Series 885 Tilting EMU for JR-Kyushu. Overall panel thickness 40mm

Fig.8. Section of typical FSW made on aluminum extrusion

Kawasaki Heavy Industries (KHI) is using friction stir spot welding (FSSW) to attach stringers to roof panels (see Figures 9 and 10) for the new Fastech 360Z train. They developed a new aluminium car body design, which is assembled by this method.^[4] KHI report a number of benefits for the FSW approach, not least the fact that it improves the flatness and visual appearance of the skin panels because of the lowheat input.

Fig.9. The FSSW robot from Kawasaki

Fig.10. Kawasaki uses FSSW for making aluminium roof panels for the Fastech 360Z ^[4]

Nippon Sharyo has been using friction stir welded panels produced by Sumitomo Light Metal Industries^[5] for the floor panels of the new Shinkansen (Figure 11 and 12). Some of these trains operate at speeds up to 285km/hour. Nippon Light Metals have also made use of friction stir welding for subway rolling stock. By 1998 they reported that over 3km of welds had been produced.^[6] The weld quality was confirmed to be excellent based on microstructural, X-ray and tensile test results.

Fig.11. Trainsets with FSW floor panels of Sumitomo Light Metal operate on the Shinkanen in Japan

Fig.12. Friction stir welded floor panel produced by Sumitomo Light Metal for Shinkansen trains

The use of friction stir welding for the fabrication of aluminium rolling stock continues to grow around the world and the use of a double-skinned extrusion design for the construction of rail vehicles is prevalent.

3 Friction stir weld integrity

3.1 Characterising flaws and defects in FSW

The solid-state nature of FSW immediately leads to several advantages over fusion welding methods since any problems associated with cooling from the liquid phase are avoided. Issues such as porosity, solute redistribution, solidification cracking and liquation cracking are not an issue during FSW. In general, FSW has been found to produce a very low concentration of defects and is very tolerant to variations in parameters and materials; however some distinct defects or flaws can exist if welding parameters or welding procedures are incorrect.

There is often some confusion between the use of the terms flaw and defect. These definitions can be defined as follows:

A defect is defined as an imperfection that has been shown to compromise the integrity of the structure and its presence is, therefore, intolerable. A flaw is defined as an imperfection whose significance has not been established, and which could possibly be tolerated in the structure.

There are two main categories of flaws in friction stir welds; volumetric flaws and joint line flaws. This has been further extended in the draft ISO standard (ISO/DIS 25239-01) which is still under review and subject to comment. The ISO standard makes use of the term 'imperfection', which takes both flaws and defects into account.

Some flaws may be acceptable to design codes, in which case they can be termed acceptable flaws. Others, depending upon size and severity, may be unacceptable, in which case the flaw can be termed a defect. Typical flaws that can be found in FSW are categorised as follows:

Fig.13. Categories of FSW flaws ^[11]

3.2 Typical flaws associated with FSW

3.2.1 Volumetric flaws (voids)

Flaws in a friction stir weld due to a lack of material consolidation are termed volumetric flaws, or voids. These terms are often qualified by descriptors such as buried, surface breaking, continuous, clustered etc. It is incorrect to describe these voids as porosity. Porosity is caused by gas bubbles or by shrinkage at the time of solidification and neither of these processes occurs in friction stir welding. Voids may be caused by inadequate material flow either due to insufficient tool features, or the selection of an excessive welding speed. Voids may also be caused by inadequate consolidation of the softened material due to a reduced forging pressure, inadequate material clamping (plates separate during welding) or the presence of a significant gap along the joint line because of poor fit up. Examples of typical voids are shown in Figures 14 and 15.

Fig.14. Macro section showing typical void and incomplete penetration in a friction stir weld ^[10,11]

Fig.15. Micro section of friction stir weld in 6mm AA5754-H111 highlighting typical imperfections such as voids, lack of penetration (LOP) and a joint line remnant (JLR)

3.2.2 Joint line flaws

In FSW the original joint-line can sometimes still be traced and can be present as a joint line remnant, as shown in Figure 15. These artefacts are sometimes referred to as 'kissing bonds'. Detailed examination shows that this type of flaw usually consists of semi-continuous oxide particles distributed around the original joint-line. The severity of this imperfection depends upon the extent of the oxide particles and these flaws may have a significant effect on mechanical performance. However, substantial metal to metal bonding can take place between the oxide particles, and experience has shown that failures in bend, tensile and fatigue testing may not follow the path of the oxide particles in some cases. Thus, the presence of a joint line remnant should not automatically be considered a defect.

In butt welds the most serious joint line flaws are usually those located at the weld root, commonly called lack of penetration flaws, as shown in Figure 15. The extreme condition is a complete lack of bond, due normally to inadequate tool probe length. Flaws in the root can also be caused by poor control of tool position/force, local variations in plate thickness, or by local cooling caused by the heat sink effect of the welded component, or the tool-bed. This type of flaw can be identified by a root bend test or from a metallographic section, and in extreme cases may also be detected by visual examination of the root of the weld.

Overall the FSW process is very robust and can produce sound welds in all common grades of aluminium alloys. The 6xxx series aluminium alloys can be considered to be the least difficult to friction stir weld and are more tolerable to variations in tool design and parameter settings. This is mainly due to its alloy composition and good formability/extrudability characteristics at elevated temperatures. Weld imperfections can readily be avoided when an appropriate tool design is used with the correct set of welding parameters, however these parameters must also be accurately controlled throughout the weld.

In lap welds another form of joint line flaw, termed 'Hook' in the draft ISO standard, may exist. An example of this type of flaw is shown in Figure 16. Due to the asymmetric nature of the FSW process these Hook flaws may exist on either the advancing side (AS) or retreating side (RS) of a lap weld. Fatigue failures can be expected to initiate at this notch on either side of the weld and therefore careful consideration must be given to joint design, tool and welding procedures used.

Fig.16. Macro section showing typical hook flaw in a FSW lap weld

It is fundamentally important to understand the notch morphology associated with friction stir lap welds. In conventional tooling the Hook feature typically 'curls' upwards, as shown in Figure 16. Recent improvements in tool designs and motions such as Skew-stir^TM and Com-stir^TM have illustrated that this direction can be reversed or minimised.^[7] The advantage of doing this is that an appropriate lap weld joint design can then be used to minimise the impact of any flaws that may occur in this area. Figure 17 illustrates two lap weld configurations where the direction of traverse and tool rotation determines the advancing or retreating side.

Fig.17. Lap weld joint configurations
a) Advancing side near the top sheet edge (ANE)
b) Retreating side near the top sheet edge (RNE)

During tensile testing of lap welded specimens it is expected that the samples will twist in directions similar to that indicated in Figure 18. This is due to the joint configuration and method of loading. The bending stresses will add to test severity, regardless of whether the test is being used to generate static (strength) or dynamic (fatigue) data.

Fig.18. Regions of tensile stress for simple lap welds

a) Advancing side near the top sheet edge (ANE) b) Retreating side near the top edge (RNE)

Figure 19 illustrates the fractured specimen of a friction stir welded fatigue sample. This weld was made with a Skew-Stir^TM tool using 6mm thick AA5083-H111. Figures 19b and 19d show that the Hook feature was curling downwards (towards the bottom plate) and that its direction did not contribute to the location of failure.

Hook features on the corners of the FSW nugget can assist failure in the maximum stressed regions, depending on the test direction and direction to which the Hook feature is curling. For example, if the Hook feature is curling upwards (towards the top plate) on the AS and RS then it is evident that this may contribute towards mechanical failure, but vice versa, this can be prevented if the direction of the Hook is reversed.

Fig.19. Fatigue tested lap weld (RNE configuration) made using Skew-stir ^TM technology ^[8] a) Macrosection. (A millimetre scale is shown);

b) Detail of fracture, bottom sheet side, with the notch tip arrowed;

c) Detail of fracture top sheet advancing side;

d) Detail of the form of the notch at the edge of the weld - advancing side, with the notch tip arrowed

From these Hook features it is important to note the difference between the ANE and RNE conditions and that both of these conditions must be tested and taken into consideration during the procedure development. The joint geometry and strength requirements must be designed accordingly to take this effect into account when the origin of the Hook flaw for a specific tool and parameter set is not clearly understood.

Generally the quality of lap welds can be more difficult to optimise than butt welds, but good quality lap welds can readily be reproduced when the correct tool design, welding procedure and joint configuration is implemented. Some initial development work and preliminary testing of the welded joint is normally recommended to verify weld quality and the orientation of the Hook feature.

4 Assessment of friction stir welded joints

4.1 Overview of commonly applied test methods

Butt welded joints may be tested to generate tensile, fatigue, fracture mechanics, shear, bend, hardness, impact and many other forms of data. Metallurgical examination under a light microscope normally also forms part of the typical weld assessment procedure to verify weld quality. Tensile and fatigue testing may be transverse or longitudinal with respect to the weld. Specimens may also be extracted from certain regions to generate specific data. The term'joint efficiency' is readily used to relate mechanical joint strength to parent material. Joint efficiency is defined as the ratio between the ultimate tensile strength of the joint and the ultimate tensile strength of the base material, expressed as a percentage. The ultimate tensile strength of the base metal must be obtained in the same direction (i.e. longitudinal or transverse to the plate rolling direction) in which the joint is tested.

Testing to generate fracture mechanics data, such as crack growth rates, is commonly performed using the compact tension specimen. A notch is introduced which may be aligned with any direction of interest to generate region-specific data. Fracture toughness is a critical input parameter for fracture-mechanics based on fitness-for-service assessments. Although fracture toughness can sometimes be obtained from the literature or materials properties databases, it is preferable to determine this by experiment for the particular material and joint being assessed.

Ultimate shear strength (UTS) data can be generated by single shear testing, using a specimen that contains side notches to guide fracture along a shear path. Bend testing is a relatively inexpensive method that is used to provide qualitative information about longitudinal and transverse joint ductility. Lap and spot joints may be tested in tension-shear, to generate strength or fatigue data. Metallurgical examination under a light microscope can relatively easy detect volumetric defects (voids) and lack of weld penetration. Sections through the weld are normally made and the surface etched with Keller's reagent to identify various regions of interest.

Impact data is typically generated using the Charpy (V-notch) specimen but root or surface defects cannot be accurately assessed with this method. Other test methods such as the Izod and more sophisticated Crack Tip Opening Displacement (CTOD) test methods can also not directly assess these defects because the notch cut into the surface of the specimen eliminates the region of interest. TWI has therefore developed a new technique to evaluate surface and root regions of friction stir welded specimens. ^[8]

4.2 The new 'through-hole' impact test method

The 'through-hole' impact testing technique can be used to evaluate how the fracture toughness of friction stir welded specimens is influenced by surface imperfections, localised notches and root defects. The feature that characterises the 'through-hole' impact test is that specimen weakness is achieved by a hole in the neutral axis perpendicular to the impact direction (see Figure 20).

Fig.20. Section of FSW sample with Ø3mm precision reamed hole

The use of a precision reamed hole instead of a 'V', 'U' or 'keyhole' notch means that the characteristics of different types of surface layers on the substrate material can be evaluated. Moreover, comparisons can be made of the effect that partial penetration; lack of penetration or other comparable root or surface imperfections has on the structural integrity of the weld or component.

Preliminary results reported by Thomas et al.^[8] suggest that this impact test method can be useful for evaluating 'lack of penetration' defects in FSW. Figure 21 shows the fractured 'through-hole' impact test specimens (tested at ambient conditions) from parent material, bead on plate, full penetration and lack of penetration FSW samples.

Fig.21. 'Through-hole' impact tests on 6082-T6 samplesa) Parent material results that gave an average of 31 Joules and showed little bending had occurred;

b) Full penetration bead on plate weld specimens that gave an average of 60 Joules and bent further than the parent material samples;

c) Welds with lack of penetration that gave an average of 25 Joules and showed the least amount of bending and did not fracture through the hole;

d) Full penetration welds that gave an average of 52 Joules and bent further than the parent material or welds with lack of penetration and bent almost as much as the 'bead-on-plate' welds

In the lack of penetration samples, the un-bonded region, as shown in Figure 22, guided the fracture path towards the HAZ region on the advancing side. Despite a reduction in the cross-section area in the weld region(resulting from the presence of the hole), failure occurred in the HAZ close to the extensively deformed un-bonded region. Compared with impact tests taken from the full penetration weld, the impact energy of partial penetration welds was reduced by 48%. Even when compared with similar impact tests taken from the parent material the impact energy of partial penetration welds was reduced by 20%. Compared with impact tests taken from parent material the impact energy of the 'bead-on-plate' test samples gave 93% improvement.

Fig.22. Macrosection showing lack of penetration and plastic deformation in the root region

From the preliminary results discussed above it is evident that the 'through-hole' impact testing method can be valuable for the assessment of friction stir welded samples.

4.3 General status of NDT for FSW

FSW technology requires a quality assurance system especially when it is used for critical component manufacture. At present there are no non-destructive testing (NDT) inspection codes that apply to FSW or any established inspection techniques, although some individual companies have developed their own quality assurance (QA) procedures and specifications. The lack of an established NDT method for FSW makes it more difficult to implement on high integrity applications. TWI has conducted some research in this area, including a European sponsored project 'Qualistir'^[9] and a Group Sponsored Project^[10] to investigate the capability of ultrasonic and electromagnetic NDT methods for FSW.

Both ultrasonic and eddy-current inspection methods have inherent limitations in this application. Eddy-current techniques are generally best for thinner material, while ultrasonic techniques provide a volumetric inspection and are best suited to thicker sections. Based upon this reasoning electromagnetic techniques have generally been applied to material less than 3mm thick and ultrasonic techniques are normally applied to material greater than 3mm thick. However, because electromagnetic techniques have proved to be less than fully effective at detecting sub-surface flaws on thin material, ultrasonic trials can be extended to evaluate welding joints on thin material, with some success.

One of the most promising general techniques for NDT of friction stir welds is phased array ultrasonic inspection. The main benefits of this technique have been described by Caravaca et al,^[11] and can be summarised as follows:

• Improved flaw detection and sizing capability.
• Pictorial presentation of data.
• Shorter inspection times through electronic scanning.
• Permanently recorded results.
• Opportunity for post acquisition data manipulation (e.g. statistical analysis).

One important restriction to inspection sensitivity is imposed by the surface finish of the component under test. The rougher the surface finish of the component, or the more uneven the surface, the greater the interference with the ultrasound signal, and in turn, the less sensitive the inspection.

In general, the body of opinion is that only open flaws, such as volumetric flaws or complete lack of bonding can be reliably detected. However, previous work by Bird et al at TWI^[12] concentrated on the detection of joint line remnant flaws. No direct detection was possible but there was some success in using measurement of the signal pattern to indicate the quality of the weld.

The main conclusions drawn from this work were:

• With carefully designed inspection systems, volumetric flaws can be readily detected in friction stir welds.
• Joint line remnants can be indirectly detected in welds by measurement of material grain noise in the root of the weld. These measurements can also be correlated with the position of the welding tool within the material.
• The phased array ultrasonic NDT method is able to detect lack of penetration defects down to around 0.3mm, which have shown to cause a reduction in the welds fatigue performance.
• Further work is necessary on the detection of very small lack of penetration defects, where phased array ultrasonic testing may not be fully effective.

4.4 Crashworthiness testing of FSW and MIG welded joints

TWI managed a project on crash performance of FSW and MIG welded structures, which was funded by The Rail Safety and Standards Board, Angel Trains Ltd and HSBC Rail (UK) Ltd^[3]. Part of this programme was to undertake and evaluate small-scale static and dynamic tests to compare weld impact and tensile strengths between FSW and MIG welded samples. The aluminium alloys used for this evaluation were 6005-T6 and 6082-T6. Both alloys were procured in two forms, 3mm thick flat rolled sheet or extruded strip, and extruded box section with a wall thickness of 3mm.

During this work it was found that conventional drop weight impact tests were not appropriate for testing welded specimens in rapid strain tensile loading. A special test was developed and validated for small-scale tests that successfully forced the weld region into tension, as it is thought occurred in the Ladbroke Grove accident. The idea was to try and simulate the welds 'unzipping'. It was however recommended that large-scale tests be performed to validate the results of this preliminary work.

The main conclusions drawn from this work are summarised below:

• Friction stir welds have a narrow ductile heat affected zone surrounding the weld whilst MIG welds are surrounded by a wider, softer region. The narrowest MIG weld HAZ was wider than those of any of the friction stir welds tested.
• Friction stir welded specimens tested in tension had higher proof and ultimate stress values than comparative MIG welds. All fractures occurred in the heat softened regions around the weld.
• Full scale or large-scale testing is required to establish a true comparison between MIG and FSW joints.

Investigations into crashworthiness of rail vehicles have since progressed by doing full-scale impact testing. TWI has worked with a number of rolling stock manufacturers to investigate fast weld fracture similar to that observed in rail vehicle accidents. Welded full-scale aluminium joints in double skin extrusions of AA6005-AT6, AA6008-T6 and AA6008-T7 alloys have been produced using conventional MIG, hybrid laser-MIG and friction stir welding processes. A widerange of welded extrusion specimens have been impact tested in specially designed and constructed test rigs, which simulate crash conditions.

The full scale welded joint results confirmed once again that welded joints in 6xxx series aluminium alloys will inevitably fracture in the weld heat affected zone. The basic cause for such fracture is the inevitable strength reduction in welds produced by the currently established fusion and non-fusion welding processes in heat treatable aluminium alloys. It is therefore recommended that the joint area of the welding region be optimised to increase its load carrying capacity enough for fracture to occur in the parent material. Minimum ratios of weld zone to parent material thickness have been established during work at TWI, but due to confidentiality, these exact values cannot be disclosed.

5 Summary

5.1 Applications of FSW in the rolling stock industry

• FSW has found commercial application in the manufacture of aluminium rolling stock around the world and several machine manufacturers can provide suitable welding machines.
• Crashworthiness considerations for the manufacturing of rail vehicles now play a more important part in the design.
• The standards in Europe relating to the fabrication of rail vehicles are changing and manufacturers will need to be able to demonstrate that their welders, welding facilities and systems have been approved to the new standards.
• The use of a double-skinned design for the construction of rail vehicles is prevalent. These are constructed from hollow aluminium extrusions joined along their length by FSW. Positive feed-back from industry has indicated that FSW results in accurate tolerance, low distortion and excellent mechanical properties when compared to MIG welding.

5.2 Friction stir weld integrity

• The solid-state nature of FSW immediately leads to several advantages over fusion welding methods since any problems associated with cooling from the liquid phase are immediately avoided.
• FSW is commonly a very robust process where good quality welds can be reliably achieved.
• The most common flaws that can occur in FSW are either characterised as volumetric flaws (voids) or joint line flaws, including lack of weld penetration (LOP) or joint line remnants (JLR).
• It is fundamentally important to understand the notch morphology associated with friction stir lap welds. It is important to note the difference between the ANE and RNE conditions and to design accordingly.

5.3 Assessment of friction stir welded joints

• The 'through-hole' impact testing technique developed by TWI can be used to evaluate how the fracture toughness of friction stir welded specimens is influenced by surface imperfections, localised notches and root defects.
• Under laboratory conditions, phased array ultrasonic NDT methods are able to detect lack of penetration defects down to around 0.3mm, which have shown to cause a reduction in the welds fatigue performance. Further work is however necessary for the detection of very small lack of penetration defects, where phased array UT may not be fully effective.
• Full scale impact testing on FSW, MIG and hybrid laser-MIG welded double skin extrusions have been performed using 6xxx series alloys. From this work it was evident that weld over-sizing should be considered in the design of these structures to account for the reduction in properties that result in the welded area.

5.4 Concluding remarks

Wherever possible in critical areas, the geometry of a welded joint should be such that the weld and HAZ strength is matched to parent material strength. Where this is not possible, the joint should be designed to be stronger than the adjacent parent part. Alternatively, the joint can be reinforced with additional members attached across the welds.

Friction stir welding technology has been widely recognised for its ability to provide greatly improved weld properties and low distortion in a wide variety of aluminium structures. The technical and economic benefits of the FSW process have led to rapid expansion and wide scale use of the technology in many industrial applications.

6 Acknowledgments

The authors would like to acknowledge the contributions of the persons and companies who were involved with the respective work programmes, as noted in the text.

7 References

1. Ohba H., Ueda C., Agatsu K.: Innovative vehicle - The 'A' train, Hitachi Review Vol. 50, No.4, 2001. Available on website:
   www.hitachi.com/ICSFiles/afieldfile/2004/06/08/r2001_04_102.pdf
2. Sylva G., Edwards R., MTS Systems Corp., Sassa T., Nippon Sharyo Ltd.: A feasibility study for self reacting pin tool welding on thin section aluminium, Proceedings of 5th International Friction Stir Welding Symposium, Metz France, 14-16 Sept, 2004.
3. Nunn M., Betaso D., Xu W.: Comparison of Friction Stir and MIG welding - Preliminary small scale and dynamic tests, TWI Report No: 14357/1/04, June 2004.
   Available on website: www.rssb.co.uk/pdf/reports/research/T035_rpt_final.pdf
4. Koga S., Seto M., Fujimoto M., Ohkubo K., Inuzuka M.: Application of new welding and joining technologies, Kawasaki Technical Review, ISSN 0387-7906, No.164, May 2007, pp22-25.
5. Davenport J., Kallee S. W., Wylde J. G.: Europe follows Japan into friction stir welding, Railway Gazette International, November 2001, pp.777-780.
6. Hori H., Makita S., Hino H.: Friction Stir Welding of rolling stock for subway, Proceedings of 1st International Friction Stir Welding Symposium, 1000 Oaks, CA, USA, 14-16 June 1999.
7. Thomas W. M., Gittos M. F., Tubby P., Staines D. G.: Friction Skew-stirTM lap welding 5083-H111 aluminium alloy - process development and mechanical/fatigue evaluation, TWI Core Research Report No: 12624.01/2004/1215.3, January 2007.
8. Thomas W. M., Norris I. M., Staines D. J., Lucas W.: The evaluation of root defects in FSW by 'through-hole' impact testing - preliminary studies, TWI website www.twi.co.uk, published July 2005.
9. Qualistir^TM : The on-line control if FSW, EC Project No: CRAF 1999-70641.
10. Bird C. R: Development of non-destructive techniques for the detection of defects in friction stir welds, TWI Report No: 14732/4A/04, 2004.
11. Caravaca D. S., Bird C. R., Beamish K., Maddox S.: Correlation of phased array inspection and fatigue performance of FSW joints, 26th International Conference on Offshore Mechanics and Arctic Engineering, San Diego, Calirfornia, USA, June 10-15.
12. Kleiner D., Bird C. R.: Signal processing for quality assurance in friction stir welds, Insight Journal, Vol. 46 No 2, February.