Weld tool travel speed effects on fatigue life of friction stir welds in 5083 aluminium

https://doi.org/10.1016/S0142-1123(03)00061-6Get rights and content

Abstract

This paper reports the results of a study into the influence of weld tool travel speed (in the range 80–200 mm/min) on the occurrence of ‘onion-skin’ forging-type defects (similar to the root defects known as ‘kissing bonds’) in single pass friction stir (SP FS) welds, and on the effect of these defects on fatigue crack initiation and overall life. Results indicate that such defects are generally not associated with fatigue crack initiation, but may act to reduce fatigue life by providing easy linking paths between two fatigue cracks. It is likely that their influence on fracture toughness of SP FS welds would be higher, as they occur more readily when growth rates and levels of plastic deformation are higher.

Introduction

Friction stir welding (FSW) is a relatively new solid-state joining technique, which offers high joint quality and good fatigue performance. Other advantages include little, or no, joint preparation, relatively few defects and little requirement for post-weld dressing. In FSW, a cylindrical, shouldered tool with a profiled probe, or pin (slightly smaller in length than the plate thickness), is rotated (typically at around 500 rpm) and slowly plunged into the joint line between two abutting pieces of plate or sheet that are to be joined. The parts must be clamped to prevent the joint faces being forced apart. Adiabatic shear and friction generate heat between the wear resistant tool and the plate. The material softens and flows around the tool as it is progressed along the joint line. Forging occurs under the pressure applied by the tool shoulder, which is inclined at a slight angle (around 2.5°) to the horizontal. The process approximates a solid-state keyhole welding process, in that a hole to accommodate the pin is generated, then filled as the weld is made.

The recrystallisation of the microstructure that occurs under the high levels of plastic strain induced in the weld, leads to a very fine-grained structure in the weld nugget, as shown in Fig. 1 for the 5383-H321 alloy. Typical FSW terminology is also indicated in the figure. It is useful to distinguish between advancing and retreating sides (advancing representing the side in which rotational direction and weld travel direction are additive) because of differing residual stresses and strains at these locations [1], which influences crack initiation and crack growth [2]. It is usual to refer to thermo-mechanically affected zones (TMAZs) in FSW, as well as heat-affected zones (HAZs).

As stated before, defect levels are generally low in FSW, compared with typical fusion welds. A number of types of defect are known to occur, however, and because they can occur in any orientation and at any angle, may be difficult to detect with directionally specific techniques such as radiography and ultrasonics [3]. The known defects in FS butt welds include lack of penetration (tool length too small for the plate thickness), voids and root defects, which are also known as ‘kissing bonds’. These occur when the root of a single pass (SP) weld achieves only partial bonding and some influences of their effect on fatigue strength are covered in another article in this issue of the journal [4]. The occurrence of kissing bonds appears to be alloy specific and in particular, in the limited range of alloys considered, 5038-H321 is known to be more susceptible to these defects than either 5083-O or 6082-T6 alloys [4]. The relative difficulty of detecting defects in FS welds makes it imperative to fully understand the influence of any defects on fatigue crack initiation and total life. As yet, however, there are few reported studies on the fatigue performance of FS welds in the open literature, and those that are reported are often preliminary in nature [5]. Thus, there is an absence of detailed information, particularly regarding internal defects, their origin and their influence on fatigue crack initiation and life.

Work in our laboratory on 5383-H321 and 5083-H321 aluminium alloys using SP and double pass (DP) welds, has indicated that occasional defect-like features do occur on the fracture surfaces of fatigue specimens. They may occur in groups, reminiscent of the ‘onion-skin’ rings in the weld nugget, may be up to several millimetres in length, and occur particularly in the case of SP welds. The fact that defects occur internally, apparently associated with artefacts of the plastic flow, is perhaps unsurprising considering the chaotic nature of the plastic deformation in the weld nugget [6]. Bendzsak et al. [6] in their paper on 3D modelling of FS welding, state that the chaotic flow region formed below the tool shoulder, on the advancing side of the weld, contains flow singularities. These singularity regions in a weld correspond with defect locations and are likely to be affected by tool rotational speed, travel speed during welding and geometry of the tool [6].

It is therefore expected that tool travel and rotational speeds would be prime parameters governing the occurrence of internal defects in FS welds. As these parameters are usually determined from empirical knowledge and trial runs on test plates, it would be advantageous to know the effect of variations in these parameters on the defect population and fatigue life. The effect of these parameters on fatigue life has been considered in another article in this issue of the journal [7], which studied FS welds in 4 mm plate of 6082-T6 and 6082-T4 aluminium alloys, and concluded that it had no major effect in that alloy. That study used tool travel speeds of 700 and 1400 mm/min, and rotation speeds of 2200 and 2500 rpm, respectively. Inspection of their SN data, however, indicates that an effect of weld speed on slope of the data is present, and that the data sets are too small to draw statistically meaningful conclusions.

The present paper reports the results of a study of the effect of tool travel speed on defect formation in 8 mm thick plates of 5083-H321 aluminium alloy, and their influence on fatigue crack initiation and life. Values of tool rotational and travel speeds used in the present study are significantly lower than those pertaining to the work of Ericsson and Sandström [7], the results obtained indicate a definite effect of travel speed on fatigue life.

In future work, it would be desirable to perform a more complete study of the influences of tool design, rotational and travel speeds on internal FSW defects. The initial work performed on SP FS welds had observed internal defects in specimens that had been subjected to four-point bending fatigue. The defects were not associated with crack initiation, but generally occurred towards the centre of the specimens, often in the region of fast fracture (Fig. 2). The origin of these defect-like features was not clear, but their appearance was usually consistent with partial forging along onion-skin layers in the weld nugget (Fig. 2(a)). Some defects could arise also from a fluid dynamics voiding effect (Fig. 2(b)), either associated with the singularities mentioned before, or possibly due to vortex generation behind the tool probe as it makes the weld.

The remit of this work was therefore to examine the effect of weld travel speed on the frequency of occurrence of internal FS weld defects, for a single tool rotation speed. The objective was to ascertain the influence of the defects on fatigue crack initiation and life in 5083-H321 aluminium alloy, which was known to be relatively susceptible to kissing bonds [4]. A secondary objective was to determine whether the defects became more like voids as tool travel speed increased. This was intended to shed some light on their origin, as being either vortex shedding or singularity induced. In this work specimens were loaded in tension, rather than bend, so as to uniformly stress the complete specimen cross-section, and hence be more likely to activate incipient defects as crack initiation sites.

Section snippets

Material and experimental conditions

Plates of rolled 5083-H321 aluminium alloy, 8 mm thick, 1000 mm long and 500 mm wide were used in this investigation. Two such plates were welded together to form the stock from which fatigue specimens were machined. Parent plate properties are given in Table 1, where σ0.2% is the 0.2% proof strength and σTS is the tensile strength. Alloy 5083 is a weldable, strain hardening structural alloy suitable for marine applications, and 8 mm is the common thickness used in shipbuilding. The H321

Fatigue (SN) data

Fig. 4, Fig. 5 show the SN results obtained, while Fig. 6, Fig. 7 provide representative macrographs of the fracture surfaces of as-welded and polished specimens for all four tool travel speeds, respectively. For the as-welded specimens, crack initiation is always associated with surface tool rotation marks, while for the case of polished specimens crack initiation reflects either slip band cracking or small internal voids.

The real area of interest to structures is the long life regime (Nf>106

Fractography

Examination of the fracture surfaces at low magnification, using stereo-binoculars, did not indicate that defects were playing any greater role in crack initiation in these tensile specimens than previously observed for the case of the bend specimens. Fracture surface features, consistent with incomplete forging along onion-skin layers, were observed in a number of specimens, but were almost never associated with fatigue crack initiation. They tended to occur either towards the end of the

Conclusions

This study of influence of tool travel speed on the fatigue strength and occurrence of onion-skin partial-forging defects in single pass friction stir welds in 5083-H321 aluminium alloy has yielded the following conclusions.

  • 1.

    Generally, there is a decrease in endurance limit stress (at 107 cycles) in both as-welded and polished specimens, as travel speed increases from 80 to 200 mm/min. The decrease has a maximum value of about 11% for polished and 19% for as-welded specimens. More in-depth

Acknowledgements

This work was supported by Corus Research, Development and Technology, and the assistance of S. Webster and Dr T.J. Hurd is gratefully acknowledged. The assistance of TWI with the welding is also acknowledged.

References (7)

  • P.J Webster et al.

    Synchrotron X-ray residual strain scanning of a friction stir weld

    Journal of Strain Analysis for Engineering Design

    (2001)
  • C Dalle Donne et al.

    Effect of weld imperfections and residual stresses on the fatigue crack propagation in friction stir welded joints

  • A Lamarre et al.

    Ultrasound phased array inspection technology for the evaluation of friction stir welds

There are more references available in the full text version of this article.

Cited by (114)

  • A review on aluminium metal matrix composites

    2020, Materials Today: Proceedings
View all citing articles on Scopus
View full text