After obtaining a degree in Mechanical Engineering, followed by a PhD for work on the stress analysis of discontinuous pipe bends, Graham joined TWI in 1975.
He worked in the Fatigue Department on a wide range of projects, including many associated with the fatigue design of offshore structures. In 1985 he joined the Edison Welding Institute as Manager of Engineering and was responsible for setting up the Engineering Department and establishing mechanical testing facilities in the laboratories in Columbus, Ohio.
After returning to Abington, he was appointed Manager of the Marketing and Membership Group. This group covers membership development, member services, marketing and information, and corporate communications.
Fatigue cracking represents one of the most common modes of service failure in welded fabrications. This is still true despite many advances in our understanding of fatigue cracking in welded joints and the availability of proven fatigue design rules. Graham Wylde takes a look at where and why fatigue cracks occur, and how to repair them.
In general, the occurrence of fatigue cracking can be attributed to a number of factors:
- Underestimation of service stresses;
- Underestimation of the number of applied stress cycles;
- Unforeseen sources of cyclic loading;
- Overworking of a structure;
- Imperfections in the fabricated joint;
- Lack of appreciation of the need to design against fatigue failure.
If the design life of a cracked fabrication has not been reached, there will be an incentive to attempt to repair it and thereby extend its working life - whatever the cause of cracking. Weld repairs can sometimes be carried out effectively, providing a useful extension in life. In other cases, a modification to the design may be necessary to obtain any benefit. In either case, care will be necessary to ensure that the repair is adequately carried out.
Why is fatigue cracking likely to occur at a welded joint?
Two major factors enable us to appreciate why welded joints are likely to suffer from fatigue cracking problems:
- The presence of stress concentrations;
- The presence of slag intrusions at weld toes.
As load is applied to a welded component there will be a general flow of stress through the section. If one considers the stress flow through a typical joint it is easy to appreciate from the flow lines that there will be a bunching of the stress contours, giving rise to a stress concentration, at the weld toe and the weld root (Figure 1). Figure 2 shows points of stress concentration in a range of joint types.
Welded steel joints contain discontinuities at weld toes in the form of slag intrusions, sharp pieces of slag which penetrate into the plate surface by up to 0.5mm. The precise origin of these slag intrusions is uncertain, but it is thought that they may arise from inclusions in the parent material which rise to the weld toe as the molten weld pool solidifies.
The presence of sharp discontinuities at the point of maximum stress concentration can result in a development of fatigue cracking at relatively low levels of applied stress and after only a very few cycles of load. Indeed, it is normally assumed that the crack initiation period can be ignored for a welded joint and that the total fatigue life is occupied in crack propagation. This contrasts with the fatigue mechanism in an unwelded component where the majority of the total fatigue life will be occupied in the initiation of the crack.
Where do fatigue cracks occur?
Fatigue cracks may propagate from defects in poorly made welds. For welded joints which have been soundly made there are two principal locations for fatigue cracking: the toe of the weld and the weld root.
Failure from weld defects
Fatigue cracking may initiate at defects in the weld and propagate through it at right angles to the applied stress. The most serious types of defect are planar, i.e. lack of penetration, lack of fusion and cracks, and those which either break, or are close to the surface.
Failures from the weld toe
Fatigue cracking is usually associated with the weld toe in the case of a joint stressed in the transverse direction, or with the weld end in the case of a joint stressed longitudinally. Figures 3 and 4 show typical examples of these initiation sites. In general, the crack will initiate at the very edge of the weld metal and will propagate first through weld metal, then through the heat affected zone and finally through parent material as shown in Figure 5. Frequently the crack will propagate away into the plate remote from the welded joint at right angles to the direction of the applied cyclic stress.
Failure from the weld root
In load-carrying fillet or incomplete penetration butt welds, if the weld is incorrectly proportioned, either through bad design or through faulty manufacture, the stress across the weld throat may be sufficient to initiate a failure from the weld root. Usually, such cracks will propagate through the weld metal and break the surface near the centre of the weld face (Figure 6). In some circumstances, especially when the plate has poor through-thickness properties, a crack initiating at the root may open up a lamination and first become visible near the weld toe.
Detection of fatigue cracking
From the appearance of a fatigue crack when seen only from the surface it is not always possible to determine whether the crack has initiated from an internal defect or from the weld toe or root. For this reason it is often desirable to investigate the cracking in more detail before attempting a repair. For example, a boat sample could be taken to include either part or all of the crack. Examination of the fracture surfaces and metallographic sectioning will normally enable the crack initiation site to be identified.
In engineering structures, fatigue cracks are seldom detected while they are small. In the majority of cases the crack will have propagated to a significant length, resulting either in a change of stiffness of the component or in complete failure of the section. The reason for this is that fatigue cracks tend to be relatively small and tight for the majority of their life. Moreover, weld toes tend to form traps for oil, grease and dirt which can cover cracks and hinder detection. Fatigue cracks can also grow to a significant degree under a paint coating prior to breaking the surface. They are therefore unlikely to be detected at an early stage by casual visual inspection of a component.
If one knows where to look for a crack perhaps because of similar cracking in other structures - detection becomes relatively simple. If the weld area is cleaned with a wire brush, fatigue cracks may be detected using either magnetic particle inspection or dye penetrant.
Both of these methods are capable of detecting fatigue cracks less than 10mm long, although magnetic particle inspection can give a spurious linear indication at a weld toe simply as a result of the weld profile. In such circumstances a light burr grinding of the toe can be used to establish whether the indications are indeed the result of fatigue cracks.
Neither of these techniques will provide an indication of crack depth, the more important dimension for the assessment of load carrying capacity. If necessary, crack depth may be estimated using the alternating current potential drop technique.
Repair of fatigue cracks
Each repair situation must be treated individually by establishing a suitable repair procedure compatible with the materials involved and the size of the repair required. If the crack has not substantially weakened the component, making a weld repair impracticable, a repair may be effected by removing the crack - either by grinding or gouging - and repair welding.
If gouging is chosen, the wall of the gouge should be ground smooth. Before repair welding, one of the crack detection techniques, preferably magnetic particle inspection, should be used to confirm that the crack has been fully removed.
When the preparation is satisfactory, the weld repair may be carried out. In choosing a suitable welding procedure, give careful consideration to all the circumstances for the repair, including the additional restraint often imposed on a weld in a repair. Make sure also that any pre-heat requirements are specified.
If a crack has penetrated through the full thickness of the plate, every effort should be made to effect the repair from both sides. Take care when removing the crack not to leave an excessive root gap. A balanced procedure should be adopted. The root of the weld should be ground back from the second side and sealing runs made from this side to complete the repair.
When there is no access to one side, a small area of fatigue crack may be left at the bottom of the excavation to act as a land for the root of the repair. When making the repair it will be necessary to fuse through this land completely to remove the fatigue crack, and it is recommended that trials be carried out to determine the depth of the land suitable for the repair welding technique selected. Any small area of fatigue cracking not fully removed during a repair will certainly reduce the fatigue life of the repair weld.
After completing the repair, the toes of the repair weld should be ground. As fatigue damage may have been incurred by weld metal adjacent to the original fatigue crack although these areas are not actually seen to be cracked - it will be beneficial if the grinding is extended beyond the region of the repair weld.
Weld repairs must often be made in more arduous conditions than those under which the original weld was made. There is therefore a high risk that defects may be introduced while making the repair. Consequently it is recommended that consideration be given to ultrasonic or radiographic inspection of the completed repair to establish its integrity. The significance of any defects found in the repair weld should then be assessed on a fitness-for-purpose basis.
Likely performance of repair welds
After a satisfactory repair the structure may be returned to service. However, unless the source of cyclic loading can be removed for example by the use of dampers to eliminate the effects of vibration or resonance - and assuming that the original weld was properly manufactured, there will be a strong possibility that further fatigue cracking may occur.
Tests on transverse and longitudinal fillet welds containing repaired fatigue cracks have highlighted two particular problems with repaired joints. First, if the pre-existing crack is not completely removed when making the repair, the repaired joint is likely to have an exceptionally low fatigue strength, far below that of the original joint. If the crack is fully removed and a sound repair made, the fatigue strength may be similar to that of the original joint. However, there will be a risk of fatigue cracking from adjacent weld joints which will have accumulated cycles during the initial period of service but not actually developed a visible crack. When the part is returned to service, cracking may occur from such details after a relatively short time.
For these reasons it may be desirable to reduce the fatigue loading on the repaired component. This may be achieved by derating the structure to reduce the maximum stress range, or by reducing the duty cycle. if neither of these options is available, consideration should be given to reducing the applied stress in the areas of concern by providing an alternative load path. Regular routine inspection of the repaired structure is also strongly recommended.
If the original failure was from a defect in the weld, or because the weld was under size, it is probable that a sound, adequately-proportioned repair will be sufficient to prevent further instances of fatigue cracking.
Conclusions
- Fatigue cracking in a fabricated structure is likely to occur at the welded joints because they possess an inherently low fatigue strength in relation to nonwelded components.
- The most common sites for fatigue cracking in sound welded joints are the weld toe, the weld end and the weld root.
- Fatigue cracks may escape detection until extensive damage has occurred because of their location and relatively narrow crack opening.
- Repair of fatigue cracks may be possible by grinding and re-welding but care must be taken to ensure that the crack is completely removed and that the repair weld does not contain any defects.
- Further fatigue cracking may occur from a repair or from adjacent details in a structure unless the fatigue loading on the area containing the repair can be reduced.
