Peter Tubby is Head of the Structural Assessment Section in the Engineering Department. After graduating in metallurgy from the University of Manchester in 1975 he was employed by Lucas Industries at their Research Centre in Solihull researching the fatigue performance of materials and components for aerospace and automotive applications, and the mechanical properties of engineering ceramics.
Since joining TWI in 1979 he has been involved in many aspects of the fatigue design, analysis and testing of welded structures and has published a number of technical papers in these areas. His particular research interests include the behaviour of joints under variable amplitude loading with special reference to the recording and analysis of service loads, the effects of multiaxial stressing on the fatigue performance of joints, and repair of fatigue cracks.
A full weld repair is frequently the solution when fatigue cracks are found in offshore structures, but what performance can be expected from the repaired joint? Peter Tubby reports.
Fatigue cracking in offshore structures as a consequence of severe wave and wind loadings is not uncommon, especially in the North Sea. While considerable care is taken at the design stage to avoid fatigue cracking at tubular nodes, this is not always achieved, and TWI has been involved in many assessments of fatigue damage to and repair of both fixed and mobile offshore units.
In view of the strict confidentiality that most oil companies apply to instances of cracking, the problem may in fact be more widespread than is currently appreciated. Furthermore, as many structures in the North Sea move into middle age it is likely that the problem will be encountered more frequently in future.
Although the majority of structures can function for a time with fatigue cracks present without performance or safety being impaired, cracked joints must be repaired eventually. Offshore repairs are invariably difficult and expensive to perform and may result in significant downtime, but the cost will generally be less than that of replacing the structure.
Repair methods include repair welding and use of mechanical or grouted clamps which provide an alternative load path. The choice of method depends on many factors such as water depth for submerged members, ease of access, member size and thickness and suitable weather windows.
Repair welding underwater entails added complexity because a purpose-built habitat is needed in which diver welders operate in dry conditions. Nevertheless, a full weld repair is the best long term solution in most instances, and at current prices, for a submerged member, this is likely to cost several million pounds.
In deciding the most appropriate repair method an important consideration is the fatigue performance of the repaired member during the remaining service life and, surprisingly, this has received little attention in the past.
To fill this gap TWI conducted an extensive Group Sponsored Project, completed in 1987, in which tubular joints were repaired by a number of alternative methods, and the fatigue performance was compared with that of unrepaired joints.
The findings have been published recently by the Department of Energy in their Offshore Technology Report Series. [1]
This article summarises the results for joints repaired by welding; a second article will describe two alternative repair methods, namely hole drilling, which attempts to retard a growing crack by introducing holes at the crack tips, and 'remedial grinding' in which part wall cracks are removed by grinding and the resulting excavations left unrepaired.
Testing
The tubular specimens were T joints, see Fig.1, fabricated by manual metal arc welding in a C-Mn structural steel, and fatigue loaded in the out-of-plane bending mode, which produced cracking at the weld toe in the chord member, see Fig.2. The strategy of the programme was to grow fatigue cracks in the joint under constant amplitude cyclic loading (in air), carry out a weld repair and then continue fatigue cycling at the same load range, thus simulating the situation for a repair in a real structure. The life after repair was then compared with that to produce the original cracking. Two crack depths were studied (approximately 50% and 100% of the chord wall thickness.)
Repairs were made by excavating the crack using a combination of disc and burr grinding, followed by dye penetrant inspection to ensure complete crack removal, and finally rewelding using MMA. To avoid having to bridge wide gaps in through-thickness repairs, a ligament of 1-2mm thickness was left unexcavated at the root to provide a suitable root face. The majority of repairs were left as-welded, but in a number of specimens the repair welds were burr ground to improve their fatigue strength, see Fig.3. All joints were extensively instrumented with foil resistance strain gauges to establish the principal stress distribution before and after repair. (Stress distributions in tubular joints are complex. In fatigue experiments a standard method of interpretation is adopted [2] to evaluate a characteristic stress adjacent to the failure location - the 'hot spot' stress. Fatigue results and design curves for tubular joints are plotted in terms of this stress parameter.)
Results and discussion
Thirteen repair welded joints were investigated: eight as-repaired and five burr ground. The fatigue results are shown in Fig.4 in terms of hot spot stress range versus endurance before and after repair. Figure 4a also includes fatigue precracking data for a further twelve identical specimens which were subsequently repaired by the other two methods outlined in the introduction.
The results are expressed in terms of both the first visually detected cracking (N 1,) and through-thickness cracking (N 2). (In Fig.4a some of the N 2 datapoints represent final crack depths of 25 or 50% of the wall thickness, as indicated.)
S-N curves from Reference 2 are plotted for comparison, namely the mean and mean minus two standard deviations curves for through-thickness cracking in joints of 16mm chord thickness, and the appropriate design curve for tubular joints, known as the T curve.
Since the tests were conducted under fully reversed loading (R=-1), cracks occurred on both sides of the brace giving two valid N 1 datapoints for each joint, which are plotted at the hot spot stress range measured at the appropriate point. In most joints only one crack broke through-thickness, hence only one N 2 value is plotted for each specimen.
The N 2 data for unrepaired joints compare favourably with the published mean line for joints of 16mm thickness, indicating that the fatigue characteristics of the joints were consistent with the established database.
Failure in as-welded repairs occurred predominantly at the repair weld toe in the chord, while in ground repairs, cracks developed in the ground weld face; three of the ground joints did not fail after either two million or five million cycles and the tests were ended.
On the basis of the N 2 data of Fig.4a and 4b it can be concluded that, where a joint is subjected to the same load range after repair, the life will be broadly similar to that before repair.
No significant difference was observed between the results for through-thickness and part-wall repairs. This is an interesting observation since in general the form of the root penetration bead in through-thickness repairs was poor and, more significantly, in some cases part of the fatigue precrack remained unfused. Figure 5 shows an example where failure occurred at the repair weld toe in the chord despite the presence of an unfused crack at the root, indicating that for the type of joint studied, weld root quality was not critical in the repair.
Grinding the repair weld markedly improved the life after repair, see Fig.4b, and, in view of the small additional effort involved, is strongly recommended. It is recommended also to grind unrepaired areas of the joint since these may have accumulated fatigue damage not detectable at the time of the repair.
Conclusions and recommendations
- When fatigue loading on the structure continues with the same range, the usable life (to through-thickness cracking) of a sound repair weld, on average, will be similar to that of the original joint.
- Burr grinding significantly improves the life after repair and is strongly recommended, especially in view of the small additional cost. The entire joint, including the weld toes and face of both the original and repair welds, should be treated.
- In nodal connections, where the stress on the outside is much higher than that inside, the root quality of through-thickness repairs is not critical. Defects should, of course, be assessed for fitness for purpose, but it is unlikely that small root flaws will impair fatigue performance.
References
- Tubby PJ: 'Fatigue performance of repaired tubular joints'. OTH 89 307, Department of Energy, HMSO, London, 1989.
- 'Background to new fatigue guidance for steel welded joints in offshore structures'. Department of Energy, HMSO, London, 1984.
