R J Pargeter and D P Baxter
Paper presented at Corrosion 2009, Atlanta, Georgia, USA, 22-26 March 2009. Paper # 09090.
High strain, low cycle, fatigue at girth welds needs consideration for subsea flowlines carrying hot fluids which operate in 'lateral buckling' mode, which is becoming more common. With corrosive fluids, the concern at the weld root is corrosion fatigue, and weld caps generally operate in sea water, with cathodic polarisation.
Fatigue testing in production fluids (sweet and sour) and seawater for steel catenary riser design, has used frequencies which are not relevant to lateral buckling. Exploration of the effects of frequency, (low frequency commonly makes corrosion fatigue more severe) using fracture mechanics specimens, must allow for potential differences between deep crack behaviour, and the conditions pertaining for the majority of life. This may be particularly important under scaling conditions, and/or in the presence of inhibitor.
The production environment varies both with position and time. The most highly corrosive conditions may not be most aggressive for corrosion fatigue. However, the environment for project specific fatigue testing needs to be that at the time when straining is occurring, rather than the most conservative. For end of life, the critical combination of K and KISCC and/or KIH needs to be considered.
A practical approach based on recent TWI experience is presented.
Keywords: Fatigue, Corrosion, Lateral buckling, Pipeline, Girth weld, Sour, Seawater, Inhibitor, Fracture toughness.
Subsea flow lines designed to carry hot production fluids will suffer cyclic stressing due to thermal expansion and contraction associated with variation in flow during life (including shutdowns). Typical service variations are shown in Figure 1. A common design solution for such pipelines is to lay them in a configuration, and possibly with sliding saddle support in critical areas, to encourage controlled bending along their length, referred to as 'lateral buckling'. Such bending results in significant cyclic loading across girth welds on the extrados and intrados of each buckle, and the possibility of cyclic strains beyond the linear elastic limit needs to be considered in design. In many cases, modelling of material behaviour during repeated buckling indicates that although initial cycles involve straining to just beyond the elastic limit, this does not persist beyond say the first 10 or so cycles. A purely elastic analysis may therefore be justifiable, although the applied stress ranges will still be high, and the fatigue design process very challenging. Where welds are exposed to corrosive environment (such as produced fluids at the weld root, or seawater with cathodic polarisation at the weld cap) environmental or corrosion fatigue performance can be a critical factor in design.
Fig.1. Typical service temperature variations in a subsea flowline
A certain amount of work on corrosion fatigue in corrosive production fluids (both sweet and sour) has been performed in recent years for steel catenary riser design, [1-6] but the frequencies used (typically 0.2 - 1Hz) are not relevant to lateral buckling, and it is commonly observed that reducing frequency makes corrosion fatigue more severe. [3,7] Techniques for exploring the effects of frequency commonly involve growth rate testing, but there are possible differences between deep crack behaviour in such specimens, and the shallow crack behaviour for the majority of life.  This may be particularly important under scaling conditions, and/or in the presence of inhibitor.
A complicating feature of such flowlines is that the internal environment (ie temperature pressure and fluid chemistry) can vary both with position and time. Where project specific testing is to be performed, selection of an appropriate environment is difficult. It clearly needs to be conservative with regard to corrosion fatigue, but the most highly corrosive conditions may not be the most aggressive for corrosion fatigue.  Furthermore, the environmental conditions need to be those in the prevailing service environment at the time when straining is occurring and not based on severe environments commonly found in standardized test methods ( e.g. NACE TM0177).
The variation in environmental conditions is also relevant to definition of end of life. This is typically when a fatigue crack reaches a size sufficient to precipitate unstable crack growth. A measure of 'fracture toughness' determined in the appropriate environment and/or for hydrogen charged material needs to be considered.
For long established technologies, such as transmission gas pipelines in non-corrosive service, there is sufficient service experience to allow design guidance to be provided in codes and standards. For example, the effect of seawater and cathodic polarisation on the performance of sub-sea pipelines has been extensively researched, and there is furthermore extensive service experience. Thus, fatigue endurance curves and growth rate data can be found in standards such as BS7608, DNV-RP-C203 or BS7910, which are appropriate to many situations. The internal environments in pipelines are much more varied, however, and even though there has now been quite a lot of research on the corrosion fatigue behaviour of steel catenary riser (SCR) welds in particular, [1-6] the likelihood of finding existing data of direct relevance to a new project is low. Furthermore, even if data generated in an appropriate environment did exist, the relevance of typically low strain, high cycle data generated at around 0.2-1Hz would be questionable for a lateral buckling situation. Thus, design data almost invariably have to be generated for each project. When setting out to do this, careful consideration needs to be given to both engineering and environmental aspects of testing.
There are essentially two types of fatigue testing available, namely endurance testing, in which the number of cycles to failure from a weld detail representative of that anticipated in service is determined, and fatigue crack growth rate (FCGR) testing, in which the rate of advance (mm/cycle) of a growing fatigue crack is determined. In theory, there is a close link between the two, and endurance performance should be predictable using growth rate data, providing an initial flaw size can be postulated. Constant amplitude endurance tests in air usually exhibit a fatigue limit, which corresponds to the threshold observed in crack growth rate tests. With corrosion fatigue however, there may not be a fatigue limit, and the correspondence between growth rate and endurance is less reliable than in air. It is believed that this is largely because of the difference between corrosive conditions in a deep crack by comparison with those at an open surface, and the dominance of the early (shallow) crack growth, on total life which cannot easily be assessed with FCGR testing. [8,9] There may also be concerns over the efficacy of inhibitors down a deep crack. However, although inhibitor effects are operator dependent, subject to coverage issues - top to bottom of the line - and bearing in mind that the formulation and application may vary during the life of a project based on decisions not related to corrosion fatigue, it is arguable that inhibitors should not be included in project specific testing. Nevertheless, as mentioned above, unconservative results may be generated if conditions are too highly corrosive.  Differences in microstructure between material close to the weld root toe, and that further into the weld, may also result in apparent discrepancies, as material hardness has been shown to influence the rate of crack growth in sour environments. 
Thus, it could be argued that endurance data are essential for lateral buckling design, for example to ensure that incorrect deductions (regarding the influence of cyclic loading frequency for example) are not drawn solely from FCGR data. Nevertheless, there is a place for FCGR testing. In the first place, the technique known as 'frequency scanning', in which sequential blocks of growth are carried out at constant ΔK, but different cyclic frequencies, can be used to explore the effect of frequency over a much wider range, and in a more acceptable time frame, than corresponding endurance tests. Although there is generally a strong effect of frequency on corrosion fatigue crack growth rate, the crack growth rate per cycle is usually found to reach a plateau as the frequency is reduced, so that there is no additional increase as the frequency is reduced further, below a critical value. It should be noted that testing at frequencies higher than this value can be justified for SCR design for example, where they can be shown to be representative of service conditions. The frequencies involved in lateral buckling scenarios are so low, however, that an appropriate higher frequency cannot be identified in that case and the plateau frequency should be used for testing. It is prudent to confirm that the test frequency selected on the basis of FCGR frequency scanning is conservative for endurance testing with limited lower frequency endurance tests. So far, in the authors' experience, reducing the frequency below the plateau indicated by FCGR testing has not resulted in reduced life, so long as frequency scanning FCGR tests have been conducted at a relatively high value of ΔK, which appears to maximise the influence of frequency. This is illustrated in Figure 2
which shows frequency scanning FCGR data for C-Mn steel cathodically protected in seawater, 
at two values of applied ΔK. At high ΔK the onset of a plateau is delayed until a much lower frequency is reached.
Fig.2. Fatigue crack growth rate data, showing effect of frequency at two levels of ΔK
In the second place, FCGR data are also required when a fracture mechanics based engineering critical assessment (ECA) is to be carried out to determine defect acceptance criteria to be used during pipe lay. In this instance the flaw sizes considered may be large enough such that 'deep crack' FCGR data are appropriate.
A major problem with regards to lateral buckling is that environmental conditions vary throughout the loading cycle. The most significant variable will be temperature, and this poses a problem for testing, as it is difficult to predict the most severe condition with regard to corrosion fatigue. The issues involved are discussed in more detail in. [5,9,10] As a further complication, when a shutdown occurs, the product is typically replaced with an inert fluid such as diesel. While this should mitigate corrosive effects at the extremes of straining cycles, a judgement needs to be made on the effects of such flushing on water which may have collected at low points in the line, or condensed water in gas lines.
If flushing cannot be guaranteed to remove a sour environment from the pipe surface, the possibility exists of straining in an environment where the temperature is low enough for hydrogen embrittlement to occur. Corrosion fatigue in such sour environments is known to be particularly severe. [1,2] If some uncertainty remains, but the number of such full shutdown cycles is predicted to be low, it may be possible to account for the effects by a conservative reduction in life, and to concentrate testing on the hotter environments which will be present for the majority of the loading cycles.
It is, thus, essential to give careful consideration to the variation in environmental conditions in relation to the loading cycle. Test temperature is often chosen to be approximately mid-way between the maximum and minimum of typical temperature cycles. If, however, the majority of life will be spent under scaling conditions, and the resultant test temperature does not result in scaling, the test data would not be representative. Consideration also needs to be given to inhibition. It has been shown in previous work, that testing in highly corrosive conditions can give unconservative results.  Thus, testing with inhibitor which would in any case be more representative of true conditions, may also be more conservative than testing without. Nevertheless, it must also be recognised that the effects of partial inhibition, or interruptions in inhibition are not well understood, and full design levels may not be most appropriate either.
End of life
Fracture mechanics based fatigue and fracture assessments assume a critical final flaw size, beyond which the pipe will fail by unstable crack growth. This flaw size is based on fracture mechanics calculations, and thus requires appropriate fracture toughness data. If the pipeline is operating in sour service, the steel will be charged with hydrogen, and this will degrade its toughness at near ambient temperature.
Depending on the manner in which stress and temperature vary during a shutdown, it may be difficult to specify a 'worst case' condition to determine end of life. It may be necessary to evaluate toughness (or KISCC) as a function of temperature so that a conservative (but not overly conservative) estimate of maximum tolerable flaw size can be determined. Furthermore, a distinction needs to be made between buried and external flaws, and internal surface breaking flaws where production fluids have direct access to the highly strained region of material at the crack tip. For the former, fracture toughness data for hydrogen-charged material, determined at an appropriate temperature, are required. For the latter, the final failure mechanism which needs to be assessed is stress corrosion cracking, rather than fracture. At present, assessments based on conventionally measured KISCC data tend to be highly conservative, but until ongoing research  develops an improved methodology, there is no reliable alternative. Certainly, assessments based on fracture toughness measured in the absence of a corrosive environment would be unconservative. For both hydrogen-charged fracture toughness, and KISCC, it is easy to generate highly conservative data, whereas testing under conditions which are closer to those in service (for example, using single sided exposure) is commonly more complicated and expensive. If the more conservative data allow a practical design, then it may not be worth refining the test techniques. Sufficient time for a staged approach to testing and design will usually provide a more cost effective and reliable result. The potential impact of environmental conditions should therefore be considered as early as possible in the design process.
The design of pipelines to operate under lateral buckling conditions requires project specific corrosion fatigue data, generated at a suitable low frequency. The predicted service conditions need to be closely scrutinised to identify the appropriate environment for testing. If facture mechanics based fatigue and fracture assessments are to be carried out, hydrogen charged fracture toughness data and/or KISCC data will also be required.
- S J Maddox, R J Pargeter and P Woollin: 'Corrosion fatigue of welded C-Mn steel risers for deepwater applications: a state of the art review', OMAE2005-67499, Proceedings of OMAE 2005: 24th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2005).
- J Buitrago and M S Weir: 'Experimental fatigue evaluation of deep water risers in mild sour service', Deep Offshore Technology Conference, New Orleans, (November 2002).
- D P Baxter, S J Maddox and R.J. Pargeter: 'Corrosion fatigue behaviour of welded risers and pipelines', OMAE2007-29360, Proceedings of OMAE 2007: 26th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2007).
- F McMaster et al: 'Sour service corrosion fatigue testing of flowline and riser welds', OMAE2008-57059, Proceedings of OMAE 2008: 27th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2008).
- R J Pargeter, B Holmes and D Baxter: 'Corrosion fatigue of steel catenary risers in sweet production', OMAE2008-57075, Proceedings of OMAE 2008: 27th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2008).
- J Buitrago, S Hudak and D Baxter: 'High-cycle and low-cycle fatigue resistance of girth welds in sour service', OMAE2008-57545, Proceedings of OMAE 2008: 27th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2008).
- J Buitrago et al: 'Effect of loading frequency on fatigue performance of risers in sour environment', Paper 51641, Offshore Mechanics and Arctic Engineering Conference (OMAE 2004), Vol.III Materials Engineering, ASME, (2004).
- C M Holtam et al: 'The behaviour of shallow cracks in a pipeline steel operating in a sour environment', OMAE2008-57083, Proceedings of OMAE 2008: 27th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2008).
- P Woollin, R J Pargeter and S J Maddox: 'Corrosion fatigue performance of welded risers for deepwater applications' Proceedings of Corrosion 2004, NACE International, New Orleans, LA, (2004), paper 04144.
- R I Hammond and D P Baxter: 'Corrosion fatigue of simulated C-Mn steel HAZs in sour produced fluids', OMAE2008-57149, Proceedings of OMAE 2008: 27th International Conference on Offshore Mechanics and Arctic Engineering, ASME, New York, (2008).
- TWI Confidential Joint Industry Project 13136, 'Fracture Mechanics Based Weld Flaw Acceptance Criteria for C-Mn Steel Pipelines in Sour Service' and Phase 2 Proposal PR13102