Steel catenary risers in sour deepwater environments

TWI Bulletin, May - June 2010

Part ll. The results

Colum Holtam is a Principal Project Leader within the Fatigue Integrity Management section of the Structural Integrity Technology Group. His main activities are related to consultancy or research and development, with the majority of his clients being from within the oil and gas sector. Colum currently manages several R&D projects related to fatigue design and flaw assessment, with particular emphasis on the corrosion fatigue behaviour of pipeline steels exposed to sour environments. He is a Chartered Engineer and a member of the IMechE and ASME.

Steel catenary risers in deepwater oil and gas environments came under the spotlight of TWI's Colum Holtam in the last issue of Bulletin. He explained that when the produced fluids are sour (ie contain water and H₂S), higher fatigue crack growth rates are expected, and this can have a significant effect on defect tolerance. With the objective of providing guidance for performing ECAs on internal surface-breaking defects in SCRs part one of his work examined the approach to the project, the input parameters, the weld geometry, the stresses considered and the fatigue crack growth rate data used in the engineering critical assessment calculations. Now part two examines the results of Holtam's work, including the fracture and fatigue assessments, and demonstrates the effect of three critical parameters when performing sour ECA calculations.

Results

Calculations have been performed using TWI's software package CRACKWISE 4, which is fully compliant with the latest version of BS 7910.

Static assessments

The first assessment carried out was a static fracture assessment to demonstrate the effect of fracture toughness (or K_ISCC) on sour defect assessments under static loading conditions. A sensitivity analysis was performed whereby maximum tolerable flaw height (maintaining a constant aspect ratio of 0.1) was determined as a function of fracture toughness, for yield strength magnitude residual stress. Figure 6 shows the resulting plot of maximum tolerable flaw height against fracture toughness. This shows that for an assumed value of K_ISCC of 3160Nmm^-3/2 (100MPam^0.5), and yield strength magnitude residual stress, a maximum tolerable flaw size of 4.5 x 45mm might be tolerated. A 3 x 30mm flaw is of the order of magnitude that could be reliably detected by automated ultrasonic (AUT) inspection while at the same time not triggering an unfeasibly large number of repairs. Repeating the analysis for such a flaw shows that K_ISCC could be as low as 2500Nmm ^-3/2 (80MPam ^0.5 ).

Figure 6 also demonstrates the effect of welding residual stress. Based on an assumed value of K_ISCC of 3160Nmm^-3/2 (100MPam^0.5), if welding residual stress could be reliably shown to be only half yield strength magnitude, then the maximum tolerable flaw size would increase from approximately 4.5 x 45mm to approximately 8.6 x 86mm. Similarly, if this could be shown to be the case then the minimum value of K_ISCC for a 3 x 30mm flaw to be tolerable would decrease to approximately 1600Nmm^-3/2 (50MPam^0.5).

Fracture and fatigue assessments

Combined fracture and fatigue assessments were performed using an assumed initial flaw size (for example 3 x 30mm), a K_ISCC value of 3160Nmm^-3/2 (100MPam^0.5) and yield strength magnitude welding residual stresses that were allowed to relax under applied load. Two different fatigue spectra were used representing VIV loading at the TDP and top weld, as described in Figure 4. At both the TDP and the top weld, three assessments were performed using the following fatigue crack growth laws:

1. In air conditions, based on the two stage upper bound curve for steels in air from BS 7910.
   
   
2. A two stage upper bound sour environment curve, a factor of 30 higher than the two stage upper bound curve for steels in air from BS 7910 but with the same threshold ΔK_TH.
   
   
3. A four stage sour environment curve; the first two stages correspond to the upper bound curve for steels in air and the final stage the upper bound sour environment curve (Figure 5).

The results of the above assessments are summarised in Table 3. For the TDP there is no difference between using the upper bound curve for steels in air and the four stage law for a sour environment, with both achieving the desired fatigue life of 600 years. This is because the stresses are so low that they never result in a value of

Table 3 Results of fracture and fatigue ECA calculations at the TDP and top weld positions for in air conditions, a four stage sour environment curve and an upper bound sour environment curve

For both the TDP and top weld, initial flaw sizes have also been calculated based on no safety factor on fatigue life in the ECA (target fatigue life 30 years), a safety factor of five (target fatigue life 150 years) and a safety factor of 10 (corresponding to a target fatigue life of 300 years) (Table 3). Despite the sour upper bound curve producing short estimates of life for an initial flaw of 3 x 30mm, the TDP could tolerate an initial flaw of 2 x 20mm with a safety factor of 5, which is still close to a typical AUT inspection limit. At the top weld however a safety factor of five on fatigue life results in a critical flaw size of approximately 1.3 x 13mm.

Discussion

Toughness has been shown to be a critical parameter in performing sour ECAs under static loading conditions. With a yield magnitude residual stress, the minimum K_ISCC value to tolerate a 3 x 30mm flaw is approximately 2500Nmm^-3/2 (80MPam^0.5)(Figure 6). If K_ISCC was lower this might be expected to lead to a significant repair rate during fabrication. It is also clear that an accurate estimate of K_ISCC is essential as conservative assumptions, in the absence of actual experimental data, could easily lead to excessively conservative results (eg indicating that a harmless flaw was unacceptable). Even when experimental data are available there can be considerable scatter in results from nominally identical tests. It should also be noted that K_ISCC may be strongly influenced by temperature so, for example, K_ISCC may be lower during shutdown (at ambient temperature) than during operation (at elevated temperature).

Although K_ISCC has been shown to be a critical input parameter under static loading conditions, it is interesting to see whether this remains the case within a fracture and fatigue assessment. Examining the upper bound case at the TDP for a fatigue life of 150 years, the initial flaw size is 2.0 x 20mm (Table 3) for a value of K_ISCC of 3160Nmm^-3/2 (100MPam^0.5). If K_ISCC is increased to approximately 4700Nmm^-3/2 (150MPam^0.5) the initial flaw size remains at 2.0 x 20mm to achieve a design life of 150 years. If the value of K_ISCC is reduced to approximately 2200Nmm^-3/2 (70MPam^0.5) the initial flaw size is 1.9 x 19mm. In this regard the value of K_ISCC has little effect on the initial flaw size and the assessment is dominated by fatigue. Similarly for the top weld, increasing the value of K_ISCC to 4740Nmm^-3/2 (150MPam^0.5) the initial flaw size remains unchanged at 1.3mm x 13mm and reducing the value of K_ISCC to 1600Nmm^-3/2 (50MPam^0.5) reduces the initial flaw size to 0.8 x 8mm. The results of the above assessments investigating the influence of K_ISCC within a fatigue and fracture assessment are summarised in Table 4.

Table 4 The influence of K_ISCC on initial flaw size within a fatigue and fracture assessment at the TDP and top weld positions using a two stage upper bound sour environment FCGR curve

It is therefore apparent that for SCRs subject to VIV, the assumption made regarding fatigue crack growth law may be more significant than the assumption made regarding either K_ISCC or welding residual stress. Indeed the assumed fatigue crack growth law, particularly at low ΔK, is likely to be a factor which influences whether a C-Mn material can be used as opposed to a corrosion resistant alloy (CRA) or a clad material, particularly at critical locations such as the TDP and top weld.

The test data illustrated in Figure 7 suggest that the influence of environment at low ΔK is far less severe than at high ΔK, and this has similarly been reported in other work. At low ΔK the data approach the standard curve for steels in air. However, it should be noted that these data were derived from a test conducted under conditions of decreasing ΔK, so that low ΔK data were determined at the end of the test, when the flaw was relatively deep. This is in contrast to the real situation in which a flaw will grow under conditions of increasing ΔK, and will be subject to low ΔK cycling when the flaw may be relatively shallow.

Published data for flaws growing under these conditions suggest that at low ΔK, shallow flaws may grow substantially faster than deeper flaws, and this casts doubt on whether FCGRs in a sour environment genuinely approach those seen in air at low ΔK. It is clear that additional data, at lower ΔK, are needed to provide a better indication of material behaviour under these conditions. The test data illustrated in Figure 7 were also derived from tests in a low pH ambient temperature sour environment, and similar tests in a more realistic operating environment (perhaps high pH and elevated temperature) are also required.

Conclusions

This report has demonstrated the effect of three critical parameters when performing sour ECA calculations; the value of K_ISCC, the magnitude of welding residual stress and the assumed FCGR law. The current case considered these in relation to a typical SCR operating under VIV fatigue loading.

• The value of K_ISCC has a significant effect on the maximum tolerable flaw size, particularly under static loading conditions. However, for the case of an SCR under VIV fatigue loading, the value of K_ISCC becomes less significant as the ECA becomes dominated by fatigue, and the assumption made regarding fatigue crack growth law becomes more significant.
• The effect of welding residual stress is also particularly important for sour (ie low toughness) ECAs, where a high tensile residual stress can provide a significant part of the crack driving force. For thick walled pipe the assumption of yield strength magnitude tensile residual stress may be overly conservative, but a combination of measurement and modelling is recommended to justify any less conservative assumption. However, where fatigue dominates, this is similarly expected to have less influence on defect tolerance.
• The selection of the assumed fatigue crack growth law is therefore critical when conducting ECA calculations for SCRs subject to VIV. There are relatively few published FCGR data for pipeline steels in a sour environment, particularly at low ΔK. Possible differences in behaviour, in this regime, between shallow flaws (under conditions of increasing ΔK) and deeper flaws (under conditions of decreasing ΔK) should also be considered.

Recommendations

Additional FCGR data in a sour environment, at lower ΔK, are needed to provide a better indication of material behaviour under these conditions, and confirm whether FCGRs genuinely approach those seen in air at low ΔK. Future tests should also be performed in more realistic operating environments.

The results reported here are material-environment specific. It is not clear whether the guidance resulting from the above work will be directly transferable to HAZ microstructures and weld material. In higher hardness materials (eg welds) one might expect higher crack growth rates. It is therefore recommended to investigate the influence of hardness and microstructure on shallow crack behaviour by testing actual pipeline girth welds and simulated HAZ material in a sour environment under cyclic loading conditions. This work is currently underway at TWI.

Acknowledgements

The work was funded by Industrial Members of TWI, as part of the Core Research Programme.