Proceedings of the 30th International Conference on Ocean, Offshore & Artic Engineering
June 19 - 24, 2011, Rotterdam, The Netherlands
Amir Bahrami, Anais Bourgeon, Mohamad Cheaitani
Failures of ferritic-austenitic stainless steel due to hydrogen induced stress cracking (HISC) have been very costly and raised concerns regarding subsea system integrity, some of which remain unresolved. The susceptibility to HISC crack initiation shows a strong correlation with austenite spacing and tests performed on smooth samples have shown that coarse-grained microstructures, with large austenite spacing, such as in forgings, are more susceptible to HISC than fine grained structures, eg as in pipe . In all reported failures, cracking has been independent of the presence of fabrication flaws, even though welds were typically present, and initiated at external stress concentrators, so the importance of flaws remains undetermined. There is no well established method for determining fracture toughness values applicable to flaws in duplex stainless steel in the presence of hydrogen and hence reliable data do not exist, leading to a lack of understanding of the criticality of flaws and whether fine austenite spacing provides any benefit in resistance to extension of flaws. This paper provides new data from fracture toughness tests conducted on duplex pipe and forging parent materials, to explore the effect of product type/ microstructure and strain rate on fracture toughness under active charging in seawater under cathodic polarisation. This is part of ongoing work aimed at the development of an engineering critical assessment (ECA) approach for assessing flaw tolerance under hydrogen charging conditions.
Concerted efforts by TWI, DNV, Sintef and other organisations have generated data on resistance of a range of ferritic-austenitic materials to HISC [1-4], which have been incorporated into DNV Recommended Practice RP-F112 . This document was validated by comparison with the results of full-scale tests on superduplex hub material from the Foinaven manifold [6, 7].
Most of the work carried out on the subject of HISC to date has been concerned with identifying the levels of stress and strain at which HISC initiates. Small-scale constant-load tests in seawater with cathodic protection (CP), either round tensile or notched rectangular specimens, have been used to characterise the HISC susceptibility of duplex and superduplex materials in terms of crack initiation. This susceptibility has been shown to have a strong correlation with austenite spacing and tests have shown that coarse-grained microstructures are more susceptible to HISC . Constant-load, tensile HISC tests on hub materials from the Foinaven field, showed that if the superduplex material was loaded to an initial strain of 0.5%, cracks would initiate and propagate very quickly although failure would occur at a substantially higher strain, as a consequence of low temperature creep . Tests on full scale hubs showed that once cracks had initiated, they could propagate through about 15mm thickness in about 10 days without further increase in applied load.
These and subsequent investigations have led to the re-evaluation of the fitness-for-service of duplex and superduplex subsea components. In all reported failures, cracking has been independent of the presence of fabrication flaws. However, evaluation of the risk of failure from fabrication flaws, particularly ones that are not surface breaking but to which hydrogen may diffuse over a long period in service, is necessary as it impossible to ensure freedom from flaws in large, complex structures.
However, the absence of a well established method for determining fracture toughness applicable to flaws in duplex stainless steel in the presence of hydrogen and hence reliable data, has led to the adoption of potentially very conservative approaches, such as that in DNV-RP-F112, which implies that flaws cannot be tolerated under hydrogen charging conditions, which seems at odds with the practical experience. Improving such approaches requires a better understanding of the behaviour of crack-like fabrication flaws and is crucial to the safe and economic operation of subsea duplex and superduplex components.
Fracture toughness tests carried out in air on superduplex weld metal containing a high level of hydrogen from exposure to cathodic protection during service  showed that toughness decreased at high hydrogen contents and low strain rates and the effect of strain rate persisted to very low hydrogen levels, suggesting an effect of low temperature creep. This paper provides new data on fracture toughness of duplex pipe and forging parent materials tested under active charging conditions and highlights, in particular, the influence of strain rate. The paper further examines whether the fracture toughness is influenced by austenite spacing in a similar manner to the stress for crack initiation in smooth specimens.
Two types of materials were tested:
Material A was a 22% Cr duplex stainless steel (UNS S31803) pipe section, with 12” outer diameter and 12mm wall thickness. Austenite spacing 31μm
Material B was a piece from a 22% Cr duplex stainless steel (UNS S31803) forging. Austenite spacing 44μm
Austenite spacing was measured using a line intercept method, in the direction of crack propagation, on specimens after test. Four areas were analysed, and four locations measured in each area.
Fracture toughness tests
Tests were conducted on eight single-edge-notch-bend (SENB) specimens. One test was conducted in air at 20°C on Specimen A-01 from material A, which was not hydrogen pre-charged. The seven other tests were conducted in natural seawater at 4°C on specimens which were hydrogen charged prior to (and during) testing by exposure to cathodic polarization (CP) at 1100mV vs. SCE. The seven specimens included four specimens from material A (A-02 to A-05) and three from material B (B-01 to B-03).
The test specimens were extracted parallel to the pipe/forging axis and notched and fatigue pre-cracked perpendicular to this axis. They had a square (B°B) cross section, with a nominal value of B of 12mm, and were prepared (including machining and fatigue pre-cracking) according to BS 7448:Part 1. Given that there is no relevant codified guidance on hydrogen charging or testing (including loading rate), a range of these variables was considered. Three specimens (A-01, A-02 and B-03) were tested under displacement rates giving stress intensity factor (K) rates within the limits specified in standards such as BS 7448:Part 1. The other specimens were tested under lower displacement rates (giving non-standard K rates) in order to determine the influence of this parameter on fracture toughness.
After test the fracture faces were cleaned and examined in a scanning electron microscope (SEM), to characterise the failure mechanism(s) and to relate them to the microstructure of the material, its hydrogen content and the test strain rate.
In addition, metallographic sections were extracted through the fracture surfaces of the specimens to relate the crack path to the microstructural features. They were prepared using standard metallographic techniques, and etched using 20% sulphuric acid with 0.2g/NH4CNS. Micrographs were taken to record pertinent features.
Fracture toughness tests
The fracture toughness test results for the different specimens and test conditions are given in the table 1, which includes information on the following parameters:
Test environment (Specimen A-01: air with no hydrogen charging; all other specimens: seawater with hydrogen charging prior to and during the test under a potential of 1100mV)
Pre-charging duration (length of charging with hydrogen prior to testing)
Hydrogen content measured in a 1.6mm thick slice taken adjacent to the specimen surface. This value is representative of the hydrogen content ahead of the fatigue pre-crack tip since the specimens were pre-cracked before being pre-charged. The hydrogen contents varied as the specimens were pre-charged together and taken off one at a time when they were tested
Test displacement rate and time to maximum load
Fracture toughness results expressed in terms of CTOD (crack tip opening displacement) and J (J integral)
In all tests, the CTOD and J results are maximum load values (ie classed as °m/Jm and represent the fracture toughness at the maximum load reached during the test). For the pipe material A (austenite spacing 31µm), the highest fracture toughness was obtained for specimen A-01 (°=0.85mm, J=1403N/mm), which was tested in air under a standard displacement rate and was not hydrogen-charged. The lowest result was obtained for pipe material, specimen A-05 (°=0.03mm, J=22.0N/mm) which was hydrogen charged, had the longest pre-charge and hence the highest hydrogen content among the pipe material A specimens (8ppm) and was tested under a very low (non-standard) displacement rate (0.002mm/hr).
For the forging material B (austenite spacing 44µm), the highest fracture toughness was obtained for specimen B-03 (°=0.30mm, J=361N/mm), which was tested under a standard displacement rate and was hydrogen-charged prior to and during testing. The lowest result was obtained for forging material, specimen B-02 (°=0.03mm, J=19.9N/mm), which was hydrogen charged, had the second highest hydrogen content among the material B specimens (13ppm) due to an extended pre-charging time and was tested at the lowest displacement rate (0.002mm/hr).
The fracture toughness results (in terms of CTOD and J), for the hydrogen-charged specimens and both the pipe material A and forging material B, decrease as the test displacement rate is reduced. For example, the results for specimens A-02 and A-03 show that for the same duration of hydrogen pre-charging and hydrogen content, reducing the displacement rate by around a factor of 70 can lead to a significant reduction in the fracture toughness, by a factor of 8-12. Similarly, the fracture toughness results for the forging material B decreased by about a factor of 10-20 as the test displacement rate was reduced by three orders of magnitude.
The results of the fracture toughness tests are plotted in Figure 1 in terms of CTOD (δm) against displacement rate. It can be seen that at displacement rates lower than 0.006mm/hr, the toughness values do not change significantly with displacement (or strain) rate. It is also clear that at such slow strain rates moderate variations in the hydrogen content do not significantly influence the material’s toughness. Most importantly, no discrimination can be made between fracture toughness performances of pipe material A (31µm austenite spacing), compared with the forging material B (44µm austenite spacing) at very slow strain rates.
The fracture faces of pipe specimens A-01 (tested in air) and A-02 (tested with CP), which contained 2ppm residual hydrogen and 6ppm hydrogen from charging respectively and were tested under a conventional strain rate, both consisted of microvoid coalescence, with microvoids of various sizes (from a few microns to about 50µm in diameter). This morphology is consistent with a ductile failure mechanism. It is noted that the charged specimen A-02 had the shortest pre-charge of all specimens tested.
The fracture face in pipe specimen A-03, which contained 6ppm hydrogen and was tested under CP and at a low strain rate (0.30mm/hr), consisted of facets that were approximately 100µm in diameter and had ridges, as shown in Figure 2. This morphology is consistent with a quasi-cleavage failure mechanism, commonly observed for hydrogen assisted cracking. The ridges probably correspond to austenite islands. The fracture faces in pipe specimens A-04 and A-05, which were tested under CP at very low strain rates (0.006 and 0.002mm/hr respectively) and which contained 6 and 8ppm hydrogen respectively, exhibited morphologies comparable to that shown in Figure 2, but with coarser facets (up to several hundreds of microns in diameter).
Forging specimen B-03 had the highest hydrogen concentration, 17ppm as a result of the longest pre-charge, and was tested under a standard strain rate. The fracture face morphology was a mixture of microvoid coalescence and quasi-cleavage facets that were approximately 100µm wide, Figure 4.
The microstructure in pipe material A consisted of elongated austenite grains (following the length of the pipe) in a ferritic matrix. Finer and more equiaxed grains of austenite were also visible, the latter presumably formed during final solution heat treatment, whilst the alignment of the coarse austenite indicates that it was present during forming. The fracture face profiles of pipe specimens A-01 and A-02, tested at the standard, relatively high strain rate, contained evidence of plastic deformation ahead of the crack tip, as shown in Figure 5. The crack propagated through both the ferrite and austenite phases. Microvoid formation was observed on and adjacent to the fracture face, which is consistent with the SEM observations that were made on the fracture faces.
The profiles of the fracture faces in pipe specimens A-03 to A-05 and forging specimens B-01 and B-02, tested with hydrogen charging, were largely similar. Contrary to the profiles described previously, they were mostly flat and there were no microvoids. A representative micrograph is shown in Figure 6. The crack propagated through both the ferrite and austenite phases, however the profile of the fracture face was predominantly smooth and flat in the ferrite, but not in the austenite. No indication of plastic deformation was visible. The primary austenite islands in forging material B were coarser than in pipe material A, and the coarse primary austenite grains were more equiaxed in the forging.
The profile of the fracture face in forging specimen B-03 (Figure 7), tested at the standard, higher strain rate under CP showed similarities to both previous groups of specimens, ie microvoids as seen in pipe specimens A-01 and A-02 (Figure 5), tested at the higher strain rate, and flat quasi-cleavage facets, as seen in pipe specimens A-03 to A-05 and forging specimens B-01 and B-02 (Figure 6). The metallographic section in forging specimen B-03 was taken through an area of the fracture face that consisted mainly of microvoids. However, due to the mixed fracture mode, if this section had been taken through an area with more quasi-cleavage facets, the profile of the fracture face would have been likely to be comparable to the profile observed in forging specimens B-01 and B-02.
Fracture toughness results
Effect of displacement rate
Lower fracture toughness was obtained as the displacement rate was reduced and the effect of displacement rate diminished at very slow displacement rates (δ0.006mm/hr). This observation is based on the fracture toughness results for pipe specimens A-04 (°=0.03mm, J=23.3N/mm) and specimen A-05 (°=0.03mm, J=22.0), which are very similar although the latter specimen was tested under a displacement rate three times lower than the former specimen. Similarly, the fracture toughness results obtained for forging specimen B-01 (°=0.03mm, J=22.7N/mm) are very close to those for forging specimen B-02 (°=0.03mm, J=19.9N/mm) although the latter was also tested under a displacement rate three times lower than the former. It is also worth noting that both of the specimens tested under the lowest displacement rate (A-05 and B-02) had higher hydrogen contents than the specimens tested under the slightly higher displacement rate due to longer pre-charging and test durations (A-04 and B-01). This reinforces the previous conclusion since it indicates that lowering the displacement rate (to 0.002mm/hr) had an insignificant effect on fracture toughness (in comparison with the results obtained at 0.006mm/hr for specimens A-04 and B-01) despite the higher hydrogen content. Additional testing at displacement rates in the range 0.006 to 0.3mm/hr would be required to determine the displacement rate below which the fracture toughness becomes largely independent of strain rate, although a value of around 0.02/mm/hr may be estimated.
It should be noted, however, that the above conclusions refer to single-point maximum load results (ie results classed as °m/Jm), which represent the fracture toughness after a certain amount of crack extension (that cannot be determined from the present tests). A more comprehensive understanding of the effects of displacement rate on fracture toughness, and the dependence of this on crack extension, requires that resistance curves be determined under a range of displacement rates and hydrogen-charging conditions, which was beyond the scope of the present study (see also Section 4.1.5).
Effect of hydrogen content under standard displacement rate
Comparing the results for pipe specimens A-01 and A-02, which were tested under a relatively high displacement rate (corresponding to a K rate within the range specified for standard tests according to BS7448:Part 1), indicates that the higher hydrogen content in A-02 had an insignificant effect on its fracture toughness at this strain rate (CTOD and J for specimen A-02 are very close to those obtained for A-01 which had not been hydrogen-charged). This conclusion, obtained under standard displacement rates, applies to the pipe material A but not necessarily to the forging material B, which had slightly coarser austenite spacing. All the tests on forging B were conducted on hydrogen-charged specimens and only one test (specimen B-03) was conducted under a displacement rate comparable to those for pipe specimens A-01 and A-02. However, the fracture toughness for specimen B-03 was significantly lower (δ=0.30mm, J=361N/mm) than those for specimen A-01 and A-02. It is unknown whether this is due to the fact that specimen B-03 had a significantly higher hydrogen content (17ppm) and/or a coarser microstructure than specimens A-01 and A-02, although given the similarity of the austenite spacings it seems most likely to be due to the hydrogen contents.
Effect of hydrogen content under low displacement rate
Figure 1 shows that for displacement rates lower than 0.006mm/hr, the fracture toughness for both pipe and forging materials was largely unaffected by moderate variations in the hydrogen content (6 to 8ppm for pipe material A, and 7 to 13ppm for forging material B), arising from different pre-charge and test durations. Since no testing was conducted under low displacement rates on specimens containing widely different levels of hydrogen, ie well below 6ppm or well above 13ppm, it is unknown whether the current results apply to hydrogen contents outside of those considered in the present tests. It is likely that under slow strain rate conditions, hydrogen from CP during testing has sufficient time to concentrate to a high level at the crack tip, driven by the high local stresses and the presence of a fresh surface created by the propagating crack. This would explain why the ‘bulk’ hydrogen concentration in the material has only an insignificant effect on the fracture toughness (at these concentrations).
Effect of microstructure
Figure 1 shows that for displacement rates lower than 0.006mm/hr, broadly identical results were obtained for the pipe material A (austenite spacing 31µm, specimens A-04 and A-05) and forging material B (austenite spacing 44µm but with a coarser primary austenite structure, specimens B-01 and B-02). However, given the relatively limited scope of the present study, it is not known whether the same observation applies over a broader range of microstructure, eg down to very fine austenite spacings of <10µm, such as are found in thin walled pipe/tube, or very coarse spacings >70µm, such as are found in very thick walled products, or over the full range of strain rates, eg under standard (higher) displacement rates. It is noted that the stress and strain for crack initiation in smooth tensile specimens does not vary significantly over this range of austenite spacing .
With regard to the mixtures of coarse and fine austenite seen in the present materials, the similarity of measured overall austenite spacing (despite different primary austenite structures) and of measured toughness for the two product forms suggests that the fine austenite does play a role in determining toughness under hydrogen charging, which is contrary to a previous suggestion that fine equiaxed austenite has no beneficial effect on HISC .
Relatively low fracture toughness results, δ=0.03mm and J = 20N/mm, were obtained for specimens with a hydrogen content of 6ppm or higher when the test was conducted with active charging during test and under low displacement rates, eg less than 0.006mm/hr. The low fracture toughness values corroborate the current industry practice of not allowing any measureable surface flaw in duplex and superduplex components used for subsea applications under CP. These are maximum load results, which represent the fracture toughness after a certain amount of crack extension. Given that the results were obtained under relatively low displacement rates, they are relevant for the assessment of subsea components subjected to constant loads or loads applied at a very low strain rate. The results represent the fracture toughness for hydrogen-charged materials containing around 6ppm of hydrogen in material up to about 1.6mm ahead of the crack tip. Ideally, the fracture toughness should be determined in terms of resistance curves since these would allow the assessment to be conducted using the toughness at the onset of crack extension or, alternatively, a tearing analysis to be conducted incorporating the influence of crack extension on driving force and fracture toughness. However, generating such resistance curves for materials where crack extension is time and strain-dependent is far from straightforward and requires further research. Additional research is also required to assess the influence of specimen geometry (mainly ratio of notch height to specimen width) and loading type (ie bending vs tension) on fracture toughness.
Fractography and metallography
Facets visible on the fracture faces were observed to have propagated through the ferrite phase, whereas the ridges were observed to correlate with the austenite. On the other hand, where the appearance of the fracture face indicated a ductile fracture mechanism (tearing), deformation of the microstructure and microvoid formation in both phases was observed.
Evidence of tearing was only visible in the specimens that were tested under a standard (conventional) strain rate. However, the fracture mechanism in the specimen which contained the highest hydrogen content (B-03) was a mixture of tearing and brittle mechanisms, presumably because the test was conducted at a conventional, higher strain rate.
Besides, all the specimens that were tested under CP and at a strain rate that was lower than the value specified in fracture toughness standards, like BS 7448:Part 1, exhibited fracture faces containing exclusively quasi-cleavage facets. The amount of hydrogen (from 6 to 13ppm) in these specimens did not seem to affect the morphology of the fracture face, but as discussed previously, this is probably due to the free availability of hydrogen from CP during testing for the slower strain rate condition.
The two materials, i.e. pipe and forging, exhibited consistent fracture toughness results and fracture face morphologies when tested under low strain rates, except that the size of the quasi-cleavage facets were coarser in the forging material. This probably reflects less alignment of the primary austenite islands in the forging material but it is interesting that the larger facet size did not translate into lower toughness. Therefore in this instance the type of material did not seem to affect significantly the fracture toughness values under low strain rates. However, no similar conclusion can be drawn with certainty for standard (conventional) strain rate, because no pipe material was tested with comparable pre-charge time and hence hydrogen content, although the pipe specimen tested under CP at conventional strain rate had significantly higher toughness than the forging specimen.
For the duplex stainless steels tested in this study, hydrogen resulting from cathodic protection and low displacement (or strain) rates both promoted crack extension by a quasi-cleavage mechanism and are associated with lower fracture toughness values
Fracture toughness decreases as the rate of applied strain is reduced. However, the effect of displacement rate on fracture toughness seems to diminish at very slow strain rates and the curve of CTOD or J against displacement rate reaches a plateau at a displacement rate of around 0.02-0.006mm/hr
Under standard displacement rates there was a significant difference between the fracture toughness determined on hydrogen-charged specimens for the finer grained pipe and coarser grained forging materials, even though the austenite spacings were only moderately different (31 and 44µm respectively). However, as there was also a large difference in the hydrogen contents the difference in the results may be due at least partly to the hydrogen content
For the fine-grained pipe material A, the lowest fracture toughness results were °m=0.03mm and Jm =22.0N/mm. These were obtained for the specimen that had the longest pre-charge and the highest hydrogen content among the pipe specimens (8ppm) and was tested under a relatively very low (non-standard) displacement rate (0.002mm/hr).
For the forging material B, the lowest fracture toughness results were °m =0.03mm and Jm =19.9N/mm. These were obtained for a specimen which had the second longest pre-charge and the second highest hydrogen content among the forging specimens (13ppm) and was tested under at the lowest (non-standard) displacement rate (0.002mm/hr)
At very low strain rates, the type of material, i.e. pipe or forging and associated microstructures, did not have a noticeable effect on the fracture toughness, although it is acknowledged that the range of microstructural coarseness was limited. However, the cleavage facets were larger for the coarser grained forging material, reflecting the coarser primary austenite structure
Woollin P and Gregori A, ‘Avoiding hydrogen embrittlement stress cracking of ferritic austenitic stainless steels under cathodic protection’, Proceedings of OMAE 2004 Conference, paper 51203.
Bahrami A and Woollin P, ‘Hydrogen induced stress cracking of duplex stainless steel subsea components’, Proceedings of OMAE 2010 Conference, paper 20088
Johnsen R, Nyhus B, Wästberg S, Lauvstad G Ø, ‘New improved method for HISC testing of stainless steels under cathodic protection ‘, CORROSION 2007, Nashville, NACE International
Bourgeon A, Ali M and Woollin P, ‘Effect of hydrogen and strain rate on superduplex stainless steel weld metal fracture toughness and fracture morphology’, Proceedings of the Duplex World 2010 Conference, October 2010
Recommended practice DNV-RP-F112, ‘Design of duplex stainless steel subsea equipment exposed to cathodic protection’ April 2006
Vargas PM, Wästberg S, Woollin P, ‘Stress based design guidelines for hydrogen induced stress cracking (HISC) avoidance in duplex materials’, Proceedings of 28th Int Conference on Offshore Mechanics and Arctic Engineering (OMAE 2009), Honolulu, 2009
Taylor T S, Pendlington T and Bird R: ‘Foinaven super duplex materials cracking investigation’. Proceedings of the Offshore Technology Conference (OTC) 1999, Houston, 1999, paper OTC 10965, 467-480
The authors gratefully acknowledge Paul Woollin and Marcio Milititsky for their valuable advice. The authors also thank Lijuan Zhang and Ashley Spencer for their assistance in the technical work.