A. Bourgeon1, M. Ali1 and P. Woollin1
1TWI Ltd, Granta Park, Great Abington, Cambridge, UK
Paper presented at Duplex Stainless Steel International Conference. Beaune, France, 13-15 October 2010.
The fracture behaviour of ferritic-austenitic stainless steels with surface flaws in seawater with cathodic protection is fairly well understood. However, the behaviour of subsurface flaws in material with a fairly uniform distribution of bulk hydrogen from either extended subsea service or welding is less well understood. The paper quantifies the fracture toughness of superduplex weld metals with bulk hydrogen of 1 to 31ppm and the effect of a reduction of strain rate by up to a factor of 100 compared to standard testing rates. The fracture morphology was studied and correlated with the testing conditions. The work indicates an effect of strain rate at a very low hydrogen content of 1ppm, suggesting an effect of low temperature creep, in addition to that of hydrogen.
Ferritic-austenitic stainless steels undergo time-dependent plastic strain ('low temperature creep') at stresses above about 80% of the 0.2% proof stress. This can continue for several weeks under sustained load. Hence, on-going plastic strain is likely ahead of the tip of a flaw under load and strain rate may be expected to affect resistance to fracture, ie the fracture toughness of a structure under sustained load or with a very slow rising load might differ from that measured in a standard fracture toughness test at a fairly high strain rate, eg as defined in BS EN 7448. 
Furthermore, weld metal contains hydrogen from welding, which escapes very slowly due to the low diffusion coefficient. Such hydrogen reduces fracture resistance but might also introduce a degree of strain rate sensitivity to measured values. Whilst hydrogen from welding is rarely of practical concern, additional hydrogen charging of material in subsea service occurs due to cathodic protection, leading to an increase of hydrogen content over time and in extreme cases hydrogen induced stress cracking (HISC) has developed leading to failure. 
The behaviour of subsurface flaws in subsea equipment, to which hydrogen may diffuse over long periods in service is a concern addressed in DNV RP F112 but due to a lack of data this RP takes a conservative approach and states that 'there is a high vulnerability to defects in general, and surface breaking defects in particular'. This view is supported by testing of fracture toughness specimens with cracks actively charged by cathodic polarisation during test but not reflected by in-service failures from buried flaws. This paper addresses this issue by examining fracture of specimens with 'bulk' hydrogen charging prior to test, at levels representative of welding and from subsea service under CP, and explores the effect of strain rate.
2. Experimental approach
2.1.1 Superduplex stainless steel weld with hydrogen from subsea service (weld A)
Weld A was a duplex stainless steel pipe (UNS S31803) which contained a TIG girth weld made with Zeron 100X superduplex (SDSS) consumables, Table 1. The pipe had been in service subsea with cathodic protection for several years, which had caused hydrogen ingress into the material.
Table 1. Weld metal chemical analyses
||24.0 to 26.0
||3.5 to 4.0
||9.0 to 10.0
||0.5 to 1.0
||0.5 to 1.0
||0.2 to 0.3
Six BxB single edge notched bend (SENB) specimens (10x10x60mm) were machined from the pipe, perpendicular to the girth weld. The specimens were notched through-thickness at the weld centreline and fatigue pre-cracked from an electro-discharge machined (EDM) notch. Three specimens were tested in the as-received condition, ie containing 31ppm hydrogen (W01-01 to 03), fairly uniformly distributed, Table 2. The three other specimens (W01-05, 08 and 09) were baked at 350°C for 120 hours, to allow most of the hydrogen to escape (1ppm hydrogen remained). The hydrogen contents were measured by an inert gas fusion method.
Table 2. Weld metal hydrogen analyses
|Material||Hydrogen content, ppm|
|Weld A (ex-subsea service), as received
|Weld A, after baking
|Weld B (as-welded), as received
|Weld B, after baking
2.1.2 Superduplex stainless steel weld with hydrogen from welding (weld B)
Weld B was extracted from a superduplex pipe which contained a TIG girth weld made with Zeron 100X superduplex consumables. This contained 3ppm hydrogen from the welding operation, presumably mostly from the consumable wire, Table 2.
Twelve Bx2B SENB specimens (12x24x150mm) were machined from this second weld. The specimens were notched through-thickness at the weld centreline and fatigue pre-cracked from an EDM notch. Six specimens were tested in the as-received condition (W02-01 to 06), i.e. with hydrogen from welding and the other six specimens (W02-13 to 18) were given a baking treatment (250°C for 90 hours), to allow hydrogen escape. The hydrogen content was measured by an inert gas fusion method.
2.2 Fracture toughness tests
2.2.1 Single point tests of weld metal with hydrogen from subsea service (weld A)
Fracture toughness tests were carried out according to BS7448-2. Three specimens from each of the as-received (31ppm hydrogen) and baked (2ppm) conditions were tested. These specimens were tested under displacement control at an initial K rate of 0.01MPam1/2s-1, except specimen W01-01, which was tested at an initial K rate of 0.06 MPam1/2s-1. Each test lasted several hours. These loading rates are around 10-100 times below those used in conventional fracture toughness tests (BS7448-2 specifies that the K rate has to be between 0.5 and 3 MPam1/2s-1), later referred to as 'slow'. The data were obtained in the form of force versus load line displacement and used to determine CTOD and J integral values at maximum load.
2.2.2 R-curve tests of weld metal with hydrogen from welding (weld B)
Fracture toughness R-curves were generated according to BS 7448: Part 4. A set of six specimens from the as-received condition and six specimens baked at 250°C was tested. For each set, three specimens were tested at standard K rates (between 0.8 and 1.1 MPam1/2s-1), and the other three specimens were tested at a 'very slow' strain rate (approximately 4x10-4 MPam1/2s-1), in order to obtain three point R-curves for each condition. Tests carried out at standard strain rate lasted typically less than an hour, whereas at slow strain rate, tests lasted up to 5 days.
2.3 Post-test examination
The fracture faces of the specimens were examined using scanning electron microscopy (SEM) in order to characterise the failure mechanisms and to relate them to the hydrogen content of the weld metal and the test strain rate.
Two metallographic sections were extracted from the specimens with highest hydrogen (weld A), to relate the crack path with the microstructural features:
- Met3 was extracted through the fracture surface of an as-received specimen containing 31ppm hydrogen.
- Met8 was extracted through the fracture surface of a baked specimen containing 2ppm hydrogen.
The metallographic sections were prepared using standard metallographic techniques. They were electro-etched using 20% sulphuric acid with 0.1g/l NH4CNS as appropriate to the microstructural features present. Micrographs were taken to record pertinent features.
3.1 Fracture toughness tests
3.1.1 Single point tests of SDSS weld with hydrogen from subsea service (weld A)
The graph in Figure 1 shows the load versus crack mouth opening displacement (CMOD) curves for each specimen and Figure 2 and Table 3 summarise the maximum load CTOD/J values for each specimen. The baked specimens exhibited higher δmax values, from 0.184 to 0.258 mm, compared to those obtained for the specimens containing a high level of hydrogen (0.049mm to 0.100mm). The high hydrogen specimen that was tested at a faster strain rate exhibited a higher toughness.
Figure 1. Load versus load line displacement records for weld A (service subsea). Strain rate was 0.01MPa.m0.5.s-1 except where indicated otherwise.
Figure 2. Toughness values measured on the weld A (service subsea)
Table 3. Fracture toughness of weld metal of weld A
|Specimen||Condition||Hydrogen, ppm||Initial K rate|
|δmax, mm||J, N/mm|
||As retrieved from seabed
Moreover, the maximum loads for the specimens containing a high level of hydrogen were reached after a CMOD of approximately 0.4mm, which is shorter than for the specimens with a low level of hydrogen (>1mm). This indicates that, under similar testing conditions, the crack extended faster in the high hydrogen material than in the baked material. The initial slopes of the curves for specimens with high hydrogen level were lower than for the baked specimens. Figure 1 also shows that the high hydrogen specimen that was tested at a faster strain rate (6 times faster, but the test still lasted several hours) exhibited better toughness than the other similar specimens.
3.1.2 R-curves for weld material with hydrogen from welding (weld B)
The results of the fracture toughness tests from weld B are presented in Table 4 and Figure 3. Overall, the values of the fracture toughness (J) were similar for the material containing 1ppm (baked) and 3ppm (as-welded condition). However, for both hydrogen levels (as-welded condition and after baking), the material's fracture resistance was significantly lower when the test was carried out at a low strain rate, Figure 3. The data presented in Table 4 are the maximum load CTOD (δmax) values measured but not all tests achieved δmax.
3a) Hydrogen content of 3ppm
3b) Hydrogen content of 1ppm
Figure 3. Effect of strain rate on fracture toughness of Weld B at two hydrogen concentrations (as-welded and baked)
Table 4. Fracture toughness values of the weld metal of weld B
|Initial K rate|
|δmax, mm||J, N/mm|
||Baked at 250°C for 91 hours
NA=Not applicable, because δmax was not reached (unloading tests to produce an R-curve)
3.2.1 Weld A which was in service subsea
Figures 4a and b are SEM images of the fracture faces of specimens from weld A tested under the same slow strain rates (each test lasted several hours) but the material in Figure 4a contained 31ppm hydrogen and the material in Figure 4b contained 2ppm.
4a) Weld A (31ppm hydrogen), CTOD tested area
4b) Weld A (2ppm hydrogen), CTOD tested area
4c) Weld A (2ppm hydrogen), liquid nitrogen fracture
Figure 4. Fracture faces of weld A specimens tested at slow strain rate
The fracture face of the hydrogen-rich material consists of facets (which are approximately 40-60µm diameter), consistent with a transgranular fracture mechanism, Figure 4a. Within the facets, series of straight, approximately parallel features are visible. These features are probably related to the elongated austenite islands in the ferrite matrix. It is noted that very few ductile microvoids are present indicating that there was little tearing through austenite. On the other hand, the fracture face of the material with low hydrogen content exhibits mostly dimples, consistent with a ductile failure mechanism, Figure 4b. Flat, featureless areas (approximately 30µm
After the CTOD test, the specimens were broken open after cooling in liquid nitrogen, which produces a conventional transgranular cleavage fracture. An SEM micrograph of this part of the fracture face is shown in Figure 4c, for comparison purposes. The fracture face consists of facets bigger than those observed in Figure 4a (≥100µm diameter), again with linear features within the facets.
3.2.2 Weld B containing hydrogen from the welding operation
This material contained quite a low level of hydrogen (3ppm), and it was tested at standard and very slow strain rates.
Figure 5a shows the fracture face of a specimen that was tested at standard strain rate. The fracture face consists of dimples, consistent with a ductile failure mechanism. Unlike the fracture face in Figure 4a (slow strain rate), no flat, featureless areas are visible.
5a) Weld B (3ppm of hydrogen), tested at standard strain rate, CTOD tested area
5b) Weld B (3ppm hydrogen), tested at a very slow strain rate, CTOD tested area
5c) Weld B (3ppm hydrogen), tested at a very slow strain rate, CTOD tested area
5d) Weld B (3ppm hydrogen), liquid nitrogen fracture area
Figure 5. Fracture faces of specimens tested at standard and very slow strain rate
Figures 5b and 5c show the fracture faces of specimens that were tested under a very low strain rate (each test lasted several days). Many dimples were visible, Figure 5c, but numerous small facets (approximately 30µm diameter) were also observed. The fracture face also contained a few large facets (≥100µm diameter), as shown in Figure 5b.
An SEM image of the fracture face produced after chilling in liquid nitrogen is shown in Figure 5d, for comparison. The fracture face consists of large facets containing approximately parallel features presumably related to the austenite islands.
3.3 Crack path
Figures 6a and b show the profiles of the fracture faces in a hydrogen-rich specimen (31ppm, Weld A, Figure 6a) and a specimen containing a low level of hydrogen (2ppm, Weld A, baked, Figure 6b).
6a) Weld A (Specimen W01-03, 31ppm hydrogen)
6b) Weld A, baked (Specimen W01-08, 2ppm hydrogen)
Figure 6 Micrographs of sections through test specimens
In the hydrogen-rich material, the profile of the fracture face consists of flat facets, and the crack was transgranular through both ferrite and austenite, which is consistent with a cleavage fracture mode. Some crack branches are
In the material containing a low level of hydrogen, the profile of the fracture face was smoother, and no flat facets were visible, although the crack path was still transgranular.
4.1 Effect of hydrogen
Figure 7 shows the CTOD (δmax) for specimens containing various levels of hydrogen (1, 2, 3 and 31ppm) and tested at three different strain rates: standard (test duration of less than an hour), slow (duration of several hours), and very slow (duration several days). The CTOD values of the specimens containing low levels of hydrogen (1-3ppm) are a factor of 3-7 times higher than for specimens containing 31ppm hydrogen. The results of the slow strain rate tests exhibited a lot of scatter but the associated δmax values were lower than those obtained from tests at standard strain rate. The results also indicate an effect of hydrogen content over the range 1-3ppm, although it is noted that there is significant scatter in the data.
Figure 7. CTOD (δmax) as a function of hydrogen content and strain rate
Therefore, extended service subsea may dramatically reduce the fracture toughness of superduplex material, by a factor of five times as measured in the slow strain rate, but otherwise conventional, fracture toughness test used
The fracture mode varied significantly depending on the level of hydrogen in the material. The hydrogen-rich material (31ppm) failed in a fully brittle manner (cleavage fracture), whereas the material with a low level of hydrogen exhibited a fracture face which consisted of mixed ductile dimples and facets, but no evidence of cleavage was visible. Therefore the introduction if hydrogen, at levels that may be encountered in subsea service, encourages a brittle cleavage mechanism.
Whilst these effects are in line with the widely documented detrimental effects of hydrogen on duplex stainless steels, it is noted that in practice this has not apparently led to in-service problems, although toughness testing of materials intended for subsea equipment would not normally include hydrogen charging.
4.2 Effect of strain rate
The R-curves obtained from Weld B (3ppm hydrogen from welding), together with the single point CTOD tests on the hydrogen-rich (31ppm) Weld A from subsea service both demonstrated that the fracture toughnesses of the materials were significantly lower when testing under a low strain rate. The toughness increased by up to a factor of two when CTOD was measured at standard strain rate, even for 1 and 3ppm hydrogen.
Weld B (3ppm hydrogen) was tested at three different strain rates (standard, slow and very slow), which enabled the effect of the strain rate on the fractography to be analysed. It was observed that the fracture face of the specimen tested at a standard strain rate consisted entirely of microvoids, indicating a fully ductile failure mechanism. The fracture face from the test at slow strain rate consisted mainly of dimples, but some cleavage facets, similar to those from tests on Weld A (31ppm hydrogen), were also observed, which indicates that a second failure mechanism was introduced as a result of reduced strain rate. Finally, dimples were again present on the fracture face tested at very slow strain rate but more evidence of cleavage mechanism was observed. Therefore the lower the strain rate introduces cleavage even at a low level of hydrogen.
The time-dependent, strain rate effect is presumably related to the additional time available for hydrogen diffusion but it is somewhat surprising that the magnitude of the effect is similar at 31 and 3ppm hydrogen. This suggests that there may be another factor involved in the strain rate effect and low temperature creep is the perhaps the most likely second contributing factor.
Although failure from buried flaws in structures that become hydrogen charged has not been reported, there is concern over the criticality of such flaws as service lives increase and materials become increasingly hydrogen-charged. When an engineering critical assessment is required, there is no accepted approach to generating suitable fracture toughness values. Currently, if testing is required for service subsea under cathodic protection, it is most likely to be under a moderately slow strain rate, immersed in artificial seawater and with an applied potential. However, these tests are not representative of buried flaws where hydrogen permeates to the flaw slowly over many years, and the flaw probably experiences a very slow strain rate or a constant load over a long period of time. There is a need for a method that more closely matches the service situation. Effects of both hydrogen, even at low levels, and low temperature creep may be expected and any test method should account for both.
- Hydrogen distributed fairly evenly through-thickness, from welding or extended subsea service, reduces the fracture toughness of superduplex stainless steel weld metal by a factor of 3-7 times over the range 3-31ppm.
- Reducing strain rate reduces fracture toughness by up to about 50% at both 3 and 31ppm hydrogen. The degradation of fracture toughness with decreasing strain rate is associated with the introduction of brittle, cleavage facets inmaterial with 3ppm hydrogen.
- The level of hydrogen and the strain rate both affect the fracture toughness of superduplex weld metal. There is a need for a method to reliably measure fracture toughness applicable to buried flaws in ferritic-austenitic structures designed for extended service subsea.
- Boniardi M, La Vecchia G M, Roberti R, 'Stress relaxation behaviour in duplex and superduplex stainless steels', Proc conf 'Duplex Stainless Steels 94', Glasgow, TWI, paper 110.
- BS 7448-2: 1997, Fracture mechanics toughness tests. Method for determination of KIc, critical CTOD and critical J values of welds in metallic materials., 15 August 1997.
- Taylor T S, Pendlington T and Bird R: 'Foinaven super duplex materials cracking investigation'. Proc conf 'Offshore Technology Conference (OTC) 1999', Houston, 1999, paper OTC 10965, 467-480.
- Recommended practice DNV-RP-F112, 'Design of duplex stainless steel subsea equipment exposed to cathodic protection', 2008.