Hydrogen Induced Cracking of Subsea Duplex Stainless Steel

TWI Ltd
Cambridge, UK

Paper presented at 29th International Conference on Offshore Mechanics and Arctic Engineering (OMAE 2010), Shanghai, China, 6-11 June 2010.

Abstract

A small number of duplex and superduplex stainless steel components have failed in subsea service due to hydrogen induced stress cracking (HISC). The significance of these failures has led to research to define critical loading conditions for HISC, to allow confident design of components in future. Data relating to the Foinaven superduplex hub failures were published at OTC in 1999 and NACE Corrosion conference in 2001 and data from TWI Group Sponsored Projects were published at OMAE in 2004. DNV RP F112 has been based on these and other data, to provide a conservative approach to design.

There are a number of gaps in the published literature and in the data available when DNV RP F112 was prepared, related to differences between small scale specimens tested at typical seabed temperature and actual components in operation, ie the operating temperature and pressure, and data from full-scale tests on pipe material with fine austenite spacing and hence good resistance to HISC.

The paper presents new data on these issues and indicates where technology gaps remain.

Introduction

A small number of duplex and superduplex stainless steel components have failed in subsea service due to hydrogen induced stress cracking (HISC). The significance of these failures has led to research aimed at defining critical conditions for HISC, to allow confident design of components in future. Data relating to the Foinaven superduplex hub failures were published at OTC in 1999^[1] and the NACE corrosion conference in 2001^[2] and data from other twi investigations were published at OMAE in 2004. ^[3]

The susceptibility to HISC shows a strong correlation with austenite spacing and tests performed on smooth samples have shown that coarse-grained microstructures are more susceptible to HISC.^[3] Small-scale constant load tests in seawater with cathodic protection (CP) are generally used to characterise the HISC susceptibility of duplex and superduplex materials.^[2] Constant load, tensile HISC tests on hub materials from the Foinaven field, at -1100mV_SCE showed that if the superduplex material was loaded to an initial strain of 0.5%, creep and crack initiation would ensue very quickly and failure would eventually occur at a substantially higher total strain, as a consequence of low temperature creep.^[1] Tests on full scale hubs have been reported,^[1,2] which indicated that once cracks had initiated, they could propagate through thickness in about 10 days without further increase in applied load. Hence it was concluded that initiation of HISC must be avoided if failure is to be prevented, and the assessment criterion should be related to the threshold initial stress or strain for initiation of HISC. ^[1]

DNV RP F112^[4,5] has been based on these and other data, to provide a conservative approach to design. However data are available to show that some product forms including pipes are substantially more resistant to HISC than coarse grained forgings, due to finer austenite spacing. RP F112 does not allow advantage of this finer spacing to be taken reliably, as it is based on a measurement of austenite spacing, which is not the subject of a recognised standard with anecdotal evidence from industry that it is not reproducible and hence open to mis-interpretation. There were also a number of other gaps in the available data available when DNV RP F112 was prepared, related to (i) differences between small scale specimens tested at typical seabed temperature and ambient pressure and actual components in operation, ie at elevated operating temperature and pressure, and (ii) full-scale tests for fine grained pipe material. This paper presents data on these issues and indicates where technology gaps remain.

Experimental procedure

Introduction

Small-scale HISC tests were performed on tensile specimens taken from as fairly coarse grained superduplex stainless steel. Tests were of 30-day duration and performed at a potential of -1100mV_SCE. Comparative tests were performed at 20 and 80°C (1 bara), and 1 and 100bara (20°C).

Large-scale four-point bend HISC testing was carried out on fine-grained girth-welded and fillet-welded seamless pipes for a maximum duration exceeding six months. Residual stress measurements were taken prior to testing. Testing was performed in seawater under cathodic protection at -1100mV and strain was recorded during testing at different locations on the welded pipe. Visual inspection, dye-penetrant examination, metallographic and fractographic studies were performed.

Materials

The microstructural characteristics of the five materials are summarised in Table 1. Material A had a fairly coarse aligned austenite structure, with fairly consistent austenite island size. Material B had a 'primary' coarse, aligned austenite structure and finer, random equiaxed 'secondary' austenite islands in between the coarse units. The measurements of austenite spacing were made in this material (i) including all austenite, and (ii) to reflect only the coarse primary austenite spacing, ie ignoring the fine secondary austenite.

Table 1 Materials examined

Effect of temperature and pressure on small scale HISC tests

Two series of constant load tensile HISC tests were performed on material A (25%Cr superduplex bar) with specimens machined in the longitudinal direction; pre-charging and testing were at -1100mV_SCE and 1 bara pressure at 20 and 80°C, in natural seawater. A salt bridge was used, so that the Ag/AgCl reference electrode could be kept cool.

Additional constant load HISC tests were performed at -1100mV_SCE, 20°C in natural seawater on material A (specimens machined longitudinally) with pre-charging and testing at 100 bara. The autoclave was pressurised with nitrogen. An Ag/AgCl reference electrode was used. Specimens were stressed to around the threshold stresses for crack propagation and initiation in 30 days, as established at room temperature and pressure (749 and 553 MPa respectively). Allowance for the internal autoclave pressure was made to the measured applied stress on the specimens, as described in EFC 17. ^[6]

After test the specimen hydrogen contents were measured by vacuum hot extraction and crack numbers and depths were measured on metallographic sections through specimens that had not failed at the end of test.

Full scale welded pipe tests

Pipe samples in material B (22%Cr duplex), which were 4m long, with 15mm wall thickness and 168mm outer diameter were used for full scale testing in a purpose-built, four point bend load frame. Two specimen weld geometries were examined:

(i) a pipe with a girth weld at mid-length.
(ii) a pipe with a fillet weld to a circular patch at mid-length, simulating an anode attachment pad. The girth-welded pipes, designated GW1 and GW2, were welded by mechanised TIG (with the pipe horizontal and rotated) employing a Zeron 100X superduplex filler wire. The fillet-welded pipe, designated FW1, was welded by manual TIG, employing similar wire.

Residual stresses were measured prior to testing using the centre-hole drilling technique in the weld toe/HAZ areas. The locations were chosen to minimise impact on subsequent test. The welded pipes were tested with a cell mounted around the weld area, containing natural, flowing seawater at a temperature of 10°C with a potential of -1100mV_SCE applied by potentiostat. The pipes were pre-charged, without load applied, for seven days, prior to test. The load applied to pipe GW1 and FW1 was then increased incrementally to identify the threshold load for macroscopic crack development. Each loading step was maintained for seven days and the pipe was examined for the onset of cracking employing a binocular microscope. After seven days exposure the applied load was increased and the procedure repeated until cracking initiated. In order to record the strain during testing, the pipes were strain gauged. Following determination of an approximate threshold load from the first two tests, the girth-welded pipe GW2 was tested with loading to give 0.5% total strain, as measured on strain gauges away from the weld and held for a period of 6 months. The welded pipes were examined by dye-penetrant inspection (DPI) at the end of the test to identify any fine cracking at the weld and sections were taken through relevant areas identified by DPI.

Stress concentration factors (SCFs) for the weld toes were estimated based upon comparison of the weld geometries and previous finite element analysis performed at TWI.^[7] The two weld geometries examined had the following estimated SCFs:

(i) girth-welded pipes. The SCF was about 2.6 ±0.2. The errors quoted allow for the variability of geometry in real welds compared with FE models. The main variables determining the SCF at the toe of a butt weld are the angle at the weld toe and the overall profile of the weld. For the girth welds, the toe angle was measured to be about 45degrees and the weld overfill was considered to be in the form of a circular arc.

(ii) fillet-welded pipe. The SCF was estimated as 2.8 ±0.2. The main variables here are the toe angle and the ratio between the weld leg length and plate (pipe) thickness. The toe angle was 20-25° and the leg length and plate (pipe) thicknesses were 8mm and 15mm, respectively.

Results

Effects of temperature and pressure

HISC testing at 20 and 80°C Figure 1 presents the results of tensile HISC testing of 25%Cr superduplex material A at 20 and 80°C, 1 bara pressure and -1100mV_SCE. In terms of applied stress, the threshold for specimen failure in 30 days was lower at 80°C than at 20°C by around 3% but no reduction in the stress for crack initiation (ie formation of small cracks that did not propagate through thickness) in 30 days was found, Figure 1a. The equivalent data are plotted in terms of strain in Figure 1b, which shows a similar small reduction in strain for cracking at 80°C. However, the 0.2% proof stress of material E was around 13% lower at 80°C than 20°C (530 and 608MPa respectively), indicating that the HISC crack initiation and propagation thresholds at 20°C were lower than at 80°C, when considered in terms of normalised stress, ie applied stress divided by 0.2% proof stress.

HISC testing at 1 and 100 bara Figure 2 present the results of small scale tensile HISC testing of 25%Cr superduplex material A at 1 and 100bara, 20°C, and -1100mV_SCE in seawater. A reduction in the stress threshold for crack propagation, by about 4%, was noted at 100bara but no reduction in threshold stress for crack initiation was noted. However, it should be noted that fewer tests were performed at 100bara, such that the threshold values were less precisely determined than at 1bara. Hence, it is possible that a small shift in crack initiation threshold exists but it could be no more than 5%.

Post-test characterisation Substantially higher hydrogen pick-up was observed at 80°C compared to 20°C, by a factor of 5 to 8, whilst measurements indicated an approximate factor of two increase in hydrogen pick-up at 100bara compared to 1bara (both at 20°C). It was noted that there were many more cracks formed at 80°C than at 20°C but increasing temperature to 80°C seemed to have reduced crack depth. Raising pressure increased crack depth for a given strain.

Full scale HISC tests on welded pipes

Residual stress A maximum tensile residual stress of 453MPa was measured in the HAZ of girth-welded pipe GW1 about 2mm from the fusion line. Maximum circumferential and axial tensile residual stresses of 447MPa and 309MPa were observed in the HAZ of pipe GW1. A maximum tensile residual stress of 392MPa was measured in the HAZ of pipe GW2 and maximum circumferential and axial tensile residual stresses were 389MPa and 279MPa. For the fillet welded FW1 pipe a maximum tensile residual stress of 404MPa was measured in the HAZ about 1.5mm from the fusion line.

Threshold stress and strain for HISC Figure 3 shows the 'crack' and 'no crack' applied stress levels from girth-welded pipe GW1, as calculated from the longitudinal strain measurements away from the weld and the stress-strain curve for the 22%Cr parent material B, indicating the approximate threshold load. The global strain, ie away from stress concentrations, during testing was constant and no straining due to low temperature creep was observed at the gauges.

No cracking was observed after seven days at 0.63% strain, corresponding to a nominal longitudinal stress of 584MPa. The equivalent normalised stress (with respect to the 0.2% proof stress) was 1.05. Cracking occurred after less than 12 hours at a strain level of 0.85% (measured far from the weld toe) corresponding to a nominal stress of 601MPa (normalised stress = 1.08). Failure in pipe GW1 occurred at the weld toe and propagated through the HAZ (Figure 4) into the parent material.

The applied load for the pipe GW2, tested at one load for six months without cracking, is indicated by the dashed line (measured 125mm from weld cap toe - gauge 6). The maximum total strain measured far from the weld toe was a little over 0.5%, corresponding to a stress of 571MPa and a normalised stress of 1.03.

Discussion

Operating conditions

Increasing temperature to 80°C and increasing pressure to 100bara both tended to reduce the threshold stress for specimen failure in 30 days but little or no effect on small crack initiation was found. When the reduction of proof stress at 80°C is taken into account, cracking behaviour is improved in terms of normalised stress but it is concluded that sensitivity to HISC is not substantially changed at 80°C. The mechanism of the effect of pressure on hydrogen pick-up is not immediately apparent, although one effect might be to inhibit recombination of H atoms. It appears that conditions that act to increase hydrogen pick-up tend to enhance specimen failure, ie crack propagation, but have little effect on crack initiation, at least for the durations studied. The fact that increased surface hydrogen content did not significantly affect HISC initiation, suggests that the surface hydrogen level is already adequately high for easy crack initiation at 20°C and 1bara. However, a greater but still quite small, effect of pressure on crack propagation was noted. This is consistent with an increased hydrogen level subsurface, arising from a higher surface concentration (the hydrogen diffusion coefficient being unchanged).

The effect of increasing temperature is not a simple one. The surface hydrogen concentration at 80°C is expected to be much higher than at 20°C, as higher charging current densities are found at higher temperatures,^[8] and this might account for the greater number of surface cracks at 80°C compared to 20°C. Hydrogen diffusion is also faster at 80°C, leading to higher subsurface hydrogen contents. However, this did not lead to greater crack depths at 80°C. One potential explanation of this could be that propagation is controlled, at least in part, by the austenite structure rather than the rate of hydrogen penetration ahead of the crack alone.

Full-scale pipe behaviour

The work showed that full scale welded duplex stainless steel pipes, with fine austenite spacing, cathodically protected at -1100mV_SCE can tolerate a global stress of 1.03 times the 0.2% proof stress, equivalent to a total strain of 0.5%, both measured well away from any stress concentrator prior to the onset of HISC at a weld with and SCF of 2.6-2.8. Only small differences were observed between fillet-welded and girth-welded pipes. The pipe test failure loads are substantially greater than for coarse grained superduplex forgings, which failed under an applied strain of only 0.25% at a stress concentrator.^[2] The initiation and subsequent propagation of HISC apparently occurred at very similar stress in the large scale samples. The stress and strain at the weld toe are not known, as there is no simple relationship between the SCF and the local conditions at the weld toe, which will be affected by local low temperature creep. When the work was originally undertaken, there was no opportunity or strong need to analyse the results further but there is now an opportunity to model the stress and strain at the weld toe and to compare with the RP F112 allowable loading.

Remaining issues: materials with improved hisc resistance and sub-surface flaws

RP F112 represents a big step forward in design to avoid HISC. However, it may be more conservative than necessary for some product forms, e.g fine-grained wrought pipe, partly because it is primarily validated against data for very coarse grained superduplex forging material and partly because it makes conservative assumptions regarding the effect of residual stress. This has led to reports from industry that some components are being over-designed, leading to production of components that are thicker, heavier and more expensive than necessary. However, some product forms, including pipes, are substantially more resistant to HISC than coarse grained forgings, at least partly due to finer austenite spacing. Whilst RP F112 allows advantage of this enhanced HISC resistance to be taken for material with austenite spacing <30µm, the qualification process is not robust, as it is based on a microstructural measurement of austenite spacing, which is not the subject of a recognised standard, and other factors such as the alignment and aspect ratio of the austenite are not considered. There is evidence that the austenite spacing measurement procedure described in RP F112 is not reproducible and hence open to mis-interpretation, eg in relation to the treatment of 'fine equiaxed austenite' clusters. Development of a robust qualification test for materials with improved HISC resistance would be beneficial.

RP F112 recommends that welds in new designs, areas of uncertainty and 'critical' items must be free of flaws. Unfortunately many items fall into these categories and the necessary inspection is typically difficult to undertake and the need to eliminate flaws may force unnecessary repairs. The concern is that such components could pass hydrotest and early service loading, eg if they are located outside of the initially shallow hydrogen-charged, stressed zone, particularly for components which are mainly subject to hydrogen ingress from the external surface. However there is a risk of failure in the medium to long term if such flaws become fully or partially encapsulated in a hydrogen-charged, stressed, zone near. As hydrogen diffuses further into a material with time, the risk of failure from such embedded flaws will increase.

RP F112 takes a conservative approach to dealing with flaws, as there is an absence of reliable data on the fracture toughness applicable to buried flaws in duplex stainless steel as hydrogen diffuses to them over a long period of time.

Slow strain rate fracture toughness tests on duplex/superduplex welds exposed to charging in seawater with CP give CTOD values as low as 0.05mm. However, the fact that no HISC failures from buried flaws have been reported indicates that these values are not appropriate and that further studies are required to quantify the fracture toughness relevant to such flaws. Contributing factors include the hydrogen concentration at a buried flaw, low temperature creep and crack tip acuity.

Conclusions and recommendations

1. Susceptibility of superduplex stainless steel to HISC is broadly similar at 20 and 80°C, despite substantially higher hydrogen pick-up at 80°C, which is about 5 to 8 times higher than at 20°C.
2. Increasing pressure to 100bara gave a modest decrease in threshold stress for HISC crack propagation but had no effect on crack initiation. Hydrogen pick-up was about twice as high at 100bara compared to 1bara, at 20°C.
3. The threshold stress for HISC of large-scale welded 22%Cr pipes, with fine austenite spacing, was 1.03 times the 0.2% proof stress, measured away from the weld, equivalent to a strain of 0.5%. There is an opportunity to re-analysethese data to compare with RP F112 allowable loads. Comparison with previous full scale tests on coarse grained superduplex forgings would assist further understanding of the effects of microstructure.
4. There is a need for a reliable and reproducible method of qualifying a material with improved HISC resistance, associated with austenite spacing <30µm.
5. If a flaw is sufficiently close to the outside surface such that hydrogen can diffuse to one of its tips over time, then theoretically this may lead to HISC. However, no failures from such features have been reported, suggesting a discrepancy between testing and the service situation. The likelihood of failure in the future from such flaws therefore needs to be quantified and assessed.

References

1. 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.
2. Woollin P and Murphy W, 'Hydrogen embrittlement stress cracking of superduplex stainless steel', Proc conf Corrosion 2001, NACE International, paper 01018.
3. Woollin P and Gregori A, 'Avoiding hydrogen embrittlement stress cracking of ferritic austenitic stainless steels under cathodic protection', Proc conf OMAE 2004, paper 51203.
4. Recommended practice DNV-RP-F112, 'Design of duplex stainless steel subsea equipment exposed to cathodic protection', 2008.
5. Vargas PM, Wastberg S and Woollin P, 'Stress based design guidelines for hydrogen induced stress cracking (HISC) avoidance in duplex materials', Proc conf OMAE 2009-79504.
6. EFC publication 17, 2nd Edition, 'Corrosion resistant alloys for oil and gas production: guidance on general requirements and test methods for H₂S service', Maney Publishing, 2002.
7. Gurney T R: 'Finite element analyses of some joints with the welds transverse to the direction of stress'. TWI Research report E/62/75, March 1975.
8. Turnbull A, Griffiths A and Reid T, 'Hydrogen embrittlement of duplex stainless steels - simulating service experience', Proc conf 'Corrosion 99', NACE International, paper 148.