Testing of superduplex stainless steel for sour service (March 2003)

Paul Woollin, TWI and Manuel Maligas, FMC Corporation

Paper # 03132 presented at Corrosion 2003, San Diego, CA, USA, 16-21 March 2003

Abstract

Superduplex stainless steels are widely used by the oil and gas industry for their combination of corrosion resistance, chloride stress corrosion cracking resistance, high strength and weldability. Where such materials are approved for sour service by NACE MR0175, ^[1] maximum hardness limits are stated, together with allowable material condition, limits of H ₂S partial pressure and in some cases limits on chloride ions, pH, temperature and elemental sulfur. This paper gives results of sour service testing of three different UNS superduplex grades and three batches of one UNS superduplex grade in environments representing conservative worst case conditions. Unstressed pitting corrosion tests were performed on all steels examined. Sulfide stress cracking (SSC) tests were performed on two steels that did not corrode in the unstressed tests. Two test stressing methods, i.e. C-ring testing and slow strain rate tensile (SSRT) testing, were compared. The results indicated that some superduplex steels may be resistant to sulfide stress cracking at higher partial pressures of H ₂S than currently allowed by NACE MR0175, ^[1] if other environmental or material conditions are favourable. However, different batches of superduplex steel, either within a UNS composition range or from different UNS compositions, may show quite different resistance to pitting and SSC in sour environments and there is no single maximum partial pressure of H ₂S to which superduplex steels are resistant. In particular, higher PREN/PRENW values are beneficial but coarse grain size with high ferrite content and nitride precipitates may be detrimental. Levels of primary alloying elements, other details of the service environment (e.g. pH and Cl ^- level) and the mode of stressing (e.g. load or displacement control) may all affect resistance to SSC of superduplex steels. Constant load, dead weight tensile testing is preferred to constant deflection C-ring or bend tests where constant load conditions may exist in service. SSRT testing is very severe for evaluating SSC performance of superduplex steels and a reliable pass/fail criterion is required.

Introduction

As the produced fluids encountered in the oil and gas industry become increasingly corrosive due to rising levels of CO ₂, H ₂S and Cl ^- and temperature, the industry is becoming more dependent on corrosion resistance alloys (CRAs). A broad range of CRAs is available, including martensitic/supermartensitic, ferritic, austenitic/superaustenitic and duplex/superduplex stainless steels in addition to nickel alloys. The use of duplex and superduplex grades has become widespread, as they offer an attractive combination of very good general and pitting corrosion resistance, chloride stress corrosion cracking resistance, strength, toughness and weldability, combined with lower cost than nickel alloys. However, the qualification of duplex and superduplex stainless steels for sour service has necessitated changes in methodology compared with carbon and low alloy steels, as many factors influence performance.

NACE MR0175 ^[1] covers the selection of materials for resistance to sulfide stress cracking (SSC). This document was originally developed for carbon steels, alloy steels and stainless steels, which were simply defined as materials having ≥10.5%Cr. These stainless steels were approved using the same criteria and test procedures developed for carbon steels, i.e. limiting hardness such that the material does not crack in the presence of hydrogen generated by corrosion in the sour medium. With increasing use, it became apparent that qualification of stainless steels for sour service could not be reliably achieved by following carbon steel practices. Corrosion resistant alloys are generally selected on the basis that they will not corrode in service. If this is achieved, then SSC will not occur, as no hydrogen is generated. If CRAs do corrode, they generally show localised pitting or crevice corrosion and it may then be possible for SSC to develop at the site of local corrosion. Limiting hardness of the underlying steel, so that it does not crack in the presence of the hydrogen generated locally, may nevertheless prevent SSC. Hence, there are two factors controlling the sensitivity of a CRA to SSC, (i) resistance to localised corrosion and (ii) resistance to cracking in the presence of hydrogen. Consequently, the qualification of CRAs for sour service needs to reflect environmental and metallurgical variables contributing to pitting/crevice corrosion resistance in addition to those contributing to cracking in the presence of hydrogen. The former is influenced by composition (in particular Cr, Mo, W and N), presence of third phases, the surface finish, and the specific environment (e.g. Cl ^-, pH, H ₂S and temperature). The factors affecting the latter are less well defined but would be expected to be related to hardness and perhaps also to grain size, any third phases and cold work.

Recognition of this complicated situation is reflected in the wording of later editions of NACE MR0175, ^[1] which includes limits of hardness, material condition (e.g. solution annealed or cold worked), temperature, chlorides, elemental sulfur and H ₂S. Further recognition of this is given in the development of EFC 17, ^[2] which gives specific guidelines for testing of CRAs for sour service as an appropriate means of qualification. This paper illustrates that the individual NACE MR0175 ^[1] H ₂S limits for specific UNS superduplex steels should not be considered in isolation but that they should be considered in relation to the various other environmental and material variables relating to a particular application. Consequently, it may be possible to operate superduplex steels at H ₂S levels above the partial pressures quoted in MR0175, ^[1] for example if other environmental or material factors are well within the limits quoted in the appropriate section of MR0175. ^[1] However, appropriate qualification testing would be required. The paper also seeks to illustrate the effects of other microstructural variables that are not addressed in MR0175. ^[1]

To assess applicability for service under specific conditions, it is necessary to perform SSC tests with conservatively severe environment and stress. Two test methods are compared here (i) static, constant deflection C-ring tests and (ii) dynamic, SSRT tests, in addition to unstressed, pitting corrosion tests. The C-ring tests are representative of many service loading situations; whilst SSRT tests are designed to detect any sensitivity to SSC, even beyond the proof stress and up to the UTS, in a fairly short time. Test environments were representative of a combination of worst case service conditions for a particular application, although such a severe combination would not be expected in practice.

Experimental work

Batches of superduplex steel meeting UNS S32760 were procured from three different suppliers (labelled A to C) and nominally solution annealed in the same way, Tables 1 and 2. In addition, a batch of solution annealed superduplex steel meeting UNS S39277 and a batch of solution annealed UNS S32750 steel were procured from two other suppliers and tested for comparison, Tables 1 and 2. The materials were procured in bar form of varying diameters. These three superduplex steels have similar compositions but with small variations in the specified levels of Ni, Mo, W, Cu and N. The relevant MR0175^[1] limits for each grade are given in Table 3. An example of each material was sectioned to allow examination for third phases, measurement of the ferrite content by point counting and estimation of ASTM grain size by comparison with standard charts. The widths of the ferrite and austenite units were compared with the standard charts, not their lengths. Comparison with standard grain size charts does not allow an accurate measurement of grain dimensions or any indication of the directionality of the microstructure but allows a first order comparison of the grain structures. Rockwell C and Vickers (10kg) hardness were measured.

Table 1. Specified composition ranges of major elements in 25%cr superduplex stainless steels and nickel alloy tested

* also %Cr + 3.3%Mo + 16%N ≥ 40
** also Al = 0.1-0.5, Cb = 0.5 max, Fe = 22 min, Ti = 1.9-2.4
NS = not specified

Table 2. Chemical analyses of 25%Cr superduplex stainless steels tested

Table 3. NACE MR0175 Limits applicable to 25%Cr superduplex stainless steel grades tested

NL = no limit specified, SA = solution annealed, CW = cold worked, * no elemental sulfur allowed for cast N09925

Unstressed pitting corrosion coupons from all superduplex batches were exposed at 186°F (85°C) and 212°F (100°C) in 10%NaCl with 3psi (0.2bar) H ₂S and 600psi (40bar) CO ₂ for 720 hours, Table 4. Sulfide stress cracking tests were performed in two environments on one of the UNS S32760 batches that was resistant in the pitting corrosion test (supplier A) and the UNS S39277 grade, which was also resistant in pitting corrosion testing. The SSC test environments were 3.3% and 10% NaCl solutions acidified with 30psi (2bar) H ₂S and 1200 and 1000psi (80 and 67bar) CO ₂ respectively at 186°F (85°C), Table 4. The SSC tests involved (i) C-ring specimens machined from the bars and stressed to 100% of the actual measured 0.2% proof stress in both environments for 720 hours and (ii) SSRT tests, on specimens taken longitudinally from the bars, at a strain rate of 1x10 ^-6s ^-1 in the 3.3% NaCl solution, Table 4. A nickel alloy UNS N09925 sample was tested under the same SSRT test conditions for comparison. The pH during test was calculated as 3.0 in each case. ^[3] The H ₂S partial pressures in each stressed SSC test were well above the maximum values stated in MR0175 but it should be noted that the chloride ion concentrations tested were lower than the maximum values quoted in NACE MR0175 ^[1] and that no cold worked material was tested.

Table 4. Summary of test conditions examined

* calculated using CORMED ^TM software ^[3]

Results

Materials characterisation

Table 5 gives details of the microstructural characterisation of each sample examined and Figure 1 illustrates typical microstructures. One of the steels (UNS S32760, supplier A, which was a 25mm diameter bar) had noticeably finer grain size than the others, approximately ASTM 6, whilst one steel (UNS S32760, supplier C, which was from 100mm diameter bar) had coarsest grain size, approximately ASTM 3, and contained some chromium nitrides within the ferrite grains. There was a general correlation of increasing grain size with increasing bar diameter. Measured ferrite content ranged from 37% (UNS S32760, B) to 59% (UNS S32760, C), although both extremes showed substantial variation from one field of view to the next.

Fig. 1. Microstructures of the various superduplex stainless steel samples examined (length of bar vertical)

Fig.1a) UNS S39277 (175mm bar)  

Fig.1b) UNS S32760, supplier A (25mm bar)  

Fig.1c) UNS S32760, supplier B (50mm bar)  

Fig.1d) UNS S32760, supplier C (100mm bar)  

Fig.1e) UNS S32750 (75mm bar)

Table 5. Summary of microstructural examination and chemical analyses

* PREN = %Cr + 3.3%Mo + 16%N, PRENW = %Cr + 3.3 (%Mo + 0.5%W) + 16%N

Pitting corrosion test results

Two materials (UNS S32760 from supplier C and UNS S32750) showed evidence of corrosion in both unstressed pitting corrosion tests, whilst the other three steels showed no corrosion in either case, Table 6.

Table 6. Results unstressed of pitting corrosion testing of 25%cr superduplex steels

Sulfide stress cracking test results

In the C-ring tests, neither material (UNS S39277 and UNS S32760, supplier A) showed corrosion or cracking in the 10%NaCl, 1000psi CO ₂ test but, in the 3.3%NaCl, 1200psi CO ₂ test, one specimen of UNS S32760 showed localised corrosion and subsequent cracking, whilst the other specimen and the two UNS S39277 samples showed no corrosion or cracking, Tables 7 and 8. In the SSRT tests, the UNS S39277 sample performed better than the UNS S32760 sample (time to failure ratios of 0.25 and 0.13 respectively were obtained for the sour environment compared to air). However, neither superduplex steel performed well, with both materials showing secondary cracking on the gauge section and apparently being susceptible to SSC in this severely stressed test, Table 9. By comparison, nickel alloy N09925 showed substantially better performance than the superduplex steels in the SSRT tests (time to failure ratio of 0.89, with no secondary cracking on the gauge section, which indicates that it is practically immune to SSC ^[4] ) but still showed some reduction of ductility in the sour environment compared to testing in air.

Table 7. C - Ring test results: 30psi H ₂S, 1200psi CO ₂, 3.3% NaCl, 186°F, 720 hours, 100% of 0.2% proof stress, pH = 3.0

Table 8. C - Ring test results: 30psi H 2S, 1000psi CO ₂, 10% NaCl, 186°F, 720 hours, 100% of 0.2% proof stress, pH = 3.0

Table 9. SSRT test results: 30psi H 2S, 120psi CO ₂, 3.3% NaCl, 186°F, pH=3.0

Discussion

Pitting corrosion performance

One of the UNS S32760 steels examined pitted in the both test environments, which had H ₂S partial pressures equal to the limiting value in MR0175 ^[1] . The H ₂S partial pressures were beyond the limit for UNS S32750, which also pitted in the tests. Of the two batches of superduplex steel (S32760, supplier C and S32750) that were attacked in the pitting corrosion tests, one (S32760, supplier C) had the coarsest grain size (ASTM 3), highest ferrite content (59%), lowest Mo content and lowest PREN/PRENW (40.1 and 41.1) values of all steels examined (although one steel that did not corrode had very similar Mo content). The S32760, supplier C steel also had chromium nitride precipitates, which would tend to remove Cr and N from solid solution and hence reduce the pitting corrosion resistance. However, its hardness (23.5HRC, 266HV10) was towards the low end of the range measured. The other steel batch that corroded (UNS S32750), had the joint lowest PRENW value (41.1) but moderate PREN (41.1 also, as there is no W), grain size (ASTM 4.5) and ferrite content (43%), no nitrides and the lowest hardness of the steels examined (22HRC, 254HV10). It was also the only batch of steel to have no deliberate addition of Cu or W, although its Mo content was the second highest of the steels examined. The N contents of the two batches of steel that corroded were in the middle of the range examined. The materials that corroded were also from two of the three largest diameter bars tested (100mm and 75mm diameter respectively for the S32760, supplier C and S32750 grades). The other large diameter bar (175mm), which was not attacked, had higher PREN and PRENW (42.0 and 43.3 respectively) and 1.93%Cu. The three batches that did not corrode (S32760 A and B, and S39277), had a range of grain size (3.5 to 6), ferrite content (37-49%) and hardness (23.5-26HRC and 265-278HV10), no nitrides, PREN in the range 40.7-42.0, PRENW in the range 41.6-43.3 and 0.54-1.93%Cu. This indicates a compositional influence on pitting resistance in sour media, with pitting occurring in the two materials with lowest PRENW, one of which was the only material tested without Cu. There may also be a microstructural influence, with coarse grain size, high ferrite content and associated nitride precipitation (which are to some extent linked) probably being detrimental, although there are insufficient data here to prove this, as only one of the two batches of steel that corroded had these features. Intermetallic phase precipitation would also be expected to be detrimental to pitting performance in sour media but none of the steels examined here contained intermetallic phases.

Sulfide stress cracking performance

In the C-ring tests, the S32760 from supplier A performed worse than the S39277 in the 3.3%NaCl test environment (no cracking was found in the latter material) and in the SSRT tests in the same environment, although both showed susceptibility to SSC in the SSRT tests. This reflects the greater severity of the SSRT test compared to the C-ring test. Also, it indicates that the S32760, supplier A sample had greater susceptibility to SSC than the S39277 sample. As the former had a finer grain size, and similar ferrite content and hardness it seems that the influence here is a compositional one. The PREN and PRENW values for the S39277 steel are about 4% higher than for the S32760, supplier A steel and the former grade also has 1.93%Cu compared to 0.70% in the latter. The fact that only one sample of S32670, supplier A pitted and cracked in the two series of C-ring tests suggests that the material was near its limit in this particular test environment. The fact that this specimen failed in the 3.3% NaCl solution and that no specimens failed in the 10% NaCl solution suggests that this increase in chloride level was not significant in the present tests. It may be noted that the lower chloride solution was associated with a higher partial pressure of CO ₂ but the calculated pH in the two solutions was essentially the same, i.e. very close to 3.0.

Sulfide stress cracking test methods

The SSC test media employed had H ₂S partial pressures of 30psi, i.e. well in excess of the limits specified in NACE MR0175, ^[1] but the chloride levels used and the hardnesses of the materials tested were below the MR0175 ^[1] limits. The C-ring tests indicated that, provided the service stress is suitably reproduced by a C-ring specimen, i.e. displacement controlled and no higher than the 0.2% proof stress, then superduplex stainless steels may be used at H ₂S partial pressures up to 30psi, i.e. exceeding the MR0175 ^[1] H ₂S limits, provided that either the material condition or other environmental factors (such as chloride concentration, temperature or pH) are sufficiently favourable and within MR0175 ^[1] limits. Appropriate testing is required to demonstrate suitability for service when such high partial pressures of H ₂S are involved. It should be noted that the samples did not contain any welds and that, if present in practice, welds should be tested as they locally alter microstructure, hardness and composition. One other aspect that should be considered when testing to qualify for service is the direction of stressing relative to the directionality of the microstructure. Assuming that cracks will be impeded as they try to travel perpendicular to the microstructural alignment, it may be noted that the C-ring tests performed here might have been more onerous than, for example, bend tests or tensile tests performed on longitudinal specimens due to less favourable orientation of the stress with respect to the microstructure in the former case. The direction of applied stress with respect to microstructural alignment was different in the C-ring and SSRT tests performed here but it was not possible to determine what the effect of this was due to the different magnitudes of the applied stresses and the fact that the SSRT test is dynamic whilst the C-ring test is static.

Ferritic-austenitic steels undergo low temperature creep over a range of stress starting below the 0.2% proof stress ^[7] and this will tend to give stress relaxation in a C-ring test to a level below the initial applied stress, as recognised in EFC 17. ^[2] In a dead weight, constant load test, stress relaxation cannot occur, but dynamic straining will occur and will tend to encourage cracking and corrosion, as the passive surface film will be continually disrupted. Some dynamic straining will occur during the early part of a C-ring test but this will tend to diminish fairly rapidly as the stress relaxes. Testing of superduplex stainless steels under cathodic protection at -1100mV _SCE has demonstrated a very significant difference between results of constant load and constant deflection tests. Under constant load, failure may occur in a few tens of hours at 100% of the 0.2% proof stress, whereas failure in constant deflection tests required several percent plastic strain and a time of around 30 days. ^[5] The difference between constant load and constant deflection SSC tests for sour service has not been fully explored but the mode of loading likely to be encountered in service should be recognised and reproduced in SSC tests as far as possible. Where service will involve constant load conditions or where there is any doubt about the mode of service loading, constant load tests are recommended. This will be particularly important if trying to qualify material for service under severe conditions, where limiting conditions are being approached, as was the case here. The SSRT tests demonstrated that the superduplex steels examined were not immune to SSC in the C-ring test environments and so presumably the C-rings did not crack because they were at a stress below the threshold for SSC initiation.

Table 10 compares the stress severity of the various SSC test methods. SSRT is the most severe SSC test method, owing mainly to the high stresses applied and choice of a dynamic strain rate controlled to maximise the opportunity for SSC. The testing time is short, of the order of 24 hours, making it an extremely rapid compared to statically loaded tests. The high stress and critical strain rate are not relevant for practical service but the speed of the test makes it attractive. However, particularly for materials that are unlikely to be completely immune to SSC in the test environment, such as superduplex steels in sour environments, if such tests are to be used for qualification purposes, a reliable pass-fail criterion is required. Some authors have argued that a fairly high level of ductility (>65% of the inert atmosphere value) must be retained in an SSRT test if the material is to be considered practically immune or mildly susceptible to SSC, ^[4] whilst other authors have argued that, at least for 13%Cr steels, materials are acceptable if they demonstrate 50% loss of ductility in an SSRT test and may be fit for service even when they demonstrate 65% loss of ductility in an SSRT test. ^[6] In the absence of such a proven pass-fail criterion, the SSRT test alone cannot be used to qualify materials that are likely to be near their environmental limits or to identify limiting service conditions, although they can be used to identify materials that are highly resistant/immune to their environment, e.g. if the ductility in the environment is >90% of that measured in an inert atmosphere and there is no evidence of cracking. ^[4]

Table 10. Comparison of different test techniques in relation to severity of stressing

* stress falls from initial level due to creep over initial part of test
PS = Proof stress
UTS = Ultimate tensile stress

Environmental factors

NACE MR0175 ^[1] gives conservative limits of H ₂S partial pressure to avoid SSC in superduplex steel and also some associated limitations with respect to chloride ions, pH and temperature. However, fairly small shifts in pH and reductions in chloride ion content can significantly improve SSC resistance for a given H ₂S partial pressure. It is also significant that in practice in the oil and gas industry, low pH is usually associated with low chloride levels and vice versa. Hence, tests at high chloride level and low pH, such as are used to qualify materials for NACE MR0175 ^[1] , may be unrealistically conservative for most practical applications. ^[7]

The presence of bicarbonate ions gives an increase in the pH of produced fluids with CO ₂ and H ₂S compared to condensed water without bicarbonate. An amount of 10 meq/l will increase the pH by around 1 unit. The presence of fairly low levels of bicarbonate ions and the associated increase in pH will reduce the SSC tendency of duplex and superduplex stainless steel. ^[7] Previous work has demonstrated that the tolerable partial pressure of H ₂S is increased by more than one order of magnitude when the pH value is increased from 2.5 to 3.9. ^[8] The presence and positive effect of bicarbonate ions in practice in high chloride produced waters have also been reported by Barteri et al ^[9] and Francis et al. ^[7]

Although no effect was found over the narrow range explored here, the performance of CRAs is also particularly affected by the chloride concentration in the process fluid, ^[7-10] as the chloride ion is particularly active in the pitting process, which is a precursor to SSC. The limiting H ₂S partial pressure of a particular grade may increase as the chloride concentration decreases. However, the large spread of pass/fail data available in the literature makes the setting of absolute limits for individual factors difficult. Testing using C-rings stressed to 100% of the actual yield strength indicated that a decrease of the chloride content by a factor of ten increases the H ₂S limit also by about a factor of 10. ⁸

Also, it has been recognized that the temperature for least resistance to cracking varies by alloy group. For duplex and superduplex stainless steels the critical temperature is generally found to lie in the range of 70 to 110°C ^[7] , which covers the range used for testing here. At higher or lower temperatures, H ₂S partial pressure limits would be expected to be greater, although it should be recognised that a chloride stress corrosion cracking mechanism may become limiting at higher temperatures.

Conclusions

1. Laboratory tests with 30psi H ₂S on solution annealed superduplex material have shown that, for constant deflection at nominally 100% of the 0.2% proof stress, some superduplex steels may be resistant to sulfide stress cracking at higherpartial pressures of H ₂S than currently allowed by NACE MR0175 ^[1] , if other environmental or material conditions are favourable. However, resistance to the sour test media varied significantly between the batches of steel tested and one pitted in an environment with H ₂S partial pressure equal to the appropriate MR0175 ^[1] limit. Consequently, testing as recommended in EFC 17 ^[2] is appropriate to qualify materials for sour service where one or more variables meet or exceed the NACE MR0175 ^[1] limits.
2. Different batches of superduplex stainless steel, either within a specific UNS composition range or from different UNS composition ranges, may show distinctly different resistance to pitting and sulfide stress cracking sourenvironments. In particular, higher PREN/PRENW values are beneficial while coarse grain structure with high ferrite content and nitride precipitates, as found in the largest diameter UNS S32760 bar tested, may be detrimental. No effectof hardness over the range 22-26HRC was found for the solution treated materials tested.
3. It is not possible to define a single maximum partial pressure of hydrogen sulfide to which superduplex steels are resistant. Factors that may affect pitting and SSC performance in sour media include (i) fairly small variation inlevels of primary alloying elements within the specification limits (i.e. Cr, Mo, N and W, as recognised in PREN/PRENW expressions and possibly Cu), (ii) microstructure (e.g. grain size, phase balance and presence of chromium nitridesor intermetallics, in addition to hardness and cold work), (iii) other details of the service environment, in particular pH, Cl ^- and temperature and (iv) the type of stressing (e.g. load or displacement control).
4. C-ring tests at 100% of the 0.2% proof stress showed no cracking in an environment that led to a reduction of ductility of ≥75% in SSRT tests on two superduplex steels. Slow strain rate testing is a very severe test for evaluating performance of superduplex steels in sour service.If it is to be used for qualification for service, a reliable pass/fail criterion is required. For service under constant load conditions, dead weight constant load tensile tests are recommended for qualification purposes.

References

1. NACE MR0175-2002, 'Standard material requirements for sulfide stress cracking resistant metallic materials for oilfield equipment', NACE International, 2002.
2. EFC Publication 17, 'Corrosion resistant alloys for oil and gas production: guidance on general requirements and test methods for H ₂S service', The Institute of Materials, 1996.
3. Sociètè Nationale Elf-Aquitaine (production), CORMED ^TM Software, V1.1, copyright SNEA, 1990.
4. DR McIntyre, RD Kane and SM Wilhelm, 'Slow strain rate testing for materials evaluation in high pressure H2S environments', Corrosion, 1988, 44(12), 920-926.
5. P Woollin and W Murphy, 'Hydrogen embrittlement stress corrosion cracking of superduplex stainless steel', Proc conf 'Corrosion 2001', NACE International, paper 01018.
6. B Kermani, P Cooling, JW Martin and PI Nice, 'The application limits of 13%Cr tubular steels for downhole duties', Proc conf 'Corrosion 98', NACE International, paper 98094.
7. R Francis, 'The Role of Environments and Metallurgical Variables on The Resistance of Duplex Stainless Steels To Sulfide SCC', Proc conf 'Corrosion 97', NACE, paper 12.
8. J Oredsson and S Bernhardsson, 'Performance of high alloy austenitic and duplex stainless steels in sour gas and oil environments', Materials Performance Vol. 22, 1983, No 1, 35-42.
9. M Barteri, R Bruno, A Tamba and G Montagna, 'Critical conditions for application of duplex stainless steels in sour environment'. Proc conf 'Pipe Technology', AIM, 1987, 525-542.
10. M Eriksson, 'The applicability of duplex stainless steels in sour environments', Proc conf 'Corrosion 90', NACE, paper 64.