Briony K Holmes and Stuart Bond
Paper presented at Corrosion 2010. San Antonio, Texas, USA, 14-18 March 2010.
To address the widely held concern that ISO15156/NACE MR0175 limits for sour service cracking resistance of Type 316/316L stainless steel in oil & gas production environments were excessively conservative, a program of laboratory studies was undertaken testing parent materials in compliance with the ballot requirements of this standard. In addition, typical weldments were tested under the same conditions to establish whether there were significant differences in cracking resistance. Whilst the existing parent material limits have been shown to be overly conservative and thus can be relaxed to more aggressive conditions, the data submitted to ballot for changes to the limits were based only upon cases where weldments also passed.
In some more extreme conditions weldments were observed to fail the criteria applied whereas parent materials passed, indicating caution is needed in further extending bounding limits. Recommendations are given for the testing of corrosion-resistant alloys where the intention is to ballot ISO15156/MR 0175 for limits applicable to all equipment forms (which may be welded), that new data are generated which include assessment of typical weldment performance alongside parent materials to provide confidence that these limits remain within boundary conditions demarking lower resistance of weldments compared with parent material.
Keywords: austenitic stainless steels, corrosion testing, GTA welding, H2S, ISO, Mo additions, nitrogen, sour service, standards, stress corrosion cracking, pipes, welded joints, hardness, yield strength
Welded pipes, pipework and components made from the Type 316/316L grade of austenitic stainless steel are widely used in the oil and gas industry to handle sour fluids (i.e. containing H2S). This material is susceptible to stress corrosion cracking in sour brines, and thus the use of this material is restricted according to the limits detailed in ISO 15156/NACE MR0175, which are detailed in Table 1. The limits for 316/316L were considered to be conservative by industry but adequate data were not available to modify them. However, the standard allows for limits to be updated based on generated data, thus corrosion-resistant alloy (CRA) materials may be qualified for sour service, or the current limits for a material may be changed, in accordance with Annex B of ISO15156-3.
Table 1 sour service limits for 316/316l according to ISO 15156-3:2003/CIR.1:2007 
|Chloride, mg/l||Temperature, °C (°F)||p,H2S bar (psi)||pH|
||No limits set
||No limits set
||No limits set
* = Limits for UNS S31600 and UNS S31603
A testing program was thus performed to produce results that were then used to ballot for changes to ISO 15156/MR0175. Testing was performed in accordance with the standard, which requires only parent material to be tested. The ISO limits are set for parent material alone, and whilst some applications such as downhole tubulars and some components in valves etc are used with material in the unwelded condition, welding is commonly required for pipework, pipeline/flowline and pressure vessels etc. This work has sought to define extended limits for parent 316/316L material (typical commercial material available now is often called 'dual certified' complying with 316 strength and 316L compositional requirements) and to compare the behavior of welded and parent material, as features associated with welds ie weld oxide, surface roughness and weld profile, can affect a material's resistance to corrosion and SCC.[1,2] Therefore, since pipes always require welding together, weldments were also tested (the results of the weld testing are not reported here other than to note that for conditions reported, typical weldments did not perform worse than parent material).
In addition, due to concerns over the effect of nitrogen levels in the parent material on the corrosion resistance and thus environmental cracking resistance of the material, parent materials containing two levels of nitrogen (0.05 and 0.1wt%) were tested.
Three 10 inch (273mm) diameter, 12.7mm wall thickness pipes of three different heats (in accordance with ISO 15156), all complying with UNS S31603 chemical composition and UNS S31600 strength were used (Pipes 1-3). Analyses were performed using optical emission spectrometry (OES) and inert gas fusion. Molybdenum content varied from 2.04 to 2.08wt% in Pipes 1-3. An additional pipe, Pipe 4, of lower N content (0.05 vs 0.10-0.11wt%N in pipes 1-3), and 2.00wt% Mo was also used for some tests. In order to assess any possible effect of composition on pitting corrosion resistance, the pitting resistance equivalent number (F PREN ) was calculated for each of the four pipes using the formula from ISO15156-3, shown below in Equation 1.
FPREN = wCr + 3,3(wMo + 0.5wW ) + 16wN (1)
wCr is the mass fraction of chromium in the alloy, expressed as a percentage of the total composition;
wMo is the mass fraction of molybdenum in the alloy, expressed as a percentage of the total composition;
wW is the mass fraction of tungsten in the alloy, expressed as a percentage of the total composition;
wN is the mass fraction of nitrogen in the alloy, expressed as a percentage of the total composition.
Girth welds were produced in each pipe by gas tungsten arc (GTA) welding using UNS S31683 filler metal. It is important to test weld surfaces representative of those that will be in contact with corrosive fluids in service, and so the welds were manufactured using typical industry procedures, and tested in the as-welded condition. The welds were not pickled, which is typical for exploration and production applications.
Vickers hardness measurements were performed, using a 10kg load, on a section taken through each pipe.
The 0.2% proof strength of the material was measured on round tensile specimens at various temperatures to provide data for loading the specimens for the environmental cracking tests (in compliance with paragraph B.3.4, ISO 15156-3, 2003).
Firstly, Annex B, paragraph B2.4, ISO 15156-3 sets out the 'Requirements of use of laboratory testing as a basis for proposing additions and changes to Annex A' (Annex A being the tables of recommended environmental cracking limits). Testing was performed in line with these requirements and the further requirements of section B3 'General requirements for tests'. Annex B, Table B.1 suggests that stress corrosion cracking (SCC), sulphide stress cracking (SSC) and galvanically-induced hydrogen stress cracking (GHSC) testing may all be required. However, only SCC tests were performed here because industrial experience and an earlier ballot had been successful in demonstrating that SSC and GHSC were not applicable to Type 316/316L stainless steels. SCC tests were performed in accordance with ISO 15156-3, EFC17 and NACE TM0177. Of these, ISO15156 gave the test requirements, EFC17 gave details of the four-point bend test method, and NACE TM0177 gave additional testing details.
Environmental cracking tests were performed on four point bend specimens in order to test the weld roots intact; parent material specimens were fully machined. Machined surfaces of all specimens were ground to a 320grit finish for testing. The specimen dimensions were 210x25x10mm. The specimen thickness was chosen to be as close as possible to the wall thickness of the pipe following machining. It was felt to be important to test the full wall thickness as taking thinner specimens would have been less representative of the actual residual stress distribution in the pipe and presently the industry is debating validity of thin specimens. Specimens were taken longitudinally from the pipes. An example of some parent material specimens (after testing) is shown in Figure 1.
Fig.1. An example of parent specimens after test
The specimens were loaded in constant displacement, using bolt loaded jigs made from UNS N10276, to 100% of the 0.2% proof stress of the parent material at test temperature (as required by ISO 15156-33), which was the mean of the data from Pipes 1-3. The jig material was chosen for its corrosion resistance and its mechanical stiffness, and the jigs were electrically isolated from the specimens. Figure 2 shows a set of welded specimens in the jigs. Strain gauges were attached to the tensile face of the parent specimens when applying the strain. The strain gauges were applied to the compressive face of the welded specimens in order to leave the weld oxide intact; the strain being applied using calibration of the measured tensile response from a specimen with gauges on both surfaces.
Fig.2. Welded specimens in the four-point bend jigs
In converting the 0.2% proof stress to a strain to apply to the specimens, the following formula, equation 2, was used:
σ0.2% = 0.2% proof stress
E = Young's modulus
εapp = applied strain, measured in units of microstrain.
SCC test environments
The test environments were based on 1 000mg/l chloride (nominal pH3.5, reflecting condensed water) and 50 000mg/l chloride (with some bicarbonate to give nominal pH4.5, reflecting produced water). Chloride was added as sodium chloride. Partial pressures of H2S were 0.01, 1 or 10bara, whereas that of CO2 was maintained at 10bara in all cases. Water vapor was accounted for in the total test pressures. The test temperatures were 60°C, 90°C, and 120°C and held constant for each test to ±2°C. The test solutions were deaerated before introduction into the test vessels to give an oxygen content of <10ppb. Continuous gas purge was applied throughout the tests using the lower partial pressures of H2S (0.01 and 1 bara); gas was purged weekly for the 10bara H2S tests. The pH values of the solutions under the test gas were predicted using CORMED2(1) - a commercial software product.
1CORMED2 ©, 2005 : CD-ROM, Crolet, Jean Louis
The solution pH was measured under 1bara CO2 gas purge at ambient pressure and room temperature prior to introduction to the autoclave. Following testing, and cooling in the autoclave, an aliquot was drawn off and pH measured under the test gas and/or under 1bara CO2 at room temperature for comparison with the predicted values of pH. The pH values of the solutions under 1 bara CO2 were also predicted for comparison. The difference in pH over the course of the test was calculated. EFC17 states that the pH in buffered environments shall be maintained within 0.2pH units, and this was maintained during the tests.
The pass/fail criterion for cracking was in accordance with ISO 15156, which states that 'No cracks are permissible'. The specimens were assessed visually at x10 magnification. Metallographic sectioning was performed where cracking was suspected, or one section was taken per triplicate set of specimens where no features were visible. In cases where corrosion may have occurred without evidence of cracking, the pass/fail criterion agreed with the Sponsors was that the maximum diameter of the corroded location was <0.1mm. No such criterion is stated in ISO15156.
There were some small compositional differences between the pipe materials (Table 2), but the calculated FPREN was similar in all cases. Pipe 4 had the lowest FPREN , which was to be expected because it had the lowest Mo and N contents. The effect of the lower N content was seen most obviously in the strength data for the pipe materials (Table 3, Figure 3), where Pipe 4 was the weakest at each temperature as would be expected as a material to Type 316L requirements in contrast to the 'dual-certified' materials. All the pipes showed a similar trend in decreasing strength with increasing test temperature.
Table 2 Elemental chemical analysis results for the 316/316L parent materials from pipes 1-4
| ||Element, wt% (m/m)|
* FPREN = wt%Cr + 3.3 x (wt%Mo+0.5wt%W) + 16 x (wt%N)
Table 3 tensile test results for parent 316/316l pipe material at the temperatures at which scc tests were performed
| ||0.2% proof strength/MPa|
Fig.3. Measured 0.2% proof strengths of parent materials tested
Figure 1 shows a typical parent steel sample. In each pipe, the microstructure of the parent was fully austenitic (Figure 4). Table 4 shows the hardness values for each of the pipes, which confirmed that the hardness of the pipes did not exceed the hardness limit of 22HRC (circa 250HV, but there is no known exact correlation for austenitic stainless steels) in Table A.2, ISO 15156-3:2003/MR0175. Hardness values near this maximum would not be present in commercial material so it is not viable to test material at hardness very close to this upper limit whilst complying with the materials condition requirements. Indeed similar debate has taken place on the committees which govern the ISO standard (ISO 15156 Maintenance Panel, NACE TG299 Oversight Committee and ISO TC67/WG7).
The specimens were strained in four-point bend in order to permit the weld root to be tested intact. This arrangement then meant that they were loaded in constant displacement to allow testing in an autoclave as is typical for CRA weldment testing. Following straining, the specimens demonstrated strain relaxation immediately after loading, thus they were re strained to the required strain. The phenomenon was most marked in the welded specimens.
Table 5 shows the most severe test conditions under which no cracking or corrosion of parent or welded material specimens was observed. Cracking was observed under more severe test conditions. In particular, welded material cracked under conditions where parent material had not, indicating a borderline pass/fail condition for this material. This latter data remains confidential to the Sponsors of this work.
Table 5 SCC testing results on parent material. Environment includes 10bara CO2 in each case
||Pipe 1 - pass
Pipe 2 - pass
Pipe 3 - pass
||Pipe 1 - pass
Pipe 2 - pass
Pipe 4 - pass
||Pipe 1 - pass
Pipe 2 - pass
Pipe 3 - pass
||Pipe 2 - pass
Pipe 3 - pass
Pipe 4 - pass
Notes: * = no cracking, no pitting or crevice corrosion >0.1mm diameter. - = not available.
Diagrams of the parent material data produced during this test program are shown in Figures 5-6 against the current ISO 15156-3 sour service limits for Type 316 stainless steel. The ISO limits include the corrigenda published recently by ISO including changes from data balloted by Kane. Figure 5 summarizes the test results from the 1 000mg/l chloride, nominal pH3.5 environment. Figure 6 summarizes the test results from the 50 000mg/l chloride, nominal pH4.5 environment.
Fig.5. Data A and B submitted for ISO ballot, based on parent material SCC tests. Environment contained 1 000mg/l chloride, nominal pH3.5
Fig.6. Data C and D submitted for ISO ballot, based on parent material SCC tests. Environment contained 50 000mg/l chloride, nominal pH4.5
Based upon the results of the test program, ballot tables for the following conditions were prepared, and at the time of writing await response from the ISO committee, all with 10bara pCO2:
These were all conditions where welded material had been tested under the same conditions and had passed also, so that the data can be more confidently applied to materials selection for welded products. However, specific testing will still be required to qualify materials and welding procedures for service application.
||1 000mg/l chloride, 120°C, 1bara pH2S, nominal pH3.5 (PASS)
||1 000mg/l chloride, 90°C, 10bara pH2S, nominal pH3.5 (PASS)
||50 000mg/l chloride, 90°C, 0.01bara pH2S, nominal pH4.5 (PASS)
||50 000mg/l chloride, 60°C, 10bara pH2S, nominal pH4.5 (PASS)
Further to this, it is suggested that tests include typical weldments to support the limits for the 'any application' tables. Where the results have impacted upon the potential application envelope of 316/316L, or indeed other CRA, in sour service, but where the test results are not suitable for ballot (for the accompanying reasons below), these should be reported to NACE TG299, Oversight Committee for ISO 15156/MR0175. This will benefit industry awareness, specifically regarding derivation of knowledge of potential boundary conditions through accumulation of data over time.
Effect of nitrogen and molybdenum content
There was no apparent effect of nitrogen content on the corrosion or SCC resistance of the 316/316L material in the tests performed. This was perhaps not surprising as the FPREN of Pipe 4 was only 0.5 0.7 lower than that of Pipes 1-3.
SCC resistance. Early on in the program it became clear that 316/316L material could withstand conditions much more aggressive than the ISO limits in terms of temperature and pH2S, and therefore new limits were derived based upon the parent material performance in this test program. During the extended test program (not reported here) failures were seen from both SCC and corrosion under even more aggressive environmental conditions.
The welded material was less crack resistant than the parent material, and suffered failure by cracking of heat tint oxidized HAZ/parent metal in an environment more severe than those balloted (ballot data are shown in Figures 5-6), whereas parent material passed at a higher temperature in a similar environment (all other variables the same). The welds' surface finish was rougher and more oxidized than the parent material, and had not been pickled (as is typical of industrial practice). It is generally accepted that CRA materials' cracking resistance in H2S containing media is dependent upon the robustness of the surface oxide layer (passivity case) or immunity (inherent metal resistance to the environment without an oxide) and it is likely that these two factors contributed to the initiation of the corrosion that preceded the cracking. Furthermore, this perhaps suggests a threshold chloride content for welded material under these conditions, as it is noted that there is no temperature limits in ISO for lower chloride content of <50mg/l.
The test conditions employed during this project were devised primarily to fit the objective of this work, which was to define more accurate sour service limits for 316L in oil & gas production applications and to ballot for changes to ISO 15156. In addition to meeting that objective, information has been gained on the individual impact of each of the variables (temperature, chloride content, pH2S, pH, welding) on the SCC resistance of Type 316/316L stainless steel in sour service. Within this test program, changing the chloride content and pH2S had the largest effect, but it must be noted that the change in magnitude of the chloride content was generally the greatest ie a factor of 50 (from 1 000 to 50 000mg/l), compared to factors of 10 for the pH2S, log10 for the pH, and up to a factor of 2 for the temperature. Although considerable further work is required to define the absolute and relative effects of each of these variables, this data set has shown that a significant increase in severity due to one variable can outweigh a reduction in severity of a combination of the other variables.
Corrosion resistance. Overall, no failures due to corrosion were seen during the test program, but there was some staining observed. Changing pH within the limits of this test program had a greater effect on the extent of corrosion than on the risk of SCC under these test conditions. Changing the chloride and pH2S had the largest effect within the confines of this test program. These data showed corrosion to be the precursor to SCC. Therefore, further testing in higher chloride content (and low pH2S) environments is of interest to explore the existence of a limit in terms of chloride content for corrosion and SCC resistance of 316/316L materials. Salts that provide higher chloride concentration/activity in solution, such as magnesium or calcium chloride, may be employed to assist in deriving empirical data to support this hypothesis. The work being performed in TWI's Core Research Program into 'Defining high temperature pitting resistance limits for welded corrosion resistant alloys' has already begun to produce data related to a threshold for corrosion initiation in these environments. 
Stressing of welded specimens
The work has highlighted a number of areas in which the SCC testing of welded austenitic stainless steels in sour environments is not fully defined within the current test standards which allow a range of methods derived mainly for parent material qualification. This includes the most appropriate specimen geometry, although four-point bend is generally preferred, and ways of ensuring that stress concentration and relaxation effects are controlled.
It was considered best practice in this project to use full thickness specimens where possible, to obtain a moderate through-thickness stress gradient. Also, misalignment apparently can have a significant effect on stress concentration effects at the weld toe in thin specimens, as the two sides of the weld can have quite different thicknesses. There is no evidence to suggest that this affected the results of the current work but it is noted that the misalignment was fairly low in this case (primarily because efforts were made to use specimens where the misalignment was minimal). Ideally a constant load test specimen would be used, to avoid stress relaxation effects, but the work has supported the view that it is important to test with the weld profile intact, hence implying that it would be ideal to test rectangular rather than round cross-section specimens, which would introduce significant practical difficulties.
There are also uncertainties over the validity of applying strain gauges on the test face, as this will inevitably involve removing some of the weld heat tint and could change the stress concentrating effect of the weld. It was noted also that the weld roots from different specimens had quite different geometries, with some having a smooth shape and little abrupt change in section at the fusion line, with substantial distortion of the adjacent HAZ/parent material, whilst others had a more abrupt change of geometry at the weld root, with less adjacent distortion.
Further guidance is required in the appropriate test standards if testing of welded austenitic stainless steel is to become as reproducible as testing of parent materials and presently initiatives are underway under EFC and NACE to address this issue.
Four-point bend testing on parent and welded dual-certified 316/316L stainless steel, to ISO 15156-3:2003 has shown that these materials are resistant to cracking in sour environments, under a range of chloride concentrations (1 000 and 50 000mg/l) and temperature (60-120°C) significantly beyond the current limits set in ISO 15156-3. Triplicate parent and welded specimens were used in all cases.
Welding reduced the chloride SCC resistance of the 316L dual certified stainless steel in very severe conditions. In order that future ballots to extend the environmental limits for CRA materials in the 'any equipment' category of ISO15156-3 remain applicable to welded product, it is recommended to include weldments to assist in deriving indication of conditions which may be borderline, accepting that each weld procedure must be individually qualified.
These results suggest that, for the range of environments examined, the most important variable in determining the SCC resistance in sour environments was the chloride content, followed by the temperature combined with the pH2S and pH. These results also suggest that chloride and pH were the main factors in corrosion resistance of 316/316L in sour environments.
There was no significant difference in the corrosion and SCC behavior of the low and high nitrogen 316/316L stainless steels in the tests performed.
ISO 15156 should contain information stating where welded material has not performed as well as parent material as a guide for industry.
A limit for corrosion acceptable in sour service cracking resistance tests should be defined in e.g. EFC17 and/or ISO 151561-3/MR0175.
SCC testing of welded austenitic stainless steels for sour service is not fully defined within the current test standards. For example, there is a need to consider further the distribution of strain for welded CRA specimens with intact roots, subsequent relaxation response and potential deformation of the specimen under constant deflection at elevated temperature.
More data on actual material response and implications for test parameters to allow assessment of cracking resistance of weldments is required. This is particularly relevant in considering higher temperature completions and flowline operations, and is recommended for industry review and incorporation in guidance and standards (EFC17 and ISO 15156/MR0175).
ISO 15156 states that Type 316/316L should not be used if intentionally cold worked, but cold work may occur unintentionally during installation or welding. Techniques to test plastically strained material should be explored; the only option may be to use tensile specimens, which presents difficulties in utilizing intact weldments.
This work was carried out in the Joint Industry Project 'Definition of sour service limits for welded type 316L and other cost effective CRAs', conducted 2005-2008 by TWI. The project was sponsored by BP Exploration Operating Co Ltd, Chevron Energy Technology Company, ConocoPhillips, ENI SpA, Petrobras, Saudi Aramco and Statoil who kindly provided permission to publish the results.
- Eid NMA, 'Localized corrosion at welds in structural steel under desalination plant conditions Part 1: Effect of surface roughness and type of welding electrode', Desalination, Vol.73, 1989, pp.397-406.
- Turner S, Robinson FPA, 'The effect of the surface oxides produced during welding on the corrosion resistance of stainless steels', Corrosion, Vol.45, No.9, September 1989, pp.710-716.
- NACE MR0175/ISO 15156-3: 2003, 'Petroleum and natural gas industries - Materials for use in oil and gas production - part 3: Cracking-resistant CRAs (corrosion resistant alloys) and other alloys', ©NACE/ANSI/ISO.
- Kane, 2006: 'AISI 316 (UNS31600) Final Ballot - May 14 2006', Ballot 2007-3, Honeywell Process Solutions.
- EFC17, 2002: 'Corrosion resistant alloys for oil and gas production: guidance on general requirements and test methods for H2S service (EFC 17): a Working Party report', 2nd ed., European Federation of Corrosion publications (number 17), Maney Publishing, London.
- NACE TM0177-2005, 2005: 'Laboratory testing of metals for resistance to sulfide stress cracking and stress corrosion cracking in H2S environments: Laboratory testing of metals - Revised', NACE International.
- 'Petroleum and natural gas industries - Materials for use in H2S-containing environments in oil and gas production - Part 3: Cracking-resistance CRAs (corrosion -resistant alloys) and other alloys', International Standard ISO 15156-3:2003 incorporating Technical Circular 1:2007, Published 2007-12-15, Ref. No. ISO 15156-3:2003/Cir.1:2007(E).
- Lee C-M: 'Corrosion Resistant Alloys - Improved approach to material selection and environmental limits', submitted to the 13th Middle East Corrosion Conference and Exhibition, 2010.