Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

The effect of low H2S concentrations on welded steels (March 2000)

R J Pargeter

Paper 143, Corrosion 2000, NACE International, Orlando, Florida, USA, 26-31 March 2000.


It is generally recognised that the risk of sulphide stress corrosion cracking in ferritic steels increases with hardness, and maximum hardness limits are the basis of guidelines for avoidance of cracking in many standards. It is also clear that higher hardnesses are tolerable in less severe environments. This project has investigated the effects of environments containing low levels of H 2 S on welded ferritic steels. The concern is over environments which come just above the NACE MR0175 [1] limit of 0.05psi partial pressure of H 2 S. Accordingly, tests have been performed, principally using environments with up to ten times this partial pressure, and also in the presence of different partial pressures of CO 2 . Some tests were performed using higher levels of H 2 S.

Threshold hardness levels show a clear variation with CO 2 partial pressure, and with H 2 S partial pressure over the range 0.05 to 2psi. In many mildly sour service situations, at or near the NACE MRO175-98 H 2 S limit, relaxation of hardness limits is reasonable. Maximum permissible hardness levels for welded steels directly exposed to mildly sour service environments have been presented as a function of H 2 S partial pressure and pH, and are applicable for total pressures of up to 2000psi (138 bar).


In designing pipeline and process equipment for handling sour (containing H 2 S) products, limitations are frequently imposed on material hardness. For oil field equipment, the NACE MR0175 standard [1] recommends a maximum hardness of 22HRC (approximately 248 HV) in carbon and low alloy steels to avoid sulphide stress corrosion cracking (SSCC). Although this standard does not apply to the refining industry [2] , close attention to material hardness is also necessary in process equipment as illustrated in NACE RP0472 [3] , which calls for maximum hardnesses of 200HBN (approximately 210HV).

Current guidelines given in NACE MR0175-99 are applicable for all environments with H 2 S partial pressure above 0.05psi, and total pressure above 65psi (for gas systems). It is, however, well known that the safe combination of stress and microstructure, required to avoid hydrogen cracking, is dependent on hydrogen concentration in the steel. At lower hydrogen concentrations, more severe microstructural (hardness) and stress conditions are tolerable. It is therefore anticipated that there will be a gradation of safe hardness levels with H 2 S partial pressure near the current limit, and also that there will be an effect of pH on safe hardness levels.

The present programme of work has therefore been carried out to determine the performance of C-Mn and low alloy steel welds in environments containing H 2 S levels at and slightly above the current NACE limit of 0.05psi partial pressure. While it is recognised that factors other than hardness can affect the risk of cracking, the aim of this project was to demonstrate the extent to which hardness limits can be safely relaxed under such 'mildly sour' conditions. Although one may not expect an effect of overall system pressure on environmental severity, the Sponsor Group decided that testing should be carried out at elevated pressure to enhance the credibility of the data.

Experimental programme


It is recognised that both H 2 S level and pH can affect safe hardness levels. In many hydrocarbon fluids, the chief acidising agent is CO 2 . Furthermore, most waters contain ionic species. The matrix of test environments was therefore selected to cover a range of H 2 S and CO 2 partial pressures, in a base solution of 5%NaCl. Most tests were carried out at 500psi, 1500psi or 2000psi total pressure, and 25°C.

Two preliminary pieces of work were carried out prior to the main programme, to validate the approach. First, a comparison was made between tests in an ambient pressure and a pressurised environment, in which H 2 S partial pressure and pH were matched. Differences were observed, particularly in the morphology of the sulphide scale, and the decision to test at elevated pressure was endorsed. Second, a set of screening tests was performed in NACE TM0177 method A solution (5%NaCl, 0.5% acetic acid, saturated with H 2 S) at ambient pressure to ensure that all welds tested were at least susceptible to cracking under these conditions, for which a threshold hardness of around 250HV would be expected from previous experience. This was confirmed, which also gave confidence in the test specimen and loading design.

Test samples were 3mm thick, four point bend loaded beam specimens. This geometry was selected largely because it allowed reasonably economic use of autoclave space. If under elastic loading conditions, outer fibre stress can be calculated accurately from deflection.

Parent materials

A summary of the steels used is given in Table 1. These consisted of six C-Mn steels (one pressure vessel grade, and five samples of linepipe) and one AISI 4130 alloy steel.

Table 1: Summary of test steels. All elements given in wt%.

 C-Mn SteelsAISI 4130
SMYS, MPa 355 413 358 358 448 448 -
Diameter mm - 914 408 762 510 410 35
Thickness, mm 15 22 25.4 16 20.6 12.5 -
C 0.18 0.11 0.14 0.15 0.05 0.14 0.29
S <0.002 0.005 0.003 0.021 <0.002 <0.002 0.003
P 0.005 0.012 0.008 0.017 0.009 0.010 0.008
Si 0.30 0.32 0.31 0.26 0.29 0.33 0.18
Mn 1.21 1.33 1.29 1.14 1.37 1.22 0.52
Ni 0.21 0.02 0.04 0.05 0.05 0.06 0.09
Cr 0.11 0.04 0.03 0.08 0.03 0.11 1.02
Mo 0.03 0.01 0.02 0.02 0.01 0.03 0.22
V <0.002 0.05 0.05 0.05 0.05 0.04 0.002
Cu 0.10 0.03 0.09 0.13 0.02 0.10 0.14
Nb <0.002 0.031 0.032 <0.002 0027 <0.002 <0.002
Al 0.022 0.031 0.044 0.017 0.038 0.035 0.022
B <0.0003 0.0008 <0.0003 <0.0003 <0.0003 <0.0003 <0.0003
Ca NA NA <0.0003 NA 0.0014 0.0015 <0.0003

The pressure vessel steel (1B808) was a non-microalloyed normalised steel to DIN 17 102 TStE 355. It had relatively high carbon and carbon equivalent levels (0.18 and 0.43% respectively) and had a very low sulphur content (<0.002%).

Steels 1B823 and 1C660 and 1B2995 were Nb and V microalloyed grades with relatively low sulphur contents (0.005, 0.003 and <0.002% respectively). 1B823 was submerged arc welded, X60 pipe, 1C660 was a seamless, X52 grade, and 1B2995 was submerged arc welded X65. Steel 1B2995 was calcium treated (0.0014%), and had a very low carbon content (0.05%). Steel 1B2996 was an X65 grade pipe, similar to the preceding pipe steels, with the exception that it did not contain a detectable level of Nb. It was calcium treated (0.0015%).

The pipeline material, 1B1002 was selected because of its high sulphur content (0.021%). It was a late 1960s/early 1970s, X52, submerged arc welded pipe.

The AISI 4130 steel (1B898) was supplied as 35mm diameter bar. This had a relatively low sulphur content (0.003%), which was necessary to prevent liquation cracking on welding in the heat affected zone (HAZ).


Welds were deposited in machined grooves. For the C-Mn steels, the GMAW process was used, with a BS2901 part 1 A18 wire (C-Mn-Si composition), and 80:20 ArCO 2 shielding. A range of welding conditions was employed, covering arc energies from 0.6-3.9kJ/mm, to give a range of cooling rates, and hence HAZ hardness levels, in each steel.

For the AISI 4130 steel, manual metal arc welding had to be used to avoid solidification cracking. AWS E10018 rods were deposited at an arc energy of 2kJ/mm (170A, 22V, 112mm/min), using mechanised equipment to ensure consistency. The HAZ hardness was controlled by postweld heat treatment in the range 330 to 700°C.

A hardness survey was carried out in the grain coarsened HAZ of a section from each weld. Further hardness surveys were carried out on stress corrosion tested specimens.

Sample preparation and loading

Samples were machined as sketched in Fig.1. The faces and long edges were prepared to a 320 grit finish and degreased using acetone before loading in four point bend as shown in Fig.2. Load was applied by deflecting the specimens, the required deflection being calculated from the formula given in ASTM G39 for four point loaded bent beam specimens. The accuracy of this method was checked on one sample using strain gauges. Deflection was measured using a dial gauge to an accuracy of approximately ±0.01mm. This translates to approximately ±3N/mm 2 outer fibre stress on the test specimen. Specimens were stressed to 100% SMYS, or mill sheet yield in the case of the 4130 steel. This is broadly consistent with the advice given in EFC Document 16 [4] .

Fig. 1. Extraction of test specimens from welds
Fig. 1. Extraction of test specimens from welds
Fig. 2. Four point bend loading clamp in the loading frame
Fig. 2. Four point bend loading clamp in the loading frame

Test procedure

After stressing samples in the test jigs, as above, eight jigs were loaded onto a support frame, which was then lowered into the test vessel. Each set of eight specimens tested together in an autoclave has been termed a test run, designated with an identification letter. For the majority of test runs, test solution, deoxygenated using a stripping column, was introduced to the vessel under a nitrogen blanket. The vessel, was then sealed. The stripping column design employed at TWI has been shown to achieve <10ppb dissolved oxygen, measured by Orbisphere. Following a further period of nitrogen sparging at about 75% of the test pressure, the appropriate H 2 S/CO 2 gas mixture flow was started, and the vessel was fully pressurised. A continuous gas flow was maintained throughout the test duration to give equilibrium H 2 S and CO 2 levels in the test solution. Temperature was monitored via a thermocouple inside the test vessel, and controlled to 25±3°C. The ratio of solution volume to specimen surface area was ~16:1. Test duration was nominally either 60 or 120 days.

Tested sample evaluation

On completion of the test period, samples were unloaded and washed in water, prior to sectioning. A single central section was taken transverse to the weld, mounted in bakelite, and polished to a 3µm diamond finish. A 1µm finish was used for specimens to be photographed. A record was made of the extent and nature of any cracking and pitting observed, and crack or pit depth in the HAZ was measured using a graticule in the eyepiece of the microscope. Attack was divided into three categories, viz, crack, pit or no crack, and a flow chart, summarising the criteria used for this categorisation, is given in Fig.3.

Fig. 3. Flow chart showing criteria for deciding on crack designation, crack, (C), pit (P) or no crack (N)
Fig. 3. Flow chart showing criteria for deciding on crack designation, crack, (C), pit (P) or no crack (N)

A hardness survey was carried out in the grain coarsened HAZ on each sectioned test specimen using a Vickers machine with a 5kg load.

Environmental control and monitoring

A summary of the long term elevated pressure test environments is given in Table 2. Test solutions were checked for pH prior to saturation with sour gas, and at the end of the test period. For early runs, the test sample was taken after letting down to ambient pressure and was clearly not a measurement of the pH at pressure. For later test runs, pH was measured at the test pressure, using specially designed and fabricated equipment at intervals throughout the test duration.

Table 2: Summary of autoclave test environments.

Gas composition, partial pressure, psi (bar)pH calculated according to ref 5
H 2 SCO 2N 2Without corrosionFe saturated
1 0.50 (0.03) 100 (6.9) 400 (27.6) 3.38 4.59
2 0.05 (0.003) 100 (6.9) 400 (27.6) 3.38 4.59
3 0.50 (0.03) 5 (0.34) 495 (34.1) 4.01 4.83
4 0.05 (0.003) 5 (0.34) 495 (43.1) 4.03 5.17
7 1.00 (0.07) 100 (6.9) 399 (27.5) 3.38 4.31
8 1.00 (0.07) 5 (0.34) 494 (34.1) 4.00 4.72
9 0.50 (0.03) 200 (13.8) 300 (20.7) 3.23 4.31
10 0.05 (0.003) 200 (13.8) 300 (20.7) 3.23 4.39
11 1.00 (0.07) 200 (13.8) 299 (20.6) 3.23 4.21
12 15 (1.03) 100 (6.9) 385 (26.6) 3.36 3.91
26 2.00 (0.14) 0 (0) 498 (34.3) 4.29 4.82
5 0.50 (0.03) 100 (6.9) 1400 (96.6) 3.38 4.41
22 0.50 (0.03) 5 (0.34) 1495 (103.1) 4.01 4.83
20 0.05 (0.003) 5 (0.340) 1995 (137.6) 4.03 5.17
21 0.50 (0.03) 5 (0.34) 1995 (137.6) 4.01 4.83
23 0.50 (0.03) 100 (6.9) 1900 (131.0) 3.38 4.41
24 0.70 (0.05) 100 (6.9) 1899 (131.0) 3.38 4.36
25 0.70 (0.05) 5 (0.34) 1899 (131.0) 4.01 4.78

All 5%NaCl solution, with continuous passage of gas.

The H 2 S concentrations were measured by iodiometric titration on samples of 'clear' solution, collected at the end of the test period with as little disturbance of iron sulphide sludge as possible. This was, however, not a true measure of H 2 S content at elevated pressure.

A measure of the 'severity' of the solution over the test period was also made by including a sample of each of three parent steels in each test vessel. These were analysed for diffusible hydrogen either by evolution over mercury at ambient temperature or by gas chromatography following extraction of hydrogen for 6 hours at 150°C, on completion of the test exposure. Samples were quenched in liquid nitrogen within half an hour of turning off the test gas supply, and within one minute of removal from solution, and stored cold prior to analysis.


Environmental monitoring

Measurement of pH at elevated pressure has allowed plots of pH as a function of time to be developed, as shown in Fig.4. In all cases, following a sharp initial drop as gas was introduced, the pH rose at a decreasing rate throughout the test duration, although the shift was generally less than one pH unit. The only exception to this is run 5679-L ( Fig.4c), in which the test solution was conditioned by allowing dummy samples to corrode in it, before the stressed test specimens were introduced, thus avoiding exposure to the initial dip in pH.

Fig. 4. Examples of pH vs time plots Top two, 500 psi vs 2000 psi total pressure Bottom left, so lution preconditioned
Fig. 4. Examples of pH vs time plots Top two, 500 psi vs 2000 psi total pressure Bottom left, so lution preconditioned

Upper (blue) line indicated calculated pH assuming Fe saturation.
Lower (red) line indicates calculated pH assuming clean solution

It is not known at what point in time any cracking takes place and thus what value of pH is controlling test results. A conservative approach is to use a high pH value towards the end of the test duration and, as values were in general fairly steady at around 500 hours, this level was taken as the measure of test environment pH.

The pH values at elevated pressure can be compared with predictions of pH made using the computer program described in ref 5, and pH levels calculated for clean solutions, and solutions saturated with iron, are shown on Fig.3. The pH developed towards the end of the tests at 500psi matches the predicted value for solution saturated with iron quite well, whereas at 2000psi, the predicted pH is just below the measured values. The pH value assigned to each test run has therefore been either the 500 hour measured value at test pressure, or a best estimate based on calculated values, where no direct measurement was available.

Hydrogen pick-up was measured on samples from the same three parent steels for each test run. There was no attempt to determine hydrogen pick-up in the stress corrosion test samples, the analyses being intended purely to give a measure of the relative hydrogen charging capacities of the environments. The data from the three steels have been averaged before plotting them in Fig.5 to provide a consistent relative measure for all test runs.

Fig. 5. Hydrogen pick-up, ml/100g steel
Fig. 5. Hydrogen pick-up, ml/100g steel

The data presented in Fig.5 have been grouped by environment, presented in order of increasing H 2 S (left to right) and CO 2 (top to bottom) partial pressures. It is evident that increasing CO 2 partial pressure has significantly increased hydrogen pick-up, whereas, over the range investigated, increasing H 2 S partial pressure has had a much lesser, although measurable, effect.

The effect of total pressure, in conjunction with the effect of CO 2 partial pressure, is also shown in Fig.5. The effect of CO 2 is confirmed, and an increase in hydrogen pickup is observed as the total pressure increases from 500psi to 1500psi, and from 1500psi to 2000psi.

Stress corrosion tests

Attack on the specimens consisted of both pitting and cracking, each covering a range of severity. Example micrographs are shown in Fig.6. Cracking was distinguished from pitting according to Fig.3.

Fig. 6. Examples of corrosive attack
Fig. 6. Examples of corrosive attack

There was a general increase in depth of attack with increasing HAZ hardness in the majority of the environments. In environment 4 (the lowest H2S and CO 2 partial pressures, 0.05psi and 5psi respectively at 500psi total pressure), no cracking was generated, and attack was less than 0.07mm deep on all specimens, covering the range 265-491 HV5. Increasing CO 2 and H 2 S partial pressures both increased the severity of attack and reduced the lowest hardness at which cracking was identified. These trends can be seen on a compilation of the results for the nine principal environments at 500psi total pressure, presented in Fig.7. At 2000psi total pressure, a similar pattern is apparent, although not fully defined at 5psi CO 2 , where no cracking was induced at two H 2 S partial pressures, despite hardness levels of over 400HV. At 1500psi, the effect of H 2 S partial pressure has not been explored, but, surprisingly, no real effect of CO 2 partial pressure between 5 and 100psi was found.

Fig. 7. Summary of cracking test results at 500psi total pressure
Fig. 7. Summary of cracking test results at 500psi total pressure

The effects of total system pressure are shown in Fig.8, where threshold hardnesses from Fig.7, and similar data for other environments have been presented. At a CO 2 partial pressure of 100psi, the threshold hardnesses are similar for 1500 and 2000psi total pressure, and these are lower than corresponding thresholds at 500psi. This indicates an increase in environmental severity with increasing system pressure up to a pressure somewhere between 500 and 1500psi. Valid comparisons of the effect of total pressure at 5psi CO 2 can only be made using data for 0.5 and 0.7psi H 2 S. The cracking results indicate C-curve behaviour, with the greatest environmental severity being reported at 1500psi. It is not clear whether this observation reflects experimental variability, or whether it actually represents an inhibiting effect at high total pressure. In the latter regard, no comparative study has been made of the sulphide film found under the various high pressure conditions, and resultant metal surface condition.

Fig. 8. All cracking results. Threshold hardness levels labelled for:
Fig. 8. All cracking results. Threshold hardness levels labelled for:

a) 500 psi
b) 1500 psi
c) 2000 psi



The data generated in this programme have demonstrated clear trends in safe hardness levels as a function of environmental severity. Over 350 specimens have been tested in the main part of the programme, under 18 different environmental conditions. This adds considerable confidence to the identification of safe hardness levels for service environments. Considering the data from this work in isolation, the thresholds in Fig.8 are all above the NACE MR0175 equivalent limit of 248HV, and there is an overall trend of decreasing safe hardness level as both CO 2 and H 2 S partial pressure increase. Some care was taken to separate cracking (which may be successfully controlled through hardness limits) from other attack, (for which hardness limits would be expected to be of little benefit). The success of the approach adopted is evident both in Fig.8, and in the relationship between threshold hardness and hydrogen charging capacity of each test run, as measured by hydrogen pick-up in parent steel samples.

One of the reasons for the present work was an expectation that there was not a step change in behaviour at the NACE limit of 0.05psi partial pressure of H 2 S. The trends in hardness threshold across Figs.7 and 8 support this view. The poisoning effect of H 2 S is not just a switch, although quite low levels can have a significant effect on hydrogen charging capacity, and hence risk of cracking.

Effect of total system pressure

It is often assumed that environmental severity, with regard to the risk of sulphide SCC, can be described in terms of pH and H 2 S partial pressure, and total system pressure is not generally taken into account. In this respect, it is necessary to be confident of the effects of total system pressure on pH. Measurements made in this work (see under environmental monitoring above) indicate that predictions of pH [5] at 2000psi, for a solution saturated with corrosion products, are slightly lower than the true value, whereas at 500psi, they are fairly accurate. Thus, at higher pressure, assessments made on the basis of these calculations will be slightly more conservative.

The results of this programme also allow any effects of system pressure, over and above effects of pH and H 2 S partial pressure, to be explored. In the first place, as stated in the introduction above, it was decided to test at elevated pressure following trials which indicated some differences, particularly in scale formation, by comparison with exposure at ambient pressure. Secondly, as described in section 3.2 above, it is evident that there is an effect of total pressure in the range 500-2000 psi on cracking, but that this is not simple, with either an increase in severity up to between 500 and 1500psi, or C-curve behaviour, with maximum severity at about 1500psi, depending on gas composition. However, although it is the risk of cracking which is clearly the prime concern, it is also possible to measure environmental severity in terms of hydrogen pick-up. On this basis, making the same comparisons as above, there is a clear increase in severity between 500 and 1500psi, and between 1500 and 2000psi, for all H 2 S/CO 2 combinations for which comparative data are now available. This is demonstrated in Fig.9. From this evidence, it would not be prudent to regard cracking risk to decrease at 2000psi (ie C-curve behaviour).

Fig. 9. Effect of total pressure and CO 2 partial pressure on hydrogen pick-up
Fig. 9. Effect of total pressure and CO 2 partial pressure on hydrogen pick-up

One possible cause of variation in environmental severity is variation in pH. Looking at Fig.4, the pH of a 500psi equivalent of environment 25 should lie between that of environments 4 and 8. Thus, the indication is that pH has changed very little or even increased with increasing total pressure. This is contrary to what would have been expected from cracking and hydrogen pick-up results, although would have been anticipated thermodynamically.

Application of results

As discussed above, clear trends in safe hardness levels have been demonstrated as a function of environmental severity. Although an effect of total system pressure on risk of cracking has been demonstrated, the pattern of behaviour is not either sufficiently straightforward, or sufficiently well defined, to permit separate guidelines for different total system pressures to be put forward. At this stage, it is more advisable to work on minimum threshold hardness, regardless of total pressure.

In order to make use of these data, the test conditions need to be related to service conditions. The H 2 S and CO 2 partial pressures, temperature, and total pressure have been well controlled, and these factors are also generally monitored in service. There is, however, some concern that the relatively small test solution volume to test specimen area ratio has resulted in a higher contamination of the aqueous environment than would be seen in many service situations. As such contamination tends to increase pH, it could result in unconservative hardness limits, if these are applied to H 2 S and CO 2 partial pressures alone. It is therefore preferable to set limits as a function of pH and H 2 S partial pressure. As a result in particular of pH measurements at elevated pressure, there is a good understanding of the development of pH in these tests, and the only remaining uncertainty is when cracking took place, and thus what pH was controlling the cracking.

Some experiments have been carried out, to try and determine whether the hardness limits are appropriate to the initial pH or a higher (later) value, by controlling the initial pH (before contamination with corrosion products) with bicarbonate additions, thus eliminating the initial dip in pH. The results indicated that threshold hardness was relatively unaffected by such additions (~330HV with and without bicarbonate), implying that the later pH 5 would be a more appropriate level. However, controlling to a lower pH with acetate addition failed to reduce threshold hardness, and thus overall it is not clear what the appropriate pH is. A conservative approach has been taken of using the stable (~500 hour) pH, even when lower pH conditions were experienced at the beginning of a test period.

Additional data points have been taken from the literature, and included in Fig.10. These cover a range of test methods and environ-ments, and comprise the following:-

  1. Three data points taken from work by Pargeter [6] covering testing at ambient pressure in 5%NaCl and ASTM D1141-75 standard substitute ocean waters. Values of pH and H 2 S were measured directly. Test specimens consisted of transverse weld three point bend samples, approximately 25mm square cross section, with the weld cap remaining intact. The thresholds quoted are for appliedyield stress in the outer fibres, not taking stress concentrating effects of the weld cap into account.
  2. Five data points taken from a study by Motoda and Yamane [7] . Smooth four point beam specimens were stressed to yield in 0.5% acetic acid solution with various H 2 S contents. A pH of 3.5 has been assumed. The hardness thresholds are taken from the figure presented in the paper, but may be unconservative by comparison with the present programme, as some specimenscontaining 'microcracking' around 50µm deep were deemed uncracked.
  3. Five data points taken from work by Nakuzawa and Tanimura [8] in which they tested welds in distilled water containing various levels of H 2 S. Values of pH for these environments have been calculated using CORMED.
  4. Two data points from Taira et al [9] who stressed smooth, 3mm thick, four point bend specimens machined from welds to 1.3 x yield deflection in 0.5% acetic acid solution. As for Motoda and Yamane's work above, a pH of 3.5 has been assumed. The failure criterion wasnot specified, but the loading was severe, so the data are unlikely to be unconservative, and indeed are more conservative than Motoda and Yamane's.
  5. Three data points from Vennett [10] who tested 3mm thick transverse samples with the weld cap intact in three point bend in 3%NaCl solution with a range of pH and H 2 S levels. Stressing by deflection was to nominally very high outer fibre stresses, well in excess of parent material yield, and crack detection was by metallographic examination. These data can be taken as relatively conservative.
  6. Nine data points from Kermani et al [11] . These are the data points on the basis of which the boundary lines for sour service regimes were drawn, as shown. The results are from smooth tensile tests carried out on N80 13%Cr steel and P110 carbon steel, with reported hardnesses of 263 and 294 HV respectively.
  7. Three data points from Nisbet et al [12] . Two points arise from a full scale test at 140 bar total pressure for which the pH was calculated to be 4.7-5. With 26mbar H 2 S, bead on plate and weld repairs with a total range of hardness down to 322 HV cracked. As the hardnesses of these two welds also went up to 366 and 360 HV respectively, 322 HV is a conservative value to choose. With 7mbar H 2 S, neither of these welds cracked, so the threshold is >366 HV. The third data point is from smooth tensile tests at 90% actual yield stress with 7 mbar H 2 S and 1bar CO 2 , pH 4.7-5.3, where cracking was experienced above 340 HV.
  8. Two data points from work reported by McIntyre and Boah [13] . Welds made in A516 grade 70 steel were tested under low pH/H 2 S combinations to determine the threshold hardness for each.
  9. Three points from Bond and Gooch [14] . Failures were induced in welded AISI 4130 at HAZ hardness levels of above ~550 HV5 at H 2 S partial pressures below the NACE limit. At 100ppm H 2 S (0.0015psi H 2 S) and pH 3.5, cracking was induced at 584 HV5. At 1000ppm H 2 S, (0.015psi H 2 S) cracking was induced at HAZ hardness levels of 549 HV (pH 3.5) and 584 HV (pH 5.5). These are not well defined thresholds, but do indicate the need for a hardness limit at below the NACE threshold.

The first point to note in Fig.10 is that there is a clear picture of hardness threshold being dependent on pH and H 2 S partial pressure, and that for many of the conditions exceeding the NACE limit of 0.05psia, thresholds are well above 250HV. With the amount of data incorporated in Fig.10, and bearing in mind the wide range of test methods, steels and environments covered, it is possible to draw boundaries with some confidence. The hardness levels for the boundaries which have been chosen are dependent on both the data in Fig.10, and the practical realities of welding C-Mn steels.

Fig. 10. Summary of data from the present programme and data reported in the literature. Threshold hardness and reference number (for data from the literature) marked above each point
Fig. 10. Summary of data from the present programme and data reported in the literature. Threshold hardness and reference number (for data from the literature) marked above each point

In the first place, 250HV is a low threshold which is frequently troublesome to achieve at welded joints. Even raising it to 260HV (which is about as close as can be taken as a truly measurable difference) would be of benefit, and the region above the bounding line set between 250HV and 260HV does cover quite a useful range of environments. Nevertheless, a more realistic and useful increase in hardness threshold would be to 275HV, and such a region, in fact, incorporates much of the area currently defined by EFC 16 as requiring a 260HV limit. To the left of the NACE limit (lower H 2 S) some hardness limit is evidently necessary, and indeed such would usually be imposed in any case to control fabrication hydrogen cracking. A typical level for general C-Mn steel fabrication would be 350HV. Within the NACE sour region at higher pH, an intermediate, but not too onerous, value of 325HV can be justified.

The boundaries for the EFC sour service regimes are shown on Fig.10. Regardless of precise positions of boundaries, it is evident from the data on Fig.10 that some hardness limit needs to be set above the transition domain. Allowing for this, it is of note that this work has generally allowed a relaxation of hardness limits to be proposed, with only a few areas of changes to more stringent requirements by comparison with EFC 16.

Summary and conclusion

A programme of experimental work has been carried out in which welded samples of C-Mn and low alloy steels have been stressed and exposed to pressurised aqueous environments, in contact with various partial pressures of H 2 S and CO 2 . A total of over 350 specimens has been tested and the following conclusions have been drawn.
  1. Threshold hardness levels show a clear variation with CO 2 partial pressure, and with H 2 S partial pressure over the range 0.05 to 2psi, that is from the sour service limit given in NACE MR0175-98 to 40x that limit.
  2. Threshold hardnesses at or near the MR0175-98 H 2 S limit were considerably in excess of the level permitted in NACE MR0175-98, and, in many mildly sour service situations, relaxation of hardness limits is reasonable.
  3. The maximum hardness levels given in Fig.10 of this paper may be used for welded steels directly exposed to service environments with the stated H 2 S and pH levels, for total pressures of up to 2000psi (138 bar).
  4. The risk of pitting of C-Mn and low alloy steels, particularly in the presence of CO 2 , must be taken into account, but this is not controllable through hardness limits.


Thanks are due to many staff at TWI, and in particular Ian Wallis, Mike Bennett and Glyn Hall for carrying out the bulk of the experimental work, and Trevor Gooch for valuable discussions and advice. The work was sponsored by the following companies, whose support and direction throughout the project and permission to publish, is gratefully acknowledged: Agip; BHP Petroleum; British Gas; Chevron; Norsk Hydro; Philips; Shell; Statoil; Sumitomo and the UK Health and Safety Executive.


1 NACE: 'Sulfide stress cracking resistant materials for oil field equipment'. Standard MR0175-99 NACE January 1999.
2 Minutes of NACE T-8 committee meeting, March 1985.  
3 NACE: 'Methods and controls to prevent in-service cracking of carbon steel welds in P-1 materials in corrosive petroleum refining environments'. Standard RP0472-87 NACE.
4 European Federation for Corrosion: 'Guidelines on materials requirements for carbon and low alloy steels for H 2 S containing environments in oil and gas production', EFC document No.16, 1995.
5 Crolet J-L and Bonis M R: 'An optimised procedure for corrosion testing under CO 2 and H 2 S gas pressure', Materials Performance, 29 (7) July 1990, pp81-86.
6 Pargeter R J, Gooch T G and Bailey N: 'The effect of environment on threshold hardness for hydrogen induced stress corrosion cracking of C-Mn steel welds'. Conference Proceedings 'Advanced Technology in Welding, Materials Processing and Evaluation'. Japan Welding Society, Tokyo, April 1990.
7 Motoda K and Yamane Y: 'Stress corrosion cracking of steels in aqueous solution of low hydrogen sulfide concentration' presented at 101st ISIJ meeting, April 1981, Lecture no S477.
8 Nakazawa T and Tanimura M: 'SSCC in welded parts of high strength steels', Nippon Kokan Technical Research Centre of Kakushiki Kaisha, October 1978.
9 Taira T, Tsijkada K, Kobayashi Y, Inagaki H and Watanabe T: 'Sulfide corrosion cracking of line pipe for sour gas service', NACE, Corrosion 37 (1), January 1981.
10 Vennett R M: 'Safe weld hardness levels to insure long service life in offshore pipelines', Offshore Technology Conference, May 1-3, 1972, Houston Texas, Paper OTC 1571.
11 Kermani M B, Harrop D, Truchon M L R and Crolet J L: 'Experimental limits of sour service for tubular steels', NACE corrosion 91, Paper 21, March 11-15, 1991.
12 Nisbet, W J R, Leune M, Attwood P A and Robinson J L: 'The effect of local hard zones on the sulphide stress corrosion cracking of C-Mn steels exposed to mildly sour environments', Corrosion 95, Paper 69, March 26-31, 1995, NACE.
13 McIntyre D R and Boah J K: 'Review of sour service definitions', Materials Performance 35 (8) August 1996 pp54-58.
14 Bond S and Gooch T G: 'Environmental hydrogen-induced cracking of welded steel in aqueous media'. TWI Core Research Programme report, 524/1995.


For more information please email: