Strain and Sulfide Stress Cracking in Pipeline Girth Welds

Paper presented at NACE 2005 Houston, Texas, 3-7 April 2005.

Abstract

For sour service conditions there are stringent weldment hardness requirements that must be satisfied in order to minimise the risk of sulfide stress corrosion cracking (SSC) in service. Currently in standards such as DNVOS-F101:2000 and BS4515:2000 the hardness threshold requirements remain unaltered for welds that are strained, for example, as can occur during pipe reeling and laying operations. Work has been carried out to investigate the SSC behaviour of samples, taken from a pipe girth weld, that were strained to conditions that simulated those of a typical pipe reeling operation and conventional hardness limits were shown to be unsafe. Samples were taken from a C-Mnsteel girth welded pipe, manufactured to API 5L X65, and were subjected to tensile and compressive strain to simulate the pipe installation. Four point bend sulfide stress corrosion cracking tests of 30 days duration were performed on specimens manufactured from the strained and unstrained material in NACE TM0177 solution A. The results showed that the application of such prestrain to the specimens led to failure by SSC despite the HAZ hardness remaining below the allowable limits specified in the relevant codes (e.g. ISO15156, DNV OS-F101 and BS4515).

Introduction

For sour service conditions there are stringent weldment hardness requirements that must be satisfied in order to minimise the risk of sulfide stress corrosion cracking (SSC) in service. Hardness testing is routinely carried out as part of the weld procedure qualification testing (as specified for example in BS4515:2000 ^[1] and DNV OS-F101:2000 ^[2] to ensure that a maximum limit (250HV10 in BS4515) is not exceeded. What is generally not taken into account is the effect of the significant level of strain that can be induced in a weldment during, for example, a pipe reeling and laying operation. Certain codes, such as DNV OS-F101, do require that the mechanical properties (hardness, strength and toughness) are verified in both the as-welded and strain aged conditions, to simulate the strain induced during standard installation. However, even if the effect of strain hardening has been taken into account, and the weldment hardness after pipe laying does not therefore exceed the allowable hardness limits, no account is taken of the possibility that the threshold hardness for SSC may in fact be different for strained material than for unstrained material.

This paper describes work performed under a short study programme into the sulfide stress corrosion cracking behaviour of samples taken from a pipe girth weld that were strained to conditions that simulated those of a typical pipe reeling and laying operation.

Experimental procedure

General approach and sample preparation

A pipe sample consisting of two short sections of seamless, 12.75in. OD x 20.6mm wall thickness to API 5L X65 was joined by a girth weld. The weld was made and inspected by a reputable contractor using manual weld procedures typical of those employed for pipe installation by reeling. The root and hot pass were made using cellulosic electrodes and the fill and cap deposited using low hydrogen consumables. The preheat temperature was 110°C (minimum) and the maximum interpass temperature was 300°C. The maximum heat input for the weld fill was 2.1kJ/mm.

Six cross-weld specimen blanks 60mm x 20mm x 350mm were machined from the 3 o'clock (specimen blanks CSB 1-3) and 9 o'clock (specimen blanks CSB 4-6) positions of the pipe section with the weld centreline situated at the centre ofthe blanks and with the weld root and cap remaining intact. The blanks were placed in a 100kN servohydraulic test machine with a 50mm gauge length between the jaws. Two clip gauges were attached to the sides of the blanks and were used to monitor strain over a gauge length of 40mm. Specimen blanks CSB 1-3 underwent a tensile:compression:tensile:compression (T-C-T-C) strain cycle to simulate those typical, for this size of pipe, experienced at the 12 o'clock position with respect to straining during reeling and laying operations. ^[3] The strain cycle was 0 to 2.5%, 2.5 to 0%, 0 to 2.5% and 2.5 to -0.4% strain. Specimen blanks CSB 4-6 underwent a similar strain cycle but in compression. The strain cycle in this case was 0 to -2.5%, -2.5 to 0%, 0 to -2.5% and-2.5 to 0.4% to represent a typical strain cycle at the 6 o'clock position during reeling and laying.

Six unstrained, cross-weld, four point bend sulfide stress corrosion cracking specimens were machined from the pipe at positions adjacent to the 3 o'clock and 9 o'clock positions from which the strain specimen blanks had been extracted (specimens SCC1-6). The specimens were machined such that they had a 200mm span and were 18mm wide with the girth weld at the centre of the span. The side faces were ground parallel and prepared to a 320-grit finish, but the weld root and cap surfaces were left intact.

Four prestrained cross weld four point bend sulfide stress corrosion specimens were machined from the specimen blanks CSB3 and CSB6 (SCC7 and 8 from CSB3, and SCC 9 and 10 from CSB6), and prepared in a similar way to the unstrained specimens. For resource reasons approximately 18 months elapsed between the initial prestraining of the specimens and the stress corrosion cracking tests. A summary of the specimens extracted, the strain conditions applied and the activities performed on each are shown in Table 1.

Table 1 Summary of samples extracted from the pipe girth weld and activities performed on each of them

Mechanical Property Measurement

A longitudinal 10mm diameter round tensile test specimen with a 50mm gauge length was removed from the unstrained parent pipe (T1). Similar specimens were machined from the prestrained blanks CSB2 (T2) and CSB5 (T5) and all specimens tested to BS EN 10002-1:2001, at room temperature.

Weld macrosection specimens were extracted from the prestrained specimen blanks CSB3 (M3) and CSB6 (M6). The macrosections were prepared using standard metallographic techniques to a 1µm diamond finish and etched in 2% nital to reveal the parent pipe, heat affected zone (HAZ) and weld microstructures. Vickers hardness measurements were made on the macrosections using a 10kg indenting load and were conducted 2mm from the outer and inner surfaces of the pipe to sample the parent material, HAZ and weld metal. The hardness survey was conducted such that the number and positioning of indents met the requirements of BS4515-1:2000. ^[1]

Chemical analysis

Chemical analyses of the parent pipe material both sides of the weld and the weld metal were performed on macrosection M3 using optical emission spectrometry, directly sparking onto the prepared surface.

Stress corrosion cracking tests

The four unstrained samples, SCC1-4, were prepared for testing with the weld root in tension. All tests were carried out under dead weight loading with the inner pipe surface in tension (referred to as root in tension). The test environment consisted of NACE TM0177 ^[4] solution A (5%NaCl, 0.5% acetic acid, in water, saturated with H ₂ S). The test stress levels were set at 85%, 90%, 95% and 100% of actual parent metal yield stress (as determined from tensile test specimen T1) for specimens SCC3, SCC1, SCC4 and SCC2 respectively. The test duration was 30 days.

Following the results of the tests on the unstrained specimens, prestrained specimens SCC7 and SCC9 were tested in a similar manner with the test stress level set at 100% of the actual parent metal yield stress (as determined from tensile test specimen T1). The two remaining prestrained specimens SCC8 and SCC10 were then tested under similar conditions with the test stress level set at 70% of the specified minimum yield stress (SMYS) for API 5L X65 material(SMYS = 448MPa). The test duration was the same as for the unstrained specimens but practical laboratory considerations meant that the duration for specimens SCC8 and 10 was reduced by one day.

On completion of the exposure period, the samples were subjected to magnetic particle inspection (MPI) and metallographic sections were taken transverse to the weld from each specimen, through MPI indications or at specimen mid thickness, as appropriate. The sections were examined using a light microscope to determine the presence or extent of cracking, and photomicrographs taken to record features of interest. Vickers hardness measurements were taken to characterise the weld metal and HAZ microstructures that had initiated any cracking.

Results

Material characterization

Test results for tensile and hardness measurements are summarised in Table 2. The parent pipe tensile yield stress used as the basis for the test stress levels was 477 MPa. Following the simulated pipe reel straining the measured 0.2% proof stress had increased to 518 MPa and 520 MPa for the tension-compression-tension-compression (T-C-T-C) and compression-tension-compression-tension (C-T-C-T) cycles respectively.

Table 2 Mechanical test results obtained from the parent pipe and weld macrosections pre- and post straining

Maximum measured HAZ hardness levels in the prestrained specimens M3 and M6 were 246HV10 and 233HV10 respectively In both specimens the HAZ hardness was highest near the weld cap. Hardness results obtained from sections of as stresscorrosion tested specimens are reported in the following section.

The results from the chemical analyses of the parent and weld materials are shown in Table 3. Results were typical for a carbon manganese steel produced to API 5L X65. Carbon equivalent values (CEIIW) for the parent pipes and weld metal were 0.36, 0.37 and 0.41 respectively.

Table 3 Results of chemical analysis for parent pipe and weld metal

Sulfide stress corrosion cracking tests

The results of SSC tests are summarised in Table 4, and photomicrographs of sectioned test specimens are presented in Figures 1-4. No cracking was evident in any of the unstrained specimens (SCC1-4). The specimens showed some evidence of pitting associated with the HAZ on sectioning but the pits were small (<50µm) and v-shaped, and did not constitute sulfide stress corrosion cracks ( Fig.1).

Table 4 Summary of SSCC test results

Fig.1. Photomicrographs showing pitting of unstrained specimen SCC2 following SSC testing for 720h at 100% of actual yield stress, root in tension

a) Pitting at the surface near the root pass;

b) Detail of pitting in parent material just outside the HAZ. Etched in 2% nital. Magnifications given by micron bars.

In contrast, both prestrained specimens tested at an applied test stress equal to 100% of the measured actual yield (SCC7 and 9) failed well within the test period. The specimen that had been first strained in tension in the cycling sequence failed in a shorter time than that which was strained first in compression (53.3h c.f. 127.4h). Figures 2a and 2b show photomacrographs of the cracks evident in specimens SCC7 and 9 on sectioning. Examination of the cracks showed they were characteristic of those associated with sulfide stress corrosion cracking, with an intergranular/transgranular branched appearance. Evidence of corrosion products within the cracks was also present. The cracks appeared to have initiated in the HAZ near the root weld fusion boundaries and had propagated towards theweld cap through the weld metal in both specimens.

Fig.2. Photomacrographs showing strained specimens SCC7 and SCC9 after SSC testing for 720h at 100% of actual yield stress, root in tension

a) SCC7, strain condition T-C-T-C;

b) SCC9, strain condition C-T-C-T.
Etched in 2% nital. Magnifications given in millimetre scales.

The prestrained specimens (SCC8 and 10) that were loaded at an applied stress of 70% of the SMYS for X65 pipe did not fail during the duration of the test, and non-destructive examination of the specimens using MPI revealed no obvious cracking. However, on sectioning specimen SCC8, a crack was evident within the weld metal but close to the fusion boundary, at a distance of a few millimetres from the weld root ( Fig.3a). Examination of the crack showed that it was characteristic of hydrogen induced cracking (as in SSC) with a predominantly transgranular appearance ( Fig.4). The crack did not contain any corrosion product as it was not surface breaking. Specimen SCC10 contained no evidence of cracking ( Fig.3b) and although there was some evidence of pitting, as with the unstrained samples, the pits were small, v-shaped and did not constitute cracking.

Fig.3. Photomacrographs showing strained specimens SCC8 and SCC10 after SCC testing for 693h at 70% for X65 material, root in tension

Fig.3a) SCC8, strain condition T-C-T-C;

Fig.3b) SCC10, strain condition C-T-C-T.

Etched in 2% nital. Magnifications given by millimetre scales.

Fig.4. Photomicrograph showing a detail of the crack evident on sectioning of strained specimen SCC8 following SSC testing for 693h at 70% SMYS for X65 material, root in tension. Etched in 2% nital. Magnification givenby micron bar.

Measured hardness values taken on the sections of the tested specimens at the root for weld metal and HAZ, and in positions adjacent to the cracks in the HAZ and weld metal for specimens SCC7 to 9, were all below 250HV10 (see Table 4). The weld metal and HAZ hardness was slightly higher in the strained specimens than in the unstrained specimens. The maximum values recorded in the cracked specimens were 222HV10, 237HV10 and 230HV10 (all in weldmetal) for specimens SCC7, 8 and 9 respectively. The corresponding maximum HAZ hardness levels near the cracks were 218HV10, 224HV10 and 209HV10 respectively. The maximum value recorded at the root in the uncracked prestrain specimen,SCC10, was 229HV10 in the HAZ.

Discussion

Results and analysis of four point bend sulfide stress corrosion cracking tests on girth weld specimens in API5L X65 material have shown that the application of cyclic prestrain to simulate the strains induced by commercial pipereeling operations affected the sulfide stress corrosion cracking resistance. Tests on unstrained specimens behaved as expected in view of the root hardness levels, and did not fail in 720h exposure tests at applied stresses equal tothe measured yield stress of the pipe. However, following simulated pipe reel straining, specimens tested at similar applied loads failed within 128 hours. One specimen loaded to 70% of the specified minimum yield stress for X65 pipe(approximately 66% of the measured yield stress) was also found to have cracked following metallographic sectioning at the end of the test. The delay between prestraining and SSC testing, which may have resulted in some strain ageingbut was not characterised, was not considered to have been significant as the strained samples were all confirmed to be significantly less than 250HV. Moreover, in most practical situations the time between straining and in production operation will be such as to provide a similar opportunity for ambient temperature strain ageing to occur.

The test failures of strained specimens therefore occurred despite measured HAZ hardness values being lower than 250HV10, which has been shown to be a safe threshold for as-welded C-Mn steel welds loaded to 100% yield in this test environment. ^[5] The fact that similar tests on unstrained specimens did not fail gives some confidence in the test method and the representativeness of the weld and parent material. Although both weld metals and HAZs were slightly harder in the strained condition, these results suggest that the threshold hardness for SSC is different for strained material than for unstrained material, and indeed demonstrate that it is significantly below the commonly specified level of 250HV. ^[4]

The sub-surface initiation of a sulfide stress corrosion crack, such as that present in specimen SCC8, is unusual, but can occur. The level of diffused hydrogen within the material resulting from corrosion in a fully immersed specimen will become relatively uniform, and crack initiation will occur where the combination of stress and microstructure is most severe. The step like form of the crack evident in specimen SCC9 was also a possible indication that a surface initiating crack coalesced with one that had initiated sub-surface. Although it is not possible to completely rule out a fabrication hydrogen cracking mechanism in specimen SCC8, this would not be expected with the procedure used, and no cracks were evident in any of the other untested metallographic sections. Furthermore it was understood that the weld had been subjected to, and passed, non-destructive inspection by the welding contractor.

Previous work in the literature on the effect of cold work on the SSC resistance of steels has also noted an apparent increase in susceptibility. Dvoracek ^[6] investigated the effect of cold work on the critical stress for SSC of quenched and tempered P-110 grade steels. In that study the results suggested that the critical nominal stress for failure by SSC in strained specimens (the level of cold work was not quantified) was 60% of that determined for unstrained, surface notched specimens.

More recently, work by Mack and Filippov ^[7] presented results that indicated that the SSC resistance of tubular steels in API grade X80 and P110 was significantly reduced by strain introduced by pipe expansion. The authors found that cold working the material prior to testing shortened the time to failure. In that work the applied stress during the tests was approximately equivalent to the actual yield of the material. One specimen that had undergone a 10% expansion strain failed in a time of 100hcompared with no failure in 720h for unstrained specimens. Two further specimens prestrained by 20% failed after 10 and 30h respectively under similar test conditions. These authors also found that strain ageing tended to shorten thetime to failure in SSC tests.

ISO 15156-2003 ^[8] recognises an effect of cold work on susceptibility to SSC, and specifies (in Part 2, Section A.2.1.6) that: 'Carbon and low alloy steels shall be thermally stress relieved following any cold deforming by rolling, cold forging, or another manufacturing process such that a permanent outer fibre deformation greater than 5%'. The hardness of the stress relieved component is also limited to 22HRC maximum.

The results of this short study have great significance in the use of pipe reeling and laying operations intended for sour service conditions. Further work is necessary to confirm the results and investigate the fundamental aspects of the effect of strain on sulfide stress corrosion cracking and to quantify the threshold susceptibility in terms of prestrain and hardness. Results from such a study would assist in enabling either the implementation of additional testing requirements/methods or revised property limits in weld procedure qualifications.

Conclusions

From a programme of four point bend sulfide stress corrosion cracking tests performed on API 5L X65 girth weld specimens, as welded and in a simulated pipe reeled condition, the following conclusions have been drawn:

1. The application of prestrain to a weldment to simulate the cyclic strains introduced during pipe reeling operations increased the sensitivity to sulfide stress corrosion cracking.
2. Cracking occurred in strained samples despite the HAZ hardness of the specimens being less than the allowable limits specified in the relevant codes.

Acknowledgements

The author would like to thank Ian Wallis and Glyn Hall for carrying out the experimental work and Richard Pargeter and Peter Hart for valuable discussion regarding the direction and results of the project.

References

1. BS4515:2000: 'Specification for welding of steel pipelines on land and offshore. Part 1: Carbon and carbon manganese steel pipelines'. BSI, 2000.
2. DNV OS-F101:2000: 'Submarine pipeline systems'. Det Norske Veritas, Norway, 2000.
3. Lockyer S A and Pisarski H G: 'Fracture control for installation methods introducing cyclic plastic strains: development of guidelines for reeling of pipelines. Report on Task 2 - Determination of material properties and response to cyclic straining'. TWI Group sponsored project report 12201/2c/01, September 2001. TWI, 2001.
4. NACE TM0177-1996: 'Laboratory testing of metals for resistance to specific forms of environmental cracking in H ₂ S environments'. NACE, Houston, Texas, 1996.
5. Pargeter R J: 'Factors affecting the susceptibility of C-Mn steel welds to cracking in sour environments'. Paper presented at ASTM International symposium on Environmentally-assisted cracking: Science and engineering, Bal Harbour, Florida, 9-11 November 1985. Proceedings of the symposium, ASTM STP 1049. Publ: ASTM, 1985.
6. Dvoracek L M: 'Sulfide stress corrosion cracking of steels'. Corrosion Vol.26, No.5. May 1970, pp.177-188.
7. Mack R and Filippov A: 'The effect of cold work and strain ageing on the sulfide stress cracking resistance and mechanical properties of expanded tubular steels - A laboratory study'. Paper presented at NACE Corrosion 2003. Paper 03108. Publ: NACE International, Houston, Texas, 2003.
8. ISO 15156-2003: 'Petroleum and natural gas industries - Materials for use in H ₂ S-containing environments in oil and gas production.' Part 2: cracking resistant carbon and low alloy steels, and the use of cast irons. Published NACE/ANSI/ISO 2003.