Weld Joint Cracking in Subsea Oil and Gas Systems Pt II

Department of Engineering University of Leicester, Leicester, UK.

M. Milititsky, R. P. Barnett, M. F. Gittos
TWI Ltd. Great Abington, Cambridge, UK.

Presented at Proceedings of the ASME 2013 32nd International Conference on Ocean, Offshore and Arctic Engineering OMAE2013 June 9-14, 2013, Nantes, France

Abstract

In subsea oil and gas systems, low-alloy steel (LAS) forgings need to be welded to leaner steels such as X and F-65. While the LAS needs to be post-weld heat treated (PWHT) to relieve stresses and temper the HAZ microstructure in order to avoid hydrogen cracking, the same heat cycle would, in general, result in the degradation of the properties of the leaner alloy. A buttering technique is, therefore, usually used so that the buttered LAS forging can be heat treated before the closure weld is carried out. In the case of clad components, nickel alloy filler materials such as Alloy 625 are commonly used for both buttering and closure welds.

This is an issue for subsea structures protected from corrosion by cathodic polarisation (CP) using aluminium based anodes. Whilst CP has proven successful as a means of preventing corrosion of steel components within subsea structures, failures along the dissimilar metal interfaces have been observed. This is due to hydrogen evolution as a result of CP.  To further our understanding on this issue, this paper focuses on the correlation between microstructures, obtained by changing material combinations and PWHT conditions, and the resistance to hydrogen assisted cracking.

Slow strain rate single edge notched bend (SENB) tests were carried out on the interfaces between AISI alloy 8630 and Alloy 625 buttering, retrieved from subsea service and tested in 3.5% NaCl solution under an applied potential of -1100mVsce. Retrieved specimens were pre-charged with hydrogen and tested at 4°C and 80°C, approximately in-line with commissioning/shutdown and service temperatures, respectively. In addition to the retrieved specimens, a testing programme has been developed to explore the effect of PWHT time on the performance of 8630-Alloy 625 and F22-Alloy 625 interfaces. The microstructures most susceptible to hydrogen cracking in these systems have been assessed by examination of the SENB test specimens.

1. Introduction

1.1 Background

Gas tungsten arc (GTA) nickel alloy buttering deposits are commonly deposited on low alloy steel forgings, facilitating in-shop PWHT of the assembly to remove residual stress and temper hard heat-affected zones (HAZs) - microstructures that form upon cooling following welding. During PWHT, carbon migrates from the ferritic steel to the high alloy ‘austenitic’ weld metal [1], causing hardening by martensite formation, carbide precipitation [2] and carbon supersaturation [3]. The resulting microstructures and consequent localised hardening have been associated with sensitivity of the dissimilar metal interface to hydrogen-induced cracking.

After heat-treatment, the buttering is machined, creating a bevel to allow for the completion of a closure weld in the field. The bevel of the closure weld is positioned such that the HAZ of the closure weld lies wholly within the austenitic layers of the post-weld heat treated buttering, thus not requiring any further heat treatments.

Subsea, the structures are subjected to cathodic protection (CP) against corrosion using aluminium alloy sacrificial anodes. Most of the joints made in this way have been satisfactory, however, hydrogen-assisted cracking has been observed at the interface between the PWHT buttering and the low alloy steel, in a small number of cases.

In part I of the paper the authors used transmission electron microscopy (TEM), scanning electron microscopy (SEM) and nano-indentation testing to characterise the microstructures and properties across representative dissimilar metal interfaces [4].  In part II, the resistance to hydrogen-assisted crack propagation along dissimilar metal interfaces has been investigated.

1.2 Previous Work on Dissimilar Joints

As with all dissimilar welds, there are interfacial zones with chemical composition gradients between the weld metal and the parent metal. At the edge of the melted zone is a region, known as the ‘partially-mixed zone’ (PMZ), which has been the focus of a number of studies in joints of this kind, [5] and [6]. A concensus on nomenclature has not been reached, between authors on the subject, so the  conventions, used here to describe the microstructures illustrated schematically in Fig. 1, are as follows:

1. The body centred cubic (BCC) ferritic microstructure of the parent LAS forging.
2. A mainly ferritic, grain-coarsened heat-affected zone (GCHAZ) immediately adjacent to the dissimilar interface with fingers of fused weld metal that have penetrated into prior austenite grain boundaries. Some level of decarburisation can be observed immediately adjacent to the fusion line.
3. A martensitic region at bead overlap positions (where ‘swirls’ of parent material penetrating into the weld normally occur).
4. An austenitic (FCC) and apparently ‘particle free’ area of planar solidification whid has been referred to as a ‘featureless zone’.
5. A region within the PMZ that contains higher atomic number interdendritic particles.
6. The bulk metal further away from the fusion zone that exhibits a FCC matrix with a less diluted chemistry.

2. Materials and Sample Fabrication

Three buttered joints were selected for analysis, the first of which was a retrieved specimen with approximately nine months’ subsea service. This 8630 LAS forging had been TIG-welded with ERNiCrMo-3 (Alloy 625) wire, but PWHT details were not available. Typical industry heat treatments for 8630 forgings buttered with Alloy 625 consist of 650-675°C for 5-10 hours. To supplement the retrieved material and explore the effect of PWHT time, a further 8630-Alloy 625 interface and an F22-625 buttered interface were fabricated. The F22 to Alloy 625 interface is a common alternative to 8630-625. The nominal compositions of each material and joint histories are given in Tables 1 and 2, respectively.

Table 1 – Nominal compositions of welding consumable (Alloy 625) and parent metals

	ERNiCrMo-3 (Alloy 625) Nominal Wt%	AISI 8630 Nominal Wt%	ASTM A182 F22 Nominal Wt%
Fe	0.8	Bal	Bal
Ni	65	0.85	-
C	0.015	0.33	0.15
Mn	-	1	0.45
Cr	22	1	2.25
Si	0.1	0.25	0.5
S	0.005	0.01	0.025
P	0.005	0.01	0.025
Mo	9	0.4	1
Nb	3.5	-	-
V	-	0.1	-
Cu	-	-	-

Table 2 – History and origin of dissimilar weld specimens

Sample Reference	Joint Combination (Parent metal – Buttering)	Origin
1	8630-Alloy 625	Recovered from Service (9 months)
2	8630-Alloy 625	Fabricated for test work
3	F22-Alloy 625	Fabricated for test work

3. Experimental Approach

3.1 Mechanical Performance Testing

Retrieved subsea specimens: The resistance to hydrogen-assisted cracking was investigated for retrieved specimens that had been subsea for approximately 9 months. Blank specimens were machined from retrieved 8630-625 partial joints to create samples 12x12x120mm. To ensure stress concentration and crack initiation near to the interface, the specimens were notched using electro-discharge wire machining. Due to the undulating weld profile, it is not always the case that the dissimilar interface is exactly perpendicular to the axis of the bend. While a very small angle (approximately 1° or less) can be applied to the notch in order to intersect the largest possible fraction of the interface, any larger angle would invalidate the bend test by introducing stresses in directions different to those intended. A notching method using fine gauge wire (Ø0.15mm) was developed to ensure crack propagation from the notch at the dissimilar interface. Calculated angles were applied to ensure intersection with as much of the interface as possible.

The notched samples were pre-charged with hydrogen by holding them at -1100mVSCE in 3.5% NaCl solution. The Potentiostatic charging was applied to simulate cathodic protection. A test programme, reflecting subsea commissioning (4°C) and service temperatures (80°C), was designed. Samples were then tested for each of the following conditions, one to half way and one to completion:

Table 3 – Charging and testing conditions for the retrieved 8630-Alloy 625 dissimilar interfaces

Condition	Precharging time / temperature	Testing temperature
1	1 week, 4°C	80°C
2	1 week, 4°C	4°C
3	1 week, 80°C	80°C
4	1 week, 80°C	4°C

Three point bending rigs were instrumented for single specimen unloading compliance tests, using a modified version of BS7448-4. A slow strain rate (plastic) loading rate of 0.018mm/h was used, with crack tip opening displacement measured by two knife edge clip gauges, located at the notch mouth. Ten elastic unloadings were used as instrument calibration (0.1mm/min), before the test began. To minimise testing time, an unloading rate of 0.05mm/min was selected. A number of test validation checks was performed; essentially the SENB specimens were heat-tinted and crack advancement measured from the notch-tip to the crack-tip (edge of discoloured region) at 9 points along the specimen width using a light microscope.

Specimens fabricated for test work: Joints 2 and 3 were machined to 12.5x12.5mm square sections. Furnace heat treatments were applied to create two of each sample in the 0, 1, 10 and 100 hours PWHT conditions, each with air cooling. The F22-Alloy 625 specimens were heat-treated at 650±5°C whilst the 8630-Alloy 625 specimen was heat-treated at 675±5°C. Each bend specimen was ground and polished to a 12x12mm square section to remove oxidation layers and then etched in 5% Nital solution. Notching and testing was performed as per the retrieved specimens. Samples with irregular J R-curves were discarded.

3.2 Fractography

Scanning electron microscopy: For crack path analysis, a selection of SENB specimens were cut transverse to the weld and prepared using standard polishing techniques. The remaining SENB specimens were heat-tinted, placed in liquid nitrogen and broken open, allowing for the fracture surface of either side of the interface to be compared.

Transmission electron microscopy: An FEI Quanta 3D dual beam system was used to prepare in-situ thin foils for TEM examination. When a cleavage-like surface was selected, the sample was machined perpendicularly to the surface of the cleavage facet. A Jeol 2100 transmission electron microscope operated at 200kV was used to examine the TEM wafers.

4. Results

4.1 Mechanical Performance Testing

Retrieved subsea specimens: The results of the slow strain rate testing at 4 and 80°C show considerably more resistance to crack growth, when testing at the higher of the two temperatures, Fig.2. However, there was no clear difference between the pre-charging temperatures. 

An important response is that of the simulated ‘shut-down’ condition, whereby the system may have been producing at the elevated temperature for some time, before standing at the lower temperature, 4°C. The limited test results presented here indicate that this condition is at no greater risk of cracking, despite the fact that, in theory, more hydrogen should be able to accumulate at the interface due to the faster diffusion rates associated with the higher temperature charging condition.

Specimens fabricated for PWHT investigation: Joints 2 and 3, that had experienced varying PWHT times, were tested using the same general procedure as that used for the retrieved joint. However, all specimens were pre-charged (one week) and tested at 4°C. The effect of PWHT time on the resistance to crack propagation under cathodic polarisation conditions is presented in, Fig.3. There was good agreement between the results for the retrieved and fabricated 8630-625 joints and, interestingly, it was found that all PWHTs applied to the two joint types either maintained or improved performance, compared to the as-welded conditions. A relatively short PWHT (comparable to common industry practice) was observed to be the most beneficial, even though the response depended on the material combination tested. For the heat treatment times selected, the F22-625 showed greater sensitivity to PWHT when compared to the 8630-625 combination, with the best performance associated with a 10h PWHT. A shorter heat treatment time appeared to provide only marginal improvement in the properties of the 8630-625 combination and, interestingly, any further heat treatment time appeared to reduce the resistance to crack propagation back to the levels observed in the as-welded condition.

For microstructural analysis by TEM, a sample coupon was electro-discharge machined from each of the F22-Alloy 625 and 8630-Alloy 625 joints (joints 2 and 3).  The as-welded specimens were heat-treated for 10 hours, following the same heating profile as for the SENB specimens. TEM thin foils were milled from continuous interface regions (Fig 4a and b). TEM inspection revealed the presence of predominantly chromium based M₇C₃ carbides within the planar region adjacent to the interface in the 8630 specimen (Fig.4d). For the F22-Alloy 625 interface, the planar zone did not exhibit the same precipitates (Fig. 4c).  The M₇C₃ carbides may form due to diffusion of carbon from the parent steel during PWHT. As the 8630 joint has a significantly higher carbon content, M₇C₃ is likely to nucleate faster than in the F22 material. With longer heat-treatments, a similar band of M₇C₃ might be expected within the planar region of the F22 sample.

4.2 Fractography

Post-test SEM and TEM inspection of SENB fracture surfaces: After each bend test, the SENB specimens were heat-tinted, cooled in liquid nitrogen and broken open. This enabled the portion of the fracture face resulting from environmental cracking to be identified, the oxidised surface representing environmental cracking during the test.

A number of typical fracture modes, depending on the position of the crack path with regards to the interface, are shown in Figure 5. The fractographs come from the steel side of the broken SENB specimen. A flat fracture region was identified adjacent to the steel (area 1). On close inspection, a grainy, crater-like morphology was observed. The hexagonal craters typically had ridges radiating from a distinct central node (Figure 6b).

A typical area of fracture morphology, such as that shown in region 1 of Figure 5, was selected for TEM analysis (Fig. 6c and d). The thin foil was created by sputtering a protective Pt layer on the fracture surface and milling towards the steel side of the SENB specimen.

Below the fracture surface, the TEM specimen showed a lath-like structure, elongated and parallel with the plane of the thin foil. Closer to the fracture edge it was  apparent that the orientation of the laths differed with a number running perpendicular to the foil surface. It is this microstructure that is responsible for the ridges on the faces of the crater surfaces. The existence of a narrow transition zone, at all the interfaces between BCC and FCC microstructures close to the fusion line, has been described by other authors [7, 8]. One explanation of the presence of this zone, immediately adjacent to the protective platinum cap at the fracture edge (Fig. 7b), is that the flat fracture region might lie at the interface between the planar solidification region and a ‘swirl’ or ‘discontinuous PMZ’. Figure 8, an image of the opposing fracture surface to that seen in Fig.5 (therefore the Alloy 625 side), shows that this flat fracture morphology is observed at the interface between the planar solidification region (previously called the ‘featureless zone’) and the iron-rich side of the interface, be it a swirl or not. The image displays two flat crater regions at different elevations together with the imprint of the opposing features seen in Fig.5.

The terraced cleavage-like region was selected to coincide with the presence of a step on one of the terraced faces (Fig. 9b). The observed steps are perhaps evidence of a localised plasticity process and were therefore extracted for further scrutiny by TEM, Fig 9c and d. Interestingly, at the location of the step, what appeared to be a 400x100nm precipitate was found (Fig. 10b). EDX analysis from the apparent precipitate resulted in a similar spectrum to that of the surrounding matrix, thus the step is likely to be a small grain misoriented with the adjacent and larger grain.

The cleavage-like TEM sample also exhibited a lath-like structure, but not immediately below the fracture surface, Fig. 10a. As the terraced cleavage-like failure exists at a greater distance into the fused zone (i.e. further towards the nickel alloy), one would expect the structure to be fully austenitic, as is the case in the planar region. Instead, the cleavage-like morphology observed was within an iron-rich region. The presence of ferritic grains above the upper flat fracture might be explained if this region represented a location where the fracture cut across a ‘swirl’ or ‘discontinuous PMZ’.

A region that is more typical of the cleavage-like regions observed on interface fracture surfaces is that seen in Fig.11. In this region, the cleavage-like faces are much more pronounced and likely to exist within the planar region of a continuous PMZ.

Fractography of SENB specimens from PWHT investigation: The 8630-Alloy 625 specimen (joint 2) was selected for SEM analysis in the as-welded and PWHT conditions. Though in-depth fractography of these samples is ongoing, the as-welded specimens clearly exhibited different fracture morphologies from the equivalent PWHT specimens.

Both specimens exhibited occasional regions of flat crater-like failure as described in Fig.8, however, the as-welded specimen showed a greater surface area of flat interfacial fracture. An example of this is shown in Fig.12a, an image of the steel side of the SENB fracture surface. Penetration of molten weld metal between the HAZ grains during welding, known as ‘prior austenitic grain boundary penetration’, is indicated in the tilted image, Fig.12b.

Post-test longitudinal sectioning of SENB specimens: An alternative method to the above fracture surface analysis is to section, mount and polish the SENB specimens so that the crack path can be viewed in cross-section. For the retrieved 8630-625 interface, an SENB specimen was EDM wire cut in a manner that allowed the SENB test checks to be performed but that also allowed for crack path analysis. An example of this is shown in Fig.14a. In this image, the notch is seen to intersect the interface. At this particular point, the crack runs between the Alloy 625 weld metal, at the notch, and the interface. Cracking in the weld metal may not be typical, throughout the thickness of the whole sample, in part due to the undulating weld profile and other effects subsurface of the face of the polished specimen. In the planar solidification region (previously called the ‘featureless zone’), the crack propagated with a cleavage-like failure mechanism, occasionally transitioning to ductile tearing within the steel. Also evident to the left of the Figure is that of an ‘interface disbonding’ type failure.

Where there was a discontinuous PMZ, Fig.14b, the crack was deflected slightly, towards the weld metal, but it remained within the primary planar solidification region, located at the interface between the two, with the cleavage-like cracking mechanism becoming less apparent. It is this believed that the crater-like fracture morphology corresponds to cracking at this location. This failure mode was limited to transitions across the discontinuous PMZ. The crack propagated through the discontinuous PMZ to link to the secondary planar solidification region, i.e. that which had formed on the other side of the swirl. Thus, regions of crater-like failure were generated on each side of a swirl. Whilst this observation can been made in two dimensions, it is important not to rule out the three-dimensional crack path characteristics. It is possible that cracks may link around discontinuous PMZs, as well as through them.

5. Discussion

5.1 Fracture Mechanisms and Mechanical Response

As the stress is applied, the dominant cleavage-like failure mode appears to represent a low energy crack path. In the presence of a discontinuous PMZ, or perturbation in the interface, the crack path is deflected away from the direction of applied stress. The crack remains close to the interface between the planar solidification region and steel-rich material, creating the crater-like morphology. Although still under investigation, it is apparent that the as-welded specimens showed more flat fracture, compared to those which had received PWHT. 

It is also possible that the lath-like, martensitic regions in the discontinuous PMZ acted to reduce the amount of carbon within the adjacent planar solidification regions, so that there was less M₇C₃ carbide growth and thus a reduced propensity for crack-linking along carbide-matrix interfaces.

Cracks may develop ahead of the primary crack-front, as demonstrated in part I of the paper. Nevertheless, failure occurs across the discontinuous PMZ, as a means of linking cracks in planar solidification, low-energy, carbide rich regions. As a result, crater-like failure can exist at different points; above and below discontinuous PMZ regions, as failure mode transitions from cleavage-like to flat, crater-like failure and vice versa. The observations of crack path characteristics for the retrieved joints are illustrated, schematically, in Fig.15.

Whilst this may be the case for the retrieved subsea specimens that have had a typical, commercial heat treatment, this may not be the case for the joints fabricated for the PWHT investigation that have undergone both longer and shorter PWHTs. Indeed, the resistance to crack growth for the F22-625 and 8630-625 interfaces showed a bi-modal split in response to an applied stress, as the length of PWHT is varied. In the 8630-625 joint, it was possible to see the importance of PWHT; the 1 hour of tempering almost tripled the resistance to crack growth. The low PWHT time may avoid the formation a critical concentration of carbide in the planar solidification region through diffusion of carbon from the parent metal.

In the 8630 joint, 10 and 100 hours of PWHT resulted in little difference in mechanical response. It is possible that the critical carbon concentration to form a continuous band of M₇C₃ is reached soon after 1 hour of heating. Further heating may act to widen the width of the carbon saturated band, effectively increasing the depth into the planar solidification region at which M₇C₃ carbide, and hence cleavage-like failure, can occur.

In the F22-Alloy 625 joint, 1 hour of PWHT provided a similar improvement in properties to the 1 hour of PWHT in the 8630 interface, Fig.3. 10 hours of PWHT, however, resulted in a much higher resistance to crack growth in the F22-Alloy 625 system. The F22 system has a much lower carbon content (0.15 wt%) compared to the 8630 (0.33 wt%) and is therefore likely to require longer PWHT times to achieve similar carbon concentrations within the planar solidification region. Fenske et al [7,8] compared F22-Alloy 625 and 8630-Alloy 625 butterings in the 10 hour PWHT condition by TEM. They found that the planar region in the 8630 specimen was densely packed with chromium based M₇C₃ carbide but the same was not true of the F22 buttered forging. This supports the TEM findings described in Fig.4. Thus, for the F22-Alloy 625 system, 10 hours of PWHT is sufficient to temper the HAZ without forming the critical carbides. This observation further explains the response of the F22-Alloy 625 100 hour PWHT R-curve. At such an extended PWHT, M₇C₃ carbide may have nucleated, reducing resistance to crack propagation. In order to establish whether or not this hypothesis is true, in-depth fractography must be performed on the 10 and 100 hours F22-Alloy 625 specimens, as the crack paths are likely to differ due to the phase differences within the planar region.

6. Conclusion

The following conclusions can be drawn from this work: 

1. The importance of PWHT in increasing resistance to cracking was pronounced in these joints, though “over-ageing” can prove equally harmful. The micromechanical response of each joint is a direct consequence of the thermal history. A review of the PWHT is required, particularly in 8630-625 joints, due to the formation of deleterious phases through carbon diffusion during tempering.
2. The dominant failure mechanism in the 8630 – Alloy 625 interface after 10 Hours of PWHT is through a cleavage-like failure, as opposed to interfacial failure for the as-welded condition. Discontinuities in the PMZ disrupt cleavage-like cracking along the planar solidification region. This deflects the crack-path resulting in flat, crater-like failure along lath boundaries.

Acknowledgments

This work was joint funded by TWI Ltd., Cambridge, and MintWeld, a European Commission Seventh Framework Programme (FP7) sponsored project (contract No: NMP3-SL-2009-229108). The authors acknowledge the University of Loughborough for use of their dual-beam system. Ashley Spencer, Hiroto Kitaguchi and Geoff West are thanked for their contributions.

References

1. The Effect of PWHT On The Material Properties And Micro Structure In Inconel 625 And Inconel 725 Buttered Joints. Olden V, Kvaale P E, Simensen P A, Aaldstedt S, Solberg J K. Cancun, Mexico : OMAE, 2003. OMAE 2003-37196.
2. Resistance Of Dissimilar Joints Between Steel And Nickel Alloys To Hydrogen-Assisted Cracking.  Gittos M. F. s.l. : NACE, 2008. NACE 2008-08095.
3. Experience With Qualification And Use of Stainless Steels In Subsea Pipelines. Kvaale P, Rørvik G. Vancouver, Canada : 23rd International Conference on Offshore Mechanics and Arctic Engineering, 2004. OMAE 2004-5136.
4. Environment-Induced Cracking In Weld Joints In Subsea Oil And Gas Systems – Part I. M.F. Dodge, H.B. Dong, M. Militisky, R.P. Barnett, V.F. Marques, M.F. Gittos, Rio de Janeiro, Brazil, OMAE, 2012 OMAE2012-83402
5. Subsea Dissimilar Joints: Failure Mechanisms And Opportunities For Mitigation. Beaugrand V C M, Smith L S, Gittos M F. Atlanta, Georgia : NACE, 2009.
6. Hydrogen Embrittlement Of 8630M/625 Subsea Dissimilar Joints: Factors That Influence The Performance. Beaugrand V C M, Gittos M F, Smith LS. Honolulu, Hawaii, USA : s.n., 2009. OMAE2009-80030.
7. J, Fenske. Microstructure And Hydrogen Induced Failure Mechanisms In Dissimilar Iron Nickel Weldments. Urbana, Illinois : s.n., 2010.
8. Microstructure And Hydrogen-Induced Failure Mechanisms In Fe And Ni Alloy Weldments. Fenske J. A, Robertson I. M, Raghavan A, Hukle M, Lillig D, Newbury B. Metallurgical and Materials Transactions A. Vol 43A Sept 2012-3011