Inclusions, Fracture Toughness and HAZ of Girth Welded Pipe

TWI Ltd,
Cambridge, UK.

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

Abstract

A programme of testing and research has been carried out to deliberately identify the causes of, and measure the fracture toughness of, potential local brittle zones (LBZs) in the heat affected zone (HAZ) of girth welds in a 20mm thick API 5L X70 grade pipe. This involved determining the most susceptible test temperature and notch location within the HAZ of two different welds by Charpy testing to define the T40J. Subsequent fracture toughness testing of a large set of fracture mechanics specimens notched into this region and tested at a low temperature based on T40J was able to achieve occasional low fracture toughness results. Post-test metallography identified that the inclusions in the parent metal affected the risk of brittle fracture initiation in two ways; firstly large and clustered inclusions acted as local brittle zones and/or fracture initiation sites; secondly, the large inclusions had reduced the effect of grain boundary pinning, as would be expected from well-dispersed smaller inclusions. This then resulted in significant grain coarsening in the HAZ fusion line, which could also act as a low toughness fracture initiation region. The inclusion types were chemically analysed, and recommendations were made to the steel producer in order to further improve the fracture toughness of the X70 pipe at increasingly lower temperatures.

1. Introduction

Girth welds in modern linepipe steels generally exhibit excellent HAZ toughness, when tested at ordinary service temperatures. Offshore pipelines are usually designed down to temperatures of around -20°C, while land pipelines can be designed down to -40°C. At such low temperatures (for example, land pipelines in Arctic conditions), isolated cases where material fails to meet the required toughness specification may be recorded, related to the presence of local brittle zones in the HAZ. Frequency of these failures is often very low, and standard testing schemes, based on impact toughness testing during welding procedure qualification (WPQ) only, may be unable to detect this behaviour. Pipe girth welds were provided to TWI fabricated from pipe material that was potentially susceptible to occasional incidence of brittle fracture in the HAZ at low temperatures. One was welded using gas metal arc welding (GMAW) and the other using manual shielded metal arc welding (SMAW). A research programme was proposed to capture and measure potential local brittle zones in the HAZs of these girth welds. This involved determining the most susceptible test temperature and notch location within the HAZs of the girth welds by Charpy testing to define the T40J. Charpy energy of 40J is within the ductile to brittle transition regime, and is a value often required by project specifications as a minimum impact toughness. The temperature at which 40J is achieved is an indication of the ductile to brittle transition of the material. Subsequent fracture toughness testing of a large set of test specimens notched into this susceptible region and tested at a low temperature based on T40J was expected to produce occasional low fracture toughness results, which could then be investigated to identify the cause(s) of this phenomenon.

2. Material and characterisation

2.1 Pipe and Girth Welds

The pipes supplied for testing were 20mm wall thickness 42in (1067mm) diameter API 5L X70 grade seam welded pipe. The GMAW weld was identified as W01, and the SMAW weld was identified as W02. The area between the 3 and 4 o’clock positions was used for Charpy testing. The 8-9 o’clock area was used for fracture toughness testing. Tensile specimens were extracted from the 5 o’clock position and a macro section was cut from the 7 o’clock position. These positions comply with the recommendations of Appendix C of DNV F101, 2010 edition [1]. The macro sections are shown in Figure 1.

2.2 Tensile properties

For each of the two welds, duplicate longitudinal round tensile tests were carried out on parent material and all-weld metal specimens machined adjacent to the plate surface, and tested in accordance with BS EN ISO 6892-1 [2]. Duplicate flat cross-weld tensile tests were also carried out for each weld to BS EN ISO 4136 [3]. All the tensile tests were carried out at room temperature. The SMYS (specified minimum yield strength) of X70 steel is Rt0.5% of 485MPa (from ISO 3183 [4]). Rt0.5% is defined as the yield point given by the strength at a strain of 0.5%. All but one specimen met this SMYS requirement for API 5L X70 material, and that lower result measured 483MPa (the other results were 487, 488 and 490MPa). The ultimate tensile strength (UTS) of the parent metal was between 584 and 586MPa, exceeding the specified minimum tensile strength value from API 5L X70 of 570MPa. Both welds over-matched the parent metal in terms of both yield and tensile strength. All the cross weld tensile tests failed in the parent metal.

2.3 Hardness

Vickers hardness survey traverses across the GMAW weld (W01) showed a maximum hardness of 242HV10 in the HAZ close to the fusion line. In the SMAW weld (W02) the hardness traverse gave a maximum hardness of 268HV10 in the weld metal. The highest HAZ hardness in W02 was 243HV10, adjacent to the weld metal, and similar to the maximum hardness recorded for W01. These hardness values were sufficiently low (less than 250HV10 in the HAZ) to indicate an acceptable level of weldability in these girth welds (Figure 2).

3. Toughness Behaviour

3.1 Charpy Transition Curves

Four sets of ten standard full-thickness Charpy specimens were extracted from each weld, machined from within 2mm of the weld cap surface. The Charpy specimens were machined and tested in accordance with BS EN ISO 148-1:2008 [5]. Charpy testing was carried out over a range of temperatures in order to generate Charpy transition curves (with 10 specimens per transition curve) for specimens notched at each of the following positions:

• Fusion line (FL)
• FL+0.5mm
• FL+1mm
• FL+2mm

Four Charpy transition curves were generated for each of the two welds. The Charpy transition curves were fitted by eye to the data points using a hyperbolic tangent (tanh) curve, defined as: 

Charpy toughness = A + B tanh ((T-T₀) / C)       (1)

Where A, B and C are fitting parameters around point of inflexion, T0. These curves were used to determine the temperature at which 40J would be measured using the following equation:

T_40J = T₀ + C atanh [ (40-A) / B ]       (2)

The Charpy data and fitted transition curves are shown in Figure 3 for GMAW weld W01, and Figure 4 for SMAW weld W02. The ‘worst-case’ location was determined to be the notch position associated with the highest value of T_40J.

The highest value of T_40J was -85°C in the FL+1mm location in GMAW weld W01. The T_40J for the other three notch locations in this weld was between -103 and -105°C. In the SMAW weld W02 the highest value of T_40J was -55°C in the FL+0.5mm location, although the Charpy transition curves from the FL, F+0.5 and FL+1 locations were all very similar, ranging from T_40J of -60 to -69°C. The differences between the two welds, in the notch position of the highest T_40J and its value (-85°C at FL+1mm for GMAW and -55°C at FL+0.5mm for SMAW) were due to the individual heat inputs and joint designs for the two types of weld.

3.2 Fracture Mechanics Testing

A set of 20 blank BxB single edge notched bend (SENB) fracture toughness specimens to BS EN ISO 15653 [6] were machined from each pipe girth weld. The fracture toughness specimens for welds W01 and W02 were notched into the location of highest T_40J as determined from the Charpy testing. The specimens were notched in the NQ (surface notch) orientation to a notch depth to specimen width (a/W) ratio of 0.5 (Figure 5). Local compression was applied before fatigue pre cracking to help ensure a straight pre-crack front. Surface notching was selected, rather than through thickness notching, so that a single microstructural region could be sampled across the whole precrack front. Despite this approach requiring post-test metallography to validate the notch positions, it gives data that are very location specific.

The fracture toughness test temperature was chosen to be T_40J+10°C to allow for the difference between the transition curves for Charpy impact toughness and fracture toughness. The fracture toughness test temperatures were therefore -75°C for GMAW weld W01 and -45°C for SMAW weld W02.

The values of CTOD were calculated using weld metal tensile properties adjusted for the temperature of testing, in accordance with BS EN ISO 15653 [6]. Fracture toughness in terms of J was calculated by estimating load line displacement from the double clip gauge. The specimens from the GMAW weld W01 were tested at a much lower temperature than the SMAW weld W02, and consequently the specimens from W01 tended to show much lower fracture toughness results, in addition to any effect of W01 and W02 having different notch positions. Of a total of 35 SENB tests carried out, three SENB tests showed pop-ins during test (Figure 6) with an associated value of CTOD less than 0.04mm, while a further three tests gave fracture results with a value of CTOD less than 0.09mm.

4. Post-test metallography

4.1 Confirming Crack Tip locations

The post-test metallography involved sectioning the specimens across the crack, transverse to the weld and then mounting, polishing and etching in 2% nitric acid in alcohol (nital) the side of each specimen containing the weld. From this section it was possible to measure the distance from the crack tip to the fusion line for each specimen. Unless the test result showed a brittle fracture event the specimens were sectioned at the mid thickness location. For any of the tests that showed pop-in or brittle results, the specimens were sectioned through the location of fracture initiation so that the brittle microstructure could be examined as well. The distances between the fatigue precrack tip and the fusion line were measured using an optical microscope. The specimens from weld W01 were targeting a region 1mm from the fusion line. In this weld the HAZ width was 0.8 to 1.2mm, and the coarse grained HAZ (CGHAZ) part was less than a quarter of a millimetre within this. The specimens from weld W02 were targeting a region 0.5mm from the fusion line. In weld W02 the HAZ width was 2 to 2.8mm, of which the CGHAZ was around half a mm wide. Therefore the notch location for W01 was at the outer part of the HAZ, whereas the notches for W02 were trying to sample the CGHAZ. Errors in positioning the full length of precrack tip at the specific distance from the fusion line are inherent to this test. Welds are not perfectly straight and fatigue precracks have some inherent curvature. Therefore, the actual crack tip location shows some scatter even for the same nominal notch location.

The values of fracture toughness are plotted against notch position in Figure 7 It was easier to obtain a notch position within the validity limit of ±0.5mm from the desired location in the straighter-sided GMAW weld W01, than in the SMAW weld, W02 which had a shallower bevel angle (Figure 1). This meant that slight variations in notch tip location or notch depth for W02 could result in a large change in the position of the crack tip relative to the fusion line and hence microstructure sampled. The greater scatter in the results from W02 compared to W01 can be seen Figure 7. Within the first fifteen SENB specimens for W01, thirteen results had notch positions within 0.5mm of the target location and were considered valid. The aim of obtaining at least ten valid specimens meant that the last five SENB specimen blanks did not need to be tested. For weld W02 only six valid results were obtained from 20 tests in all. However, if a distance from the target microstructure between 0.5 and 0.85mm could be considered ‘borderline’ then there were ten results that were either valid or borderline for the W02 specimens. It should be noted that the lowest value of fracture toughness corresponded to the target locations for both welds (Figure 7), but the correlation between toughness and location is subject to significant scatter in both welds.

4.2 Pop-ins

A pop-in is defined as an abrupt discontinuity in the force versus displacement record, featured as a sudden increase in displacement, and generally, a sudden decrease in force, subsequent to which, displacement and force increase to above their values at pop-in (Figure 6). All the pop-ins for this work exceeded 5% load drop and/or crack extension, and were considered to be significant for the determination of fracture toughness. Pop-ins can occur where a region of the ductile crack front becomes unstable and undergoes a change in fracture behaviour from stable ductile tearing to brittle cleavage. As the unstable cleavage fracture region arrests and reverts back to stable tearing for the remainder of the test the load / displacement recovers up to a load above that at the pop-in. Pop-ins occurred in specimens W01-50, W01-52 and W02-45. These three specimens were subjected to further metallurgical investigation to identify the cause(s) of the pop-in. The arrested cleavage fracture could be clearly seen on the fracture surface of specimens W01-52 and W02-45 (Figure 8). Specimen W01-50 showed a large split on the fracture surface but the cause of the pop-in was not so obvious from the fracture surface (Figure 9).

For all the pop-in specimens, the location of primary fracture initiation was identified using low power light microscopy and scanning electron microscopy. The location identified was marked, and a section was taken through that location. Each section was mounted in clear mounting compound and polished to a ¼µm diamond finish using standard metallographic preparation techniques such that the marked initiation region coincided with the polished metallographic specimen surface. The sections were examined using high power light microscopy before and after etching in nital to reveal the microstructure.

4.3 Specimen W02-45 Pop-in

When the section through the main initiation location of the pop-in in W02-45 was examined coarse, banded inclusions were observed and coarse grain clusters were also seen, (Figure 10). The change in local fracture mode in this region was therefore either a result of crack front intersection with a population of coarse brittle inclusions or through a cluster of coarse grains.

4.4 Specimen W01-52 Pop-in

Specimen W01-52 did not reveal clusters of coarse intermetallic inclusions near or on the main fracture face. Cleavage appeared to have initiated at regions where bands of coarser grains were present at the edge of the HAZ (Figure 11). Transgranular subsidiary cleavage was observed in regions where the crack front traversed coarse grains. Clusters of coarse grains near the centreline (mid thickness) of the specimen may have reduced the local toughness to an extent where the fracture mode changed from tearing to cleavage. Once the crack had traversed the coarse grained region, the toughness increased, hence causing crack arrest and a change in crack propagation mode to stable ductile tearing.

4.5 Specimen W01-50 Pop-in

The fracture surface of W01-50 was dominated by a change in the fracture plane (‘split’) feature oriented transverse to the direction of the precrack (Figure 9). The lack of evidence of further stable tearing around this feature indicated that it had occurred during final breaking open of the specimen, and was not itself the cause of the pop-in. Two potential initiation sites for the pop-in during CTOD testing were identified on the specimen, despite being masked by this split-like feature. Two sections were taken from this specimen, from each of the two potential initiation regions indicated by yellow dots on the fracture face shown in Figure 9. Under the SEM the region on the left side showed coalescence of large microvoids at the site of initiation, see Figure 12.

This was consistent with large voids associated with clustered inclusions. A second section was taken on the right hand side of the fracture face. The SEM image of this region is shown in Figure 13, which showed a fracture adjacent to the split. A section was taken through the initiation region on the right of Figure 9 and examined under the optical microscope. This location exhibited a split transverse to the fracture propagation direction, along with a high number and density of large inclusions (Figure 14). The blunt nature of the tip of the split, which arrested at the grain refined visible HAZ indicated that it is not due to cracking, but that during final fracture the rupture has followed a plane of weakness in that orientation. This may again have been due to the nature of the inclusions in this weld.

The fracture initiation section the left side of Figure 9 could not be linked to a specific inclusion (Figure 15), but given the distribution of large inclusions observed, and their alignment, it is likely that the interaction of the notch tip with an inclusion cluster had triggered the brittle fracture event. The parent material generally had a fine ferrite and carbide (bainitic) microstructure.

4.6 Other Brittle Fracture Results

Two other results from weld W01 and one from W02 gave brittle fracture results corresponding to values of CTOD less than 0.09mm. The microstructure at the initiation regions of these specimens was also examined. The two fractured specimens from W01 had initiated behind the fatigue precrack tip at inclusions at the edge of the visible HAZ (an example is shown in Figure 16).

This was consistent with the findings from the pop-in specimens W01-50 and W01-52. The other fractured specimen from W02 initiated from the grain coarsened HAZ adjacent to the fusion boundary, similar to the pop-in specimen W02-45 except that in this specimen the notch position was closer to the fusion boundary.

The possibility that the inclusions in the parent steel were a significant contributing factor to the pop-ins led to some further investigations into the nature of the inclusions.

5. Inclusion analysis

5.1 Inclusion Morphology

The size and dispersion of non-metallic inclusions in W01-50 can be seen from the un-etched image of parent metal (left) and weld metal (right) in Figure 17. This distribution of parent metal inclusions is similar for all the specimens tested from both welds. Figures 18 and 19 show in higher magnification the type of inclusions seen in this sample, some significantly elongated in nature. The post-test metallography sections from specimen W01-50 were subjected to semi-quantitative energy dispersive X-ray analysis (EDX) in the scanning electron microscope to determine the elements present in the inclusions. Several inclusions were examined, from elongated stringer to round pore-like inclusions and the elements present in the each type of inclusion were recorded.

5.2 Inclusion Composition

Figures 20 to 23 show the results from energy dispersive X-ray (EDX) analysis of the inclusions identified in the parent material, showing the presence of manganese sulphide, titanium and niobium (expected to be carbo-nitrides), aluminium and calcium. In the parent material these inclusions were significant microstructural features, up to 10µm in diameter, and where present as clusters, or strings, these may have a significant effect on the fracture toughness of the material, as observed in the results of test W01-50.

The compositions of the inclusions must be considered in comparison to the overall steel composition, which was measured using optical emission spectroscopy, given in Table 1.

6.Discussion

6.1 Project Objectives

The intention of this work was to find a fracture toughness test temperature which gave isolated low fracture toughness results (approximately 20% of the valid test results showed this). The objective was to identify the causes of sporadic LBZs, and isolated low toughness test results, while allowing for the inherent errors in obtaining consistent crack tip positions relative to a weld fusion line. Testing at too high a temperature would give fully ductile behaviour without any information on LBZs, whereas too low a test temperature would exhibit gross brittle behaviour which would mask any localised brittle regions. The method for choosing the test temperatures was therefore reasonably successful, since 25% of valid results for W01 and 33% of the valid results for W02 gave CTOD of below 0.1mm.

6.2 Charpy Impact Properties

The highest value of T_40J was in the FL+1mm location in the GMAW weld W01, and had a value of -85°C. The highest value of T_40J was in the FL+0.5mm location in the SMAW weld W02, and gave a value of -55°C; although the Charpy transition curves from the FL, F+0.5 and FL+1 locations were all very similar. These Charpy impact toughness results were very good, as indicated by the low temperatures determined for the T_40J for both welds.

Table 1 Composition of parent pipe material

The difference in the location of the lowest impact toughness between the two weld types was likely to be a result of the different HAZ widths between the two. The SMAW weld, W02, had been welded with a higher heat input (1.1 to 3.1kJ/mm in the fill passes), and resulted in an HAZ width of 2 to 2.8mm, whereas the GMAW weld, W01, welded with a heat input of 0.53 to 1.76kJ/mm, gave an HAZ width of 0.8 to 1.2mm. Furthermore the differences in bevel angle and thus weld bead sequence resulted in different degrees of HAZ refinement. The GMAW weld showed less coarse-grained HAZ along the fusion line than the SMAW weld.

The scatter in the repeat Charpy tests, carried out at temperatures close to the T_40J are an indication of the inherent scatter that Charpy tests show when steel is tested within the ductile to brittle transition regime, and also the scatter due to slight differences in the exact notch positions within the narrow HAZ region. Both repeat sets were considered acceptable in order to confirm the T_40J temperatures, and hence decide the fracture toughness test temperatures.

6.3 Fracture Performance

The fracture toughness results for the GMAW weld, W01 tested at -75°C showed that even where relatively high values of CTOD were obtained (greater than 0.2mm CTOD), the load versus displacement traces did not show much ductility, and most specimens failed to reach maximum load. This was shown in the results by a δc  (fracture at less than 0.2mm of tearing) or δu (>0.2mm of tearing but load displacement trace did not reach maximum load) instead of δm (fully ductile) description of the fracture toughness. The higher test temperature of the W02 specimens meant that more of the specimens showed fully ductile results (ie δm), particularly when the actual notch position was more than half a millimetre from the fusion line. The HAZs of both the SMAW and GMAW welds showed occasional pop-ins and low fracture toughness results at test temperatures of -45 and -75°C respectively.

6.4 Microstructure

From examination of the SENB specimens that showed pop-ins, the sub-critical HAZ region gave low toughness due to the parent material inclusion distribution rather than any other microstructural features. The distribution of inclusions, thought to be present for grain pinning and strengthening effects, appeared to be clustered. Regions were observed containing many similar inclusions rather than there being an even distribution of different types of inclusions in the material. This would indicate that there may be concerns over the temperature of the melt (insufficiently high to dissolve certain inclusions), or the mixing within the melt, prior to casting the original steel bloom from which the pipe was made.

The parent metal microstructure had a fine effective grain size, which in itself would give good fracture toughness behaviour. However, grain coarsening in the HAZ (for some of the specimens) may have occurred due to a localised reduction in grain pinning inclusions from the intended level. For a given thermal and processing history, the material contained a given percentage of inclusions per unit volume. Local coarsening of the inclusions would have resulted in an increase in the mean distance between the inclusions. Clustering of inclusions during processing would also have resulted in some relatively large localised inclusion‑free zones. Either mechanism would have resulted in a reduction in the grain pinning effect and an increase in the local grain size near the HAZ. The relative effect of these two mechanisms will depend on the welding procedure, the heat input of the welding process and hence the amount of grain coarsened HAZ developed in the welded joint. Hence the wide HAZ of the SMAW weld W02 was associated with occasional fracture events mainly from the CGHAZ, whereas the narrow HAZ of the GMAW weld W01 showed occasional fracture events only associated with clustered inclusions.

6.5 Composition

The composition of this pipe showed sufficient titanium was present to achieve the potential benefit in terms of GCHAZ grain refinement and hence HAZ toughness, but, the effect of titanium was inter-dependent upon other alloying additions, and it is possible that its effect could have been maximised with slightly more vanadium and less niobium. Some of the inclusions detected were consistent with TiN and modification of these inclusions may help. The level of nitrogen should be carefully controlled so that it is in a stoichiometric ratio with nitride forming elements (Ti, V and Nb) to optimise the high toughness these nitrides can provide, bearing in mind that excess nitrogen can be very detrimental.

The pipe tested in this work did not have deliberate V additions, but the presence of Nb containing inclusions is known to have a detrimental effect on HAZ toughness at higher heat inputs. This is considered to arise from Nb remaining in solution after finish rolling of the plate, and subsequently precipitating out during cooling. These precipitates can then return to solution with the high temperatures in the HAZ, and depending on the speed of cooling, can remain in solution (fast cooling) or re-precipitate (slow cooling) (Milititsky, 2010). Work by Dolby (1975) suggested a pre-precipitation clustering effect of these types of inclusion can occur and has a detrimental effect. The Ca addition of 27ppm was considered to be beneficial in HAZ toughness as complex inclusions containing Ca can be initiation sites for polygonal ferrite. However, as the Ca containing inclusions were up to 10µm in diameter, it is possible that the complex inclusions were too large to effectively nucleate polygonal ferrite throughout the HAZ.

Conclusions & recommendations

The following conclusions have been drawn from the results of this work:

1. The Charpy impact tests on the girth weld HAZs in both GMAW and SMAW welds were very good, with impact toughness in excess of 40J at temperatures as low as -100°C, with even the worst location giving T_40J of -55°C.

2. The parent material composition and grain size were both typical of X70 steel and gave no indication in themselves that local brittle effects may occur in the HAZ when welded.

3. Most of the fracture toughness tests gave reasonably high values of CTOD (above 0.2mm for the GMAW welds tested at  -75°C, and above 0.45mm for the SMAW welds tested at           -45°C). However, a number of specimens did give very low fracture toughness values (below 0.1mm) in both welds; either due to pop-ins or critical fracture events.

4. The low values of fracture toughness were attributed to the size and distribution of inclusions in the steel, and to grain coarsening in the HAZ microstructures.

5. Inclusions which were relatively large in size and/or clustered or aligned within the microstructure, may have acted as planes of weakness to allow brittle crack initiation and propagation.

6. The size and distribution of inclusions could also have reduced the effectiveness if the grain boundary pinning, such that they did not prevent grain growth within the HAZs.

The conclusions presented here would be strengthened with extra evidence from further testing and the generation of additional pop-in results for analysis, but there are no further tests currently planned for this project.

TWI recommended that modifications be made to the steel making process to reduce the number of large and elongated inclusions in the steel, and to ensure they were well distributed throughout the steel. In addition, fracture toughness testing of specimens notched into the HAZ during development of new steels and their products should be performed, so that the effectiveness of any changes in inclusion shape, size and distribution on HAZ fracture toughness can be quantified.

References

1. DNV-OS-F101:2010, ‘Submarine pipeline systems’. Det Norske Veritas, 2010.
2. BS EN ISO 6892-1:2009 ‘Metallic materials -- Tensile testing -- Part 1: Method of test at room temperature’ British Standards Institution 2009.
3. BS EN ISO 4136:2012, ‘Destructive tests on welds in metallic materials -- Transverse tensile test’. British Standards Institution 2012.
4. ISO 3183:2012 ‘Petroleum and natural gas industries -- Steel pipe for pipeline transportation systems’.
5. BS EN ISO 148-1:2009 ‘Metallic materials -- Charpy pendulum impact test -- Part 1: Test method’, British Standards Institution 2009.
6. BS EN ISO 15653:2010 ‘Metallic materials -- Method of test for the determination of quasistatic fracture toughness of welds’ British Standards Institution 2010.
7. Milititsky M, 2010. TWI Internal Communication
8. Dolby R E, 1975 ‘The toughness of high heat input welds in C-Mn steel fabrication’ Proceedings, Conference on Mechanics and Physics of Fracture, Cambridge. 6-8 January 1975. Institute of Physics; Metals Society. 1975. Paper 11.