Improved heat treatment for superduplex thin walled tubes

TWI Bulletin, November - December 1997

Robert is Plant Assessment Manager with Plant Integrity Ltd (A TWI subsidiary) and has written a PhD thesis on 'The effect of thermal cycles on the microstructure and toughness of superduplex stainless steels'.

A common problem experienced by fabricators, contractors and operators when welding superduplex stainless steels is described here by Robert Gunn in co-operation with Weir Materials Ltd (WML), with whom a work programme was undertaken to solve this problem. This article looks at the approach to the work and the results obtained.

Over the last decade, concern has been expressed by the offshore industry at the formation of intermetallic phases in welds in superduplex stainless steels. Study under TWIs Core Research Programme ^[1] had indicated that the risk of intermetallic formation depended critically on the exact heat treatment condition of the material. It showed that, if solution annealing was not correctly carried out, the material could be left in such a condition that it met all specification requirements, but subsequent welding led to the formation of intermetallic phases in the weld HAZ, at higher levels than expected. Indeed, available evidence suggests that some of the industrial problems experienced with seamless pipe circa 1990 to 1994, relating to failure in meeting the fabrication specification requirement of no intermetallic phases in the weldment, could have stemmed from this cause. As a new facility has become available, which allows for an improved thermal cycle, WML decided to re-heat

To demonstrate WML's commitment to supplying an optimum product, they requested that an exercise be undertaken comparing welding behaviour of current stock material and that which would be re-heat treated. In essence, the work comprised welding steel of 3mm thickness, both in the present '1992 heat treatment' and after re-solution annealing. Gas tungsten arc welding (GTAW) conditions covered the normal situation, with the welds evaluated in terms of microstructure and pitting resistance in a ferric chloride test.

Experimental procedure

Materials

WML forwarded two welds in thin walled pipe. The pipes were originally 6" Schedule 40 (168mm OD, 7.11mm wall), but had been machined to 165mm OD, 3.7mm wall, in order to simulate the pipe dimensions (and heat sink) that had lead to intermetallic formation in production welding. One of these pipes was welded in the as-manufactured condition (pipe A, weld W2), while the other had been re-solution annealed (pipe B, weld W1) in an alternative continuous furnace which had become available recently. The heat treatment of pipe products, especially thin walled tubes, raises the problem of ensuring an adequate soak time, efficient quench, while minimising collapse. As superduplex materials have very limited hot strength at heat treatment temperatures, pipe collapse under their own weight can render them unusable ^[2] . The new furnace is designed to address the problems of pipe collapse, allowing for a longer soak time at temperature with a rapid quench.

The information from the original inspection certificates for the two pipes is shown in Table 1a, together with the requirements of UNS specification S32760. Further, the reported mechanical properties are summarised in Table 1b and compared with the requirements of ASTM A790. Both pipes can be seen to meet the compositional and mechanical requirements of the relevant standards and also comply with normal oil industry specification requirements.

Table 1a: Reported chemical compositions of pipes supplied

Table 1b: Reported mechanical properties of pipes supplied

Welding

Previous work undertaken by TWI ^[1] involved the evaluation of GTA welds in thin walled pipe following a production procedure known to induce HAZ intermetallic precipitation. This procedure was forwarded to WML and was adopted for the two test welds. The welding parameters are summarised in Table 2.

Table 2: Summary of welding conditions

Test programme

From each weld, samples were removed from both the 6 and 12 o'clock positions for metallographic assessment. Each sample was prepared to a 1µm finish and electrolytically etched in 20%KOH. The phase balance in the parent pipes was assessed by a point count procedure ^[3] involving 16 fields of 25 points and the presence of intermetallic phases was assessed at x500 magnification.

Given these results, one pitting corrosion test sample was removed from each weld at the location of highest sigma content. The pitting corrosion tests followed TWI Recommended Practice ^[4] . The root surfaces were left in the as-welded condition, while the cap and sides were ground and pickled to reduce the risk of attack on these non-test faces. Exposure to ferric chloride solution began at 35°C and increased by 2.5°C per day, until specimen failure, or until 52.5°C was reached. The failure criteria followed that described in ref.4, ie a significant weight loss or the first signs of pitting. This temperature was defined as the critical pitting temperature (CPT).

Results

Metallography

The ferrite point count results for the two pipes were 50.5 ± 2.9% for pipe A (Fig.1a) and 56.5 ± 4.8% for pipe B (Fig.1b). These results are statistically the same, due to the relatively large 95% confidence intervals. No intermetallic phases could be seen in either pipe at x500 magnification, although, after welding, some precipitates were found in the HAZ of pipe A (Fig.2), amounting to some 0.1%. There appeared to be a greater propensity for precipitation at the 12 o'clock, as opposed to the 6 o'clock position.

Pitting tests

Both weld samples recorded a weight loss of less than 10mg after exposure at 50°C. While at 52.5°C the as-manufactured pipe weld exhibited extensive attack in the root weld metal, (Fig.3), ie a CPT of 52.5°C. Even after exposure to 52.5°C, the re-solution annealed pipe weld showed no weight loss or evidence of pit initiation, ie CPT above 52.5°C.

W1 - Re-heat treated pipe B. No attack or weight loss.
W2 - As-manufactured pipe A. Weld metal attack.

Discussion

The as-received pipe formed intermetallic phases in the HAZ during welding (Fig.2), whereas the HAZ in the re-solution annealed pipe was free of such precipitates. It is considered unlikely that the small differences in composition between the pipes could account for this effect (Table 1a). In fact, the lower nitrogen content in the reheat-treated pipe may be expected to enhance intermetallic formation. Similarly, there is no clear evidence, when optically comparing the two pipe microstructures, as to why the as-manufactured pipe should be more susceptible to HAZ intermetallic formation. This leaves the variations in prior thermal processing as the most likely cause for the heat treated tubing being more resistant to intermetallic formation.

Results of the corrosion tests showed that, even though HAZ intermetallic precipitates were present, high CPTs, in excess of 50°C, were achieved. It is believed that, as these precipitates were present in such low volume fractions, ie about 0.1%, even though some alloy depleted regions may have been removed, the dissolved volume would have been too small to sustain a pit.

The major implication of this work is that some of the restrictive welding practices currently specified for superduplex alloys were based on thin walled pipes from the early 1990s, which had not been subject to the optimum solution heat treatment. Therefore, it is considered that, provided pipe is selected from more recent production or has been re-heat treated, the current limits ^[5] on interpass temperature and arc energy could be reset at higher levels.

In this regard, control of thermal cycle may be more critical with thin wall material than with the other products. The former can often be welded in only one or two passes, so that heat flow during arc welding approaches the two dimensional situation, and cooling rate is slow. With thicker material, heat flow will be nearer the three dimensional case, and cooling more rapid, so that greater latitude will exist in selection of welding conditions. At the same time, caution is necessary, and excessive arc energy must be avoided: arc energy and interpass temperature must be carefully selected in formulating a welding procedure, recognising the specific thickness and joint geometry concerned. A further caveat is that the HAZ thermal cycle experienced will depend on the exact welding procedure, as well as on the calculated arc energy. Certainly, if root runs in pipe are made using the GTAW process, the amount of filler addition will affect the situation, HAZ cooling rates tending to be faster with greater rate of filler addition, thereby minimising the risk of intermetallic precipitation.

Conclusions

• Under the welding conditions used, the as-received pipe showed some intermetallic particles in the HAZ, while no such particles were found in a similar weld HAZ in the re-solution annealed pipe.
• The current work has demonstrated that some of the pipe produced in the early 1990s had not been given an optimum solution anneal. Further, it has been shown that, if the pipe is re-solution annealed, the resistance to HAZ intermetallic formation is increased.
• Even with a small volume fraction of intermetallic phase in the HAZ (about 0.1%), a CPT in excess of 50°C was recorded, with the weld metal being the limiting weldment region.
• The results imply that some relaxation is possible of very restrictive welding practices developed for superduplex stainless steel, particularly for thicker section products ( eg >5mm). Nevertheless, it is essential that arc energy and interpass temperature are carefully selected for a given application, recognising the specific material thickness and joint geometry.

References

Duplex stainless steels - Microstructure, properties and applications - a book edited by Robert Gunn is based on keynote papers from the international series of duplex stainless steel conferences. Published by Abington Publishing (Tel: +44 (0)1223 891358).