Silver anniversary of duplex stainless steels

TWI Bulletin, November/December 1998

Trevor Gooch graduated in industrial metallurgy at the University of Birmingham, and continued to obtain the degrees of MSc(Eng) and PhD from the University of London. Dr Gooch joined TWI in 1965. In 1980, Dr Gooch became Head of the Materials Department, embracing the welding characteristics of virtually all metallic materials. In 1994, he was made Materials Technology Manager for TWI, involved with joining characteristics of all materials used for construction.

In 1973, the first welding studies at TWI on ferritic/austenitic stainless steels were initiated. These materials have made a major impact on many sectors of industry, and Trevor Gooch looks back over the last quarter century to review the work carried out at Abington.

The revelation that the addition of more than about 10%Cr to steel bestows the material with the property of passivity, and the subsequent development of stainless steels, ranks as one of the major technological achievements of the 20th century.

The fact that passivity and associated corrosion resistance can be obtained in an iron-based system is extremely fortunate, since, by appropriate adjustment of other alloying elements, notably nickel and carbon, a wide range of microstructures can be developed, martensitic, ferritic or austenitic, together with multiphase materials. Hence, stainless steels can be produced offering a remarkable range of mechanical properties and corrosion resistance, and they are used in large quantities by most sectors of industry, whether petrochemical, power generation, transport or whatever.

The basic metallurgy of the iron/chromium and iron/chromium/nickel systems was understood by about 1940, and by the fifties stainless steels had become well standardised in specifications that have changed little since that time. Over much of their history, and certainly at the present time, the austenitic grades have been dominant and produced in greatest tonnages.

The most common austenitic alloys contain roughly 18%Cr and 10%Ni, with addition of other elements, such as molybdenum, niobium or titanium, to obtain specific corrosion resistance or mechanical properties. These alloys are very widely used for process plant of many types and functions, and play a major role in the petrochemical and power generation industries in particular. However, with all their attractive properties, they have an Achilles heel in that they are very sensitive to chloride stress corrosion cracking (SCC) in aqueous media at temperatures higher than about 50-60°C.

Even low chloride levels are sufficient to induce cracking, and the problem is a major cause of plant failure and economic loss. In consequence, extensive research was carried out internationally in the fifties and sixties, but the result of this study involved realisation that the problem is inherent in most austenitic materials. Many steelmakers therefore undertook research into alternative compositions of stainless steels that would avoid this common and expensive failure mechanism. By the early seventies, users were being offered the choice of two new types of stainless steel, namely fully ferritic grades processed to have extra low interstitial element (ELI) contents and materials with composition adjusted to give a duplex ferritic/austenitic microstructure (Fig.1).

While the new alloys were a long way from being established in the market place, their potential range of application was considerable. So in 1973, the decision was taken to initiate studies at TWI on their welding behaviour as part of the Core Research Programme, since clearly welding would be required for fabrication of any structure of consequence. In the event, the fully ferritic ELI grades have remained minor tonnage materials, largely because it is difficult to ensure high toughness at welded joints. The situation is very different for the duplex stainless steels, which now constitute a major proportion of worldwide stainless steel production. This position has not, however, been reached without alloy development specifically to obtain improved weldability.

Duplex alloy development

By 1973, ferritic/austenitic stainless steels had been sold commercially for many years, primarily as castings or high strength bar stock, typified by the AISI 329 grade. It was recognised that materials of this type transformed to ferrite in the high temperature heat affected zone (HAZ) of a fusion weld, with inevitable loss of corrosion resistance and toughness (Fig.2). In consequence, it was not expected that these initial product forms would be placed in service in the as-welded condition: even if weld repairs were needed to castings for example, a full solution treatment to restore the ferrite/austenite balance was considered essential.

A number of the new stress corrosion resistant alloys were based directly on the AISI 329 specification, with only minor compositional change to recognise the needs of welding, while retaining an alloy balance that could be processed following established procedures by the steelmaker. One commercial material, UNS S31500, was derived from the AISI 316 composition by reducing nickel to increase the amount of ferrite, together with some increase in silicon content since this element was at the time considered beneficial for stress corrosion resistance. Interestingly, most alloys contained an addition of nitrogen to enhance austenite formation on welding, although the amount was low, less than 0.1%.

Recognising the different alloy types, initial work at TWI set out conventionally to examine the response to fusion welding of representative commercial alloys. Joints were made by a range of processes, using various consumable types, with metallurgical examination and assessment of mechanical properties and corrosion resistance. The former included both Charpy and CTOD testing, while the corrosion studies were aimed primarily at sensitisation to intercrystalline attack, since in 1973 this was seen as the main corrosion effect of a weld thermal cycle. Stress corrosion cracking tests were also undertaken in various chloride media, to compare the resistance of welded joints and parent material.

Following this work, the picture emerged that the new alloys were readily weldable, in the sense that joints could be made, but achievement of an adequate austenite content in the weld area was essential to obtain reasonable properties (Fig.3).

At the time, the steels were seen as direct replacements for conventional austenitic grades in plant which had suffered stress corrosion cracking. Demands on welded joints were in consequence fairly modest; they were not expected to have high toughness below say 0°C, while the principal corrosion concern was that there should be resistance to intercrystalline attack. These requirements could be met with an austenite content of 20%, which would be considered unacceptably low by today's standards. Nevertheless, the demonstration that the new alloys could be welded and display good properties led to a general interest in many sectors of industry.

The results of the studies at Abington were released to Member Companies and then published in open literature, and a seminar was organised by TWI specifically on the new steels. Increasingly, the alloys became the subject of technical queries to TWI staff, as they were specified more frequently. A particularly unusual application was in Bush Lane House in London. Cast duplex stainless steel was used to produce a lattice which was the main load bearing structure of the building, and this remarkable design involved demonstration by TWI that adequate weld area fracture toughness could be achieved in the chosen material.

As might be expected, not all applications of the new steels were successful. Service failures were encountered not only from chloride cracking in particularly severe environmental conditions but also from intercrystalline corrosion at welds where the combination of base metal composition and weld thermal cycle had resulted in a substantially ferritic HAZ microstructure (Fig.4). This deficiency was recognised by alloy producers, and the late seventies saw the development of a grade later to be standardised as UNS S31803. The material can in many ways be regarded as the first designed specifically to give a duplex ferritic/austenitic structure with good weldability and overall properties. The nitrogen content was raised significantly from previous alloys: the UNS S31803 specification permits 0.08-0.2% nitrogen, but few if any commercial heats were made with less than 0.10% of this element. At the same time, chromium levels were increased relative to AISI316, for example, to about 22%. This, in conjunction with the nitrogen, endowed the steel with a useful improvement in corrosion resistance.

Despite the clear advantages offered by duplex steels for many uses, the market remained more sluggish than had been anticipated. The materials were widely used for purposes such as heat exchanger tubing, but few designers were prepared to make a radical change in materials used for most process plant, apparently preferring conservatively to live with the problem of SCC. However, this was about to change.

Around 1980, the oil and gas industry began looking hard at their continuing problem of CO ₂ corrosion, with the necessity for inhibition to protect carbon steel lines and other plant with associated potential unreliability. The virtual immunity to CO ₂ corrosion shown by stainless steels was recognised, together with the fairly high strength obtainable from duplex alloys in particular. Hence, despite the high material cost relative to carbon steel, serious examination began of the use of duplex steels for linepipe. This led to a flurry of activity throughout the world, not only to find the limits within which 22%Cr ferritic/austenitic steels could be used, but also to derive more highly alloyed, more corrosion resistant materials.

In consequence, a range of 'superduplex' steels emerged, achieving better resistance to environmental attack by increased chromium, molybdenum and sometimes tungsten and copper levels, in conjunction with an increase in nitrogen content to above 0.2%. In all cases, the requirement for good weldability by fusion processes was recognised, and the compositions offered were balanced to provide tolerance to a weld thermal cycle. This development has continued to the present day, and the designer and fabricator now have a range of materials available commercially in product forms appropriate to most applications.

Welding procedure development

After 10 years study, the guiding principles for formulating a welding procedure for ferritic/austenitic stainless steels were well established. Demands on welded joints had become appreciably more onerous. Good toughness at sub-zero temperatures was now a requisite, while applications in media with high chloride levels, whether for offshore service or otherwise, had led to the requirement that weldment pitting resistance should be measured and should at least approach that of the parent material.

To meet these needs, it was clear that ferrite/austenite balance should be carefully controlled, and in particular that the weld thermal cycles should entail cooling conditions sufficiently slow that substantial austenite reformation could take place (Fig.5), generally over some 30% austenite being required in the weld area. At the same time, studies for Member Companies on higher alloy grades of duplex steel had highlighted the possibility of intermetallic formation and degradation of properties if cooling times were unduly protracted (Fig.6).

These investigations led to the now well recognised requirement that arc energy should be maintained within upper and lower levels, and this remains the key point in designing a welding procedure. In this regard, the alloys share the necessity to limit maximum heat input (and interpass temperature) with austenitic grades, but the need for a minimum heat input must be clearly recognised as characteristic of duplex steels. Consumable manufacturers had appreciated the hazards of low weld metal austenite contents, and by 1980 it had become the norm to weld duplex steels with consumables having increased nickel level relative to the parent material, to facilitate reliable achievement of a satisfactory phase balance.

Furthermore, from work primarily on high nitrogen high strength austenitic alloys, it was known that nitrogen would be lost from TIG weld metal using conventional Ar shielding, leading to reduction in austenite level and corrosion resistance. Solidification cracking had been observed but found not normally to be a significant practical problem, while the possible risk of hydrogen cracking was recognised, primarily in weld metals of fairly high ferrite and hydrogen contents.

The main principles to be followed in welding ferritic/austenitic steels were outlined in a keynote paper presented by TWI at the first major conference dedicated solely to the materials, held by ASM in St Louis, USA, in 1982. With the trend to more exacting requirements on joint properties, it became necessary to quantify these guidelines more precisely. A number of studies were therefore carried out at Abington in the eighties, continuing to the present day, both to define limits on welding procedure variables for a range of alloys, and to generate data on material transformation behaviour and resultant joint properties. Work has been undertaken as part of the TWI Core Research Programme, but, significantly, direct industrial support has made a major contribution, in Group Sponsored Projects and single client investigations.

Welding metallurgy

In view of the controlling role of ferrite/austenite balance on joint properties, effort in the eighties in the Core Research Programme concentrated on obtaining a predictive system based on material composition. The conventional Schaeffler diagram was unacceptably inaccurate, but the general principle was followed of regarding individual alloying elements as promoting either ferrite or austenite. This led to formulation of the Q-factor as in Fig.7. Derived essentially for weld metals, it has been used by consumable manufacturers, but is applicable also to HAZ transformation. In 1992, the WRC-92 predictive diagram was published and has become widely used to predict weld metal microstructure knowledge of the composition, but recent study has indicated that the Q-factor offers at least equal reliability. Neither diagram is completely accurate, and in part this is because neither recognises the effect of welding procedure and cooling rate on the extent of the ferrite/austenite transformation. The situation has therefore been re-examined at TWI with a particular contribution from Koichi Yasuda, a visiting Japanese worker seconded by Kawasaki Steel. The work led to a model recognising both composition and welding procedure variables that can be run fairly simply on a PC, and enables austenite content to be predicted, at least for single pass welds, sufficiently accurately for most practical purposes.

In many practical situations the high weld completion rate offered by power beam processes or resistance welding is attractive. These methods, however, inherently apply a high power density with attendant rapid cooling. The use was studied at TWI of laser and resistance seam welding for duplex steels. As expected, weld area austenite levels with these processes are low. However, the work showed that for many applications joint properties may be acceptable, and, further, consistent with the earlier modelling, most satisfactory results are obtained with fairly high nitrogen materials which can be calculated to have a high equilibrium austenite content.

Following development and increased use of the high alloy superduplex grades, it became clear in the early nineties that - as had been known since the seventies - the materials could be subject to precipitation of intermetallic phases during a weld thermal cycle. The effect was most marked for combinations of material thickness and heat input such as to cause extended cooling times. Study was undertaken on the precipitation behaviour of various commercial alloys and cooling times for intermetallic formation were found to be generally comparable with those indicated by isothermal time/temperature/transformation investigations.

With most precipitation processes, nucleation and growth during welding tend to be rather faster than indicated by isothermal studies, and the behaviour of duplex alloys may reflect an inherent accelerating effect on precipitation of internal strain arising from differential expansion between the ferrite and austenite phases. Nevertheless, the work showed that commercial superduplex steels can be reliably welded without unrealistic restriction on welding conditions, although it was noted that the base metal must be correctly solution annealed prior to welding to minimise any subsequent intermetallic problems.

Apart from transformation behaviour, both Group Sponsored and Single Sponsored studies were undertaken to address nitrogen contents in TIG weld metal as commonly involved in root runs of pipe for example. The effects of welding conditions were explored together with material composition and also the addition of nitrogen to the shielding gas (Fig.8). In consequence of this work, a number of nitrogen-bearing gases have become available commercially for welding ferritic/austenitic steels.

Weldment properties

In the mid-1980s, a major Group Sponsored Project on 22%Cr duplex steel was undertaken, supported by over 20 companies. The work showed the potential adverse effects of metallurgical transformation during welding (Fig.9). Further, non-equilibrium partitioning of alloying elements between ferrite and austenite during a weld thermal cycle was identified, and the study pointed to the benefits of slight overalloying of the welding consumable to obtain pitting resistance equivalent to that of base metal (Fig.10). Corrosion and toughness data were then generated for a range of welding processes and procedures.

This study was followed by a similar programme on a range of superduplex alloys, again sponsored by over 20 research member companies. As before, corrosion resistance and mechanical properties of welds were assessed, but this time also with study of stress corrosion cracking behaviour in H ₂ S and dilute chloride conditions. The property levels achieved for the different steels were good, and, consistent with the high nitrogen contents, the alloys examined displayed considerable tolerance to welding conditions, although the potentially adverse effect of high arc energy or interpass temperature was again demonstrated.

By 1990, a number of requirements for weldment procedure qualification specifications had been accepted by industry. As is generally the case for welds in ferrous materials, these involved demonstration of joint soundness and mechanical properties. However, it was now the norm also to specify metallographic examination to ensure that the phase balance was within certain limits, typically 35-65% austenite, with freedom from intermetallic phases. Moreover, demonstration was required that welding had not reduced corrosion resistance to an unacceptable degree, and this normally took the form of a pitting test in ferric chloride following the precepts of ASTM G48.

Weld procedure qualification testing was undertaken by a number of laboratories, both within individual companies and in test houses, and it became clear that differences in test procedure could lead to extremely disparate results, especially with regard to metallographic examination and ferric chloride testing. Accordingly, a Group Sponsored project was undertaken to derive 'standard' test procedures that could be followed by different laboratories but give similar results. The resultant test procedures were transmitted to IIW and have found their way into oil company specifications. They are also being included as appendices to the draft revision of BS 4515 Part 2 for welding procedures for duplex steel pipelines.

Moreover, TWI has produced secondary standards for measurement of ferrite level in both nominally austenitic and ferritic-austenitic weld metals. These were prepared under the auspices of IIW Subcommission IIC, from material supplied by Mladis Co (Russia) in comparison with NPO CNIITMASH. The samples are recognised for calibration of measuring instruments in the current draft revision of ISO 8249 on ferrite determination for stainless steel weld metal.

In many oil and gas applications, ferritic/austenitic steels could be required to resist media containing H ₂ S. For some years, the NACE MR0175 standard has been the major reference document describing materials requirements for such service, but there has been debate regarding tolerable H ₂ S levels to avoid sulphide stress cracking (SSC) for a range of corrosion resistant alloys (CRAs) including ferritic/austenitic steels. A review was therefore undertaken with Group Sponsorship of published and in-house information.

From this review, conservative limits of H ₂ S partial pressure were obtained for austenitic and duplex stainless steels and nickel alloys. Recognising a number of further studies reported during the 1990s, the review has recently been updated. Data from both projects have been collated in spreadsheet form, and the technical reports and PC software can be obtained by joining the Sponsor Groups effectively on a retrospective basis. The information is also being incorporated in a database on duplex steel corrosion properties, which will be offered for sale by the National Physical Laboratory.

The results of these reviews gave encouragement that welded joints in duplex steels, made following a good welding procedure, will have sulphide cracking resistance equivalent to that of the parent steel. However, it was apparent that no data had been published on the SSC behaviour of joints made under less than optimum conditions. Again, a Group Sponsored Project has been undertaken, and an extension to this study is still in progress at Abington.

One of the problems in introducing a new material class into service is the derivation of appropriate design criteria. Certainly this is the case for toughness, and the Structural Integrity Department of TWI undertook a Group Sponsored Project to define toughness acceptance levels which can be incorporated in relevant national and international standards. These are recognised in the current draft of BS 4515 Part 2. The work also illustrated the potentially major effect of intermetallic phases on toughness (Fig.11). Nevertheless, from this and subsequent Core Research Programme study, the Charpy and CTOD toughness criteria obtained can be taken as appropriate not only to welds made with optimum procedures but also to joints containing some intermetallic precipitation (Fig.12).

In the general sense, problems still remain of avoiding intermetallic formation, primarily in superduplex grades, but in heavy section base metal as well as at welds. From study in the Core Research Programme, there is no doubt that the hazards of intermetallic formation are frequently overestimated. Nevertheless the possibility of intermetallic formation must be recognised at all stages of material fabrication. In this regard, recent work has shown that, although high intermetallic contents can greatly reduce properties, there is no direct correlation between the volume fraction of intermetallic and properties at levels of more concern with fusion welding (Fig.13).

It must therefore be recommended that weld procedure qualification, although recognising the possibility of excessive intermetallic formation, should be based primarily on direct measurement of properties of interest. Following this theme, a Group Sponsored project is in progress to assess corrosion resistance of weldments containing intermetallic phases (in both ferritic/austenitic and superaustenitic alloys) exposed to media reproducing the practical situation, as opposed to the severe laboratory environment constituted by ferric chloride tests.

From the substantial amount of production welding carried out during the eighties and nineties, it was clear that there was little risk of hydrogen cracking at welds in duplex steels. Nonetheless, cracking was found on sporadic occasions, most commonly in weld metal with a largely ferritic structure or in which hydrogen levels were unusually high. Although not a frequent problem, some instances of cracking led to considerable economic loss. Hence, a Group Sponsored project has been carried out to derive a suitable hydrogen analysis procedure for duplex weld metal, to generate data on hydrogen levels of a range of commercial consumables, and to define the relationship between ferrite content, hydrogen level and incidence of cracking. This work is in progress at the present time.

With the greatly increased use of ferritic/austenitic steels since 1980, a wide range of other studies has been undertaken at Abington, examining welding behaviour relative to one or other application, including hyperbaric welding. The properties and limits of weldments in 22%Cr steels have been well defined, so that the materials have become fully integrated into many sectors of industry, and are specified and fabricated on a routine basis. In large part, this is true also for the superduplex alloys, although the inevitable higher material cost means that they are produced and used in lower quantities. The higher corrosion resistance available is potentially attractive for diverse plant, and effort is ongoing to define the environmental limits on welded joints in various media.

Conclusions

The growth in use of ferritic/austenitic steels over the last 15 years has been dramatic, and indeed it has been reported that the 22%Cr UNS S31803/32205 grades are now third in production tonnage behind the common austenitic 304 and 316 variants. This situation exemplifies the fact that materials can be manufactured and used in welded fabrications reliably, and this would not have been possible without the considerable efforts undertaken on welding behaviour at TWI and elsewhere. As with any metallic material, not just a new class of alloy, problems remain. Nevertheless there is no doubt that the last 25 years have seen the generation of a whole new area of welding technology to the present thoroughly mature position in the market place.