The influence of steel cleanliness on HAZ hydrogen cracking - the present position

TWI Bulletin, March 1986

Peter Hart, BSc (Eng), ARSM, MIM, MWeldI, is Deputy Head of the Materials Department. This article is based on IIW Document IX-1308-84.

Information on the effect of steel cleanliness shows a consensus indicating that reduced sulphur levels increase the risk of HAZ hydrogen cracking in C-Mn and low alloy steels. Most reports consider, however, that the benefits to other properties in low sulphur steels outweigh the detrimental effect in respect of HAZ hydrogen cracking. The most commonly cited reason for the increased risk of cracking is the effect of a reduced number of inclusions increasing the HAZ hardenability.

It is now nearly 18 years since the first publications [1] concerning effects of low sulphur content on HAZ (heat affected zone) cracking. In the intervening years further papers have appeared, some reporting no influence of reduced sulphur level. This short review has been carried out to summarise the present position.

The cleanliness of a steel is usually taken to refer only to its non-metallic inclusion content, i.e. oxides and sulphides. These inclusions have well-known detrimental influences with respect to a number of steel properties including resistance to ductile fracture, hydrogen pressure cracking in wet H 2 S containing environments, and lamellar tearing. In response to the need to enhance such properties, steel producers have made substantial improvements and lowered the inclusion content of steels, especially in respect of sulphides.

There have also been associated changes to inclusion type and morphology, again principally in respect of sulphides, first by REM treatment and more recently and widespread by Ca treatment. Thus the main changes in steel cleanliness over the last 18 years, especially for Al-treated steels, have been in the sulphide content. Most published information concerning the influence of steel cleanliness on HAZ hydrogen cracking therefore has referred specifically to the effect of changing the sulphur content, as reflected in this review.

Information available

The first reference to a study of the effect of sulphur content on HAZ hydrogen cracking was that of Hewitt and Murray [2] in 1968 following a report [1] of a fabrication HAZ hydrogen cracking problem attributed to low sulphur levels. Since that time there have been a further 16 published references, [3-18] making a total of eighteen from seven different countries. The main points of all 18 references have been summarised in Table 1 for ease of comparison. Three further references [19-21] relevant to the review but not directly studying HAZ hydrogen cracking are summarised in Table 2.

Table l Summary of data in references to sulphur and HAZ hydrogen cracking

Table l Summary of data in references to sulphur and HAZ hydrogen cracking - continued

Table 2 Summary of data In references to sulphur affecting hardness/hardenablllty/transformation

Results

The summary of data in Table 1 shows that a substantial majority (some 80%) of the references report an increased risk of HAZ hydrogen cracking caused by a reduction in parent steel sulphur level.

Examples of this are shown in Fig.1 and 2. Eleven of the reports were on C-Mn steels, five on low alloy steels. Only one was concerned with a fabrication experience, all remaining reports being of laboratory investigations. Twelve of these used commercially produced heats of steel, and seven used experimental casts (25-1000kg).

Eleven of the laboratory investigations used the implant cracking test and eight a self-restraining cracking test, CTS ^[5] or Y-groove type. ^[3] One investigation used electrolytically charged simulated HAZ three point bend specimens. Two investigations ^[4,9] using the implant test also tested specimens extracted in the plate Z direction so that the plane of the notch and cracking were of the orientation for lamellar tearing. As would be expected from the known influence of sulphur on lamellar tearing, both of these reported slight ^[4] and considerable ^[9] beneficial effect of reduced sulphur on threshold stress for cracking in this orientation.

Those investigations which have considered the causes of the increased risk of cracking have indicated two main factors. First and foremost, that lowering the sulphur content has changed the continuous cooling transformation behaviour, by increasing hardenability and hence hardness (for a given cooling rate) and secondly that varying sulphur influences hydrogen diffusivity in the steel.

In only seven of the reports were one, or both, of these factors examined. In respect of increased hardenability a substantial majority (four out of five) reported an increased hardenability with decreased sulphur content. An example of this is shown in Fig.3. (If only weld HAZ results were considered in ref ^[11] , simulated HAZ results being excluded, then all five references would show an effect). Hardenability was investigated only in respect of C-Mn steels. Of the five references which looked at diffusivity, three found an increase with decreasing sulphur content.

No marked effect of steel source ( i.e. experimental heat or commercial heat) is apparent from Table 1 on the three aspects, risk of cracking, hardenability and diffusivity.

Discussion

The survey has shown a consensus reporting that a decrease in sulphur content increases the risk of HAZ hydrogen cracking. Only five of the references ^{[10,11,15-17]} have considered the broader aspect of steel cleanliness i.e. oxides and sulphides, but all of these have shown that total cleanliness is more important than sulphur content alone. The research that has been carried out strongly suggests that the principal reason for the increased risk of cracking arises directly from the influence of steel cleanliness on austenite decomposition. An example of nucleation on an inclusion in a HAZ is shown in Fig.4. An additional factor may be related to the influence of inclusions on the extent of trapping and on hydrogen diffusivity values. A general review of hydrogen trapping effects has recently been produced ^[22] but does not contain new data specific to the present topic. Hirai et al ^[11] also believe that low sulphur steels, through lowering weld pool sulphur levels, increase weld hydrogen pick up and hence risk of cracking.

Apart from the data in Table 2, which strongly support the reports of sulphur influencing HAZ hardenability, there are many other references to the ability of second phase particles to nucleate the transformation of non-martensitic products from austenite both in the HAZ [23,24] and weld metal. [25-27]

The hardenability aspect has been investigated only in respect of C-Mn steels, and would not be expected to apply to low alloy steels when welded to produce wholly martensitic microstructures. However, for levels of alloying and/or cooling rates which produce some non-martensitic structures in the HAZ, an effect would be expected. It might be expected that lowering the inclusion content of the steels would influence austenite grain growth behaviour in the HAZ, a cleaner steel producing a larger austenite grain size with enhanced hardenability. This aspect only seems to have been reported in ref. [10] where no influence on austenite grain growth of the reduction in sulphur from 0.033 to 0.005% was detected. However, it may be changes in the numbers, particularly of small inclusions, which are important. Changes in the numbers of small inclusions depend as much, if not more, on the particular characteristics of a steel making and casting route than actual level of oxygen and sulphur, so this aspect still requires examination.

An important aspect of the present author's work [10] was that the hardenability effect appeared to be linked to the number of inclusions rather than their total volume fraction. Thus, only when changes in sulphur (and/or oxygen) produce changes in the total number of inclusions, might an effect on hardenability be produced. The importance of this is that, depending on steel making and casting practices from source to source, different sulphur levels may not necessarily produce different total numbers of inclusions. This may be one reason why only 80% of the investigations reported an increased risk of cracking; the total number of inclusions (both oxides and sulphides) may not have decreased in those investigations reporting no effect. Indeed Fig.5 shows the increasingly wide range of behaviour as the sulphur level is decreased. (The range of ΔTp increases from -10 to +25°C, at 0.020%S to -50 to + 75°C at 0.002%S). Thus some steels are showing a significantly lower risk of cracking, and some a significantly higher risk of cracking, at low sulphur levels. This may again be linked to different total numbers of inclusions in the different steels examined.

The extent of the observed effects on the risk of cracking has been summarised in Table 3 for those references reporting results in terms of either an equivalent shift in carbon equivalent

and/or a change in pre-/post heat required. The Table shows that the effects are variable (in part depending on the change in sulphur and inclusion levels studied) but are nevertheless significant in terms of welding procedure.

Table 3 Summary of reported practical effect of reduced sulphur

One of the input parameters for prediction of required welding procedures for avoiding HAZ cracking is a compositional characterising parameter ( e.g. CE or P _cm ) which principally, but not solely, is assessing the hardenability of the steel. The various results in the reports surveyed show that the effect of steel cleanliness cannot be accurately predicted merely by including sulphur level. This is probably because it is the total number of inclusions (oxides and sulphides) which are important. It is therefore not currently possible to determine from a steel analysis, e.g. a product analysis as supplied by a steelmaker to a purchaser, the contribution to hardenability (and therefore to CE or P _cm ) of the inclusions. One of the causes of variation in extent of an effect in Table 3 may be because of the inclusion differences between small laboratory experimental heats and large commercial heats, although the inclusion sizes reported by the author ^[10] (experimental heat) and Hirai et al ^[11] (commercial heat) were, in those instances, similar. The values in Table 3 should therefore only be taken as an approximate guide to possible changes required in determining safe welding procedures. Furthermore, sulphur levels of some low sulphur present day steels are already well below most of the lower limits studied in the reports surveyed and the continued trend for cleaner steels may mean the data in Table 3 are not conservative.

That only one of the reports of increased risk of cracking with low sulphur refers to a practical fabrication should not be taken to mean necessarily that there is limited practical experience of such an effect. In practical fabrications, it is rare that the only difference between two situations is the steel sulphur level; usually many factors are involved, preventing a definitive comparison. Moreover, if the welding procedure used in situation 'A' with 'high' sulphur had a considerable safety margin, when applied to a comparable situation 'B' on low sulphur steel, cracking may still be avoided, albeit with a reduced safety margin.

All the reports which showed a detrimental effect of lowering sulphur level, with the exception of the first references to the problem, agree that the benefits in respect of other properties which arise from cleaner steels outweigh the disadvantages produced in respect of hardenability and risk of HAZ hydrogen cracking. In some low alloy steels, particularly the ~0.4Mn, 2.5Ni, 0.3Mo type, decreasing sulphur by reducing the risk of liquation cracking may, as a result, beneficially influence the risk of HAZ hydrogen cracking.

Although the main purpose of this review was to examine the problem of HAZ cracking, it has shown a significant degree of support for an additional effect of practical importance in clean steels. This is the increased hardenability which, while any contribution to increased cracking risk can be overcome by procedural changes, has significance in respect of an increasing trend for specification of maximum HAZ hardness requirements. However, even this aspect may be partly counterbalanced for C-Mn steels by the trend to low CE levels and by use of Ti treated steels which can reduce HAZ hardenability. ^[28]

Summary

A review of published information on the effect of steel cleanliness, particularly sulphur level, on the risk of HAZ hydrogen cracking has shown:

1. Some 80% of reports found that reducing sulphur level increased the risk of HAZ hydrogen cracking in both C-Mn and low alloy steels. However, there was general agreement amongst the reports showing an effect of sulphur, that other benefits of low sulphur clean steels outweighed the disadvantage in respect of HAZ hydrogen cracking.
2. Two main factors were considered to be causing the increased risk; an increased hardenability and an increased hydrogen diffusivity. Eighty per cent of the investigations examining hardenability found it to increase with reduced sulphur level while only 60% of those examining diffusivity found an increase.
3. An accurate prediction of the extent of the increased risk for any given steel is not possible without specific knowledge of its inclusion content. An approximate guide is that a decrease in sulphur from ~0.025% to ~0.005% may be equivalent to an increase in CE (see footnote below) of ~0-03, or a need to increase preheat by ~50- 100°C.
4. The effect of cleanliness on HAZ hardenability may be important for fabrications which have maximum HAZ hardness requirements even when the risk of hydrogen cracking is low.
5. There is a need for further work in relation to two aspects. The first is to improve methods for predicting the effect of steel cleanliness on HAZ hardenability and risk of cracking. This will require closer determination of the factors involved to see if the effect can be adequately described by reference to oxygen and sulphur levels. The second aspect is to study the reason for the apparently wide range of behaviour reported in some cases at low sulphur level, in the hope that this may lead to the development of steels with enhanced resistance to HAZ hydrogen cracking.

References