Hydrogen - a moving target

TWI Bulletin, January/February 1993

Richard Pargeter joined TWI's Materials Department in 1976 after gaining a BA from the University of Cambridge in Metallurgy and Materials Science. He became a Member of the Institute of Materials in 1980, and achieved Chartered Engineer status shortly afterwards. He was appointed Section Leader of the Ferrous Section within the Materials Department in 1988.

Initially, he worked principally on ferritic steel weld metal toughness with particular regard to submerged-arc welding. His subsequent work has included virtually all aspects of the welding metallurgy of transformable steels. More recently Richard has also been concerned with corrosion of ferritic steels with particular emphasis on stress corrosion and sour environments, and he is responsible for TWI's H ₂S testing facility.

He has carried out a number of investigations into the various forms of weld metal and HAZ cracking, and has become involved in problems of HAZ toughness in ferritic steels.

Hydrogen has a powerful effect on the fracture properties of steels. Richard Pargeter discusses the problems of incorporating this factor in testing and fitness-for-purpose assessments.

Beyond this it is often assumed that toughness will remain constant, despite a number of factors which can progressively alter its value. These factors include strain ageing, temper embrittlement, neutron irradiation damage and hydrogen embrittlement.

Hydrogen is peculiar in that its concentration may increase or decrease during service, and the engineer should ideally attempt to identify the worst case, and assess his material in this condition. There is a further complication in that toughness degradation by hydrogen is dependent on temperature and strain rate ( Fig.1), thus making the task of generating reliable data for inclusion in fitness-for-purpose assessments particularly difficult.

Sources of hydrogen

Hydrogen from welding

During the late 1970s, considerable efforts were being made to achieve high weld metal CTOD toughness. It was felt that hydrogen in weld metals being tested in an as-welded condition, often very soon after welding, was reducing toughness. It was argued that welded joints in, for example, offshore structures would not see service loads until some months after welding, and that the measured toughness was unduly pessimistic. It therefore became common practice to give test welds a hydrogen release heat treatment (typically 150°C for 40 hours) before testing.

The converse of this argument applies to pipeline girth welds, which generally see their most severe loading immediately after welding, either during lowering off onshore, or as the pipeline goes over the stinger offshore. With cellulosic manual metal arc (stovepipe) welds, considerable amounts of hydrogen are introduced into the weld, and CTOD tests performed some time after completing a test weld could give unduly optimistic results.

A programme of work to investigate this effect was carried out for the Department of Energy by TWI in 1983, and as a result it was recommended that, where field welds would see a service load within three days of completion, special precautions should be taken. First, welding of a procedure test weld should be performed within a similar time to fabrication welds, and secondly, fracture mechanics testing should be completed within three days of welding.

The allowance of three days arose because the fracture toughness of the welds tested in this programme only began to improve three to four days after welding, attributed by the author to a 'saturation effect'. Figure 2 shows the effect of hydrogen escape on the micro-mechanism of fracture, which correlated with the toughness results.

Hydrogen from corrosion

A further very clear demonstration of the effect on toughness of hydrogen from this source is given in a paper by Humphries et al. In this work, CTOD specimens made from ASTM A516 grade 70 and A285 grade C steels in various heat treatment conditions were charged with hydrogen in corrosive media before testing. It can be seen from Fig.3 that the effect of hydrogen is very dramatic. This also fits well with the observation referred to earlier, that the fracture toughness of cellulosic welds does not improve noticeably for about three days. Presumably, as suggested by the authors, above a certain level of hydrogen no further damage is done, and it takes around three days for the content to fall to the level at which toughness starts to improve.

Hot, high pressure hydrogen

Hydrogen is absorbed by steel at high pressures and temperatures. These conditions are found in many pieces of refinery equipment such as hydrocrackers and desulphurisers. As discussed below, the embrittling effect of hydrogen is much less severe at elevated temperature, but concern exists for such vessels when they are taken out of service, and need hydraulic pressure testing before being returned to service. This, naturally, is performed at near ambient temperatures.

Effects on toughness

Temperature and strain rate

The problem with assessing effects of hydrogen on fracture toughness of steel is that one is chasing a moving target. Not only is hydrogen very mobile, which makes it difficult to catch the steel in the right condition (or at least to know what condition you have caught it in) but also, its effect is dependent on both temperature and strain rate.

Both these effects were investigated in the late 1960s by Graville et al, and a summary of the results is given in Fig.1. There is a minimum in toughness, typically at around room temperature, although this is to some extent dependent on strain rate. Over the range of strain rates used, toughness decreases steadily as strain rate decreases. At higher temperatures the effect of strain rate is less pronounced, and 'bottoms out' - there being no difference in embrittlement for 0.5 or 0.05 mm/min at 100°C for example. (In the work by Humphries et al referred to above, a strain rate at ambient temperature was determined beyond which no further embrittling effect was discernible).

To be conservative, testing of hydrogen embrittled specimens is generally carried out at ambient temperature, and at as slow a strain rate as is practically possible. A compromise on strain rate is sometimes necessary, however, as if the test takes too long, sufficient hydrogen can escape from the specimen during loading to affect the results. Specimens are sometimes zinc or cadmium plated to counteract this effect, but this gives other practical problems and uncertainties.

Time dependent effects

The mobility of hydrogen at ambient temperature means that embrittlement effects change with time. As indicated above, these changes are generally caused by hydrogen escape to the atmosphere, or hydrogen absorption from the service environment. It should also be recognised, however, that hydrogen will diffuse up a stress gradient. This is perhaps more relevant to hydrogen cracking (fabrication cracking or hydrogen assisted stress corrosion), but the dividing line between cracking and local embrittlement is often unclear.

Material factors

Hydrogen cracking is generally observed only in relatively high strength materials, and there is a well recognised relationship between the risk of cracking and material strength or hardness. It is a mistake, however, to assume that soft materials are not susceptible to embrittlement. The data in Fig.1 are for steels with hardnesses in the range 120-180HV.

In conclusion

It is evident that all ferritic steels are susceptible to embrittlement by hydrogen, and that hydrogen content is as important a variable as temperature for toughness evaluation. As shown by recent practical experience, failure to take account of effects of hydrogen from welding or in operating plant can be catastrophic, and very expensive. The problems which face the engineer are determination of likely maximum hydrogen level in the steel during service, and reliable control of hydrogen level in a test specimen. The engineer then needs to ensure that strain rate is sufficiently low during testing to reveal the full embrittling effect. If effects of hydrogen are not taken into account, unsafe fitness-for-purpose evaluations are likely to be made.

An alternative approach to taking effects of hydrogen into account, which is of particular use for existing plant where hydrogen charged data do not exist, is being followed in a current TWI Group Sponsored Project. A database is being generated from which it will be possible to estimate the likely extent of embrittlement in a range of steels, HAZs and weld metals. Further information is available on this from the Project Leader Mike Gittos, or the author.