Richard Pargeter and Dave Godfrey
TWI Ltd, Granta Park Great Abington, Cambridge, CB21 6AL, UK
Paper presented at IIW International Conference 2012
1. Introduction
When welding, for repair or modification, is being considered on equipment that has been in, and may be continuing in, “hydrogen service” (service which introduces hydrogen into steel), there are two issues particular to the hydrogen service which need to be taken into account. These are the effects that hydrogen in the steel may have on the welding process, and weld quality requirements for continued “hydrogen service”. Hydrogen absorbed by the steel during prior service will contribute to hydrogen in the weld, and can have a measurable effect on the risk of fabrication hydrogen cracking. Some account therefore needs to be taken of this hydrogen in terms of increased precautions (such as higher preheat), or the hydrogen must be removed prior to welding. During subsequent service, the primary concerns are weld area hardness, and residual stress.
2. Effects of Hydrogen in the Steel on Welding
Even under severe wet, acid, sour conditions, a fully immersed steel sample will typically absorb between about 1 and 10ml/100g hydrogen (1). Thus, since safe welding procedures can be derived in many cases when using consumables of this or higher hydrogen contents, (2) it is not immediately apparent that hydrogen in the steel should have a significant effect on risk of cracking. This is, however, a misleading comparison as the consumable hydrogen content is measured on a standard single run sample, which is quenched immediately after the arc is extinguished to retain as much hydrogen as possible, and furthermore is expressed in terms of the deposited metal weight, not the fused metal weight (3). Thus, the hydrogen concentration in a real weld when it has cooled to a cracking temperature is significantly lower than the quoted hydrogen level for the consumable. One of the first significant findings of work carried out at TWI to determine the effects of hydrogen in a steel at the time of welding on the risk of fabrication hydrogen cracking, was confirmation that hydrogen in the steel can make a significant contribution to cracking risk (4). For example, it was found that for a high CEIIW steel (0.43%) with a high sulphur content (0.029%), 8.5ml/100g diffusible hydrogen in the steel led to a need for an increase in preheat of 40°C to prevent cracking, even when welding with consumables which gave 9.7ml/100g on a standard ISO 3690 test. In a 2¼Cr 1Mo steel, the effect was stronger, with 0.7ml/100g diffusible hydrogen in the steel leading to the need for an increase of 130°C in preheat to prevent cracking. In that case, the consumable hydrogen level was 4.3ml/100g. However a clean C-Mn steel, of the same CEIIW (0.43%) as the above example, absorbed much less hydrogen (1.0ml/100g diffusible hydrogen), and no increase in preheat was required to avoid cracking in the hydrogen charged condition, when welded under the same heat input conditions with the same consumables. Thus, it must be recognised that hydrogen service can charge the steel with hydrogen, and that this can affect the risk of fabrication cracking. The two possible solutions are to remove the hydrogen (bake-out) or to adjust the welding conditions (eg increase preheat). The challenge with both approaches is to know the level of hydrogen charging, and the parameters required to reduce that to a safe level and/or to compensate for it.
3. Control of cracking risk through Hydrogen bake-out
Although the usual response to hydrogen pick-up during service is to perform a bake-out or hydrogen release heat treatment prior to any welding, the required heat treatment conditions are not codified, and companies tend to have semi-empirical temperatures and times in their internal specifications. These often do not take proper account of the effects of thickness and temperature, and take no account of the initial hydrogen level or the range of diffusion coefficients in different steels. An indication of the effect of thickness is given in Figure 1, for the maximum and minimum of the scatter band of coefficient of diffusion at 300°C for ferritic steels given in Bailey et.al (5). Since welding is taking place on the surface of the steel, and a gradient in hydrogen concentration, with much lower concentrations near the surface, develops during bake-out heat treatment, the effect of thickness on times required to reduce the real risk of hydrogen effects during welding may be exaggerated, but cannot be ignored. Uncertainties due to differences in coefficient of diffusion between steels are however, accurately represented by this figure. Another factor which can make calculation of hydrogen bake-out conditions difficult is hydrogen trapping. Any hydrogen which is contained in deep traps, and in particular in molecular form in voids and blisters, has to escape from those traps before it can diffuse out of the steel. This is not taken into account in any diffusion based time/temperature calculations. Furthermore, any hydrogen which is not released by a relatively low temperature bake-out, may be released by the heat of welding and contribute to cracking risk. Overall, if hydrogen bake-out treatments are to be used, calculations based on remaining hydrogen at mid wall should give a reasonably conservative times, particularly for thick materials. If there is any indication of hydrogen blistering, hydrogen pressure induced cracking (variously known as HIC, HPIC, SWC) or stress oriented hydrogen induced cracking (SOHIC), then extended treatment and/or other precautions should be taken.
One way of managing bake-out treatments, which does not depend on a knowledge of materials diffusion properties or the level of hydrogen charging in service is to monitor the rate of effusion from the surface. Brown et al have published their experiences (6) which showed how the times required varied at different locations on a single vessel. In that paper, the bake-out was deemed complete when evolution (at 320°C) had reached <500pl.cm-2s-1, and in product literature (7), a value of 1000pl.cm-2s-1 is advised as a suitable end point for a bake-out. This is probably a conservative value, in that it essentially represents complete removal of hydrogen, whereas safe welding procedures could be devised and qualified for some intermediate hydrogen level, as discussed further below. A direct use of hydrogen effusion measurements during welding has been reported by Nicholas and Pargeter. (8). With any bake-out heat treatment which involves heating a local area, it should be remembered that hydrogen can feed into the area from material outside it. Thus, it is advisable to progress directly from bake-out to preheat to welding to PWHT (if required).
4. Control of Cracking Risk Through Preheat
In the work reported above (4), it was demonstrated that the additional risk of cracking resulting from quite severe hydrogen charging could be overcome by increasing preheat. A summary of the results is presented in Figure 2. It was advised that a preheat of 50°C, or 50°C in excess of that recommended by EN 1011-2, should provide adequate compensation for hydrogen picked up in sour service, provided the steel has CEIIW of ≤0.45%, and that welding is performed at ~0.8kJ/mm heat input with ≤ scale C consumables (ie hydrogen of ≤10ml/100g deposited metal). It is also advisable to increase precautions by using lower hydrogen consumables and higher heat input than assumed in designing the procedure, if possible.
For the 2¼Cr 1Mo steel used in that work, the results are summarised in Figure 3. In that case the magnitude of the effect of only a small amount of hydrogen in the steel was found to be such that control through increased preheat is unlikely to be practical. It also means that bake-out needs to be very thorough, and probably supported by post heating, to encourage further hydrogen escape after completion of welding. As with C-Mn steels, any other hydrogen cracking precautions, and in particular good consumable hydrogen control are to be recommended.
5. Effects of Hydrogen Service on Welds
The effects of hydrogen on a weld made for repair or modification are the same as for an original fabrication weld. Thus, for corrosive (for example, sour) service, it is particularly important to control hardness to an appropriate level, to prevent sulphide stress cracking (SSC). For severe wet sour service, at around normal ambient temperature, limiting hardness to about 250HV has been found to give good resistance to SSC. This is the limit set in ISO15156/NACE MR0175 (9) for upstream equipment. For downstream operations however, NACE SP0472-2008 (10) is applicable. This has a limit of 200HB (equivalent to 210.5 HV) for weld metal, and 248HV10 for HAZs.
Weld metal hardness is required to be tested on the actual production weld, although it is recognised that for repairs there may not be sufficient accessible material. It is also recognised that hardness impressions may in themselves cause a problem. In the work which provided the basis for the 200HB limit, ( 11) cracking was found to develop from Brinell hardness impressions, and removal of hardness impressions by grinding was advised. It has also been advised that smaller impressions than Brinell would pose less risk. Hardness testing is discussed in more detail below.
There is a further sour service cracking mechanism, namely SOHIC, which is not hardness dependent. This occurs in susceptible steels under particularly severely hydrogen charging conditions and under particularly high stresses. A review of the problem has been published by Pargeter (12), which gives some indication of materials and circumstances involving a risk of this type of cracking. With regard to repairs or modifications, these are typically very highly restrained, and this can result in a higher susceptibility to SOHIC than for otherwise closely similar welds in the same vessel. The benefits of stress relief heat treatment for controlling SOHIC are not universally accepted, (12) but in the author’s opinion, if restraint is high, and there is a risk of SOHIC, PWHT is a wise precaution.
Other, high temperature, degradation mechanisms such as high temperature hydrogen attack, are not significantly affected by welding, and are of lesser concern, although high residual stresses should be avoided as far as possible.
6. Hardness control and monitoring
Guidance on HAZ hardness control is given in NACE SP0472-2008 (10). This is primarily intended for original fabrication, but is equally applicable to repair or modification. Original weld procedures (which are required to be appropriately qualified with hardness surveys by the standard) will not necessarily be useable, however, if the repair situation results in restrictions on preheat, heat input, or post weld heat treatment. If weld procedural trials show that achieving the required hardness will be difficult, it may be advisable to confirm hardness levels by testing after welding. It is possible to assess HAZ hardness on site, using the ultrasonic impedance (UCI) technique. Comparator methods and Leeb (rebound) methods are not capable of determining the hardness of small regions such as HAZs. Even with UCI, very good surface preparation is necessary, and sufficient measurements need to be made to account for the greater scatter than is encountered with laboratory testing.
7. Guidance on weld repair
This section is intended to provide general guidance for weld repair of carbon manganese and low alloy equipment, primarily pressure vessels and piping, that is in hydrogen service. Such equipment is mainly to be found in use in refineries, sour gas processing and chemical plants. Due to the wide variety of equipment, materials and process conditions that may be encountered, this paper does not attempt to give definitive requirements, as these should be addressed on an individual basis. The intention is more to highlight the type of problems that may be encountered during repair of hydrogen service equipment and offer methods of addressing those problems.
7.1 Factors to be considered
As discussed above, when hydrogen is contained within a parent material it has a potential to cause cracking during or soon after a welding operation and this risk cannot be assessed by equating hydrogen concentration in the steel to welding consumable hydrogen levels. Therefore, the long established guidelines for avoiding hydrogen cracking during original fabrication may not be effective when repairing in service plant and the following points should be addressed:
- Condition of shutdown. Wherever possible equipment in hot hydrogen service should be taken out of service in a planned manner, which should involve slow cooling to ambient temperature. This will give time for some hydrogen to diffuse out of the vessel material. There may be times where equipment is shut down quickly, in which case more hydrogen will be trapped in the vessel material.
- Material type and composition. In addition to whether the vessel is of C-Mn or CrMo steel, it is important to determine the sulphur content. As indicated above, higher sulphur steels generally absorb more hydrogen than clean steels, and a reliable value of CEIIW for C-Mn steels is also required.
- Determine whether a pre-weld hydrogen bake out is required. For thinner sections (eg C-Mn steels of t ≤ 25mm and CEIIW < 0.45) this may not be necessary and an increased preheat may be satisfactory. For steels of higher CEIIW and for CrMo steels this stage is seen as essential.
- Determine pre-weld bake out conditions. Diffusion calculations can be performed, following guidance in Bailey et al ( 5) (see Figure 1), or direct measurement of hydrogen effusion can be used, according to references 6 and 7.
- Determine minimum preheat temperature. For C-Mn steels the use of EN 1011-2 Annex C is suggested, with an increase of 50ºC over the calculated value to be applied if omission of hydrogen release is deemed to be acceptable.
- Develop a suitable welding procedure.
- Plan the timing, sequence of operations and availability of personnel, equipment, access to repair area, and health and safety issues. It is probable that any bake out will only be applied to a local area; it is, therefore, advisable to complete all operations in a continuous cycle if at all possible to prevent ingress of hydrogen into the repair area from surrounding material.
7.2 Preparation for Repair
In general, it is helpful to carry out cutting and grinding operations before any hydrogen release heat treatment, as diffusion distances are then reduced. This is particularly relevant to deep excavations or cut-outs in thick plate.
Depending on the concentration of hydrogen within the steel and the steel composition, arc-air gouging prior to hydrogen release may lead to further cracking as hydrogen from the surrounding steel diffuses into the hardened HAZ of the gouge. Thus, this technique should be avoided if at all possible, and at least restricted to clean steels of low CEIIW (<0.43). In any case, where arc-air gouging is used, at least the final 5mm of material should be removed by grinding. For excavations of ≤ 15mm in depth it is advisable to use grinding for all of the excavation.
For excavation of grooves in CrMo material, it is suggested that all removal is by grinding in order to avoid creation of hard zones in hydrogen charged material. Following MPI, the surface of the groove should then be covered with two temper bead layers, with the body of the groove then being filled using a stringer bead technique.
Occasionally, areas containing multiple cracks may be found, and in this situation it may be necessary to cut out and replace the damaged area with new plate. In this case, the area round the cut line should be preheated prior to cutting then the cut surface immediately ground back by 3mm to remove hard zones in hydrogen charged material before they cool to a temperature at which hydrogen cracking may occur. A hydrogen bake out should then be carried out, as required. The prepared face can then be covered with two layers, following a temper bead approach, prior to fitting a new insert plate. Assuming that the vessel will be subjected to PWHT and the HAZ hardness kept below 248HV, it will not be necessary to temper bead the prepared face of the insert plate, and only the usual controls to avoid fabrication hydrogen cracking need be applied, see Figure 4.
7.3 Type of repair
Considering the defects and damage mechanisms that may be encountered, the main types are cracking, erosion and corrosion. Some of these defects may arise as a direct result of the equipment being in hydrogen service, eg SSC and corrosion of the weld and HAZ. Erosion may occur due to process flow conditions not directly linked to hydrogen service, but hydrogen may need to be taken into account when planning a weld repair. The details of repair will be largely dictated by the extent of the defect. For surface damage eg erosion/corrosion, it may be sufficient to apply two temper bead layers, whereas cracking will normally require a more extensive repair involving excavation of a groove to remove the crack.
Additionally, there are occasions where no damage has occurred to the vessel but there is a requirement to modify or alter the vessel. Although the vessel may be in good condition with no cracking or corrosion damage present, any modification involving welding should be approached in the same way as a defect repair.
7.4 Hardness control
In addition to the ‘standard’ requirements to meet parent material tensile and toughness properties, a welding procedure for repair of equipment which has been in, and is to continue in, hydrogen service must also aim to achieve sufficiently low hardness to avoid both fabrication (repair welding) cracking and subsequent in-service cracking. For sour service, the requirements of ISO 15156/NACE MR0175 (9) (for upstream equipment) or NACE SP0472-2008 (10) (for downstream equipment) should be observed. It is useful to bear in mind that for procedures designed to comply with these two standards, and which are subject to PWHT, the hardness is measured after this has been carried out. For new fabrications this is acceptable, as the reasons for applying the hardness limitations (hydrogen from sour corrosion) only come into play when the vessel enters service; however, for in-service repairs the hydrogen is already present, so a different approach is advisable, if a full hydrogen release bake out has not been applied.
7.5 The Temper Bead Technique
The temper bead technique has long been used as a method of weld repairing PWHT vessels while avoiding PWHT of the repair. Applied correctly and assuming the use of SMAW/MMA process, the repair surface is covered by a layer of overlapping weld beads, laid so that each bead overlaps the previous one by 50%. A second layer is then deposited using the same technique, but at no point does layer two impinge on the parent metal. Layer one should be deposited with 3.2mm electrodes, layer two with 4.0mm, see Figure 5. This has the effect of tempering the HAZ in a way that replicates a conventional PWHT, and should, for a C-Mn steel, result in a HAZ microstructure and hardness that is resistant to cracking without any PWHT. This technique is presented as a means of hardness control (without the need for PWHT) in NACE SP0472-2008 (10), which does make specific reference to repair of ‘minor defects in cast, forged and plate components’. Furthermore, NBIC chapter III supplement 3 (13) allows use of temper bead repair, in place of PWHT after repair, without limit on section thickness. Nevertheless, although the temper bead technique is successful in reducing HAZ hardness and refining microstructure, it will not remove the residual stress arising from the repair weld. Furthermore, considerable care and QA are needed to ensure that no small hard regions remain, and just one could be sufficient to initiate cracking in service. Thus, in the authors’ opinion, if the temper bead approach is necessary, even if it is adequate to achieve safe hardness levels, PWHT remains advisable for equipment returning to hydrogen service.
Welding consumables should be in the lowest hydrogen condition possible and this is most reliably achieved for SMAW/MMA welding by use of vacuum packed electrodes. Most manufacturers guarantee less than 5ml/100g diffusible hydrogen for either eight or twelve hours from point of opening the vacuum seal. Note that if working in conditions of high ambient temperature and relative humidity, the safe exposure time may be shorter than claimed. In this case it may be prudent to assume > 5 ≤ 10ml H2/100g deposited metal (Scale C, EN 1011-2) and to calculate the preheat accordingly.
While carrying out qualification of the welding procedure it is advisable to carry out hardness surveys of the weldment both before and after PWHT, especially if it is not possible to go straight into PWHT on completion of welding. This step will confirm the effectiveness of the temper bead technique and will indicate whether it is safe to allow the equipment to cool to ambient temperature prior to PWHT, bearing in mind that if only local hydrogen bake out has been performed, hydrogen can diffuse into a hardened HAZ from surrounding areas.
7.6 Post-Weld Heat treatment
NBIC, “Alternative methods applicable to pressure vessels”, requires that the temperature of the weldment shall be raised to 232-288°C immediately after welding and held for 4 hours. This acts as a hydrogen release treatment and is the final thermal treatment when temper bead is used for repair without PWHT. NACE MR 0175/ISO 15156 and NACE SP0472-2008 (9,10) do not have a mandatory requirement for PWHT but do require that hardness is less than 250/248HV. It is generally necessary to apply PWHT at thicknesses below the maxima allowed for as-welded construction in the vessel codes to achieve this. Furthermore, as mentioned above, if the temper bead approach is necessary to achieve safe hardness levels, PWHT remains advisable for equipment returning to hydrogen service. It is advised that the post weld heating required by NBIC is applied to repair of hydrogen vessels even when PWHT is to be applied, especially where the vessel is of greater than 25mm section thickness/CEIIW>0.43 and where PWHT cannot be applied immediately following welding.
7.7 Welding Procedure Development
Procedure qualification standards such as ASME IX and ISO 15614-1 (14, 15) put parent materials together in groups based on composition and mechanical properties. Generally, a procedure qualified on a given grade of steel is valid (according to these standards) for use on other steels in the same groups and for steels in lower groups. This confers a degree of flexibility so that a fabricator called upon to carry out a repair may possess a qualified procedure which covers the code requirements applicable to the repair, and a new qualification may not be necessary for code compliance.
For repair of hydrogen service vessels this may not always be the best option. It is quite possible that a recently qualified procedure will have been made using steel with a significantly lower CEIIW than the steel to be welded, especially if the vessel has been in service for many years. This is recognised in ASME IX, which specifically addresses temper bead repair and includes a reduction in CEV as an essential variable for qualification of temper bead repairs. No such restriction exists in ISO 15614-1 so a low CEIIW would approve welding of higher CEIIW steels within the same groups, which could lead to unexpectedly high hardness values in the HAZ of the repaired vessel.
To counter this it is desirable to make a new qualification test on steel of at least equal CEIIW to that of the vessel. If this cannot be done, eg due to time constraints or material availability, then the preheat should be calculated on the basis of the highest possible CEIIW of the material grade to be welded. Allowance should also be made for any adjacent attachments such as compensation plates, which may create greater heat sinks and/or additional restraint.
For temper bead welding, using a stringer bead technique, heat inputs for layer one should lie between 1.0 to 1.4kJ/mm, and for layer two should be 1.3 to 1.6kJ/mm, using 3.2 and 4.0mm electrodes for layers one and two respectively. It is important to achieve as closely as possible a 50% overlap of the weld passes in order to achieve the correct thermal cycle in the HAZ. The distance between the edges of the passes one removed from the parent material should be 2-3mm from the toe of the weld to minimise cap HAZ hardness, see Figure 5b.
Once the surface of the preparation has been covered by the temper bead technique, the body of the groove can be filled using conventional build up techniques. The welders should be fully briefed on the requirements prior to carrying out the repair.
When carrying out the hardness survey for procedure qualification, the indent pattern should follow the profile of the fusion boundary at a distance of about 0.5mm on the HAZ side, to ensure that the hardest microstructures are sampled.
A procedure qualified in this way should provide confidence in the soundness and integrity of the actual repair process.
8. Summary
Welding on equipment which has been operating in hydrogen service involves risks associated with cracking directly as a result of the welding, and also risks associated with continued operation of the equipment in hydrogen service. It should be noted that the vessel design codes provide minimum requirements and cannot be expected to address every conceivable situation, and for special situations such as severe hydrogen service, additional measures may be needed.
With regard to direct cracking risk, hydrogen may be removed prior to welding by heat treatment. There is very limited guidance available on appropriate conditions, (time and temperature) however. One approach to overcome this is to use direct measurement of hydrogen effusion to determine when hydrogen release is complete.
An alternative option for some C-Mn steels, is to mitigate the additional fabrication cracking risk by increasing other precautions, and in particular preheat, but this is not feasible for more highly alloyed steels.
Avoidance of cracking during continued operation in hydrogen service requires hardness control, and if very highly restrained welding cannot be avoided, PWHT may be necessary to prevent SOHIC, dependent on the parent material. If hardness control is difficult, it may be advisable to perform site hardness checks, in which case the ultrasonic impedance technique should be used, on a well prepared surface.
If weld procedure qualification is to be relied upon for hardness control, it is necessary for this to have been carried out on steel of at least as high CEIIW as the steel to be repaired.
Figure 6 presents a flow chart which may be of assistance in selecting a repair methodology for repair of hydrogen service vessels.
9. References
- Pargeter R J. “Factors affecting the susceptibility of C-Mn steel welds to cracking in sour environments” ASTM STP 1049 “Environmentally assisted cracking – science and engineering”
- BS EN 1011-2: 2001. “Welding – recommendations for welding of metallic materials – Part 2: Arc welding of ferritic steels” British Standards Institution, 2001.
- EN ISO 3690: 2000 “Welding and allied processes – Determination of hydrogen content in ferritic steel arc weld metal”
- Pargeter, R J and Wright, M D “Welding of hydrogen charged steel for modification or repair” Welding Journal Research Supplement, February 2010 pp 34s – 42s
- Bailey N, Coe, FR, Gooch, T G, Hart, P H M, Jenkins, N and Pargeter, R J. “Welding Steels without hydrogen cracking” 2nd edition (revised) 2004. Woodhead Publishing
- Brown, C N, Carroll, M J, Dean, F W H, Harrison, T J and Kettle, A “Applications of hydrogen flux monitoring to pre-weld bakeouts of steel” Corrosion 2004, paper 4478, National Association of Corrosion Engineers
- Ion Science website, www.ionscience.com
- Nicholas, J and Pargeter, R “Novel control of weld metal hydrogen cracking in the welding of thick steels” “Steely Hydrogen” Ghent, Belgium, 28-29 September 2011. OCAS
- ISO 15156/NACE MR0175 “Petroleum and natural gas industries – Materials for use in H2S containing environments in oil & gas production. Part 2: Cracking resistant carbon and low-alloy steels, and the use of cast irons.” National Association of Corrosion Engineers, 2009.
- NACE SP0472-2008 “Methods and Controls to prevent In-service Environmental Cracking of Carbon Steel Weldments in Corrosive Petroleum Refining Environments” National Association of Corrosion Engineers, 2008.
- Kotecki, D J and Howden, D G “Submerged-arc-weld hardness and cracking in wet sulphide service” WRC Bulletin No 184 June 1973
- Pargeter, R J “Susceptibility to SOHIC for linepipe and pressure vessel steels – review of current knowledge” Corrosion 2007, Nashville Tennessee, 11-17 March, 2007, National Association of Corrosion Engineers.
- National Board Inspection Code, National Board of Boiler and Pressure Vessel Inspectors 2007.
- ASME Boiler and Pressure Vessel Code, Section IX: Qualification Standard for Welding and Brazing Procedures, Welders, Brazers, and Welding and Brazing Operators, American Society of Mechanical Engineers, 2010.
- ISO 15614-1 Specification and qualification of welding procedures for metallic materials-Welding procedure test-Part 1: Arc and gas welding of steels and arc welding of nickel and nickel alloys. British Standards Institute 2008.