Welding Of Hydrogen Charged Steel For Modification Or Repair

By R J Pargeter and M D Wright

The authors were awarded the A F Davis Silver Medal for this paper by the AWS in 2011.

Paper published in Welding Journal Vol.89 No.2, February 2010 pp 34s - 42s.

Abstract

Various types of steel equipment, particularly in refinery service, absorb hydrogen during operation. Materials selection and design should ensure that this does not cause any damage, but the presence of hydrogen in the steel also needs to be taken into account if modifications or repairs involving welding are required. Hydrogen in the steel will contribute to hydrogen in the weld, and may increase the risk of fabrication hydrogen cracking, or 'cold cracking'. The usual approach is to impose a hydrogen removal heat treatment, or 'hydrogen bakeout' prior to welding, to ensure that there is no significant hydrogen left in the steel, in which case, normal welding precautions can then be taken to avoid cracking. This, however, is a costly and time consuming process, with the time taken making a direct and particularly significant contribution to the cost if it results in extended downtime of a refinery. The effects of hydrogen in the base steel have been explored in an experimental programme of work, and recommendations for safe welding procedures have been made.

Introduction

When faced with a requirement to weld on hydrogen charged steel, the welding engineer will typically perform some sort of calculation to determine the time and temperature required to remove the hydrogen. For example, the hydrogen removal curves presented in Ref ^[1] may be used. Any such calculation will require a knowledge of the coefficient of diffusion in the steel of concern, which is generally not known with any confidence, and some judgement will have to be made with regard to the amount of hydrogen which needs to be removed. One solution to these uncertainties is to employ direct measurement of hydrogen effusion, and successful use of such an approach has been reported.^[2] Nevertheless, both down time and uncertainty would be removed if the effects of hydrogen in the steel could be accommodated in the welding procedure. Welding conditions which avoid hydrogen cracking can be devised for high hydrogen consumables, and a similar approach should be possible for a hydrogen charged steel.

A further advantage of devising welding procedures which accommodate hydrogen in the steel is that trapped hydrogen which may not be driven out by relatively low temperature bakeout treatments, or therefore registered by direct hydrogen flux measurements, would be taken into account. There is some risk associated with assuming that a hydrogen bakeout treatment has been completely effective, by calculation or measurement.

Bearing the above considerations in mind, a programme of experimental work was carried out at TWI. The aim was to demonstrate and quantify the effect of hydrogen in the steel, such that appropriate modifications to welding procedures could be recommended.

Experimental programme

Approach

The issue of hydrogen charging of steels in service arises both in low temperature corrosive conditions, and when handling hot, high pressure hydrogen. For corrosive conditions the primary concern is with carbon manganese (C-Mn) steels, particularly when operating in sour (H₂S containing) environments. For hot, high pressure hydrogen, chromium molybdenum (Cr-Mo) alloyed steels, resistant to hydrogen attack, are used. In view of both the significant differences in the materials and the hydrogen charging routes, the work included tests on corrosively charged C-Mn steel, and Cr-Mo steel charged in a hydrogen autoclave.

The controlled thermal severity (CTS) test^[3] was selected as a weldability test to provide a comparison between hydrogen charged and hydrogen free steels. Data generated using this test at TWI formed the basis of the guidelines currently in BS EN 1011-2, Appendix C^[4], and in the TWI book on welding steels without hydrogen cracking.^[1] Thus it was considered to be an appropriate test, which would allow the data to be fed directly into guidelines for avoidance of cracking. It was also possible to carry out hydrogen charging and analysis with only slight modification of the test method. To facilitate hydrogen analysis, extension pieces were included on all test blocks, which were removed for analysis immediately prior to test welding.

Bearing in mind the common constraints on heat input in a repair weld situation, 'crack : no crack' boundaries were determined in terms of preheat at a single heat input typical of repair practice.

Materials

The test steels consisted of three C-Mn steels and one Cr-Mo steel. Chemical compositions are presented in Table 1, and micrographs are presented in Fig.1.

Table 1. Chemical compositions of parent materials (wt%)

	C	S	P	Si	Mn	Ni	Cr	Mo	V	Cu	Nb	Ti	Al	O	N	CE*
High CE, high S C-Mn steel, A	0.21	0.029	0.018	0.22	1.12	0.08	0.07	0.02	<0.002	0.12	<0.002	<0.002	<0.003	121	62	0.43
High CE, low S C-Mn steel, B	0.19	<0.002	0.021	0.28	1.38	0.01	0.02	<0.005	<0.002	0.005	0.024	<0.002	0.047	4	41	0.43
Low CE, high S C-Mn steel, C	0.14	0.033	0.035	0.20	1.25	0.04	0.03	0.005	<0.002	0.02	<0.002	<0.002	0.004	100	38	0.36
2¼ Cr 1Mo steel, D	0.14	0.002	0.004	0.20	0.43	0.10	2.17	0.96	0.002	0.02	<0.002	<0.002	0.024	24	34	---

* CE_IIW = C + Mn/6 + (Cr + Mo + V)/5 + (Ni + Cu)/15

Fig.1 Micrographs of test steels. All etched in 2% nital. Magnifications given by micron marks

Three C-Mn steels were used to investigate the effects of two material variables. The steels selected cover high and low carbon equivalents (CE), the parameter that best describes this class of material’s relative susceptibility to fabrication hydrogen cracking. The selected steels also cover both high and low sulphur contents. The higher sulphur level was selected to allow the effect of hydrogen trapping around inclusions (which might slow diffusion of hydrogen in the steel) to be explored.

A high CE, high sulphur C-Mn steel (A) was chosen as representative of older 'dirty' (high sulphur and oxygen) steels, likely to be encountered in repair welding operations. It conformed to the old British Standard BS 1501-221 grade 32A, and had a CE of 0.43% and a sulphur content of 0.032%.

A steel with the same CE (steel B), but with a very low sulphur content of <0.002%, provided a comparison in terms of steel cleanliness. Furthermore, this was the only Al-treated steel of the three C-Mn steels tested, and had a low oxygen content of 4ppm. This steel complied with BS 4360: 1990 grade 50D. The grain size was finer than that of steel A, but was not unusual for a steel of this grade.

Steel C was another high sulphur (0.033%) steel to BS 1501-121, but with a lower CE of 0.36%. The grain size was relatively coarse, and closer to that of steel A than steel B.

The Cr-Mo steel (B) was a 2.25Cr 1Mo type, with a tempered bainitic microstructure, and low sulphur and oxygen contents.

The manual metal arc/shielded metal arc process was selected, and part dried welding consumables were procured which could be conditioned to the desired hydrogen level. For C-Mn steels, AWS E7018 consumables were used. These were dried at 330°C for one hour, to give 9.7ml/100g hydrogen, that is towards the upper end of scale C in BS EN 10111-2. For the Cr-Mo steel, matching AWS E9018 consumables were used, dried at 250°C for one hour to give 4.3ml/100g hydrogen, that is towards the upper end of scale D.

Hydrogen charging of C-Mn steels

Hydrogen charging of the C-Mn steels was achieved by placing the modified CTS top blocks (see below) in acidified (pH3) 5%NaCl solution saturated with H₂S (standard NACE TM 01 77 solution A ^[5]), with an exposure time of 96 hours. After charging, the blocks were removed from the solution, lightly cleaned then packed in dry ice and stored in a freezer until required for testing. Results of diffusible and total hydrogen determinations on the tabs from charged CTS top blocks are included in Tables 2 to 4. Diffusible hydrogen analysis was performed by evolution and collection over mercury at ambient temperature, and residual hydrogen (to give total hydrogen by addition) by vacuum hot extraction at 650°C.

Table 2. Test results, C-Mn steel, A.

Weld No.	Heat input* kJ/mm	Preheat °C	HAZ Hardness HV10 max - min ------------- mean	No of faces showing cracking	Hydrogen, ml/100g in top block
					Diffusible	Total
W2	0.83	20	425 - 390 ----------- 406	3/6	Not charged
W3	0.83	100	376 - 354 ----------- 365	0/6	Not charged
W4	0.86	50	429 - 360 ----------- 400	0/6	Not charged
W6	0.82	38	429 - 401 ----------- 415	0/6	Not charged
W8	0.88	85	383 - 357 ----------- 377	0/6	6.3	6.4
W9	0.88	35	421 - 401 ----------- 409	0/6	Not charged
W11	0.78	60	394 - 380 ----------- 386	1/6 inclusion cracking 3/6 conventional	5.9	6.1
W12	0.81	35	394 - 376 ----------- 385	5/6 inclusion cracking 2/6 conventional	8.5	8.7
W13	0.82	77	401 - 376 ----------- 391	6/6 inclusion cracking 0/6 conventional	7.4	7.6
W15	0.88	100 top 95 base	376 - 357 ----------- 365	3/6 inclusion cracking 0/6 conventional	9.2	9.9
W17	0.72	120	366 - 342 ----------- 349	5/6 inclusion cracking 0/6 conventional	12.0	12.1
Not welded	---	As charged	---	---	14.0	14.2
Not welded	---	150	---	---	9.5	9.7

* using arc efficiency of 0.8

Table 3. Test results, C-Mn steel, B.

Weld No.	Heat input* kJ/mm	Preheat °C	HAZ Hardness HV10 max - min ------------- mean	No of faces showing cracking	Hydrogen, ml/100g in top block
					Diffusible	Total
W27	0.83	45	417 - 405 ----------- 412	6/6 conventional	Not charged
W28	0.88	110	421 - 405 ----------- 414	4/6 conventional	Not charged
W29	0.82	125	413 - 345 ----------- 376	0/6	Not charged
W31	0.83	125	429 - 413 ----------- 419	0/6	Not charged
W32	0.83	160 top 150 bottom	405 - 366 ----------- 390	0/6	Not analysed
W34	0.82	135	405 - 387 ----------- 393	0/6	0.8	1.0
W35	0.85**	135	405 - 373 ----------- 395	0/6	1.0	1.2
W37	0.88	125	376 - 357 ----------- 370	0/6	0.7	0.8
W40	0.81	110-115	417 - 394 ----------- 403	1/1 conventional	0.9	1.0

* using arc efficiency of 0.8
** estimated using V = 23 because true voltage reading unavailable

Table 4. Test results, C-Mn steel, C.

Weld No.	Heat input* kJ/mm	Preheat °C	HAZ Hardness HV10 max - min ------------- mean	No of faces showing cracking	Hydrogen, ml/100g in top block
					Diffusible	Total
W20	0.78	20	330 - 314 ----------- 322	0/6	Not charged
W21	0.78	20	330 - 306 ----------- 321	0/6	Not charged
W22	0.85	20	360 - 304 ----------- 342	2/6 conventional 2/6 inclusion cracking	6.8	6.9
W23	0.79	45	342 - 327 ----------- 332	3/6 inclusion cracking 1/6 conventional HAZ	10.0	10.2
W24	0.86	70	332 - 281 ----------- 303	1/6 inclusion crack into HAZ from parent plate	11.3	11.5
W25	0.81	70	314 - 302 ----------- 309	2/6 inclusion cracks	11.4	11.5
W26	0.90	95	289 - 283 ----------- 285	2/6 inclusion cracks	7.3	7.4

* using arc efficiency of 0.8

Hydrogen charging of Cr-Mo steel

Hydrogen charging of the Cr-Mo steel was carried out using an autoclave. In this case CTS top blocks were exposed to a hydrogen atmosphere at a temperature of 450°C and pressure of 10.3MPa (1500psi) for 48 hours. This temperature was chosen as it is approximately that at which many vessels operate and the pressure places this condition just within the 'safe' region of a Nelson curve plot for this material. The exposure time of 48 hours should be sufficient to saturate the material under these conditions. Previous published work^[6] indicates that a hydrogen content of approximately 4ppm should be obtained. The same source quotes a calculated hydrogen content of 4.8ppm for an α iron under similar conditions.

After charging, the Cr-Mo blocks were stored in the same way as the C-Mn blocks. However, in this case some loss of hydrogen is expected between charging and storage because the blocks must cool to 250°C before pressure can be released and the blocks extracted and quenched. This can take up to one hour.

Results of diffusible and total hydrogen determinations on the tabs from charged CTS top blocks are included in Table 5. Diffusible hydrogen analysis was performed by evolution and collection over mercury at ambient temperature, and residual hydrogen (to give total hydrogen by addition) by vacuum hot extraction at 650°C.

Table 5. Test results, 2¼Cr - 1Mo steel, D.

Weld No.	Heat input* kJ/mm	Preheat °C	HAZ Hardness HV10 max - min ------------- mean	No of faces showing cracking	Hydrogen, ml/100g in top block
					Diffusible	Total
W5	0.70	20	429 - 397 ----------- 410	6/6	Not charged
W7	0.78	115	417 - 401 ----------- 411	0/6	Not charged
W10	0.85	90	413 - 401 ----------- 405	1/6	Not charged
W14	0.75	65	417 - 409 ----------- 413	4/6	Not charged
W16	0.76	115	437 - 409 ----------- 422	0/6	Not charged
W18	0.76	50	425 - 405 ----------- 411	6/6	Not charged
W30	0.86	150	405 - 401 ----------- 403	4/6	0.5	0.8
W33	0.87	200	413 - 390 ----------- 403	2/6	0.1	0.4
W36	0.96	250	409 - 401 ----------- 406	0/6	0.4	0.7
W39	0.90	225	409 - 401 ----------- 404	2/6	0.5	0.8
W41	0.88	250	409 - 397 ----------- 402	0/6	Not analysed

* using arc efficiency of 0.8

CTS tests

CTS testing was performed as far as possible, to BS EN ISO: 17642-2: 2005^[3], but slightly modified to take account of the special requirements of this work. All welding was performed at a nominal heat input of 0.8kJ/mm (arc energy of 1k/mm), representative of likely practice for repair welds. Tests were carried out at different preheats to define 'crack : no crack' boundaries. The threshold preheat is defined as the highest preheat for which cracking was observed, established to within 25°C of a 'no crack' result and confirmed with a second 'no crack' result.

Test block design

The test assembly itself (shown in Fig.2) was modified so that restraint was provided by four bolts (tightened to 100Nm torque), rather than by anchor welds. This was to avoid hydrogen loss from the hydrogen charged CTS test assemblies during anchor welding. In addition, top blocks which were hydrogen charged were machined with small (10x12x75mm) tabs on the face opposite the weld. These were sawn off immediately prior to preheating and used for hydrogen determinations. The modified design was used for all tests to ensure consistent comparisons.

In order to establish the validity of this procedure, including uncharged baseline, hydrogen determinations were performed on tabs from two simultaneously charged blocks (steel A) before and after preheating to 150°C. The results of this trial are reported with the rest of those for steel A below.

Preheating procedures

In order to minimise hydrogen loss during preheating, a procedure for rapidly heating hydrogen charged top blocks and bolting these to the separately preheated bottom blocks, was developed. The larger bottom blocks and bolts were brought to the required temperature by soaking in a furnace (1hr per inch thickness min). The hydrogen charged top blocks were heated using an electric resistance heating method. The equipment used (Fig.3) was a weld thermal simulator which, for the size of specimen used in this work, i.e. 50mm CTS top block, was found to be capable of heating to a temperature of 180°C in 70 seconds. During heating and subsequent assembly of the CTS components, temperature was monitored with a portable digital thermometer and by taking readings from a (Chromel Alumel) thermocouple, located on the upper face of the top block.

Prior to carrying out the CTS tests, an investigation of temperature variations between this top face and the block centre was performed. The results showed that when heating to 180°C, the temperature measured on the top face rose more quickly, exceeding the temperature at the centre by up to 20°C. As the current was turned down (as 180°C was approached) the difference decreased. With the current off, a uniform temperature was reached after 10 seconds. The temperature measured on this upper face was therefore taken as being representative of the whole block.

Once at temperature, top blocks were transferred to the welding area and the CTS assembly bolted together ready for welding. This incurred a delay of approximately seven minutes. During trials, an assembly heated to 180°C was found to cool to 160°C in this time (obviously heat loss will be less at lower preheat temperatures). The temperature quoted for individual tests is the average temperature of the top and bottom blocks immediately prior to welding, measured using a contact thermometer. Following welding, water cooling was applied as per BS EN ISO 17642-2:2005.

Examination of CTS test welds

After welding, sectioning and examination of the CTS assemblies was also carried out to BS EN ISO 17642-2:2005, following a delay of at least 72 hours. Metallographic sections were taken as required, polished to a 3µm finish, etched in 2% nital and then examined for cracking in the top block HAZ, using an optical microscope. Following this, hardness surveys were performed using a Vickers hardness machine with a load of 10kg.

Results

High carbon equivalent, high sulphur, C-Mn steel (A)

A 'crack : no crack' preheat threshold of 20°C (room temperature) was established for this steel in the as-received condition (at a heat input of 0.8kJ/mm and with a weld metal hydrogen level of 9.7ml/100g). CTS test results for this steel are presented together with hardness survey results in Table 2. The threshold is lower than would be predicted by the nomogram in ref ^[1], apparently due to the low hardenability of this steel, as discussed in the next section.

The results of diffusible and residual hydrogen determinations performed on the tabs cut from hydrogen charged top blocks are included in Table 2. These results show a wide scatter in hydrogen contents, with a maximum total of 12.14ml/100g, a minimum total of 6.05ml/100g, and an average total of 8.47ml/100g, in the welded blocks. This variation was not expected considering previous experience ^[7] with this charging method. Although greater variations in hydrogen pick-up have been seen between different steels, this present level of scatter has not been seen with one steel. It is possible that this is due to the high inclusion content. Table 2 also shows the results of hydrogen determinations on tabs removed before and after preheating. These seem to indicate that for the two blocks in question the hydrogen level measured in a tab is reduced during heating to 150°C and subsequent cooling. However, two separate blocks were used for this trial and the differences between the two may just be experimental scatter in light of the other results for this steel. It is not therefore thought possible to draw any firm conclusions from this trial.

Results of CTS tests on this material after hydrogen charging are shown in Table 2 and included on Fig.4. It can be seen that the hydrogen content (average 8.47ml/100g) of the charged parent material has increased the preheat threshold by approximately 40°C.

In this work the 'crack : no crack' preheat threshold for charged material has been defined by the occurrence of conventional HAZ type cracking. However, cracking associated with blistered inclusions was also observed, a photomicrograph showing typical examples of both types of crack is shown in Fig.5. Examination of unwelded hydrogen charged material has shown that such cracking can exist even before preheating and therefore, although its occurrence has been noted, it has not been used to define 'crack : no-crack' thresholds. Nevertheless, this type of inclusion cracking did seem to be exacerbated by preheating, when preheated but unwelded blocks were examined and it was also noticeably worse in the HAZs of welded blocks.

High carbon equivalent, low sulphur, C-Mn steel (B)

CTS results for this steel in the as-received condition and after hydrogen charging are shown in Table 3. Hydrogen determinations on the tabs from charged specimens are also included in Table 3. The results for total hydrogen content show the maximum to be 1.15ml/100g, the minimum 0.80ml/100g and the average 0.99ml/100g. It can be seen from these results that the level of hydrogen, particularly diffusible hydrogen, in this steel is substantially lower than for the higher sulphur steel A.

All the CTS data for this steel are included in Fig.4. It can be seen from this graph that preheat thresholds are effectively the same in both conditions, at 110°C, and both are significantly higher than for the high sulphur steel A which has the same CE. In the as-received condition this is not unexpected, considering the difference in sulphur contents between the two steels and the previously reported observation that high sulphur can reduce relative cracking susceptibility (all other thing being equal)^[8]. Critical hardnesses for the two steels, are however, very close in the as-received condition, at 425 HV10 for the high sulphur and 421 HV10 for the low sulphur steel (Tables 2 and 3), consistent with the idea that high sulphur alleviates cracking risk via its effect on hardenability. As expected, no ‘blistered inclusion’ type cracking was observed in this low sulphur steel.

Low carbon equivalent, high sulphur C-Mn steel (C)

CTS test results for this steel, in both the as-received and hydrogen charged condition are given in Table 4, and included on Fig.4. Results of hydrogen determinations after charging are also included in Table 4, showing a variation in total hydrogen content from 6.88 to 11.47 with an average of 9.48ml/100g. These results are similar to those for the other high sulphur steel, A.

 

No cracking was observed in tests carried out at room temperature with the material in the as-received condition. In the hydrogen charged condition however, a preheat threshold of 45°C was established. This equates to a shift of >25°C by comparison with the as-received steel.

Blistered inclusion type cracking was also observed in this steel, but not used to define the threshold as with the other high sulphur steel A.

2.25Cr 1Mo steel (D)

CTS test results for as-received and hydrogen charged material, welded at 0.8kJ/mm heat input and weld metal hydrogen levels of 4-5ml/100g are given in Table 5. Results of hydrogen determinations on charged material are also included in Table 5. These results indicate a very small amount of diffusible hydrogen present in the tabs (0.36ml/100g). The extent to which this is representative of the charged blocks as a whole is unknown, but it would be expected that hydrogen loss during cooling in the autoclave immediately following charging will be greater from the tab because of its larger surface area to volume ratio. The average total hydrogen content was 0.68ml/100g.

Despite the low measured hydrogen content, the effect of charging on the preheat threshold was very marked as shown in Fig.6. The shift in threshold preheat being approximately 130°C, from 90°C for the as-received to 225°C after hydrogen charging.

Discussion

C-Mn Steels

It was not intended to vary charged hydrogen level in this programme. The original intention was to determine any increase in preheat necessary to prevent hydrogen cracking for the three C-Mn steels after hydrogen charging, assuming the hydrogen charging had resulted in a constant material hydrogen level. The results of the programme are presented in these terms in Table 6.

Table 6. Summary of effects of hydrogen on necessary preheat to prevent hydrogen cracking.

	High CE (0.43) High S (0.029%)	High CE (0.43) Low S (<0.002%)	Low CE (0.36) High S (0.033%)	2¼Cr 1Mo
Average total hydrogen in steel, ml/100g	8.5	1.0	9.5	0.7
Increase in preheat after charging	40°C*	Zero*	>25°C*	130°C**

* Heat input of 0.8kJ/mm and 9.7ml/100g consumable hydrogen
** Heat input of 0.8kJ/mm and 4.3ml/100g consumable hydrogen

For C-Mn steels containing high levels of sulphur, hydrogen charging using the NACE TM0177 solution A ^[5] results in total parent material hydrogen contents of approximately 9ml/100g and this in turn increased the preheat necessary to prevent cracking by approximately 25-40°C. The increases in preheat necessary for both the high and low CE steels equate roughly to what would be expected on increasing consumable hydrogen level to scale B from the scale C level actually used. This indicates that, at this level of hydrogen in parent material, the effect of hydrogen introduced by welding cannot be considered in isolation or assumed to swamp any parent material hydrogen effects.

The lack of any observed increase in cracking following hydrogen charging of the low sulphur steel (B), is probably due to the low hydrogen content, this being sufficiently low that any effect it may have is masked by the hydrogen from the consumable.

As noted above, the hydrogen charging did not result in consistent levels for all the C-Mn steels, and in some cases there was less consistency within one steel type than was expected. The scatter in levels attained for the three C-Mn steels makes it possible to plot a graph of total hydrogen content versus preheat. This graph has been presented in Fig.4 (note that it has been assumed that as-received materials contain no hydrogen). From this graph the differences in behaviour between the three types of C-Mn steel can be seen more clearly. In addition, the need for data at other hydrogen levels or, in the case of the high CE low sulphur steel (B), at a lower consumable hydrogen level, can be seen.

For the low sulphur steel B, the reported results show considerably lower diffusible hydrogen content than the higher sulphur steels, but a roughly comparable residual hydrogen content. Other unpublished work at TWI^[9] confirms this trend. In a project looking at the effect of hydrogen on fracture toughness of steels, a C-Mn steel of intermediate sulphur content (0.015%) was charged in an identical manner. Hydrogen determinations revealed a diffusible hydrogen content of 3.97ml/100g and a residual content of 0.1ml/100g. Thus it appears that diffusible hydrogen content rather than residual hydrogen content is a function of sulphur content. This is perhaps surprising as it would be expected that more sulphide inclusions would provide more deep traps, and result in a greater residual hydrogen content. It is however, thought that the split between diffusible and residual hydrogen may be misleading, as it is possible that some hydrogen present in inclusion voids may be able to diffuse out even at room temperature over the time period of the measurements. It is also worth noting that only the low sulphur steel was aluminium treated.

The effect of hydrogen in causing blistered inclusion type cracking with these high sulphur steels must also be considered. Although this form of cracking was seen in two of the unwelded parent materials (because of the severity of the corrosive charging medium and the quality of the steels) as well as in HAZs, it was observed that such cracking could also be induced or exacerbated in these steels by both preheating and welding. In view of this it is thought that a less severe charging condition, that does not result in cracking during charging, could be used to elucidate this effect in any future work. It should also be recognised that preheat and/or welding could exacerbate this type of cracking in a repair situation.

In the absence of data over a wider range of consumable and pre-charged hydrogen levels, it is unwise to propose any comprehensive guidelines for the behaviour of C-Mn steels in general. Nevertheless, this work has shown that parent material hydrogen content does indeed have an exacerbating effect on the risk of fabrication hydrogen cracking, but that the possibility exists to overcome this without the need for pre-weld hydrogen release heat-treatments. Specifically it has been shown that an increase in preheat of the order of 50°C when welding at scale C (5-10ml/100g) consumable hydrogen level is sufficient to prevent cracking at a heat input of ~0.8kJ/mm if the parent steel contains up to 12ml/100g hydrogen, as might be encountered following sour service.

2.25Cr 1Mo steel

The effect of hydrogen charging was the most marked with this steel, although the measured hydrogen level was low (average total 0.68ml/100g). However, it is thought that these hydrogen levels will be underestimates of the hydrogen present in the actual top blocks owing to the short diffusion distances in the analysis tab. It has been estimated, using a simple diffusion model, that the hydrogen levels in the main body of the blocks are probably consistent with the hydrogen levels expected (approximately 4ml/100g).

The very marked effect of the parent material hydrogen content (see Fig.6) is again indicative of an 'add on' effect as observed with the high sulphur C-Mn steels. In this case the much greater increase in threshold preheat is probably a reflection of the difference in behaviour between the Cr-Mo steel and the C-Mn steels. For the former, preheat only decreases the likelihood of cracking by enhancing hydrogen diffusion, whereas with C-Mn steel preheat also influences HAZ hardness to a certain extent, rendering the HAZ microstructure in turn, less sensitive to hydrogen. The effect of preheat on maximum HAZ hardness is shown in Fig.7. With a bulk source of hydrogen available in the steel, hydrogen level control via preheat, and hence weld cooling time, is difficult, but, at least in the size of block tested here, possible.

The implications of these results are twofold. Firstly, although for a real repair situation the possibility does exist of preventing cracking through preheat control of weld cooling times, it is probable that this will be impractical for 2.25Cr 1Mo steel in most cases due to the magnitude of the change in cooling rate required. A more practical solution may be the application of postheat to enhance hydrogen diffusion.

A second implication is that any hydrogen release heat treatments on this type of Cr-Mo steel should be combined with conservative (i.e. high preheat) welding procedures as this work has shown that even a very small amount of hydrogen remaining in the parent material will greatly increase the risk of cracking during welding.

Summary and conclusions

An experimental procedure has been developed for CTS tests on hydrogen charged parent material. Tests have been performed, following this procedure, using a modified test block assembly. Results have enabled threshold preheats to be established for three C-Mn and one 2.25Cr 1Mo steel, at a heat input of 0.8kJ/mm, both in the as-received and hydrogen charged conditions.

Specific conclusions are as follows:

1. Hydrogen charging increases the risk of fabrication hydrogen cracking in both C-Mn and Cr-Mo steels. For C-Mn steels this can be overcome by controlling weld cooling times through the application of preheat, negating the need for pre-weld hydrogen release heat treatments.
2. The high sulphur C-Mn steels of both high and low carbon equivalent required an increase in preheat to prevent cracking following charging to give hydrogen levels of up to 12ml/100g. This increase was of the order of 40°C for the high carbon equivalent steel and equal to or in excess of 25°C for the low carbon equivalent steel when welded using a scale C consumable hydrogen level at ~0.8kJ/mm heat input.
3. For the high sulphur steels, both conventional HAZ cracking and blistered inclusion cracking were observed to occur after welding in the hydrogen charged condition. There is some indication that preheat and welding exacerbates blistered inclusion cracking.
4. No change in the preheat required to prevent cracking was observed for the low sulphur high carbon equivalent C-Mn steel following hydrogen charging under the same conditions as used for the high sulphur steels. For the low sulphur steel the charged hydrogen levels were however significantly lower.
5. A significant increase of 130°C in preheat was required to prevent cracking in the 2.25Cr 1Mo steel following hydrogen charging in an autoclave at 450°C and 10.3MPa pressure for 48 hours.

Recommendations

General recommendations for welding steels containing hydrogen

When welding either C-Mn or Cr-Mo steels containing or suspected of containing hydrogen, it should be assumed that this will result in an increased risk of fabrication hydrogen cracking during repair or alterations.

Recommendation for welding C-Mn steels containing hydrogen

For C-Mn steels that have been in sour service the possibility exists of overcoming the increased risk of cracking in many circumstances by raising the preheat temperature.

In particular if the steel to be welded has an IIW carbon equivalent of ≤0.45%, and is to be welded at a heat input of ~0.8kJ/mm, with consumables of scale C, i.e. 10mlH₂/100g deposited metal (or lower) hydrogen content, then the increased tendency to crack can be compensated for by specifying a preheat of at least 50°C or 50°C in excess of that recommended in BS EN 1011-2 for a comparable hydrogen free situation, whichever is the higher.

The use of consumables of scale D or lower is strongly recommended, but an assumed consumable hydrogen level of scale C should still be used when deciding upon an initial (hydrogen free) level of preheat. In situations where a heat input in excess of ~0.8kJ/mm is to be used it is recommended that the level of preheat be decided upon assuming the heat input is ~0.8kJ/mm. If it is not possible to use a heat input of greater than or equal to ~0.8kJ/mm, then the recommendations set out above cannot be fully relied upon to prevent cracking.

Recommendations for welding Cr-Mo steels containing hydrogen

For these types of steels it appears unlikely that the increased risk of cracking can reliably be negated by the use of increased preheat, at least for those steels that have been in typical high temperature hydrogen service.

In view of this, it is recommended that where such welding operations are to be undertaken, a pre-weld hydrogen release heat treatment is performed and that this should be combined with conservative welding procedures, i.e. high preheat and low consumable hydrogen levels. Consideration should also be given to the use of a post heat to further reduce welding hydrogen levels.

Acknowledgements

This work was funded by a group of TWI member companies.

References

1. Bailey N et al: 'Welding steels without hydrogen cracking'. Second edition (revised) Woodhead publishing Ltd. 2004. ISBN 1 85573 014 6.
2. Brown C N, Carroll M J, Dean F W H, Harrison J H and Kettle A: 'Applications of hydrogen flux monitoring to pre-weld bakeouts of steel'. NACE Corrosion 2004, paper 4478.
3. BS EN ISO 17642-2:2005: 'Destructive tests on welds in metallic materials. Cold cracking tests for weldments. Arc welding processes. Self restraint tests.' British Standards Institution.
4. BS EN 1011-2:2001: 'Welding - Recommendations for welding of metallic materials - Part 2: Arc welding of ferritic steels.' British Standards Institution.
5. NACE Standard TM0177-2005: 'Testing of Metals for Resistance to Sulfide Stress Cracking and Stress Corrosion Cracking in H2S Environments'. National Association of Corrosion Engineers.
6. ‘The effect of outgassing cycles on the hydrogen content in petrochemical-reactor-vessel steels’. API publication 946, July 1981.
7. Pargeter R J: 'Factors affecting the susceptibility of C-Mn steel welds to cracking in sour environments'. ASTM International Symposium on 'Environmentally assisted cracking'. Florida 1987, STP1049.
8. Hart P H M: 'The influence of steel cleanliness in HAZ hydrogen cracking: the present position.' Welding in the World, 23, pp.230-237. 1985.
9. TWI: 'Effect of hydrogen on the fracture toughness of welded steels used for pressure containment plant'. Group Sponsored Project ref. 5610.