Hydrogen Cracking of Ferritic-Austenitic Stainless Steel Weld Metal

R N Gunn (Plant Assessment Manager, Plant Integrity, UK)
T G Gooch (Technology Manager, TWI, UK)

Presented at 'Stainless Steel World Duplex America 2000' conference, 29 February - 1 March 2000

Abstract

Ferritic-austenitic stainless steels are normally regarded as readily weldable by common arc processes. However, cases of weld metal hydrogen cracking have been reported, principally at joints produced using shielded metal arc welding (SMA); the present study was carried out to examine conditions under which the problem was likely to occur.

First, a procedure was derived for hydrogen analysis of ferritic-austenitic SMA weld metals, based on the vacuum hot extraction method. Second, self-retrained single run and multipass weld cracking tests were performed, with consumables varying in deposit hydrogen potential and phase balance. Weld metal cracking was found to occur only at high ferrite contents and hydrogen levels. The results were correlated with other published information, and a diagram was produced relating cracking to the consumable hydrogen content determined under standard conditions and the deposit phase balance. From this correlation, the risk of hydrogen cracking in practice is expected to be low, given adoption of normal welding procedure guidelines.

1. Introduction

To achieve optimum mechanical and corrosion resistance properties, welding procedures for ferritic-austenitic stainless steels must be carefully controlled to ensure that a satisfactory phase balance is achieved in the weld area with negligible formation of intermetallic phases. With such control, the alloys are normally regarded as readily weldable by common arc processes, in the sense of producing defect-free joints. However, cases of weld metal hydrogen cracking under production conditions have been reported to TWI, principally at joints produced using shielded metal arc welding (SMA). In practice, the problem is particularly associated with fairly high hydrogen levels and ferrite contents. Several workers have investigated the problem using various experimental methods ^[1-7] , but quantitative limits have not been well defined for either duplex or superduplex grades.

Conventionally, hydrogen cracking risk with transformable ferritic steels is dependent upon the diffusible hydrogen content at normal ambient temperature, and this can be determined using a standard procedure with collection over mercury ^[8] . With duplex stainless steels, overall diffusion rates are slow because of the presence of austenite ^[5] , although local diffusion in ferrite may well be as rapid as with carbon-manganese steels. Thus, collection of diffusible hydrogen at room temperature is impractical, and a number of studies have recommended extraction at elevated temperatures ^{[1-7,] [9, 10]} . However, there remains a need for definition and standardisation of a suitable analysis temperature and time.

The present work was in two phases. First, a procedure was derived for hydrogen analysis of ferritic-austenitic SMA weld metals based on the vacuum hot extraction method (VHE) ^[11] . Not all laboratories have access to VHE equipment for measuring hydrogen in a production situation. Therefore, weld metal hydrogen levels were also measured using vacuum encapsulation of a sample in quartz glass ^[3] followed by heating for hydrogen extraction and analysis by a commercial gas chromatography instrument. The aim was to identify an extraction temperature that would give consistent and complete hydrogen totals within a reasonable time period, i.e. less than 24 hours, which would allow rapid quality control of electrode batches. Second, self-restrained weld metal cracking tests were carried out, involving both single run and multipass assemblies, to examine the conditions of hydrogen content and deposit microstructure under which the problem was likely to occur.

2. Experimental Procedure

2.1 Materials

The compositions of the base materials and consumables used in the study are shown in Table 1. In Phase I, in order to produce a range of weld metal ferrite levels, three parent material/consumable combinations were explored. Most of the hydrogen analyses were performed on 25%Cr deposits on 316L-type base plate, which were chosen to approach 80% austenite on the basis that this would cover the range of practical concern and would indicate the maximum evolution time necessary for different production batches of consumables. Further deposits were produced to study the effect of lower austenite contents (i.e. 60%) on hydrogen evolution from 22% and 25%Cr materials. Deposits were then produced to assess the reliability of the analysis method for different hydrogen levels associated with varying coating type and welding process. Table 2 lists the materials employed and welding conditions used for the Phase I deposits.

Table 1 Compositions of materials employed in the study

Material Code	Thickness, mm	Phase	Element, wt%
			C	Si	Mn	Cr	Ni	Mo	N	Other
Base Plates
316L	12mm	I	0.013	0.42	1.35	17.05	10.3	2.11	0.045	-
S31803/1	12mm	I	0.020	0.45	1.65	22.0	5.8	3.03	0.165	-
S31803/2	20mm	II	0.019	0.48	1.74	21.8	5.8	2.73	-	-
S31803/3	12mm	II	0.020	0.45	1.63	22.3	5.8	1.97	0.166	-
S32750	12mm	I	0.017	0.39	0.37	25.0	7.7	4.09	0.275	-
Electrodes
22.9.3.LR*(a)		I	0.025	1.0	0.80	22.0	9.0	3.0	0.12	-
25.10.4.LR		I	0.031	0.54	0.74	25.3	10.4	3.8	0.242	-
25Mar		I	0.021	1.05	0.78	24.5	8.9	3.0	0.20	-
25Mb		I	0.032	0.65	0.97	25.4	9.9	4.12	0.23	-
22Mb		I	0.032	0.42	0.95	23.2	9.22	3.07	0.16	-
22MbL		I	0.031	0.39	0.91	23.1	9.2	3.05	0.16	-
22Mbr		I	0.024	0.4	0.8	23.5	8.7	3.1	0.22	-
W1		II	0.045	0.33	1.12	24.8	8.0	3.17	0.19	1.74Cu
W3		II	0.040	1.07	0.55	22.3	5.3	3.01	0.22	-
22.9.3.LR(b)		II	0.025	1.0	0.8	22.5	9.5	3.2	0.16	-
85% ferrite		II	-	-	-	-	-	-	-	-
Submerged arc
22S		I	0.009	0.47	1.59	22.4	8.7	3.14	0.14	-
25S		I	0.016	0.33	0.78	25.2	9.2	3.66	0.23	0.64Cu, 0.58W
Flux cored
22F (W2)		I, II	0.017	0.69	1.35	22.6	9.6	3.11	0.16	0.05Cu
*R = rutile, ar = acid rutile, b = basic, br = basic rutile

Duplex stainless steel plate, grade UNS S31803, was used for both single and multipass welds in Phase II. Plates of 12 and 20mm thickness were used for the single pass and multipass deposits respectively. The single pass welds utilised two SMA consumables. A standard 22%Cr duplex consumable was employed, together with an experimental electrode, designed to produce high weld metal ferrite levels (~85%). Both consumables were 3.2mm in diameter and had a rutile flux coating. The multipass welds studied in Phase II utilised three consumables. Two SMA consumables were used, viz a commercial 25%Cr electrode, and an experimental 22%Cr electrode with only 5.3%Ni to give a high ferrite level (62-85%), while a third weld was produced using a commercial 22%Cr flux cored wire.

2.2 Derivation of Hydrogen Analysis Procedure

Hydrogen evolution behaviour was evaluated at different temperatures. For this purpose, each weld consisted of a single pass bead-on-plate deposit of 100mm length following the guidance of ISO 3690: Part 2 ^[8] . Stainless steel base metal blanks were not degassed to maximise hydrogen content, although each electrode was baked at 350°C for 1 hour prior to welding. The welding conditions employed for each deposit type are given in Table 2, with electrodes being deposited using automatic equipment. The end pieces were removed and the ferrite/austenite content determined by point counting (Section 2.4).

 

Table 2 Materials employed and welding conditions used for Phase I deposits.
(a) Evaluating hydrogen behaviour.

Deposit code	Base plate	Electrode type	Electrode diameter, mm	Voltage, V	Current, A	Travel speed mm/min	Arc energy, kJ/mm
25/80	316L	25Mar	4.0	30	132	224	1.1
22/60	S31803	22.9.3.LR	4.0	28	147	224	1.1
25/60	S32750	25.10.4.LR	3.25	22	103	150	0.9

 

(b) Varying consumable type and welding process.

Weld code	Drying conditions	Humidity* g/m 3	Voltage, V	Current, A	Travel speed mm/min	Arc energy, kJ/mm
SMAW
25Mar	From sealed packet	10.8	30	130	205	1.1
25Mar350	350°C/2hrs	7.7	30	131	214	1.1
25Mb	From sealed packet	5.0	23	111	171	0.9
22Mb	From sealed packet	7.7	23	110	188	0.8
22MbL	From sealed packet	5.0, 7.7	23	111	182	0.9
22Mbr	From sealed packet	7.7	23	133	200	0.9
SAW
22Sab	As-received wire & flux	4.8, 8.6	30	485	293	3.0
25Sab	As-received wire & flux	8.6	26	339	188	2.8
22Sab350	350°C/2hrs	8.6	30	491	293	3.0
22Sab350	350°C/2hrs	4.8	26	354	194	2.9
FCAW
22F	From sealed packet	5.8	31	187	439	0.8
* Humidity in grammes of water per m 3 of dry air.

On completion, most of the test pieces were stored in liquid nitrogen until analysis was performed. However, one sample of a triplicate set was left under ambient conditions for four days, to study the need for liquid nitrogen storage. The hydrogen evolution behaviour of each deposit, in terms of ml/H ₂ /STP/100g deposited weld metal, was determined against time by the VHE method at different temperatures. This was generally based on individual deposits, although duplicate or triplicate analyses were carried out on selected samples to investigate the level of repeatability in the VHE analysis. If extraction required longer than six hours, the specimens were allowed to cool and then reheated under vacuum the next day. The high austenite deposit 25/80 ( Table 2) was first analysed at extraction temperatures over the range 400°C to 950°C, and the time required for hydrogen evolution was noted. Once the temperature that would lead to hydrogen analysis results in less than one day was identified, the evolution behaviour of two duplex deposits (i.e. 22/60 and 25/60 in Table 2) was explored. Analysis was then carried out using other SMA electrodes and submerged arc and cored wire welding, as in Tables 1 and 2, using triplicate samples.

Comparison between VHE and encapsulation methods of hydrogen analysis was made using three commercially available 22Cr duplex stainless steel consumables, the analyses of which are included in Table 1. All of the consumables were deposited on type S31803 plate. VHE was carried out on each of the deposits at a temperature of 700°C for 3.5 hours. Vacuum encapsulated samples were heated in a furnace at 400°C for 72 hours to allow for hydrogen evolution: the hydrogen content was then determined by gas chromatography (GC) at ambient temperature by breaking the capsule in a flow of Ar carrier gas using a Yanaco G-1006 instrument. In the event, the results obtained indicated that complete extraction into the capsule was not being achieved. Accordingly, an additional test was undertaken with titanium powder added to the capsule to getter any hydrogen evolved from the sample.

2.3 Hydrogen Cracking Tests

2.3.1 Single Run Welds

The single run tests were based on the Y-groove approach, to induce weld metal cracking as encountered in practice. Test plates were machined in accordance with JIS standard Z3158 (1993) ^[12] . In order to obtain a high ferrite (~85%) level in Y-groove test runs, two panel assemblies were buttered with the high ferrite consumable prior to machining, the electrodes having been baked at 350°C for 2 hours prior to use. Welding conditions were set to provide an arc energy of 1kJ/mm. The root gap in the Y-groove was machined to 1.6±0.2mm. Anchor welds, either side of the Y-groove, were made by electron beam welding.

To obtain different hydrogen levels, test electrodes of each type were placed in a humidity cabinet, at 28.5°C and 74% humidity, for 24 hours, or baked at 350°C for 2 hours prior to welding. Single run, Y-groove test welds were deposited using each electrode type and hydrogen potential. Welding was carried out at 22°C and 46% relative humidity. A total of four test welds was deposited, the welding conditions again being chosen to give an aim arc energy of 1kJ/mm ( Table 3).

Table 3 Welding conditions used in the Phase II Y-groove cracking tests

Weld	Consumable	Current, A	Voltage, V	Arc energy, kJ/mm
W4	85% ferrite	111	28	0.9
W5	85% ferrite	111	28	0.9
W6	22.9.3.CR(b)	111	30	1.0
W7	22.9.3.CR(b)	111	29	0.9

Hydrogen analysis test blocks were machined from the same test plate as the Y-groove samples in accordance with BS 6693 ^[13] . Triplicate sets of blocks were machined for each electrode and aim hydrogen level. Test electrodes were either baked or humidified as described above. Test weld beads were deposited at the same time as the Y-groove test runs to ensure similar testing conditions. All test runs were deposited with arc energies of ~1kJ/mm. After welding, the test blocks were cleaned to remove the slag and were stored in liquid nitrogen. The hydrogen content was then determined by VHE at 950°C for 1.5 hours.

On completion of welding, the Y-groove samples were left for 1 week prior to examination. After this time, the weld caps were examined for evidence of any surface cracks. In addition, each test weld was sectioned as detailed in JIS Z3158 ^[12] , which requires a total of five transverse sections to be taken through the weld bead at equal distances apart, and examined for evidence of cracking.

2.3.2 Cracking in Multipass Welds

The test plate was cut into portions of 400mm length x 150mm width, a 30° bevel edge preparation was applied, and the root was seal welded using superduplex consumables. The first 100mm from the stop and start of each preparation was filled, again using superduplex consumables. The test welds were then deposited in the resultant groove using the conditions given in Table 4.

Table 4 Summary of Phase II - Multipass welding conditions

Weld code	Consumable	Diameter	Humidity*, g/m 3	No. of passes	Current, A	Voltage, V	Travel speed, mm/min	Arc energy, kJ/mm	Interpass temperature, °C
W1	25%Cr SMA	5mm	10.8	10	168-202	25	170-203	1.4-1.9	<200
W2	22F	1.2mm	11.0-11.5	9	192-198	31	320-400	0.9-1.2	<150
W3	22%Cr SMA	3.25mm	11.8-13.0	21	100-110	26	170-260	0.6-0.9	<150
* Humidity in grammes of water per m 3 of dry air.

As soon as each panel cooled from welding, it was inspected using 45° and 70° angled compression wave probes, from all three orthogonal directions, including the weld root. The equipment employed was a Force Institute P-scan system with a sensitivity set using a 3mm diameter side drilled hole in a duplex test block, following BS 3923 part 1 ^[14] . This system takes ultrasonic (UT) signals and probe positions to produce a three view image of the weld volume with a detection limit of about 0.5mm. Subsequent UT inspection of each panel took place after 1, 2, 4, 8, 16 and 30 days, and then monthly, until at least six months had expired. For comparison, dye penetrant inspection was undertaken, as was radiography using 400keV X-ray equipment viewed from the cap side.

2.4 Metallography

Metallographic sections through each Phase I weld deposit and Y-groove test run were prepared in order to determine the phase balance. This was performed by manual point counting, using a 25 point (5x5) grid at X1400 magnification; a total of 16 fields was measured. Phase identification was aided by staining the ferrite phase, either by electrolytic etching in 20%KOH, or by using a 'magnetic etch' treatment, which involved immersion in a magnetic iron oxide colloidal solution, 1%Fe ₃ O ₄ in a hydrocarbon base. On the multipass welds in Phase II, the cap pass of each panel was ground flat and the Ferrite Number (FN) determined using a Ferritscope.

3. Results

3.1 Derivation of Hydrogen Analysis Procedure

3.1.1 Hydrogen Evolution

The hydrogen evolution behaviour of deposit 25/80 at different VHE temperatures is shown in Fig.1. The rate of hydrogen extraction was dependent on the temperature, with higher temperatures accelerating extraction. For instance at 950°C, extraction was almost complete after about 100 minutes, whereas over 50 hours were required for complete removal at 400°C. The temperature of extraction did not significantly affect the final level of hydrogen evolved over the range studied.

Comparison of the evolution behaviour at 700°C for different deposits and base materials is given in Fig.2. The austenite content of each deposit is shown in Table 5. Deposit 25/80 appeared to take slightly longer to reach a plateau than the other deposits, consistent with the high austenite content (81%, Table 5) reducing the hydrogen diffusion rate. Evolution was substantially complete for all deposits after about 3 ¹ / ₂ hours.

Table 5 Austenite content of Phase I weld deposits

Material code	Austenite content, %
25/80	81.2 (2.5)
22/60	59.3 (4.9)
25/60	60.8 (3.8)
Figures in parentheses are ± 95% confidence limits

Results of the analyses on a range of deposits are given in Table 6. It will be noted that the different sample types displayed a range of hydrogen content, from about 5-20ml/100g deposited metal, with good repeatability in all cases.

 

Table 6 Hydrogen contents of commercial consumables

Sample Code	VHE results, ml H 2/STP/100g				Variation from Average, %
	1	2	3	Average
SMAW
25Mar	19.8	18.8	19.3#	19.3	±2.6
25Mar350	16.4	16.6	17.2	16.7	±3.0
25Mb	4.5	5.4	5.9	5.2	±13.5
22Mb	9.0	9.0	9.2	9.1	±1.1
22MbL	8.6	8.3	9.7	8.9	±9.0
22Mbr	4.8	5.6	5.1	5.2	±7.7
W1*	23.7	21.9	25.8	23.8	±8.4
W3*	23.2	-	-	23.2	-
SAW
22S	5.3	5.4	5.5	5.4	±1.9
25S	3.7	3.8	4.0	3.8	±5.3
22Sab	11.7	13.8	12.6	12.7	±8.7
25Sab	11.1	10.2	11.1	10.8	±5.6
22Sab350	6.7	6.4	7.4	6.9	±6.8
25Sab350	5.8	6.5	4.3	5.6	±23.2
FCAW
22F (used in W2*)	15.4	15.5	14.9	15.3	±2.6
-   Not analysed/not applicable *   Employed in Phase II #  Sample left under ambient conditions prior to analysis. All other samples were stored in liquid nitrogen.

The sample stored under ambient conditions prior to analysis gave a result of 19.3ml of H ₂ compared with 18.8 and 19.8ml for the comparable samples stored at -196°C. Thus, there appeared to be no effect of storage conditions on the consequent hydrogen levels.

3.1.2 Comparison between VHE and encapsulation techniques

The results of VHE and encapsulation methods are presented in Table 7. The encapsulation technique consistently produced lower results than VHE in the deposits analysed. Therefore, selected samples, which had been encapsulated and heated, were re-analysed by the VHE method. This exercise showed that significant levels of hydrogen remained in the sample after encapsulation, while the sums of the two analyses were similar to single analysis by VHE at 700°C for 3.5 hours ( Table 7). The sample encapsulated with titanium powder showed low hydrogen content on analysis, indicating complete hydrogen evolution and pickup by the titanium.

Table 7 Summary of results for different hydrogen analysis techniques and sample types

Deposit code	Sample code	Humidity g/m 3	Preparation	Analysis technique	Extraction Conditions	Results, mlH 2/STP/100g
22Mb	1-3	7.7	-	VHE	700°C/3.5hrs	9.0, 9.0, 9.2
	4,5	7.7	Sealed in glass, 400°C/72hrs	GC	RT*	3.4, 3.6
	4	-	Re-analysis of no.4	VHE	950°C/1.5hrs	+6.5
22Mbr	6-8	7.7	-	VHE	700°C/3.5hrs	4.8, 5.1, 5.6
	9-11	7.7	Sealed in quartz, 400°C/72hrs	GC	RT	1.1, 1.3, 2.0
	9	-	Re-analysis of no.9	VHE	950°C/1.5hrs	+4.7
	12-14	7.7	Sealed in quartz, 400°C/72hrs	GC	RT	1.2, 1.8, 3.2
	15,16	7.4	Sealed in quartz with 6g of Ti, 400°C/72hrs	VHE	950°C/1.5hrs	0.86, 0.93
	17,18	7.4	Sealed in quartz, while heated to 200°C/2 hrs, 400°C/72hrs	GC	RT	1.15, 0.95
	17,18	-	Re-analysis of no.17, no.18	VHE	950°C/1.5hrs	+4.3, +3.7
22F	24-26	5.8	-	VHE	700°C/3.5hrs	14.9, 15.4, 15.5
	27, 28	13.8	Sealed in quartz, 400°C/72hrs	GC	RT	7.5, 8.5
	27	-	Re-analysis of no.27	VHE	950°C/1.5hrs	+9.6
	29-31	13.8	Sealed in quartz, 400°C/72hrs	GC	RT	9.6, 9.8, 10.2
*RT - normal ambient temperature.

3.2 Hydrogen Cracking Tests

3.2.1 Y-Groove Tests

Table 8 shows the weld metal hydrogen levels achieved with each of the consumables. The ferrite content and results of the hydrogen cracking tests are shown in Table 9. Both of the high ferrite weld beads (W4 and W5) cracked. The high hydrogen test bead (W5, 17.5ml/100g) cracked within a few hours of welding, while the low hydrogen bead (W4, 11.4ml/100g) cracked in less than three days.

Table 8 Weld metal hydrogen results - Y-groove cracking test samples.

Weld	Hydrogen level, ml/H 2/STP/100g weld metal
	1	2	3	Average
W4	11.18	11.76	11.22	11.4
W5	17.23	18.23	17.07	17.5
W6	15.00	14.36	15.88	15.1
W7	18.17	19.28	18.43	18.6

 

Table 9 Weld metal ferrite content and Y-groove cracking test results.

Weld	Volume fraction of ferrite, %	Result of Y-groove test
W4	83 (5)	Cracked
W5	82 (4)	Cracked
W6	57 (6)	Not cracked
W7	56 (4)	Not cracked
Figures in parentheses are ±95% confidence limits.

The weld metal microstructures were typical of duplex stainless steel deposits. The high ferrite welds contained some grain boundary and isolated intragranular austenite. Cracking in each of the high ferrite welds was transgranular in nature and propagated through the ferrite phase, Fig.3. Some cracks arrested at ferrite-austenite boundaries. The crack morphology was typical of that observed in practical problems of hydrogen cracking in duplex weld deposits.

3.2.2 Multipass Welds

The ferrite and hydrogen contents of the multipass welds are shown in Table 10. Panel W1 showed two root indications on the initial UT scan within one hour of welding. The nature of these indications did not change during the subsequent 9.5 months inspection, nor did any other indications appear. No flaws were detailed in panel W2 at any time during a six month inspection. Two indications were found in panel W3 directly after welding; no changes occurred in these indications after five months inspection, nor did any flaws appear. No evidence of cracking was detected by either dye penetrant inspection or radiography. Sectioning of each of the panels after inspection was completed did not reveal any cracking, although porosity was found in panel W2, which had not been detected by UT.

Table 10 Multipass weld metal ferrite and hydrogen content.

Weld	Ferrite content, FN	Hydrogen level, ml/H 2/STP/100g weld metal
		1	2	3	Average
W1	59	23.7	21.9	25.8	23.8
W2	55	15.4	15.5	14.9	15.3
W3	90.5	23.2	-	-	23.2

4. Discussion

4.1 Hydrogen Analysis Procedure

The objective of Phase I was to derive a hydrogen analysis procedure with particular emphasis on the extraction temperature that would lead to total hydrogen analysis results to be obtained within a 24 hour period. Table 11 lists the key results of this work, showing the recommended times at a range of extraction temperatures for VHE. The use of an extraction temperature of 700°C allows for welding and analysis within 24 hours, meeting the perceived practical aim, even for electrode batches giving high austenite contents. Nevertheless, it is within the capability of VHE to extract hydrogen at 950°C, which only takes 1 ¹ / ₂ hours and allows for triplicate analyses in one working day. Further, the work indicates that samples do not need to be stored in liquid nitrogen prior to analysis. This allows samples to be stored under ambient conditions without jeopardising results.

Table 11 Recommended minimum extraction times for VHE.

Extraction temperature, °C	Recommended minimum extraction time
400	7 days
500	36 hours
600	16 hours
700	3 1 / 2 hours
950	1 1 / 2 hours

Table 6 illustrates that the VHE method with extraction at 700°C is applicable to a range of consumable types, welding processes and associated hydrogen levels. As would be anticipated, the greatest variations in results from triplicate analyses was observed for low hydrogen contents, about 5ml/100g deposited metal. At higher hydrogen levels, the repeatability is consistently less than 10% of content. This behaviour is similar to conventional analysis of ferritic steel deposits, and, in the general case, is as expected for analysis of any element present at the ppm level.

The results of this work clearly indicate that the encapsulation method gives lower readings than obtained by VHE. Subsequent measurement by VHE showed that hydrogen was retained in the steel using the encapsulation method. The most likely explanation for this is that, during hydrogen evolution at 400°C in the capsule, a positive hydrogen partial pressure is developed, hindering further liberation from the weld deposit: this is consistent with the apparent complete evolution in the sample encapsulated with a titanium getter. This observation has also been made by other workers ^[16] . On the other hand, Kaçar ^[17], performed repeat analyses on duplex weld deposits using the encapsulation technique, and did not report any further hydrogen evolution on reheating. Clearly, further work is required in this area before the generally more convenient encapsulation/gas chromatography approach can be accepted for routine hydrogen analysis.

4.2 Hydrogen Cracking Tests

4.2.1 General Comments

The current work has demonstrated the adverse effect of high levels of weld metal ferrite on the risk of fabrication hydrogen cracking in duplex stainless steels. Figure 4 shows the results together with previous data generated at TWI from Y-groove SMA tests on 22%Cr duplex stainless steel ^[5] .

The programme sought to generate cracking data for SMA welds, in 22%Cr duplex stainless steel, for which the consumable hydrogen level had been determined using the VHE method. Cracking occurred in welds containing ~83% ferrite and hydrogen levels of 11.4 and 17.5ml/H ₂ /STP/100g deposited metal. The welds containing 'normal' volume fractions of ferrite did not crack. In addition, three multipass welds were made, none of which cracked.

4.2.2 Comparison With Other Work

Several workers have investigated fabrication hydrogen cracking in duplex and superduplex stainless steels using a variety of experimental methods ^{[1-7], [9,10]} . Comparison of the current results with other published work is problematic, in that other workers used a variety of welding techniques, cracking tests and hydrogen analysis procedures. The choice of welding process may affect the weld metal hydrogen content and also microstructure in the sense of inclusion levels, which may influence cracking behaviour ^[15] . The current work primarily examined SMA welding, as this is of particular practical significance in terms of fabrication hydrogen cracking. Data exist for gas tungsten arc welding (GTAW) ^{[1,2], [7],} but, because hydrogen was added via the shielding gas, they cannot be readily converted to a deposit hydrogen content for comparative purposes. Figure 5 shows the results of the current work, together with published data from other workers who used the Y-groove test to evaluate the cracking of SMA deposits. The axes of Fig.5 are deposited hydrogen content and weld metal ferrite volume fraction. The former may be determined by appropriate consumable analysis and the latter may be obtained by phase balance checks on weld procedure qualification samples: hence, both parameters can be readily employed by a welding engineer, to assess the risk of a problem arising in practice.

Lundin et al carried out Y-groove cracking tests using both duplex and superduplex stainless steels, with SMA weld deposits ^[4,5] . The hydrogen content of the deposited metal was determined using a quartz encapsulation method. The current work, supported by van der Mee et al ^[16] , has shown that the encapsulation method of measuring hydrogen may underestimate the true level, and thus the data points from Lundin's work may be raised from their position in Figure 5. The degree by which each of the points should be raised is not known, but it is unlikely that Lundin's actual hydrogen levels were considerably higher than the reported analyses. It should be noted that Lundin recorded ferrite number (FN) and not % ferrite in his work. However, reasonable correlations exist between FN and % ferrite, ^[18], for both 22Cr ^{[6], [19,20],} and 25Cr alloys ^[21]. The results in Fig.5 are presented in terms of % ferrite, with Lundin's data having been converted using relationships derived at TWI, namely:

% ferrite = 0.57FN + 8.82 (for 22Cr duplex stainless steel, ^[20];
% ferrite = 0.82FN + 3.6 (for 25Cr superduplex stainless steel, ^[21] .

The current work agrees well with the results of Lundin et al, accepting that the reported hydrogen levels of the welds are probably lower than actual.

Walker and Gooch ^[5] used a similar experimental approach to the present investigation, and carried out Y-groove cracking tests on 22%Cr duplex stainless steel with commercial SMA consumables. The deposited metal hydrogen content was determined by VHE. Welds containing ~60% ferrite and 12-16ml/H ₂ /STP/100g deposited metal did not crack. These results compare favourably with the current work.

It is possible to include in Fig.5 a boundary curve, separating cracking and no cracking conditions for SMA deposits in duplex stainless steel. It should be noted that there exists some degree of uncertainty as to the shape of the curve in Fig.5 at ferrite volume fractions between 65 and 85% because of a lack of data, and further study is desirable.

4.2.3 Practical Implications

In the current comparison of results, both microstructure, in terms of ferrite volume fraction, and hydrogen content, in terms of deposited metal, are accounted for. The graph ( Fig.5) provides a guide to prevent fabrication hydrogen cracking in SMA welded ferritic-austenitic stainless steels. To illustrate the point, the work of Beeson ^[22] represents an in-service failure of SMA welded UNS S31803 duplex stainless steel. It was reported that cracks appeared to start at a highly ferritic (over 90% ferrite) cap pass, and that the welds contained 24ppm hydrogen. This behaviour would clearly be expected from Fig.5. The present results show that fabrication hydrogen cracking may be avoided by maintaining a weld metal phase balance of less than 60% ferrite and a deposited hydrogen content of less than 18ml/H ₂ /STP/100g: this is consistent with practical experience. Whilst the majority of the data presented was generated using single pass welds, it is believed that the graph does present a basis for predicting the likelihood of cracking in multipass production welds.

5. Conclusions

1. The VHE method of hydrogen analysis is applicable to duplex stainless steel weld metals. Table 11 presents guidance as to the minimum time required to extract hydrogen using VHE equipment in the 400 to 950°C temperature range.
2. Fabrication hydrogen cracking in SMA 22%Cr duplex stainless steel weld deposits may be avoided by maintaining a weld metal phase balance of less than 60% ferrite and a deposit hydrogen content of less than 18ml/H ₂ /STP/100g.
3. A predictive diagram has been generated, using the results of the current work and other published data, to aid the prevention of fabrication hydrogen cracking in practice.

6. Acknowledgements

The authors thank their colleagues at TWI for advice and assistance throughout the programme. Grateful acknowledgement is made also to members of two Group Sponsored Projects at TWI for guidance during the programme of work, and for permission to publish this paper, viz: Chevron UK Ltd, ESAB AB, Lincoln Smitweld BV, AB Sandvik Steel and Shell U.K. Exploration and Production.

7. References

1	Fekken V, van Nassau L and Verweg M,	'Hydrogen induced cracking in austenitic/ferritic duplex stainless steels', Proc conf. 'Duplex Stainless Steels '86', The Hague, The Netherlands, 26-28 October 1986, Nederlands Institut voor Latechniek, 1986, Paper 26, 268-279.
2	Miura M and Ogawa K,	'Hydrogen embrittled cracking in duplex stainless steel weld metal', IIW Doc IX-1461-87.
3	Lundin C D, Kikudin Y and Khan K K,	'Measurement of diffusible hydrogen content and hydrogen effects on the cracking potential of duplex stainless steel weldments - Phase I Report', Welding Research Council Progress Reports, 1992, Vol XLVII No.1/2, 41-62.
4	Lundin C D, Kuroda T, Zhou G, Wang Y, Qiao P and Khan K K,	'Measurement of diffusible hydrogen and hydrogen effects on weldability of 2205/2209 duplex stainless steel', Proc conf. 'Third international conference on trends in welding research', Gatlinburg, Tennessee, USA, 1-5 June 1992, ASM International, 1992, 743-747.
5	Walker R A and Gooch T G,	'Hydrogen cracking of welds in duplex stainless steels', Proc conf. 'Duplex stainless steels '91', Beaune, France, 28-30 October 1991, Les editions de physique, Vol 2, 1992, 895-903.
6	Shinozaki K, Ke L and North T H,	'Hydrogen cracking in duplex stainless steel weld metal', Welding Journal 1992, 71, 11, 387s-396s.
7	Fang P J, Kirkwood D, Power L J and Baxter C F G,	'The resistance of superduplex steel to hydrogen cracking', Proc conf. 'Duplex Stainless Steels '94', Glasgow, Scotland, 13-16 November 1994, TWI, 1994, Paper 51.
8	ISO 3690: 1977:	'Welding determination of hydrogen in deposited weld metal arising from the use of covered electrodes for welding mild and low alloy steels', including Addenda 1 and 2: 1983.
9	van der Mee V, Meelker H and van Nassau L,	'How to avoid hydrogen cracking in (super) duplex stainless steel weldment', Proc. conf. 'Duplex stainless steels '94', Glasgow, Scotland, 13-16 November 1994, TWI, 1994, Vol.1, Paper 119.
10	Lundin C D,	'Hydrogen embrittlement of duplex stainless steel weldments', Proc Conf. '8th Annual North American Welding Research Conference', Columbus, Ohio, 19-21 October, 1992, AWS/EWI/TWI.
11	Jenkins N, Hart P H M and Parker D H,	'An evaluation of rapid methods for diffusible weld hydrogen', Welding Journal 1997, 76, 1, 1s-10s.
12		JIS Standard Z3158 (1993), 'Method of Y-groove weld cracking test'.
13		BS 6693: Part 2: 1986 'Method for determination of hydrogen in manual metal arc weld metal'.
14		BS3923: Part 2: 1972: 'Methods for ultrasonic examination of welds'.
15	van der Mee V,	private communication, 1998.
16	van der Mee V, Meelker H and van der Schelde R,	'How to control hydrogen level in (super) duplex stainless steel weldments using the GTAW or GMAW process', Welding Journal 1999, 78, 1, 7s-14s.
17	Kaçar R,	'Hydrogen in duplex stainless steel welds'. PhD Thesis, The University of Leeds, UK, October 1997.
18	Gunn R N,	'Duplex stainless steels - microstructure, properties and applications', 1997. Abington Publishing, Cambridge, UK.
19	Liljas M K, Qvarfort R,	'Influence of nitrogen on weldments in UNS S31803', Proc conf Duplex Stainless Steels '86, 26-28 October 1986, The Hague, The Netherlands, 244-256.
20	Walker R A,	TWI confidential Group Sponsored Project 5550, 'Corrosion resistance of duplex ferritic-austenitic stainless steel weldments'. November 1987.
21	Gunn R N,	TWI confidential Group Sponsored Project 5578, 'Corrosion properties of high alloy duplex stainless steel weldments'. August 1992.
22	Beeson D L,	Proc conf. 'Corrosion resistant alloys', Kuala Lumpur, Institute of Engineers Malaysia, January 1995, loose paper.