John Rothwell and David Abson
Paper presented at WELDS 2009: Design, testing, assessement and safety of high-temperature welded structures. Fort Myers, FL. USA. 24-26 June 2009
The improvement of tensile and creep rupture strength, achieved by the development of martensitic-ferritic 9-12%Cr creep-resistant steels, known as creep strength enhanced ferritic (CSEF) steels, is attractive to designers who want to take advantage of them to improve power plant efficiency and reduce component wall thickness. However, although parent alloy developments have been making good progress, weldment performance has been of increasing concern. In particular, the cross-weld creep performance for CSEF steels is an issue, with type IV cracking characterised by rupture in the outer region of the heat affected zone (HAZ) and low strain to failure being the typical features of failure in cross-weld tests. In this investigation, narrow gap TIG welding, flux-cored arc welding (FCAW) and reduced pressure electron beam (RPEB) welding were employed to produce a series of butt welds in a new European steel - FB2 - a wrought, boron-containing 9%Cr steel. These welds were subjected to metallurgical examination and mechanical testing, including toughness and cross-weld creep rupture tests. Results are compared with the well known steel grades 91 and 92. Welding process was found to have a marked effect on weld metal toughness, but not on the long term cross-weld creep rupture performance. FB2 weldments were found to out-perform those of grades 91 and 92 in long term creep tests, but they did suffer from type IV cracking and reduced rupture strength, compared with the parent steel. The paper draws conclusions on the performance of the welded joints, and makes recommendations for further work and possible application within the power industry.
1. Background
Improved thermal efficiency of power plant has been the main driver for the development of the martensitic-ferritic 9-12%Cr creep-resistant steels known as creep strength enhanced ferritic (CSEF) steels. The target operating temperature for these steels is 650°C, with a common target design life of 100,000hrs. Increasingly, the demand for efficiency is linked to efforts to reduce CO2 emissions, in order to meet environmental obligations and minimise any form of carbon tax that may be levied in the future.
The improvement of hot tensile and creep rupture strength, achieved by the development of CSEF steels is attractive to designers who want to reduce pipe wall thicknesses, and thereby minimise thermal stresses for a more reliable plant. This is increasingly important for today's power plant (which are subject to temperature cycles in an effort to respond to the peaks and troughs of demand), and to improve profitability. However, although parent alloy developments have been making good progress, attention to weldment performance has failed to keep pace. Various cracking mechanisms associated with the weldments can and do occur, and weldments can often give cause for concern in power plant fabrications. In particular, the cross-weld creep performance of CSEF steels is an issue, with creep failure typically occurring by distinctive 'Type IV', which occurs over a small region of the HAZ, and which results in little total strain across the joint. Only recently have codes such as ASME I (power boiler code) taken into account the major difference between parent and cross-weld creep performance.
Aside from performance of the steel and weldments themselves, productivity is very important, in order to meet the heavy demand for new power plant, especially in China and India. With their higher creep strength, the newer steels generally allow reduced wall thicknesses, which in turn reduce joint thickness, weld volume and welding times. Other measures to improve productivity include, replacing MMA welding with flux-cored arc welding and the use of higher productivity processes, such as narrow gap welding (including narrow gap TIG welding) and electron beam welding. New reduced pressure electron beam welding technology may offer very high quality welds at substantially reduced welding times, particularly for thick section components.
In this investigation, which was performed as part of the European Collaboration in Science and Technology (COST) programme, narrow gap TIG, FCA and RPEB welding were employed to produce a series of butt welds in a new European steel designated FB2, a forged, boron-containing steel. These welds were subjected to metallurgical examination, and to a range of mechanical tests, primarily aimed at establishing weld metal and cross-weld properties, including cross-weld creep rupture strength. The compositions of the rutile flux-cored and metal-cored wire were based on the compositions of weld metals developed in earlier studies at TWI [Abson et al 2007, Abson 2004, 2005] .
2. Fabrication and testing details
2.1 Materials
The FB2 parent steel used for this study is the most promising of a range of compositions containing 9-12%Cr that were developed in the previous phases of the COST programme, COST 501 and 522 for their creep resistance. The designation 'F' refers to compositions for forgings of the 10%CrMoVNbN type, and 'B' indicates the alloy is boron-containing. The number 2 indicates a particular composition in a series of experimental heats of similar compositions. Small additions of boron have been shown to improve creep strength [Kondo et al, 2006] . It has been reported that it combines with M23C6 precipitates, retarding their coarsening and the onset of tertiary creep [Zeilinska-Lipiec et al, 2007] . A Co addition is also present as an austenite stabiliser, to inhibit the retention of δ-ferrite, which is known to be detrimental to toughness and creep performance. The steel is a serious candidate for rotor forgings, piping and pressure vessels, albeit with different tempering treatments for the different applications. Further information on the development of CSEF steels in the COST programme have been reported by [Mayer and Masuyama, 2008] .
Testing of other FB variants from the COST programme was effectively abandoned due to their poorer creep performance, illustrated in Figure 1. The poorer performance of these variants, which contained higher Cr levels, was attributed to the early formation of modified Z-phase, (Cr(V,Nb)N), a nitride that forms via dissolution of strengthening MX precipitates [Danielsen, 2007; Mayer and Masuyama, 2008; Kozeschnik and Holzer, 2008] .
Figure 1. Creep rupture strength of the COST development steels FB2, FB6 and FB8 as a function of time to rupture at 650°C. FB6 and FB8 both suffered from short term on-set of z-phase causing the creep rupture strength to fall dramatically [Mayer and Masuyama, 2008]. Reproduced by permission of Woodhead Publishing.
The FB2 steel parent properties were not evaluated in this programme, but this has been carried out by other participants of the COST programme [Kern et al, 2006; Di Gianfrancesco et al, 2008]. Maximum service temperatures are likely to be in excess of 600°C, but may not be quite as high as 650°C. FB2 steel is already finding application with one example being in RWE's Newrath power station in Germany where it will be used for intermediate pressure (IP) modules operating at temperatures of up to 620°C [Peel et al, 2007].
Segments cut from a large rotor forging of FB2 steel approximately 1.1m in diameter were provided by Alstom for this programme of work. The steel was supplied already cut into slabs 80mm thick (to allow a simulation of the welding of a thick section rotor steel) and 35mm thick (to allow simulation of the welding of pipework).
2.2 Welding
The welding techniques used in the investigation were FCAW, narrow gap TIG (both using experimental cored wires), and reduced pressure (~5 x 10-2 mbar) RPEB welding. The panels were given a PWHT appropriate for the intended application (3 hours at 760°C for 35mm thick FCAW, simulating pipe welding, and 10 hours at 680°C for the 80mm thick material simulating rotor welding). In some instances modified 9%Cr slabs were RPEB welded on as extensions to the FB2 steel, as can be seen in Figure 2, to make the cooling characteristics of the weld more realistic.
This study is the first to use RPEB for the welding of 9%Cr steel. The RPEB gun was developed at TWI to reduce pumping time for vacuum conditions, and to provide a more robust and economically viable solution for industry, compared with previous EB welding technology. Whereas conventional EB welding requires vacuum conditions (10-3 mbar), RPEB is operational at 0.1 to 10mbar. The process has great potential for thick section, high quality, high productivity welding requirements, and for this reason it is of interest in the current project. Further details of the development of this technology have been documented by [Punshon, 2008] and [Rothwell et al, 2008].
TIG welding is generally employed where high quality defect-free welding is required. The use of a narrow gap preparation has two immediate advantages. The first is the reduction in weld metal volume and therefore welding time for the production of a weld, and the second is that the smaller volume of weld material means less distortion of the fabrication is likely to take place.
FCAW has become increasingly popular as a semi-automatic welding technique and is considered an easier and quicker fabrication and repair technique than for example manual metal arc (MMA) welding.
The consumable compositions used for TIG and FCA welding were based on a commercial consumable for grade 92 but with the addition of Co and Ti. Studies have shown that the incorporation of 1%Co can improve both toughness and creep rupture strength of the weld deposit by reducing the amount of retained δ-ferrite [Barnes and Abson, 2003; Abson, 2004]. A separate study has indicated that small additions of Ti increase weld metal creep strength, [Abson et al, 2007]. The FCAW consumable was supplied by Metrode Products Limited and the metal-cored wire for the TIG welding was manufactured by Air Liquide.
Each welded panel of FB2 was subjected, in the as-welded condition, to radiographic examination to BS EN 1435: 1997, to check for the presence of any cracking or welder-induced defects. The minimum delay between completion of the welding and the radiography was approximately one week.
Following welding and radiography, each panel was given a post-weld heat treatment (PWHT). For the 35mm thick flux-cored wire welds, which simulated pipe welding, the heat treatment was 2h at 760°C (PWHT1), Holloman-Jaffe (H-J) parameter = 21.29, while for the thicker narrow gap and TIG and RPEB welds, which simulated the possible welding of turbine rotors, a different treatment was used (PWHT2), 10h at 680°C, H-J parameter = 20.07. The cooling and heating rate for each of these heat treatments was 30°C/h.
2.3 Testing and evaluation
Weldment evaluation comprised chemical analysis (optical emission spectrometry), metallographic examination of weldments after fabrication and of welded specimens after testing, Vickers hardness testing, cross-weld room temperature tensile testing (to BS EN 10002-1: 2001), weld metal Charpy testing (to BS EN 10045-1: 1990), weld metal CTOD testing (to BS 7448: Part 2: 1997). Cross-weld elevated temperature tensile testing (to BS EN 10002-5: 1992) and creep testing (to BS 10291: 2000), were carried out at 600, 625 and 650°C.
3. Results
3.1 Weld quality and microstructure
3.2 Metallurgical examination
Butt welds were created successfully by each of the three welding processes. Although some slag entrapment and porosity were visible in places, the continuity was considered acceptable for the extraction of test specimens for the purposes of microstructure and property evaluation in an experimental programme of work. Figures 2, 3 and 4 show macro and micro images of each of the weldment types produced, after PWHT.
Figure 2. Photomacrograph and micrograph of FCAW weld after PWHT. (The backing bar, having once been joined to the weld root, was sawn off, but then re-instated for this section. The EB welds towards each side join the FB2 parent steel containing weld FCW3 to the ends of the extension slabs). Etched in 2.5%HCL, 2.5%Picric dissolved in ethanol: a) Weld cross section; and b) HAZ;
Figure 3. Photomacrograph and micrograph of TIG weld 1 after PWHT. Etched in 2.5%HCL, 2.5%Picric dissolved in ethanol: a) Weld cross section; and b) HAZ;
Figure 4. Photomacrograph and micrograph of RPEB weld W43 after PWHT. Etched in 2.5%HCL, 2.5% Picric dissolved in ethanol solution: a) Weld cross section; and b) HAZ;
Both parent and weld metal microstructures consisted of a tempered martensite with prior-austenite grain boundaries clearly visible in the parent microstructure and columnar structure of the RPEB weld. The weld metal beads in the FCA welds were suitably staggered to avoid continuous regions of HAZ within the weld metal, which can lead to preferential failure of the weld metal [Cerjak and Mayer, 2008]. Considerable refinement was observed in the TIG weld metal, due to the build up of layers on top of one another. All HAZs produced featured a much more refined prior austenite grain size than the parent material, and the definition between different regions in the HAZ was far from obvious. The TIG weld in particular seemed to have a particularly dark-etching region in the HAZ which is believed to be due to the arrangement of thin layers causing a high degree of tempering. Delta ferrite was not apparent in the RPEB weld inspected. However, very small quantities directly beneath the capping layer and also ~10-20mm from the root tip were observed in the TIG weld. In the FCA weld, more areas of delta ferrite were observed at various locations within the weld metal.
3.3 Chemical analyses
The target compositions were largely achieved for the FCA and TIG welds, while the autogenous RPEB weld reflected the parent steel composition more closely; Table 1 shows the target composition alongside the analyses. The TIG and RPEB welds have substantially lower oxygen contents than the flux-cored weld, and the RPEB weld has a much lower nitrogen content compared with both of the other welds.
3.4 Hardness
The results for Vickers hardness (HV10kg) measurements in the parent steel (after PWHT) and for the micro-hardness (MHV200g) traverses across the fusion boundaries for all the weld types are shown in Figure 5. An obvious difference in parent steel hardness is observed as a result of the different PWHTs applied. For PWHT2 applied to the TIG and RPEB weldments, the average hardness was ~270HV, whilst for PWHT1 applied to the FCAW the average hardness was ~230HV. The H-J value is higher for PWHT1 and the hardness is lower, as a consequence of the greater extent of tempering compared with PWHT2.
Figure 5. Micro-hardness traverses across the fusion boundaries of the weldments after PWHT.
The microhardness results for the traverses across the joints highlight differences in the HAZ between the different welding processes. Notably the TIG weld appeared to have a particularly soft HAZ region, which corresponds to the darker-etching heavily tempered region visible in the corresponding photomicrograph (Figure 3). The FCAW and RPEB welds did not display such a dip in hardness.
Hardness values for creep tested specimens from the TIG and FCA welds are considerably lower than the heat treated specimens. The base hardness value for creep specimens appears to be of the order of 190HV for both welds (Figure 6).
Figure 6. Micro-hardness traverses across the fusion boundaries of the weldments after PWHT and cross weld creep tests for TIG and FCA welds.
3.5 Room and elevated temperature tensile results
The room temperature yield (0.2% proof) strength and tensile strength, for the FB2 parent steel, were measured as 467 and 637MPa (after PWHT), respectively. By contrast, weld metal yield and tensile strength values were all above 690 and 830MPa, respectively. The elevated temperature strength of the parent steel and that of cross-weld specimens fell with increasing test temperature in the range 600 to 650°C. Table 2 summarises the tensile test results. All cross-weld samples broke in the parent metal, thereby proving the strength of the weld metal to be sufficient.
Table 1 Manufacturers' all weld metal deposit and welded joint compositions. TWI analysis reference numbers S/07/83, S/06/84, O/N 07/8 and O/N 06/31.
| Identity | Elements, wt% |
|---|
| C | Si | Mn | P | S | Cr | Mo | Ni | Al | Co | Cu | Nb | Ti | V | W | B |
|---|
| AW |
Metal-cored wire** (DHD506
P92GTAW005) |
0.13 |
0.35 |
0.64 |
0.007 |
0.010 |
9.8 |
0.49 |
0.42 |
0.013 |
1.14 |
<0.010 |
0.041 |
0.065 |
0.29 |
1.42 |
0.0033 |
Flux-cored wire Supercore F92
(FZF145)** |
0.12 |
0.34 |
0.76 |
0.019 |
0.006 |
9.86 |
0.52 |
0.49 |
0.007 |
0.88 |
0.06 |
0.032 |
0.061 |
0.23 |
1.56 |
0.0030 |
| WJ |
Target composition* |
0.13 |
0.34 |
0.67 |
~0.018 |
~0.005 |
9.6 |
0.46 |
0.44 |
0.011 |
1.00 |
0.04 |
0.04 |
0.060 |
0.26 |
1.40 |
~0.003 |
| FCAW Cap+ |
0.12 |
0.34 |
0.84 |
0.019 |
0.006 |
10.4 |
0.57 |
0.49 |
0.02 |
0.94 |
0.05 |
0.02 |
0.08 |
0.26 |
1.44 |
0.005 |
| FCAW Root+ |
0.12 |
0.32 |
0.83 |
0.018 |
0.005 |
10.2 |
0.68 |
0.44 |
0.01 |
1.05 |
0.05 |
0.03 |
0.08 |
0.27 |
1.26 |
0.006 |
| TIG cap+ |
0.12 |
0.32 |
0.62 |
0.010 |
0.008 |
9.72 |
0.52 |
0.42 |
<0.01 |
1.11 |
0.01 |
0.04 |
0.06 |
0.28 |
1.26 |
0.007 |
| TIG root+ |
0.13 |
0.28 |
0.56 |
0.010 |
0.006 |
9.64 |
0.71 |
0.36 |
<0.01 |
1.15 |
0.02 |
0.04 |
0.04 |
0.27 |
1.04 |
0.009 |
| RPEB cap+ |
0.13 |
0.10 |
0.31 |
0.011 |
0.001 |
9.12 |
1.40 |
0.17 |
<0.01 |
1.29 |
0.03 |
0.05 |
<0.01 |
0.22 |
<0.05 |
0.012 |
| RPEB root+ |
0.13 |
0.10 |
0.30 |
0.012 |
0.001 |
9.14 |
1.39 |
0.20 |
<0.01 |
1.30 |
0.03 |
0.05 |
<0.01 |
0.22 |
<0.05 |
0.012 |
| FB2 steel++ |
0.13 |
0.09 |
0.34 |
0.007 |
0.001 |
8.93 |
1.45 |
0.14 |
<0.01 |
1.20 |
0.03 |
0.05 |
0.14 |
0.21 |
<0.05 |
0.008 |
Table 1 Manufacturers' all weld metal deposit and welded joint compositions. TWI analysis reference numbers S/07/83, S/06/84, O/N 07/8 and O/N 06/31 continued.
| Identity | Elements, wt% |
|---|
| N | O |
|---|
| AW |
Metal-cored wire** (DHD506
P92GTAW005) |
- |
- |
Flux-cored wire Supercore F92
(FZF145)** |
0.04 |
- |
| WJ |
Target composition* |
- |
- |
| FCAW Cap+ |
0.041 |
0.0635 |
| FCAW Root+ |
0.039 |
0.0574 |
| TIG cap+ |
0.031 |
0.003 |
| TIG root+ |
0.034 |
0.004 |
| RPEB cap+ |
0.015 |
<0.001 |
| RPEB root+ |
0.016 |
<0.001 |
| FB2 steel++ |
0.022 |
- |
Note: **Manufacturers batch reference number. *As <0.01; Sn <0.012; B ~ 0.003. +As, Sn ≤ 0.009. ++As< 0.01, Sn ≤ 0.005.
AW = All weld metal deposits. WJ = composition of weld metal in welded joint.
Table 2 Tensile data.
| Specimen description | Test
temperature,
°C | 0.2% offset
proof strength,
MPa | Tensile
strength,
MPa | Elongation,
% | Reduction
of
area, % |
|---|
| Material | Type | PWHT |
|---|
*FB2 parent
steel |
Parent |
None |
Room |
713-715 |
834-835 |
15-16 |
56 |
| |
|
|
|
|
|
|
|
FB2 parent
steel |
Parent |
680°C, 10h |
Room |
467 |
637 |
24.3 |
- |
| Parent |
680°C, 10h |
600 |
393 |
477 |
18.5 |
77 |
| Parent |
680°C, 10h |
625 |
359 |
436 |
21.0 |
80.5 |
| Parent |
680°C, 10h |
650 |
321 |
408 |
22.0 |
81.5 |
| |
|
|
|
|
|
|
|
| FCAW |
AWM |
760°C, 2h |
Room |
712 |
836 |
13.5 |
- |
| AWM |
760°C, 2h |
Room |
717 |
841 |
13.2 |
- |
| X-weld |
760°C, 2h |
600 |
- |
399 |
17.5 |
- |
| X-weld |
760°C, 2h |
625 |
- |
361 |
21.5 |
- |
| X-weld |
760°C, 2h |
650 |
- |
319 |
22.5 |
- |
| |
|
|
|
|
|
|
|
TIG cap,
AWM |
AWM |
680°C, 10h |
Room |
694 |
836 |
19.0 |
62 |
TIG root,
AWM |
AWM |
680°C, 10h |
Room |
730 |
867 |
18.0 |
60 |
TIG mid
thickness |
X-weld |
680°C, 10h |
600 |
- |
420 |
23.5 |
- |
TIG mid
thickness |
X-weld |
680°C, 10h |
625 |
- |
371 |
14 |
- |
TIG mid
thickness |
X-weld |
680°C, 10h |
650 |
- |
340 |
14.5 |
- |
| |
|
|
|
|
|
|
|
RPEB cap,
AWM |
AWM |
680°C, 10h |
Room |
748 |
884 |
20.0 |
49 |
RPEB root,
AWM |
AWM |
680°C, 10h |
Room |
765 |
895 |
19.0 |
62 |
| RPEB cap |
X-weld |
680°C, 10h |
600 |
- |
468 |
13.5 |
- |
| RPEB mid |
X-weld |
680°C, 10h |
625 |
- |
424 |
16 |
- |
| RPEB root |
X-weld |
680°C, 10h |
650 |
- |
386 |
14.5 |
- |
AWM = all weld metal specimen. X-weld = cross weld specimen. * Properties advised by Alstom [electronic communication, 2006] .
3.6 Charpy absorbed energy
Lower shelf impact energy values were ≤11J, and 27J temperatures ranged between 20 to 76°C; see Figure 7. Upper shelf toughness increased, in line with decreasing weld metal oxygen content, in the order flux-cored, TIG and RPEB welds. RPEB HAZ Charpy energy was found to be similar to that of the weld metal, below 20°C (Figure 8). Above 20°C the weld metal displayed higher Charpy absorbed energy. HAZ Charpy energy of the other weld types was not investigated, due to a shortage of material.
Figure 7. Charpy energy results for FCW, TIG and RPEB weld metal (note PED = Pressure equipment directive 2005);
Figure 8 Charpy energy results for RPEB weld metal and HAZ together.
3.7 CTOD toughness
At typical temperatures of concern during start up (20-80°C) there is not a significant separation between the best fit line curves although RPEB and FCAW appear marginally tougher Figure 9. However, at 120°C and higher, the CTOD values were better for the TIG and RPEB welds than for the FCA welds.
Figure 9. Weld metal CTOD results for all three weld types
3.8 Creep rupture
Figure 10 shows the creep rupture data determined at 650°C. In all cases, the cross-weld creep rupture strength is less than that of the FB2 parent steel (as measured in the COST 522 programme), and the parent steel subjected to PWHT2 in this study.
Figure 10. Creep rupture data determined at 650°C
For short test durations, the FCA weldments have appreciably lower creep rupture strength than that of the TIG and RPEB weldments and nearly all the fracture face was located in the parent material. At longer durations the performance of different weldment types is comparable.
Figures 11 and 12 show the relative performance of FB2 compared with Grades 92 and 91 respectively. FB2 parent material and particularly weldments appear to out-perform parent materials and weldments respectively for both Grades. Comparing FB2 weldments with those of Grade 92, a 21% improvement in rupture stress (at 11,340hrs) and 68% improvement in rupture time (at 70MPa) for cross-weld specimens was observed.
Figure 11. Creep rupture plot showing the performance of FB2 parent material and weldments relative to Grade 92 data. The allowable stress (S) for grade 92 (9Cr-2W) seamless pipe, and S multiplied by weld strength reduction factor (w) for N+T (0.77) and a sub-critical PWHT (0.5) are illustrated. The w multiplier was introduced in 2008 in recognition of the lower cross-weld creep rupture strength of CSEF grades.
Figure 12. Creep rupture plot showing the performance of FB2 parent material and weldments relative to Grade 91 data. The allowable stress (S) for grade 91, and S multiplied by weld strength reduction factor (w) for N+T (0.77) and a sub-critical PWHT (0.50) are illustrated. The w multiplier was introduced in 2008 in recognition of the lower cross-weld creep rupture strength of CSEF grades.
3.9 Post-test investigation of creep rupture specimens
Many of the creep tested specimens were sectioned to determine failure locations. Microstructures of the creep tested specimens were noticeably more tempered, and the presence of creep voids particularly concentrated along what appeared to be prior austenite boundaries was observed for the longer running specimens as shown in Figure 13. In most cases, it was possible to identify whether the failure occurred in a particular region of the HAZ, parent or weld metal, but in some cases failure spanned different areas of the weld, and it was more difficult to define the region of initiation, fracture and cracking (Full details of fracture location and appearance reported by [Abson and Rothwell, 2009]). Two of the TIG specimens featured a partial fracture through the weld metal, but neither of the failures could be attributed solely to the fracture of the weld metal.
Figure 13. a) Macro image of Type IV failure for TIG specimen tested at 650°C, 80MPa and a rupture time of 5700hrs. mm scale shown. Etched in Picral; b) close up of HAZ region which didn't suffer complete failure showing void damage in the outer region of the HAZ; c) close up of void damage as indicated in b; and d) close up of void damage as indicated in c
The transverse metallographic section (Figure 13) shows creep cavitation that would have preceded failure, and that the damage accumulated in the outer edge (intercritical or fine grain region) of the visible HAZ (Figure 13d). This is the typical appearance of a Type IV fracture and void damage preceding failure in the HAZ.
The longer duration tests, especially those at high temperatures and low stresses, were clearly Type IV failures. Figure 14 shows a plot of the temperature and stress test conditions, and separates those specimens that obviously failed by the Type IV mechanism and those that did not. This is an attempt to define stress and temperature fracture map.
Figure 14. Plot of test temperature versus stress showing the failure mechanism and an estimate of the region for failure type (shaded) in terms of stress and temperature ranges.
4. Discussion
4.1 Weld quality and composition
It is expected that all three processes could be improved to reach a defect level acceptable to appropriate standards. The 35mm slabs, welded with FCAW, simulated pipe welding, and the PWHT applied was appropriate to this application. Hence, the process could be used for the welding of tubular members, and also for general fabrication in steam power generation applications. The other two welding processes were used for the welding of 80mm thick slabs. These two processes could find application in the welding of thick wall pipe and pressure vessels or rotor forgings. Despite this being the first instance of RPEB technology used for welding boron-containing 9%Cr steel, the results are promising, with very little porosity or defects having been detected, and the overall performance of the welds being somewhat superior to those created using the other processes.
4.2 Charpy impact energy and CTOD toughness
Both the 27J Charpy temperature and the upper shelf Charpy toughness improve progressively from the FCA to the TIG to the RPEB weld. Also, the RPEB weld has the best CTOD toughness and the flux-FCA weld the worst. It is likely that the most important contributory factor is the progressive change in oxygen content, and thus the number density of oxide inclusions [Suenaga et al, 2008; Chovet et al, 2008; Abson and Pargeter, 1986] . As oxide inclusions are capable of initiating both ductile and cleavage fracture, the higher oxygen, and therefore the likely higher inclusion content of the cored wire welds, detailed in Table 1, would have been detrimental to their toughness. The RPEB weld also features less delta ferrite, which is known to be detrimental to toughness (this may be due in part to the higher Co content). The low W content of the parent steel, and thus of the RPEB weld metal may have also contributed to higher toughness, although the lower Ni content may have been expected to counter this to some degree, as it is generally thought to increase toughness in W-free 9%Cr weld metals [Barnes, 2002; Abson, 2004] . If it is considered desirable to improve the toughness of the RPEB weld further, this could probably be achieved by introducing a metal or alloy as a shim between the two faces of the joint, so that it is incorporated into the weld. Ni and Co have previously been used effectively to improve toughness in weld metal of arc welds [Barnes, 2002; Abson, 2004] .
Although upper shelf energy values were markedly different between the welds, the absorbed energy values obtained at 20°C, which are generally the basis of toughness requirements, are quite close together. With respect to the basic European Pressure Equipment Directive (PED) guidance of 27J at 20°C [Department for Business Enterprise and Regulatory Reform, 2005] , the results were borderline, with some individual values displaying absorbed energy values far in excess of the requirement while all trend lines were found to lie on or below 27J at 20°C. With respect to the values reported for the parent steel at 20°C, which ranged between 30-56J [electronic communication, 2006] , the weld metal and HAZ data were generally slightly lower; see Figures 7 and 8. The deposited weld metal Charpy data at 20°C were close to those reported for weld metal containing a similar amount of Ti in a previous weld metal study [FCAW deposit study, Abson, 2005, Abson et al, 2007] , on which the present compositions were based. However, the addition of Co, which was made in the light of studies indicating a beneficial effect on impact toughness [Barnes and Abson, 2003] , appears not to have made a significant improvement in this study.
Despite the Charpy absorbed energy at 20°C being below 27J, this does not necessarily preclude the welds from application, since an ECA could be used to assess integrity. Commercial FCAW consumables such as Metrode 'Supercore F92' have reported typical values of 25J after PWHT at 760°C/4-6hrs [Metrode Products Ltd, 2007] .
The benefit of high upper shelf energy obtained by RPEB and TIG welding compared with that of the FCAW welds may be of little merit if the greatest risk of cracking is during hydrotesting or start up, i.e. when temperatures close to 20°C are experienced. At 20°C Charpy impact values range between 24 and 7J (according to the best fit lines) for all weld types. In this respect there is a very limited advantage of RPEB and TIG welds over FCAW deposits. However, the comparatively better CTOD toughness and Charpy energy of TIG and RPEB welds at higher temperatures are of greater value if problems with cracking at temperatures above 20°C are of concern. Where toughness and quality are paramount, TIG welding is still the process of choice. RPEB welding has been shown to be superior in nearly every aspect of this study. However, unfamiliarity with the process and the capital expense of equipment is off-putting to most fabricators.
4.3 Creep rupture performance
The steel FB2 was developed within the European COST programme (522) with an operating temperature of 650°C in mind, and also to succeed the popular alloys available at the time, such as Grades 91 and 92. Inspection of Figures 11 and 12, show that FB2 parent alloy is similar in rupture strength to Grade 92, but is a definite improvement over Grade 91. However, the main benefit of FB2 appears to be its cross-weld creep rupture stress performance, which exceeds that of Grade 92 for longer term tests (>1000 hours), and is comparable to that of Grade 91 parent material.
For the higher strength CSEF steels developed such as Grade 92, although the parent properties are appealing, the divergence of cross-weld properties from those of the parent for long duration tests are severe. The divergence from parent rupture strength is so bad that at 100,000hrs Grade 92 weldments have been estimated to perform no better than Grade 91 weldments [Kimura et al, 2008] . Up until 2008, there was no obligation in commonly used power plant construction codes, such as ASME section I, to include the effect of welding on creep strength in any calculations for determining pressure vessel thicknesses. This inadequate provision has undoubtedly resulted in premature failures. However, in 2008, a 'weld strength reduction factor' was introduced into ASME section I which takes into account the poor creep performance of weldments in CSEF steels, increasing component thickness requirements and effectively reducing the allowable stress value permitted (in many cases halving it).
The creep strength of the cross-weld specimens lies below but almost parallel to that of the FB2 parent steel. The fact that the departure from the parent steel properties is not increasing significantly at longer durations may help designers to seek exemption from the newly introduced codes, and allow them to apply a less restrictive 'weld strength reduction factor'. Studies have shown that if boron and nitrogen are present in the right quantities, cross-weld creep strength is similar to the parent creep strength (Figure 15). This has been found to depend on the B:N ratio being such that large BN precipitates are fully dissolved, Figure 16. This preferable ratio and performance is characteristically accompanied by a HAZ that has not been refined by the welding process (Figure 17), which may explain to some degree the improved performance of these steels. Whilst FB2 is a boron-containing steel, the boron and nitrogen contents are such that BN is not fully dissolved at the solution treatment temperature. So, although boron appears to have improved creep strength overall and given the steel a competitive advantage over other grades, the boron and nitrogen ratio has not impacted significantly on susceptibility to Type IV cracking (a diffusion-controlled failure mechanism occurring in the intercritical and fine grain region of the HAZ). The effect of boron and nitrogen are discussed further in the next section.
Figure 15. Creep rupture data of [Abe, 2005] for parent steel and cross-weld specimens of 92 grade and a boron-containing steel tested at 650°C.
Figure 16. Solubility limit for BN at 1150°C for high Cr steel, with the corresponding information for the Fe-B-N system, derived by [Fountain et al, 1962]. superimposed. Also shown is the composition specification for P92. The figure has been adapted from [Sakuraya et al,2006] .
Figure 17. Microstructure in steel NPM1 before and after welding thermal cycle simulation showing that the location and orientation of features remains the same, [Mayr and Cerjak, 2007] . Reproduced by permission Dr P Mayr and Professor H Cerjak.
The poorer performance of FCAW joints at short test durations is likely to be due to the difference in heat treatment, with the higher temperature (760°C) relieving residual stress and tempering the parent microstructure more extensively than for the other weldments subjected to a PWHT at a lower temperature (680°C) with a corresponding lower H-J parameter value. This indicates that the higher H-J parameter of heat treatment PWHT1 is detrimental to rupture strength of the parent material in the short term.
Although two of the failed TIG creep rupture specimens featured partial failures within the weld metal at relatively high stress levels, it can be reasoned that the experimental weld metal designed and conceived within this project has been an overall success in terms of creep performance. No weld metal failures were observed in longer duration tests where Type IV failure became the dominant mechanism.
4.4 Practical implications
The flux-cored wire welding procedure may be preferable for on-site fabrications and repairs due to ease of use and high productivity, but the weld metal in this study suffered from poor upper-shelf toughness in particular. The TIG and RPEB welding procedures are more suited to welding thicker components that require greater integrity. The ability of RPEBW to produce thick section high quality welds in a single pass may offer a way to improve productivity and reliability for components that are usually welded using SAW or narrow gap TIG welding processes.
Where new plant has been designed to old ASME Section 1 rules, it may be acceptable in some instances to replace for example Grade 91 material with almost the same thickness of FB2 material to bring the plant into line with the new rules, since the weldment creep strength for FB2 is almost the same as for the Grade 91 parent material.
At present there is no commercially available grade of CSEF steel that does not suffer from Type IV cracking and the associated poor cross-weld creep performance. However, experimental steels containing carefully controlled amounts of boron and nitrogen could offer a way forward. The martensitic boron-nitrogen (MARBN) steels, 9cr-3W-3Co-VNb-B, developed at The National Institute for Materials Science (NIMS) do not appear to suffer from reduced cross-weld creep strength, at least for the time scale reported; see Figure 15. [Abe, 2005] and [Abe et al, 2007] have shown that if the boron and nitrogen levels are controlled within 9%Cr steels, it is possible to avoid nucleation of coarse BN precipitates and greatly enhance the creep properties of parent steels and their weldments. [Sakuraya et al, 2006] have determined the limits for boron and nitrogen levels for the avoidance of coarse BN, as illustrated in Figure 16, although exceptions to this diagram have been found. When coarse BN precipitates are avoided, there is a strengthening effect from boron, which is understood to combine with M23C6 precipitates and prevent their coarsening. Another interesting consequence of a higher B:N ratio is the unusual appearance of the HAZ region; the parent steel microstructure appears virtually unchanged from its state prior to welding except for darkening due to tempering, i.e. almost the exact parent microstructure that existed prior to application of a welding thermal cycle is present after heat treatment, down to the location of prior austenite grain boundaries and martensitic laths (Figure 17) even though transformation to austenite and back has taken place [Mayr and Cerjak, 2007] .
Although boron has been introduced into recent steels such as Grade 92 and FB2 to improve creep strength, the boron to nitrogen ratio was not subject to appropriate control. One of the next steps discussed within the European COST consortium has therefore been to alter the B-N ratio of FB2 steel to enhance the properties. It is also intended to produce a larger heat of a steel created at T-U Graz called NPM1, which is similar to the successful Japanese compositions, already mentioned (MARBN steels). The incorporation of boron seems to be an important step towards achieving the desired creep properties. However, reduction in toughness is also associated with this addition, and therefore optimisation of toughness and creep properties needs to be considered. For these type IV failure-resistant steels, higher strength weld metals will need to be developed. The incorporation of Ti into weld metals, as demonstrated in this study and previously, may offer a guide for new weld metal compositions. However, while Ti is beneficial for creep strength, it has an adverse effect on toughness [Abson 2005, Abson et al 2007] , and thus an appropriate balance must be achieved.
In summary, FB2 is the first of a series of experimental steels produced within the European COST programme to show some advantage over existing coded steels of the same family, such as Grade 92, especially in terms of creep strength (although longer term cross-weld rupture data is desirable). However, the welding trials of the present programme have shown that type IV cracking is still the main creep failure mechanism operating at long durations, and an obvious, but lesser difference between parent steel and weldment creep strength still exists.
5. Conclusions
- The boron-containing 9%Cr steel FB2, developed in the European COST programme, has been welded successfully by FCA and TIG welding, using experimental cored wires and by RPEB welding.
- Weld metal Charpy impact and CTOD values were close to the lower shelf boundaries at 20°C. In this respect only marginal advantage of one weld type over the other in resistance to brittle fracture is expected. However athigher temperatures, RPEB and TIG welds demonstrate superior performance, which is associated with their lower oxygen content in the weld metal.
- The location of failure for the short term creep specimens was predominantly in the parent metal. Long term specimens mostly failed in the outer region of the HAZ, and were characteristic of Type IV failures. The onset duration forType IV failure was strongly dependent on the test temperature, with the duration decreasing as the temperature increased.
- No failures were observed in the weld metal for long-term cross-weld creep specimens, indicating that the experimental TIG and FCAW consumables and the autogenous weld metal resulting from RPEB welding used were of adequate creep strength for use with FB2 steel.
- No obvious effect on long term creep rupture performance due to welding process type was apparent in this study. In the short term, TIG and EB weldments performed better than the flux-cored welds. This is most likely to be due tothe differences in PWHT and consequential effect on the parent microstructure. However, there is no obvious advantage of one welding process type over another apparent in the longer term (>2,000hrs). The influence that PWHT has onthe results decreases with test duration, as the effect of metallurgical changes occurring during prolonged exposure at the test temperature becomes the most significant factor in governing the life of the specimens.
- At the target application temperature for FB2 steel (650°C) a ~20% improvement in creep strength and ~70% improvement in rupture life were observed for welded joints, when compared with Grade 92 data at long durations(>6,739hrs). The creep strength of FB2 weldments was found to be comparable to that of Grade 91 parent material.
6. Recommendations
- In view of the good creep rupture performance, of both the FB2 parent steel and the weldments, this steel should be considered as a candidate for elevated temperature service, to replace Grade 91 and 92 steel, particularly if FB2variants with an improved balance of boron and nitrogen have a reduced sensitivity to Type IV cracking.
- Where plant has been designed to code that has not included a weld strength reduction factor, eg ASME section 1 2007, the use of FB2 steel should be considered as a replacement to satisfy new code requirements without the need to increase thickness values appreciably. However, greater confidence in long term cross-weld creep performance is preferable, which necessitates the need for lower stress creep tests.
- An extended test programme for welds lasting 50,000-100,000 hours should be carried out to confirm the superior performance of FB2 weldments, for the expected lifetime span of power plant components. When more test data become available, alternative extrapolation methods could be employed.
- Further work to improve welding procedures and the room temperature toughness of the weld metals should be carried out, to make the steel more attractive and acceptable to industry.
7. Acknowledgements
The work was undertaken as part of TWI's Core Research Programme, funded by Industrial Members of TWI, with funding from the COST 536 project provided by the UK Technology Strategy Board. The authors are grateful to Messrs B Scarlin and M Staubli of ALSTOM, Switzerland for arranging provision of the experimental rotor steel, and initial cutting. The rutile flux-cored wire was supplied by Metrode Products Ltd, and the metal-cored wire by Air Liquide. The assistance of colleagues at TWI particularly R L Jones, C S Hardy, D Patten, T P Mitchell, C S Punshon, D Jefferies, J C Godden, S Fewell, M S Ali and I Hadley, is gratefully acknowledged.
8. References
Abe F, 2005: 'High performance creep resistant steels for 21st century power plants', Proc. of The 1st Conf. on Super High Strength Steels, Italy, Nov.
Abe F, Tabuchi M, Kondo M and, Tsukamoto S, 2007: 'Suppression of Type IV fracture and improvement of creep strength of 9Cr steel welded joints by boron addition', International Journal of Pressure Vessels and Piping, vol. 84, pp 44-52.
Abson D J, 2004: 'The effects of composition on the toughness and creep behaviour of weld metals for advanced high temperature 9-13%Cr steels', TWI Research Report for Industrial Members 788/2004, February.
Abson D J, 2005: ''The Influence of Ti, Al and Nb on the Toughness and Creep Rupture Strength of Grade 92 Steel Weld Metal', TWI Research Report for Industrial Members 833/2005, September.
Abson and Pargeter, 1986: 'Factors influencing the as-deposited strength, microstructure and toughness of manual metal arc welds for C-Mn steel fabrications', International Metals Reviews, 1986, 31 (4), 141-194.
Abson D J and Rothwell J S: 2009 'Mechanical Properties of High Productivity Weldments in the Advanced 9% Chromium Steel FB2', TWI Research Report for Industrial Members, 15914.02/2009/1375.3, September.
Abson D J, Rothwell J S and Woollin P, 2007: 'The Influence of Ti, Al and Nb on the Toughness and Creep Rupture Strength of Grade 92 Steel Weld Metal', Proc, IOM Conf. on Integrity of high temperature welds, 24-26 April 2007, London, IOM Communications, Editors D J Allen et al., 2007, 129-138. (See also Abson D J, 2005: 'The influence of Ti and Al on the toughness and creep rupture strength of Grade 92 steel weld metal', TWI Research Report for Industrial Members, 833/2005, Sept).
Barnes A, 2002: 'The effect of composition on the microstructural development and toughness of weld metals for for advanced high temperature 9-13%Cr steels', TWI Research Report for Industrial Members 744/2002, May.
Barnes A M and Abson D J, 2003: 'The effect of composition on the microstructural development and toughness of weld metals for for advanced high temperature 9-13%Cr steels', 2nd Int Conf Integrity of High Temperature Welds, IOM, London.
BS EN 1435: 1997 'Non-destructive examination of welds. Radiographic examination of welded joints', British Standards Institution, 1997.
BS EN 10002-1: 2001: 'Tensile testing of metallic materials. Method of test at ambient temperature', British Standards Institution, 2001.
BS EN 10002-5: 1992: 'Tensile testing of metallic materials. Method of test at elevated temperatures', British Standards Institution, 1992.
BS EN 10045-1: 1990: 'Charpy impact test on metallic materials. Test method (V- and U-notches)', British Standards Institution, 1990.
BS EN 10291: 2000 'Metallic materials. Uniaxial creep testing in tension. Methods of test', British Standards Institution, 2000.
BS 7448: Part 2: 1997: 'Fracture mechanics toughness tests. Method for determination of KIc, critical CTOD and critical J values of welds in metallic materials', British Standards Institution, 1007.
Cerjak H and Mayr P, 2008 'Creep strength of welded joints of ferritic steels' in Creep resistant steels, Abington Publishing, Great Abington, UK, Eds. F Abe, T-U Kern and R Viswanathan, 472-503.
Chovet, C, Galand E, Leduey B, 2008: 'Effect of various factors on toughness in P92 SAW weld metal', Welding in the World, Volume 52, Issue 7/8 July/August, IIW doc II-C341-07.
Danielsen H, 2007: 'Z phase in 9-12%Cr Steels'. PhD Thesis, The Technical University of Denmark.
Department for Business Enterprise and Regulatory Reform (replacing former Department of Trade and Industry (DTI)), April 2005, 'Guideline to PED (URN 05/1074).
Di Gianfrancesco A, Cipolla L, Venditti D, Neri S and Calderini M, 2008: 'Creep Behaviour and microstructural analysis of FB2 trial rotor steel', Proc of the Fifth International Conference on Advances in Materials Technology for Fossil Power Plants, Marco Island, FL ,USA, ASM International, pp. 366-376.
Electronic communication between Marc Staubli of Alstom and David Abson of TWI 17/8/2006.
ECCC data sheets 2005.
Fountain R W and Chipman J, 1962: 'Solubility and Precipitation of boron nitride in Iron-Boron alloys' Transactions of the Metallurgical Society of AIME, vol. 224, 599-605.
Kern T-U, Staubli M, Mayer K H, Donth B, Zeiler G and Di Gianfrancesco A, 2006: 'The European effort in development of new high temperature rotor materials - COST 536', in Materials for Advanced Power Engineering 2006, Proc. 8th Liege Conf, 18-20 September, Research Centre J?lich, Ed Liege, Belgium. Ed: J Lecomte-Beckers, M Carton, F Schubert and P J Ennis. Publ: D-52425 Julich, Germany; Forschungszentrum Julich GmbH; 2006, Vol.53, Part 2, 843-854.
Metrode Products Limited, 2007: 'Welding Consumables Technical Handbook', Rev 9.
Kimura K, Tabuchi M, Takahashi Y, Yoshida K and Yagi K, 2008: 'Long term creep strength and strength reduction factor for welded Joints of ASME Grades 91, 92 and 122 type steels', Proc IIW Int Conf on Safety and reliability of welded components in energy and processing industry, Gratz, Austria, July, pp.51-57.
Kondo M, Tabuchi M, Tsukamoto S, Yin F and Abe F, 2006: 'Suppressing Type IV failure via modification of heat affected zone microstructures using high boron content in 9Cr heat resistant steel welded joints', Science and Technology of Welding and Joining, 11(2), 216-223. Kozeschnik E and Holzer I, 2008: 'The development of creep resistance' in Creep resistant steels, Abington Publishing, Great Abington, UK, Eds. F Abe, T-U Kern and R Viswanathan, 305-349.
Mayer K-H and Masuyama F, 2008: 'The development of creep resistance' in Creep resistant steels, Abington Publishing, Great Abington, UK, Eds. F Abe, T-U Kern and R Viswanathan, 15-77.
Mayr P and Cerjak H 2007 'The heat affected zone of boron alloyed creep resistant 9% chromium steels and their susceptibility to Type IV Cracking.' Proc EPRI - 5th International Conf on Advances in Materials Technology for Fossil Power Plants, Marco Island, Florida, October 3 - 5.
Peel P, Scarlin B and Vanstone R, 2007: 'From materials development to advanced steam turbines', Proc. 7th Int. Charles Parsons Conf, Glasgow, IOM Communications Limited, 389-402.
Punshon C, 2008: 'Reduced Pressure Electron Beam Welding Development of a Prototype Local Vacuum System', TWI Research Report for Industrial Members 17241.01/2007/1321.3, March.
Rothwell J S, Punshon C S, Dance B G I and Abson D J , 'Practical and Experimental Applications of Electron Beam Welding for the Power Sector'. Proc IIW Int Conf Safety and reliability of welded components in energy and processing industry, Gratz, Austria, July.
Sakuraya K, Okada H and Abe F, 2006: 'BN type inclusions formed in ferritic heat resistant steel', Energy Materials, volume 1, No.3, pp158-166.
Suenaga K, Okazaki Y, Ishida H and Hidaka T, 2008: 'Influence of Oxides on Structure and Notch Toughness of Weld Metal with Flux-cored Wires for High Strength Steel', Proc of International Welding Symposium '8WS - Innovations in Welding and Joining for a New Era in Manufacturing' Japan Welding Society, Kyoto International Conference Center, Kyoto, Japan, November.
Zeiler G, Mayer W, Spiradek K and Wosik J, 2004: 'Experiences in manufacturing and long-term mechanical and microstructural testing on 9-12% chromium steel forgings for power generation plants', Proc and Fourth International Conference on Advances in Materials Technology for Fossil Power Plants Hilton Head Island, South Corolina USA, ASM International, pp. 506-519.
Zeilinska-Lipiec A, Spiradek-Hahn K, Bryla K and Czyrska-Filemonowicz A, 2007: 'Boron in high chromium steels for USC steam power plant'. Proc of the seventh Int. Charles Parsons Turbine Conf, Glasgow, UK, IOM communication 11-13 September, pp.489-498.
Zhang Z, Farrar J C M and Barnes A M, 2000: ''Weld metals for Grade 91 - tough enough?' Proc 4th EPRI Conf on 'Welding and repair technology for power plants', 7-9 June, Naples, Florida, USA.