Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more ➡
Download
Standard view
Full view
of .
Add note
Save to My Library
Sync to mobile
Look up keyword
Like this
4Activity
×
0 of .
Results for:
No results containing your search query
P. 1
Effects of Weld Preheat Temperature

Effects of Weld Preheat Temperature

Ratings: (0)|Views: 366|Likes:
Published by hdyoon3379891

More info:

Published by: hdyoon3379891 on Jun 06, 2011
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See More
See less

11/05/2012

pdf

text

original

 
Effects of weld preheat temperature and heatinput on type IV failure
 J. A. Francis
*
1
, G. M. D. Cantin
2
, W. Mazur 
2
and H. K. D. H. Bhadeshia
3
Type IV cracking refers to the premature failure of a welded joint due to an enhanced rate of creepvoid formation in the fine grained or intercritically annealed heat affected zone. A great deal ofresearch effort has been directed at understanding the underlying mechanisms for this type offailure, but most have approached the problem from a metallurgical standpoint, andcomparatively little effort has been directed at understanding the effects of welding variables.Here the effects of parameters such as the preheat temperature and heat input on the tendencyfor type IV failure in 9–12%Cr steels have been quantitatively estimated. These calculations havesubsequently been verified experimentally to form the first systematic study of weldingparameters on type IV cracking. The joint geometry and preheat temperature have been foundto ameliorate type IV failures, while the effect of heat input is less significant.
Keywords:
Creep, Heat affected zone, Steel, Type IV, Welding parameters, Welding procedures
Introduction
It clearly is useful to enhance the efficiency of fossil firedpower plants, and significant gains can be realisedthrough increases in steam temperature and pressure.This requires better steels capable of sustaining theharsher conditions and there is considerable work inprogress to design alloys for service temperatures andsteam pressures of 630
u
C and 35 MPa respectively.
1,2
Whereas the overall properties of such alloys are clearlyimportant, welding introduces localised changes in thestructure which become the life limiting factor. Inparticular, type IV cracking is a phenomenon in whichthere is an enhanced rate of creep void formation in thefine grained and intercritically annealed heat affectedzone (HAZ).
3
A number of researchers have studied the metallurgyof type IV failures in ferritic power plant steels.
4–11
Theauthors have approached the problem from a differentperspective, by interpreting the results of publishedcross-weld creep tests using neural networks in aBayesian framework.
12
This analysis identified quantita-tively and for the first time the effects of many weldingparameters, including the heat input and preheattemperature, on the tendency for type IV failure. As aresult of this analysis, a programme of creep testing wasinitiated that systematically interrogated the influence of these parameters on type IV cracking tendency. Theresults of the experiments support the findings of theneural network analysis, so that the authors are nowable to report the effects of preheat temperature andheat input on type IV failures in 9–12%Cr steels.After a brief description of the predictions from theauthors’ published neural network model, this paperdescribes the experimental programme initiated to assessthe predictions, and concludes with a discussion of theimplications of the work on weld design and integrity.
 Analysis of published data
Details of the Bayesian neural network method havebeen described elsewhere,
13,14
but it is important to notethat the technique is associated with two measures of uncertainty. When conducting experiments, the noiseresults in a different output for the same set of inputswhen the experiment is repeated. This is because thereare variables which are not controlled, so their influenceis not included in the analysis. The second kind dealswith the uncertainty of modelling; there might existmany mathematical functions which adequately repre-sent the same set of empirical data but which behavedifferently in extrapolation. This latter uncertainty isuseful in assessing the significance of calculations carriedout in input domains where knowledge is sparse.The data used to create the original model wereobtained from Refs. 4 to 9. It is the rupture stress that ismodelled as a function of the test temperature, creep lifeand welding parameters. The justification for usingthe rupture stress as the output is mathematical,
15
i.e.because it always has a finite value, whereas the creeplife tends to infinity as the stress is reduced. Thedatabase on which the model was created is summarisedin Table 1, consisting of data from 53 experiments inwhich type IV failure occurred in 9–12%Cr steels and forwhich details of the composition of the steel, the heattreatment, welding parameters, post-weld heat treatmentand creep test conditions had all been reported.
1
School of Materials, University of Manchester, Grosvenor Street,Manchester M1 7HS, UK
2
CSIRO Materials Science and Engineering, Australia
3
Materials Science and Metallurgy, University of Cambridge, PembrokeStreet, Cambridge CB2 3QZ, UK
*
Corresponding author, email john.francis@manchester.ac.uk
ß
2009 Institute of Materials, Minerals and Mining Published by Maney on behalf of the InstituteReceived 16 December 2008; accepted 27 January 2009
DOI 10.1179/136217109X415884
Science and Technology of Welding and Joining 
2009
VOL
14
NO
5
436
 
The significance perceived by the neural network, of each input variable in explaining changes in the cross-weld creep strength, is shown in Fig. 1. This quantity isanalogous to a partial correlation coefficient, and itshould be noted that it does not indicate the sensitivityof the output to the input. The logarithm of the rupturelife (log
t
r
) and test temperature have reasonably beenidentified as having the strongest correlations with therupture stress. The well known effects of normalisingtemperature,
15
tempering temperature and tungsten con-tent
2
have also been correctly identified. Interestingly,the preheat temperature during welding has also beenperceived to have a significant effect on rupture stress,but not so the weld heat input. The sign of the effect isindicated in Fig. 1 for each variable that has beenperceived to have a high significance. In summary,increases in preheat temperature are predicted totranslate into an increase in the rupture stress for agiven creep life and test temperature.
Cross-weld creep tests
All previous individual studies on type IV cracking havekept the welding conditions constant but the authorswere able to perceive, using the neural network, theeffects of these conditions by combining data fromdifferent sources. This is the first time that a programmeof cross-weld creep testing was initiated to investigatesystematically the effects of welding parameters.
Welding procedures
A section of P91 pipe supplied in the normalised andtempered condition, with an outer diameter of 356 mmand a wall thickness of 53 mm, was used as the basematerial in all of the welded joints in this study. Thismaterial was cut so that in all cases a weld was formedbetween two 180 mm long sections of pipe. The chemicalcomposition of this pipe section was measured by opticalemission spectroscopy and is summarised in Table 2.The welding experiments were planned in parallel withthe compilation of the neural network database;
12
theauthors decided in the first instance to follow themanufacturer’s recommendations
16,17
relating to preheattemperature. As such, four welded joints were manu-factured to investigate the effects of heat input and jointpreparation on type IV failures, noting that an effect of the joint preparation angle has been reported.
6
At a laterstage, the results of the neural network analysis inspiredthe manufacture of a fifth joint, in order to confirm theinfluence of preheat temperature on type IV limitedcreep life.The welding parameters used to fabricate each of thefive joints are summarised in Table 3. A single Veepreparation with an included angle of 30
u
was used forthe welds numbered 1–3, as is illustrated in Fig. 2
a
.Weld no. 4 (Fig. 2
b
) incorporated a single bevel at 45
u
to assist in interrogating the effect of the joint pre-paration angle. Joint no. 5 was fabricated to test animproved welding procedure, utilising a higher preheat
Table 1 Range in composition, heat treatment, weldingparameters and test conditions covered by 53sets of published data on type IV failures inRefs. 4–9
 Variable Min. Max.
C, wt-% 0
?
09 0
?
13N, wt-% 0
?
041 0
?
078B, wt-% 0 0
?
003Cr, wt-% 8
?
45 12
?
0Mo, wt-% 0
?
34 0
?
96Nb, wt-% 0
?
05 0
?
13W, wt-% 0 2
?
21Mn, wt-% 0
?
40 0
?
81Si, wt-% 0
?
02 0
?
35Cu, wt-% 0 3Normalising temperature,
u
C 1050 1080Normalising time, h 0
?
5 2Tempering temperature,
u
C 760 820Tempering time, h 1 6Heat input, kJ mm
2
1
0
?
8 3
?
8Preheat temperature,
u
C 100 250PWHT temperature,
u
C 740 760PWHT time, h 0
?
25 8Internal pressure test (0/1) 0 1Test temperature,
u
C 600 700Test duration, h 113 11 220Rupture stress, MPa 40 150
1 Significance of each input variable in explaining varia-tion in type IV rupture stress in cross-weld creep testson 9–12%Cr steels, as perceived by Bayesian networktrained on data in Refs. 4–9: perceived significance isanalogous to partial correlation coefficient, but doesnot necessarily indicate magnitude of effect; sign ofeffect for input variables that were perceived to havesignificant correlation with rupture stress is indicatedin figure; alloying additions are in wt-% and refer tobase metalTable 2 Composition of base material used in current work as determined by optical emission spectroscopy, wt-%
C Si Mn P S Al Cr Ni Mo V Nb N
0
?
09 0
?
28 0
?
40 0
?
01 0
?
01
,
0
?
005 8
?
5 0
?
15 0
?
89 0
?
21 0
?
06 N/A
Francis et al.
Effects of weld preheat temperature and heat input on type IV failure
Science and Technology of Welding and Joining 
2009
VOL
14
NO
5
437
 
temperature (350
u
C) and incorporating a joint prepara-tion with an included angle of 10
u
(Fig. 2
c
). It isimportant to note the way in which the joint preparationangle has been defined in Fig. 2
a
and
c
, i.e. as theincluded angle within the Vee preparation, rather thanthe bevel angle that has been applied to either side of the joint.In all cases, one sided welding was carried out in asemimechanised manner, with the pipe being rotatedduring welding, while the torch was maintained in thedownhand position for all weld passes. Preheating wascarried out with resistance heating blankets and inter-pass temperatures were generally maintained within20
u
C of the specified preheat temperature. After weld-ing, it was recommended that the entire joint be allowedto cool to a temperature below 100
u
C to complete thetransformation to martensite. A post-weld heat treat-ment procedure was necessary, and all welds were heattreated at 760
u
C for 2 h.For joints 1–4, the root runs and one additional ‘hotpass’ were made by manual metal arc welding(MMAW). The filling passes were completed usingflux-cored arc welding (FCAW) in all cases. For joint 5,gas tungsten arc welding (GTAW) was used for the rootpass and hot pass, while FCAW was used for fillingpasses. In all cases, the weld filler metal was chosen tomatch the composition of the base material.
Extraction of specimens
Fourteen creep specimens were extracted from the fivewelded joints. In all cases, the specimens had a gaugediameter of 11
?
3 mm and a gauge length of 62 mm(Fig. 3). Care was taken to ensure that all specimenswere extracted exclusively from the region of the weldcorresponding to the filling passes, which were depositedusing FCAW in all of the joints. Table 3 lists thespecimens that were extracted from each joint, andFig. 2 provides a schematic representation of thelocation from which each specimen was extracted.Most specimens had only one HAZ within the gaugelength, although for two specimens, there were twoHAZs within the gauge length (specimens 11 and 12 in
Table 3 Summary of cross-weld creep testing conditions in current work
Joint no.(Fig. 3)Welding process(root/fill)Heat input,kJ mm
2
1
Preheat temperature,
u
CCreep specimenIDJoint angleincluded,
u
Test temperature,
u
CTest stress,MPa
1 MMAW/FCAW 0
?
8 250 1 30 620 932 30 620 812 MMAW/FCAW 1
?
6 250 3 30 620 934 0
*
620 933 MMAW/FCAW 2
?
4 250 5 30 620 936 30 620 814 MMAW/FCAW 1
?
6 250 7 0 620 938 90 620 935 GTAW/FCAW 1
?
6 350 9 0
*
620 9310 0
*
620 8111 10 620 9312 10 620 8113 30
*
620 9314 30
*
620 81
*
An equivalent single Vee included angle that was achieved by rotating the axis of the creep specimen relative to the axis of the pipe(Fig. 2).
2 Schematic representation of joint preparations for eachweld in current work, and locations of extracted creepspecimens: with exception of samples 11 and 12(Table 3), all specimens had only one HAZ withingauge length; samples 11 and 12 had two HAZs withingauge length; in all cases, the HAZs were
y
15 mmfrom nearest shoulder of specimen; note that differentorientations between HAZ and loading axis of creepspecimen could be achieved by rotating axis ofextracted specimen with respect to axis of pipe3 Configuration of creep test specimens in current work:all dimensions are given in mm
Francis et al.
Effects of weld preheat temperature and heat input on type IV failure
Science and Technology of Welding and Joining 
2009
VOL
14
NO
5
438

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->