Effects of degradation on the mechanical properties and fracture toughness of a steel pressure-vessel weld metal

Effects of degradation on the mechanical properties and fracture toughness of a steel pressure-vessel weld metal

International Journal of Pressure Vessels and Piping 80 (2003) 807–815 www.elsevier.com/locate/ijpvp Effects of degradation on the mechanical propert...

359KB Sizes 0 Downloads 83 Views

International Journal of Pressure Vessels and Piping 80 (2003) 807–815 www.elsevier.com/locate/ijpvp

Effects of degradation on the mechanical properties and fracture toughness of a steel pressure-vessel weld metal S.J. Wu, J.F. Knott* Department of Metallurgy & Materials, School of Engineering, The University of Birmingham, Edgbaston, Elms Road, Birmingham B15 2TT, UK Received 15 February 2001; revised 9 January 2003; accepted 10 January 2003

Abstract A degradation procedure has been devised to simulate the effect of neutron irradiation on the mechanical properties of a steel pressurevessel weld metal. The procedure combines the application of cold prestrain together with an embrittling heat treatment to produce an increase in yield stress, a decrease in strain hardening rate, and an increased propensity for brittle intergranular fracture. Fracture tests were carried out using blunt-notch four-point-bend specimens in slow bend over a range of temperatures and the brittle/ductile transition was shown to increase by approximately 110 8C as a result of the degradation. Fractographic analysis of specimens broken at low temperatures showed about 30% intergranular failure in combination with transgranular cleavage. Predictions have been made of the ductile –brittle transition curves for the weld metal (sharp crack) fracture toughness in degraded and non-degraded states, based on the notched-bar test results and on finite element analyses of the stress distributions ahead of the notches and sharp cracks. The ductile – brittle transition temperature shift ðDT ¼ 110 8CÞ between non-degraded and degraded weld metal at a notch opening displacement of 0.31 mm was combined with the Ritchie, Knott and Rice (RKR) model to predict an equivalent shift of 115 8C for sharp-crack specimens at a toughness level of 70 MN/m3/2. q 2003 Elsevier Ltd. All rights reserved. Keywords: Steel; Weld metal; Embrittlement; Degradation; Fracture toughness

1. Introduction It is well known that neutron irradiation can cause degradation to the mechanical properties of steels used in nuclear reactor pressure-vessels (RPVs). The shift in transition temperature of a pressure-vessel weld metal, DT; as a function of neutron irradiation, is governed by three factors [1]: (a) an increase in yield stress resulting from ‘matrix damage’ due to point-defect or cluster hardening, (b) an increase in yield stress due to Cu precipitatehardening, (c) a decrease in (local) fracture strength due to the segregation of P to grain boundaries. Since performing real irradiation tests is both complex and expensive, investigations have been made to try to simulate the effects of irradiation on material properties using an appropriate degradation procedure. It was initially suggested that simulation of the irradiation effect could be achieved * Corresponding author. Tel.: þ44-121-414-6729; fax: þ 44-121-4147468. E-mail address: [email protected] (J.F. Knott). 0308-0161/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijpvp.2003.01.003

through a specially designed heat treatment procedure [2]. The transition temperature shift (45 – 50 8C) in the work of Wiesner [2], however, results mainly from grain-coarsening during heat treatment rather than from matrix damage. The matrix damage from neutron irradiation usually results in an increase in yield stress sy and a decrease in work-hardening rate, ds=d1; whereas coarsened grain size (and carbide thickness) usually results in a decrease in sy and an increase in ds=d1: A simple grain-coarsening heat treatment is not, therefore, appropriate for simulating the effect of neutron irradiation. It should also be noted that, although P segregation to grain boundaries during temper embrittlement heat treatment can cause a decrease in local fracture strength sF ; it does not alter the sy and ds=d1 values as does matrix irradiation. Thus, it has little influence on the spread of strain, although fracture may occur at an earlier stage. There is another possible way to simulate the neutron irradiation effect. It has been shown that it is possible to simulate the macroscopic effects of point-defect (cluster) hardening on the yield stress and work-hardening characteristics by the application of cold prestrain. In fine detail,

808

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

the effects will not be identical to those of point defects, but it should simulate the effect of irradiation on sy and ds=d1 values more closely than alternative heat treatments. Work has been carried out on effects of cold prestrain on the shift of transition temperature, DT; and the local critical fracture stress, sF ; in wrought mild steel [3]; and on the lowering of the crack-tip opening displacement for the initiation of ductile fracture, di ; and the ductile fracture resistance curve in mild steel, HY80 and A533B [4,5]. Cold prestrain produces increases in sy and decreases in ds=d1: The present paper describes research carried out to explore the effect of a ‘degradation procedure,’ which combines cold prestrain and an embrittlement heat treatment. The aim is to simulate both the mechanical aspects of change in yield and flow properties and the effects of segregation that occur as a result of neutron irradiation.

2. Experimental 2.1. Materials The material used in this work was a double V multipass butt weld, designed to reproduce original Magnox RPV welds. The nominal section thickness of the weldment was 100 mm. The manufacturing procedure of this reproduction weld is given in Ref. [6], and the chemical analysis and a hardness survey have been carried out and presented in Ref. [7]. The chemical composition of the weld metal is given in Table 1. Material in the ‘as-received’ condition was supplied as machined notched testpieces cut from weld, which had experienced a stress-relieving postweld heat treatment at 600 8C for 6 h. A further block of weld was provided in the as-received condition and was used to manufacture ‘degraded’ specimens. The degradation procedure employed three steps: (a) a grain-coarsening heat treatment and temper embrittlement sequence (1150 8C, 4 h, furnace cool to 500 8C, 48 h, and then air cool to room temperature), (b) machining into single-edge-notch bend (SENB) specimen blanks, and (c) applying 8% uniform prestrain, along the S-direction (Fig. 1b), to the whole side of the specimen blanks. The application of prestrain was monitored using a clip-gauge. The amount of prestrain was determined by matching the yield strength to that of irradiated material, based on information provided by Lidbury [8]. The microstructures of the reproduction weld in both as-received and degraded states consisted of ferrite and pearlite. The average grain sizes were measured as 8 mm for as-received and 20 mm for degraded weld metals. This indicates that

grain growth occurred during heat treatment prior to prestraining. Subsidiary experiments showed that the direct cool to 500 8C did not produce any component of copper precipitation-hardening. 2.2. Specimens Hounsfield No. 13 tensile specimens were used to determine the basic tensile properties of the as-received and degraded welds, while blunt-notch four-point-bend specimens were chosen to determine the fracture stress sF for the Ritchie, Knott and Rice (RKR) [9] model. Specimen blanks were cut from the two weld blocks. The orientation and geometry of the blanks are shown in Fig. 1. Hardness measurements and optical observation showed that the first three ‘layers’ from the weld cap have consistent hardness values and similar microstructures. Thus, specimens from these three layers can be used in tests. The distance from the cap to the weld centre is about 37 mm as illustrated in Fig. 1a. In the actual experiments, specimens from the first two layers were used in both tensile and four-point bending tests, and some further specimens from the third layer of the degraded material were used in tensile tests. The geometry of the blunt-notch bend specimens is shown in Fig. 1c in which the notch root radius was 0.025 mm for the asreceived specimens and 0.2 mm (the design intent) for the degraded specimens. To meet the programme schedule, it was necessary to carry out tests on those specimens, despite the differences in radius. Elastic/plastic finite element stress analyses were obtained for both geometries [10] to enable the stress – strain distributions ahead of each notch to be calculated, given experimental values of yield stress and work-hardening exponent at the appropriate test temperature. There was a concern that the different radii might affect the sampling volume, but subsequent analysis of the values of critical local fracture stress (see later) gives confidence that this effect is not significant in the present case. 2.3. Mechanical tests To determine the basic mechanical properties of the materials, tensile tests were carried out at different temperatures using an Instron Testing Machine with a loading speed of 1 mm/min. The blunt-notch four-pointbending tests were carried out at different temperatures using a 200 kN ESH Hydraulic testing machine with a loading rate of 0.5 mm/min. The distance between the two loading points ðSmin Þ is 20 mm and the span between the two holding points ðSmax Þ is 50 mm. Curves of load versus Notch-Mouth Opening Displacement (NMOD) were

Table 1 Chemical composition of the reproduction weld metal [7] Element Weight (%)

C 0.10

Si 0.50

Mn 1.66

P 0.036

S 0.025

Cr 0.06

Ni 0.06

Cu 0.19

Sn 0.01

As ,0.005

Al 0.007

O 0.042

N 0.01

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

809

Fig. 1. (a) Orientation and (b) geometry of the specimen blanks, arrow indicates the prestrain direction (S-direction), and (c) geometry of the SENB specimens.

recorded. The nominal COD ðdnom Þ values were calculated according to the British Standard (BS 7448).

behaviour of the as-received weld metal. The temperature effect can therefore be considered to arise simply through the variation of the yield stress of the material. For the degraded weld metal, however, the shape of the stress – strain curves can be divided into two groups based upon the test temperature, i.e. (1) above 2 30 8C with an average strain hardening exponent ðNÞ of 27.3, and (2) below 2 30 8C with an average N of 17.6 (Table 2). Therefore, finite element analyses and predictions of the transition temperature shift need to incorporate different parameters above and below 2 30 8C. It should be noted that, as shown in Table 2, the values of ultimate tensile strength (UTS, suts ) for the degraded material are lower than those for the as-received material. This decrease is attributed to the grain growth during heat treatment prior to prestraining. Optical microscope observation reveals that the average grain size, d; of the degraded weld metal (20 mm) is larger than that of the asreceived weld metal (8 mm). Consider the Hall –Petch equation [13,14]

3. Test results and discussion 3.1. Tensile tests Tensile tests for as-received and degraded materials have been carried out at different temperatures. The test results are presented in Table 2, which show that the yield stress of the degraded material is about 10 – 15% higher than that of the as-received material. The strain hardening exponent, N; defined by the Ramberg –Osgood equation ð1=1y ¼ s=sy þ aðs=sy ÞN ; where a is a material constant), is also given in Table 2. Comparison of the values of yield strength with initial results being obtained by Bolton on Trawsfynydd cutouts [11] and with values quoted by Knott and English [12] suggest that the increase is equivalent to that which would be produced as matrix damage by a neutron dose of order 9 £ 1024 dpa (displacements per atom) at an irradiation temperature of 190 8C. Welds in service also experience an increase in yield strength due to copper precipitation hardening. It can be seen from Table 2 that the strain hardening behaviour of the weld metal after degradation is different from that in the as-received state. The stress – strain curves of the as-received weld metal have a similar shape at different temperatures with an average strain hardening exponent ðNÞ of 6.2. This indicates that the test temperature does not have a significant influence on the strain hardening

sy ¼ si þ ky d21=2

ð1Þ

with ky ¼ 0:7 MN=m3=2 (2.2 kg/mm3/2) [15]. From the yield stress (426 MPa) of the as-received weld metal at room temperature, the value of si can be calculated as 179 MPa, and the equivalent as-received yield stress for the coarse grain material is then deduced to be 336 MPa. The effect of prestrain is similar to that of unloading and reloading during a tensile test. While it changes the yield stress significantly, prestraining to a level less than that corresponding to the UTS of an as-received specimen

Table 2 Yield stresses, ultimate tensile stresses (MPa) and strain hardening exponents at different temperatures Test temperature (8C) Yield stress Ultimate stress Harden. exp., N

As-received Degraded As-received Degraded As-received Degraded

70

45

464

473

498

515

27.5

26.2

20 426 485 591 505 6.3 31.4

0

497 554 25

215 441 513 604 559 6.0 26.6

230

520 565 21.7

245 458 526 626 584 6.0 17.1

260

554 611 17.9

280

2100

2120

2196

507 574 672 630 6.4 18

542 600 707 668 5.9 17.3

604

939 1014 973 1039

735 6.8

810

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

does not have a large effect on the ultimate strength. The strain at the UTS for the as-received specimens was over 15% compared with 8% applied as cold prestrain. If it is assumed that for non-prestrained fine- and coarse-grained materials the term ðsuts 2 sy Þ remains constant, the ultimate strength of the degraded weld metal may be estimated as

This is close to the average experimental result (505 MPa) and supports the conclusion that the lower value of UTS in the degraded material is due to the coarser grain size. The significance of this, compared with effects on UTS of neutron irradiation, is discussed later.

facets were only occasionally found on the as-received fracture surfaces (Fig. 3a), there were about 30% intergranular facets co-existing with the cleavage facets on the fracture surfaces of the degraded specimens (Fig. 3b). This also contributes to the differences in ductile – brittle transition behaviour between the as-received and degraded weld metals. The intergranular failure on fracture surfaces of the degraded materials may be attributed to P-segregation to grain boundaries during the embrittlement heat treatment at 500 8C. Inclusions have been found to be responsible for the brittle cleavage fracture of both as-received and degraded weld metals (Fig. 4). The cleavage fracture was triggered by the fracture of an inclusion in the plastic zone ahead of the notch root, in agreement with previous observations [16].

3.2. Four-point-bend tests

3.3. Experimental estimation of DT shift

Eighty-one Charpy-size blunt-notch specimens ða=W ¼ 1=3Þ; 37 from as-received material and 44 from degraded material, have been tested under four-point-bend loading at different test temperatures. Typical curves of load ðPÞ versus NMOD at different temperatures for as-received and degraded materials are shown in Fig. 2. It can be seen that NMOD values at failure decrease with decrease of test temperature. Macroscopic observation showed the characteristics of brittle failure on all the fracture surfaces of the degraded specimens tested below 45 8C. For as-received material, specimens tested below 2 60 8C show brittle fracture at the macroscopic level, while those tested above 2 30 8C show ductile failure. Microscopic features of the fracture surfaces were examined in a JEOL 5410 Scanning Electron Microscope, operated at 20 kV. It was found that significant (. 0.2 mm) ductile tearing occurred prior to cleavage on fracture surfaces of as-received specimens tested at and above 2 60 8C, while it occurred at and above 45 8C for degraded specimens. This may be taken to imply a transition temperature shift DT of order 105 8C based on crack initiation toughness d0:2 : One difference in failure behaviour between degraded and as-received specimens results from their different yield stresses and strain hardening characteristics. Observations revealed that, whereas grain boundary

Based upon the curves of load versus NMOD obtained from the blunt-notch specimens, nominal toughness values, ðdnom Þ; can be calculated in a manner similar to that used in the British Standard (BS7448) [17] for fracture mechanics toughness ðdÞ tests. The variation of dnom with temperature is shown in Fig. 5. It should be noted that the effect of the degradation procedure is both to shift the bottom of the curve to higher temperatures and to decrease the level of the ‘upper shelf’. This closely parallels the effect of neutron irradiation on transition-curve behaviour in notched specimens (Ref. [5]). Although the absolute values of ductile – brittle transition temperature for the as-received and degraded weld metals are likely to differ from those determined from (sharp crack) fracture mechanics toughness specimens, the transition temperature shift ðDTÞ due to material degradation may be taken as a rough estimate of that based on fracture toughness values. The experimental results show that the tests at a temperature of 2 80 8C give a good dC2 -type trace (see Ref. [17]) for as-received material, which corresponds to a fracture toughness KIC value of about 70 MN/m3/2 [18]. Taking the average dnom of 0.31 mm at 2 80 8C as equivalent to KIC ¼ 70 MN=m3=2 ; as shown in Fig. 5, the ductile – brittle transition temperatures for as-received and degraded weld metals are about 2 80 and 30 8C, assuming

suts ðdegradedÞ ø sy ðcoarse grainÞ þ ðsuts ðas-receivedÞ 2 sy ðas-receivedÞÞ ¼ 501 MPa

Fig. 2. Load versus notch-mouth opening displacement curves of (a) as-received and (b) degraded weld metals.

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

811

Fig. 3. Fracture morphology of (a) as-received blunt-notch specimen tested at 2100 8C and (b) degraded blunt-notch four-point-bend specimen tested at 260 8C.

that the degradation procedure does not affect the relationship between nominal dnom and fracture toughness. Then, the estimated transition temperature shift DT is about 110 8C. This is more than twice as large as that obtained through a grain-coarsening heat treatment alone [2]. 3.4. RKR prediction The RKR model can be expressed analytically by using the Hutchinson [19], Rice and Rosengren [20] small scale yielding solution for the plane strain stress – strain distribution around a crack tip in a non-linear elastic material. By applying the RKR criterion that the stress intensity factor K ¼ KIC when the principal stress syy . sF over a characteristic distance x ¼ l0 directly ahead of the crack tip, the RKR expression for the fracture toughness can be written as [21] ððNþ1Þ=2Þ ððN21Þ=2Þ =sy Þ KIC ¼ b2ððNþ1Þ=2Þ l1=2 0 ðsF

ð2Þ

where

b ¼ f ðNÞðð1 2 n2 Þ=IÞ1=ðNþ1Þ

ð2aÞ

is the amplitude of the stress singularity, sy is the yield stress, n is Poisson’s ratio, N is the Ramberg – Osgood hardening exponent, I is a numerical constant, and f ðNÞ is the normalised angular function at u ¼ 0: The numerical values of b (Eq. (2)) corresponding to different strain hardening exponents N can be obtained as follows [21], for as-received weld metal, N ¼ 6:2; b ¼ 4:32; for degraded weld metal, below 2 45 8C, N ¼ 17:6;

b ¼ 3:4; at about 2 30 8C, N ¼ 21:7; b ¼ 3:3; above 2 15 8C, N ¼ 27:3; b ¼ 3:21: Thus, use of the RKR model for predicting the variation of the lower shelf KIC values with temperature involves determining the local fracture stress sF and employing a suitable characteristic distance l0 representative of the microstructure and fracture micromechanisms. 3.5. Local critical fracture stress sF Critical local fracture stress ðsF Þ values for both asreceived and degraded weld metals were determined using values of fracture load in combination with a finite element analysis (FEA) of the stress –strain distribution ahead of the notch root [18]. As shown above, the as-received weld metal has a similar strain hardening behaviour over the whole test temperature range, and that of the degraded weld metal is treated as falling into two regimes divided at 2 30 8C. The true-stress/true-strain curve at 2 80 8C for the as-received material has been employed in the FEA, whilst curves for degraded material obtained at 0 and 2 80 8C were used to represent the two different regimes. The results of the stress distribution analysis showed that the stresses were highest directly ahead of the notch root. The variation of the maximum principal stress normalised by yield stress, smp =sy ; with distance below the notch is plotted in Fig. 6 for various loads. It can be seen that the largest (peak) maximum principal stress occurs within the plastic zone ahead of the notch root. The largest value of smp =sy at each applied load ratio is shown in Fig. 7. The load ratio is designated as Papp =PGY ; where the general yield load PGY is

Fig. 4. Cleavage fracture initiator (inclusion) on fracture surfaces of (a) as-received specimen tested at 2100 8C and (b) degraded specimen tested at 280 8C.

812

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

Fig. 5. Variation of nominal notch root opening displacement ðdnom Þ with temperature. Dotted line corresponds to KIC of 70 MN/m3/2.

calculated from the yield stress and dimensions of the testpiece and the loading-span: in the present case, PGY ¼ 21sy (where the number 21 has dimensions). Assuming that the largest maximum principal stress corresponding to the failure load of the specimen is the critical local fracture stress, sF ; the sF values for both degraded and as-received materials at various temperatures can then be obtained, based on the failure load of the blunt-notch four-point-bend specimens, and the value of yield stress at the test temperature. Test results showed that the sF values were achieved at about 100 mm ahead of the notch root for the asreceived specimens and at about 550 mm for the degraded specimens, which coincides with the location of the cleavage fracture initiators (inclusions) such as those shown in Fig. 4. The variation of the local fracture stress with temperature is demonstrated in Fig. 8. It is clear that sF values are almost independent of test temperature when the temperature is comparatively low, below 2 60 8C for as-received weld metal and below 2 10 8C for degraded weld metal. Above these temperatures, the sF values decrease slightly as temperature increases. This may be related to the plastic zone size ahead of the notch root, which becomes larger at higher temperature. The probability of sampling a particularly large inclusion in a high stress region is then also higher. Based on the Griffith expression for fracture stress, as interpreted by Curry and Knott [22,23] and Tweed and Knott [16], cleavage fracture initiated at large inclusions can occur at a lower local fracture stress. For a penny-shaped microcrack, the Griffith criterion can be written as 2

sF ¼ ½pEgp ð1 2 n ÞC0 

1=2

ð3Þ

where gp is the ‘effective surface energy’, E is Young’s modulus, n is Poisson’s ratio, and C0 is the diameter of the penny-shaped microcrack. Clearly, a larger C0 will result in a lower sF : Since the sizes of the cleavage-initiating inclusions in the highly stressed test volume may vary from specimen to specimen, scatter in the sF and KIC values at a given temperature is possible. Statistical analysis of the sF

Fig. 6. The variation of the maximum principal stress directly ahead of the notch root of the blunt-notch four-point-bend specimens. (a) As-received specimens (notch root radius r ¼ 0:025 mm), (b) degraded specimens ðr ¼ 0:2 mmÞ above 230 8C, and (c) degraded specimens ðr ¼ 0:2 mmÞ below 230 8C.

values, however, shows that both sets of results conform to a normal distribution. The means are 1560 MPa for asreceived material and 1418 MPa for the degraded condition, but, in both conditions, the standard deviation is small: 35 and 33 MPa, virtually identical for the two conditions and only 2.3 and 2.4% of the respective means. This is no greater than that expected from random experimental errors. It is therefore concluded that the different notch radii do not, in these experiments, affect the sampling of crack nucleation sites (inclusions).

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

813

transgranular cleavage in mild steel, sF and gp values increase with the amount of prestrain. The difference between these observations and the current findings may be attributed to the intergranular failure resulting from the additional embrittlement heat treatment, which causes grain-coarsening and P segregation to grain boundaries, which reduces the local work of fracture. Measurements of the diameters of cleavage-initiating inclusions such as that shown in Fig. 4 are about 2.2 mm for as-received specimens and 2.5 mm for degraded specimens. From Eq. (3) together with the values of sF (Fig. 8), the effective surface energy gp at fracture can be calculated as 7.2 J/m2 for as-received specimens and 6.9 J/m2 for degraded specimens. This is a 4% decrease compared with a 25% increase after 8% prestrain as found in Ref. [3]. For 8% prestrain, a 25% increase would, for the present steel, give a value for gp of approximately 9 J/m2. This implies that the effect of the intergranular embrittlement heat treatment is greater than that of prestrain on sF ; due to the reduction in gp : The sF values chosen for the RKR analysis in the present work are the average value below 2 70 8C for the asreceived material and the average value below 2 60 8C for the degraded material, i.e. sF ¼ 1560 MPa for as-received, and sF ¼ 1418 MPa for degraded material. The value for as-received material is similar to those quoted in Ref. [24]. 3.6. Characteristic distance, l0

Fig. 7. The variation of the largest maximum principal stress with applied load for (a) as-received, (b) degraded (.230 8C) and (c) degraded (,230 8C) specimens.

In Fig. 8, it can also be seen that the sF values for the degraded material are lower than those for the as-received material. A different trend for sF values with prestrain was obtained by Groom and Knott [3] who showed that, for

Fig. 8. Test results of fracture stress and yield stress test temperature. D— degraded material, U—undegraded (as-received) material.

The RKR analysis relates the characteristic distance l0 to the stress intensification ðsmp =sy Þ ahead of the crack tip. In the original paper, it was equated microstructurally to approximately two grain diameters for a sharp crack. Later, Curry and Knott [22] found no simple relationship between the critical distance and the grain size as such. The dimension must be regarded as essentially an empirically obtained quantity, although reflecting the spatial distribution of fracture-initiating particles [23]. Microscopical observations in the present work have revealed that cleavage fracture was triggered by cleaved inclusions (Fig. 4) in agreement with the findings in Ref. [16]. Since the inclusion distribution is not significantly changed by heat treatment and prestraining, the same characteristic distance value was chosen for both degraded and as-received weld metals. Note that, if fracture had initiated at carbides, this assumption would be unlikely to hold, because the coarser grain size associated with the degraded condition would be associated with coarser, and hence more widely spaced, carbides: see Ref. [22]. Rice and Johnson [25] and Tracey [26] have analysed the stress distribution ahead of a sharp crack in plane strain under ‘small scale yielding’ conditions for powerlaw hardening material. Based on their finite element solutions for the stress distribution function f ðsyy =sy ; NÞ in

814

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

the following equation

syy ðxÞ x ¼ fð ; NÞ 2 sy ðKI =sy Þ

ð4Þ

the characteristic distance x ¼ l0 can be determined for a material, if its fracture toughness KIC or COD, yield stress sy ; and strain hardening exponent N are known. For the purpose of determining the characteristic distance l0 ; three fracture toughness specimens of degraded material were tested at 0 8C and an average COD value of 0.05 mm was obtained. The characteristic distance l0 was then determined as 0.5 mm and this same value was assumed to hold for both as-received and degraded weld metals. 3.7. RKR predicted KIC values Using Eq. (2), KIC values can be calculated using the different values of yield stress and hardening exponent at different temperatures for both as-received and degraded materials. As fractographic observations revealed, there is obvious ductile tearing (. 0.2 mm) involved in the fracture of as-received specimens tested above 2 60 8C and of degraded specimens above 45 8C, which corresponds to a KIC value of 80 MN/m3/2. The curves for the predicted fracture toughness KIC versus temperature (Fig. 9) above 2 60 8C for as-received material and above 45 8C for degraded material are therefore affected by the onset of the ductile tearing prior to cleavage failure. As a result, the RKR prediction of the transition temperature shift, DT; when KIC . 80 MN=m3=2 ; deviates significantly from the actual behaviour. Up to KIC ¼ 80 MN=m3=2 ; however, the RKR prediction (Fig. 9) should give a reliable result since ductile tearing was not involved in fracture. The ductile – brittle transition temperature shift DT at 70 MN/m3/2 between as-received weld metal and the degraded weld metal is predicted as 115 8C. This is in close agreement with a substantial data-set of fracture toughness test results [18] and with our own experimental estimation ðDT ¼ 110 8CÞ;

Fig. 9. RKR prediction of the transition temperature shift ðDTÞ between asreceived and degraded materials. Broken lines show where the RKR predictions no longer hold. The open and filled circles represent test results for as-received and degraded materials, respectively.

as shown in Fig. 5, based on the equivalence of dnom ¼ 0:31 mm (for the notch) and KIC ¼ 70 MN=m3=2 (for a sharp crack). Note that, although the degradation treatment has increased the yield strength by an amount comparable with that induced by neutron irradiation, the requirement to generate an amount of intergranular fracture has reduced the UTS (by increasing the grain size). In irradiation surveillance samples, both yield stress ðsy Þ and UTS ðsuts Þ are observed to increase, although the sy =suts ratio also increases. In the present work, the change in shape of the stress –strain curve from as-received to degraded material is indicated by both the increase in sy =suts ratio and the increase in N-value (Table 2). The sy =suts ratio is greater than that observed for irradiated surveillance samples. By definition, the UTS relates to the onset of plastic instability in a ductile, tensile testpiece and, as such, is not relevant to the conditions pertaining to cleavage fracture in a region ahead of a blunt notch or sharp crack where the plastic strains are only slightly higher than the elastic strain. The local tensile stresses generated in such regions do depend on the work-hardening rate (as well as yield strength), most critically on the hardening rate for strains just greater than the yield strain. High hardening rates (low N values) allow high local stresses to be developed for low strains. It could therefore be argued that the present degradation treatment might underestimate effects on transition shift for cleavage fracture (on grounds of work-hardening alone) although this should be balanced against the associated decrease in sF : The net shift (, 110 8C) is greater than that produced by prestrain alone (, 60 8C at general yield)) [3] or graincoarsening alone (, 50 8C) [2].

4. Conclusions 1. Test results showed that the degradation procedure applied to the weld metal produced an increase in yield strength and a decrease in work-hardening rate ds=d1: The magnitudes of these effects are similar to those which would be expected for the matrix damage associated with a level of neutron irradiation of order 9 £ 1024 dpa at an irradiation temperature of 190 8C. 2. The blunt-notch four-point-bending test results give a rough estimate of the transition temperature shift ðDTÞ between as-received and degraded weld metals as 110 8C, based upon the nominal crack (notch) opening displacement calculated from curves of load against notch-mouth-opening-displacement. 3. The critical local fracture stress ðsF Þ values are almost independent of test temperature when the test temperature is comparatively low, below 2 60 8C for as-received weld metal and below 2 10 8C for degraded weld metal. Above these temperatures, the sF values decrease slightly as temperature increases. An area fraction of approximately 30% intergranular facets was observed on

S.J. Wu, J.F. Knott / International Journal of Pressure Vessels and Piping 80 (2003) 807–815

the fracture surfaces of the degraded weld metal. This reduces the value of sF by about 10%. 4. There is significant ductile tearing (. 0.2 mm) prior to cleavage on fracture surfaces of as-received specimens tested above 2 60 8C and of degraded specimens above 45 8C. Therefore, the cleavage-based RKR model can only predict fracture toughness values up to these temperatures, corresponding to a maximum KIC level of 80 MN/m3/2. 5. At 70 MN/m3/2, the RKR predicted ductile – brittle transition temperature shift ðDTÞ between as-received and degraded weld metals is 115 8C, which is similar to that observed in the fracture toughness test results obtained in Ref. [18] and to the experimental estimate of 110 8C for blunt-notches at dnom ¼ 0:31 mm: Note that the value dnom ¼ 0:31 mm for as-received material corresponds to a test temperature of 2 80 8C for which KJC has been calculated as approximately 70 MN/m3/2.

Acknowledgements The authors would like to thank BNFL Magnox Generation and AEA Technology for the supply of the reproduction weld metal. Dr D.P.G. Lidbury and Dr W.G. Xu are gratefully acknowledged for valuable discussions and provision of private information.

References [1] Knott JF. Characteristic microstructural features of different types of fracture. In: Proceedings of the Second Griffith Conference on Micromechanisms of Fracture and their Structural Significance, The Institute of Materials; 1995. p. 3– 14. [2] Wiesner, CS. Validity of crack arrest arguments for magnox RPVs, TWI report 220422/1/95; 1995. [3] Groom JDG, Knott JF. Cleavage fracture in prestrained mild steel. Met Sci 1975;9:390 –400. [4] Thompson HE, Knott JF. Fracture control of engineering structures. In: Proceedings of ECF6; 1986. p. 1737–49. [5] Knott JF. Effects of microstructure and stress-state on ductile fracture in metallic alloys. In: Proceedings of ICF7; 1989. p.125– 38.

815

[6] Billing P. Manufacture and testing of a magnox reactor pressure vessel weld. Nuclear Electric Contract NE/BED/TD25247; 1991. [7] Abson DJ, Jones RL. Comparison of original magnox welds and subsequent (full-size) reproduction welds. TWI report 29259/2/94; 1994. [8] Lidbury DPG. Private communication; 1998. [9] Ritchie RO, Knott JF, Rice JR. On the relationship between critical tensile stress and fracture toughness in mild steel. J Mech Phys Solids 1973;21:395 –410. [10] Xu WG. Private communication; 1998. [11] Bolton CJ. (BNFL/Magnox) Private communication; 2000. [12] Knott JF, English CA. Views of TAGSI on the principles underlying the assessment of the mechanical properties of irradiated ferritic steel reactor pressure vessels. Int J Pressure Vessel Piping 1999;76: 891–908. [13] Hall EO. Proc Phys Soc 1951;64B:747. [14] Petch NJ. J Iron Steel Inst 1953;174:25–8. [15] Knott JF. Strength and toughness of steels. In: Advances in the Physical Metallurgy and Applications of Steels, The Institute of Materials; 1982. p. 181–98. [16] Tweed JH, Knott JF. Micromechanisms of failure in C –Mn weld metals. Acta Metall 1987;35:1401–14. [17] British Standard Fracture Mechanics Toughness Tests, BS 7448; 1991. [18] Beardsmore DW, Lidbury DPG, Sherry AH, Burdekin FM, Xu WG, Burstow MC, Howard IC, Knott JF, Wo SJ. Constraint based methodology for assessing transition temperature behaviour in fullsize components. AEAT report PC/GNSR/5025; 1999. [19] Hutchinson JW. Singular behaviour at the end of a tensile crack in a hardening material. J Mech Phys Solids 1968;16:13 –31. [20] Rice JR, Rosengren GF. Plane strain deformation near a crack tip in a power-law hardening material. J Mech Phys Solids 1968;16:1–12. [21] Ritchie RO, Server WL, Wullaert RA. Critical fracture stress and fracture strain models for the prediction of lower and upper shelf toughness in nuclear pressure vessel steels. Met Trans 1979;10A: 1557– 70. [22] Curry DA, Knott JF. Effects of microstructure on cleavage fracture stress in steel. Met Sci 1978;12:511 –4. [23] Curry DA, Knott JF. Effect of microstructure on cleavage fracture toughness of quenched and tempered steels. Met Sci 1979;13: 341–5. [24] Reed PAS, Knott JF. Investigation of the role of residual stresses in the warm prestress (WPS) effect. Fatigue Fract Engng Mater Struct 1996;19:485 –90. [25] Rice JR, Johnson MA. The role of large crack tip geometry changes in plane strain fracture. In: Inelastic behaviour of solids; 1970. p. 641 – 72. [26] Tracey DM. Finite element solutions for crack tip behaviour in small scale yielding. J Engng Mater Technol 1976;98:146–51.