atigue Fracture of Engineering Materials Structures doi10.11114602695201001496X Assessment of creep rupture properties for dissimilar steels welded joints between T92 and HR3C YI GONG, JIAN CAO, 1 LI-NA Jl, 1 CHAO YANG, CHENG YAO, 1 ZHEN-GUO YANG, JUN WANG, I XIAO-MING LUO, 2 FU-MING GU, 2 AN-FANG QL SHANG-YUN YE and ZHENG-FEI HU4 I Department of Materials Science, Fudan University, Shangbai 200433, PR China, Shangbai Institute of Special equipment Inspection d Technical Research, Shanghai 200062, PR Cbina, Shangbai Boiler Works Ltd, Shangbai 200245, PR China, +School of Materials Science and Engineering, Tongji University, Shangbai 200092, PR Chim Received in final form 4 May 2010 ABSTRACT Dissimilar steels welded joints, between ferritic steel and austenitic stainless steel, are always encountered in high-temperature components in power plants. As two new grade ferritic steel and austenitic stainless steel, T92(9Cr0.5Mo2WVNb)and HR3 TP310HCbN), exhibit superior heat strength at elevated temperatures and are increas- ingly applied in ultra-supercritical (USC) plants around the world, a complete assessment of the properties for T92/HR3C dissimilar steels welded joints is urgently required. In this paper, metallographic microstructures across the joint were inspected by optical mi- roscope. Particularly, the creep rupture test was conducted on joints under different oad stresses at 625C to analyse creep strength and predict their service lives, while their fractograph were observed under scanning electron microscope. Additionally, finite element method was employed to investigate residual stress distribution of joints. Results showed that the joints were qualified under USC conditions, and T92 base material was commonly the weakest part of them Keywords creep rupture; dissimilar steels welded joints; finite element method; HR3C INTRODUCTION research. Consequently, heat-resistant steels, mainly re- Vith the worsening of global energy crisis an ferring to ferritic steels and austenitic stainless steels, have ronmental pollution, higher energy utilization and lower been pursued simultaneously since the emergence of USC CO2 emission are presently the two prior criteria for de- boilers for applications in boiler components, including signing fossil power plants, ' which will still serve as the superheater, reheater, header, turbine, steam piping and dominant energy form for at least 20 years. 2-As for fossil so on. Subsequently, for the sake of reliable services in power boilers, it is well known that the enhancement of USC boilers, a wealth of research has been carried out steam parameters can facilitate improvement in thermal on these heat-resistant steels as F12(X20CrMoV12. 1), efficiency, which can result in reduction in not only the fossil costs but also the COz, SO2 and NOx emission. vestigate characteristics of their base materials and perfor- Concretely, compared with supercritical(SC) boilers(24 mance deterioration after long-term services. 1-29 Among MPa, 565oC), modern ultra-supercritical (USC)boilers them, the T92(9Cr0 5Mo2WVNb), approximately simi- operating around 30 MPa and 600 C can lead to both lar to T9I but with a little modification in chemical com- an increase of 8-10% in thermal efficiency(from 37% positions for preferable high temperature properties than to 45-47%)and a reduction of 20-25% in COz emis- T91, and the HR3C (TP310HCbN) are the two typical representatives of new grade ferritic steels and austenitic Development of USC boilers with increasingly highe stainless steels for their superior heat strength above prospect in forthcoming USC boilers. However, appli- cations of T92 and HR3C are currently a bit constrained Correspondence:Z.-G.Yang.E-mail:zgyang@fudan.edu.cn due to the lack of sufficient literature and experience on @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
doi: 10.1111/j.1460-2695.2010.01496.x Assessment of creep rupture properties for dissimilar steels welded joints between T92 and HR3C YI GONG, 1 JIAN CAO, 1 LI-NA JI, 1 CHAO YANG, 1 CHENG YAO, 1 ZHEN-GUO YANG, 1 JUN WANG, 1 XIAO-MING LUO, 2 FU-MING GU, 2 AN-FANG QI, 3 SHANG-YUN YE 3 and ZHENG-FEI HU4 1Department of Materials Science, Fudan University, Shanghai 200433, PR China, 2Shanghai Institute of Special Equipment Inspection & Technical Research, Shanghai 200062, PR China, 3Shanghai Boiler Works Ltd., Shanghai 200245, PR China, 4School of Materials Science and Engineering, Tongji University, Shanghai 200092, PR China Received in final form 4 May 2010 ABSTRACT Dissimilar steels welded joints, between ferritic steel and austenitic stainless steel, are always encountered in high-temperature components in power plants. As two new grade ferritic steel and austenitic stainless steel, T92 (9Cr0.5Mo2WVNb) and HR3C (TP310HCbN), exhibit superior heat strength at elevated temperatures and are increasingly applied in ultra-supercritical (USC) plants around the world, a complete assessment of the properties for T92/HR3C dissimilar steels welded joints is urgently required. In this paper, metallographic microstructures across the joint were inspected by optical microscope. Particularly, the creep rupture test was conducted on joints under different load stresses at 625 ◦C to analyse creep strength and predict their service lives, while their fractograph were observed under scanning electron microscope. Additionally, finite element method was employed to investigate residual stress distribution of joints. Results showed that the joints were qualified under USC conditions, and T92 base material was commonly the weakest part of them. Keywords creep rupture; dissimilar steels welded joints; finite element method; HR3C; T92. INTRODUCTION With the worsening of global energy crisis and environmental pollution, higher energy utilization and lower CO2 emission are presently the two prior criteria for designing fossil power plants,1 which will still serve as the dominant energy form for at least 20 years.2–4 As for fossil power boilers, it is well known that the enhancement of steam parameters can facilitate improvement in thermal efficiency, which can result in reduction in not only the fossil costs but also the CO2, SO2 and NOx emission.5 Concretely, compared with supercritical (SC) boilers (24 MPa, 565 ◦C), modern ultra-supercritical (USC) boilers operating around 30 MPa and 600 ◦C can lead to both an increase of 8–10% in thermal efficiency (from 37% to 45–47%) and a reduction of 20–25% in CO2 emission.1,6–10 Development of USC boilers with increasingly higher steam parameters is an added incentive for boiler material Correspondence: Z.-G. Yang. E-mail: zgyang@fudan.edu.cn research. Consequently, heat-resistant steels, mainly referring to ferritic steels and austenitic stainless steels, have been pursued simultaneously since the emergence of USC boilers for applications in boiler components, including superheater, reheater, header, turbine, steam piping and so on. Subsequently, for the sake of reliable services in USC boilers, a wealth of research has been carried out on these heat-resistant steels as F12 (X20CrMoV12.1), T91 (9Cr1MoVNb), TP347H (18Cr10NiNb), etc. to investigate characteristics of their base materials and performance deterioration after long-term services.11–29 Among them, the T92 (9Cr0.5Mo2WVNb), approximately similar to T91 but with a little modification in chemical compositions for preferable high temperature properties than T91, and the HR3C (TP310HCbN) are the two typical representatives of new grade ferritic steels and austenitic stainless steels for their superior heat strength above 620 ◦C, and will certainly have a broad application prospect in forthcoming USC boilers. However, applications of T92 and HR3C are currently a bit constrained due to the lack of sufficient literature and experience on c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96 83 Fatigue & Fracture of Engineering Materials & Structures
84 Y GoNG et al Table 1 Chemical compositions and heat treatment conditions of T92 and HR3C samples(wt%) Elements P s Si N 0.21 0.011.63 ASME 0.07-0.138.50- 0.30-0600.15-0.250.040.09≤0.40 0.30-0.60<0.020<0.010<0.500.03-0.07<0.041.50-2.000.001-0.006 SA-213 12400120.0010390.24 ASME 3.00-2700 0.20-0.601700-23.00≤2.000.030≤0.030≤1.500.15-0.35 TP310HC T92: 1050C x 20 min(normalizing)+760C x 60 min(tempering). HR3C: solution-treated at 1110.C minimum reep rupture performances of them and their welded (a) joints, let alone the dissimilar steels welded joints between them. Therefore, a thorough assessment of the compre hensive properties, particularly the creep properties of T92/HR3C dissimilar steels welded joints, seems pretty urgent. In this paper, besides various conventional mechanica tests including tensile test, bending test and hardness sur- ey, optical microscope(OM)was also applied to inspect the metallographic microstructures across the dissimilar steels welded joint between T92 and HR3C. Moreover, creep rupture test was particularly employed under dif- ferent load stresses at 625 oC to investigate the creep features of the joints, whose fractograph was then ob- served by using scanning electron microscope (SEM)as ell. Furthermore, the residual stress distribution of the element method(FEM), which was a tentative approach (b) to evaluate residual stress of the dissimilar steels welded joint between these two novel materials through the com the degradation curves of T92/HR3C dissimilar steels yelded joints were reported, but also the mechanism of voids initiating creep rupture was concretely discussed which may have critical significance in both service-life prediction and future heat-resistant steels preparation fo boiler components EXPERIMENTAL 20题 Tested materials were nominal T92 and hr3C heat- Fig. 1 Metallographic microstructures of tested base materials (a) resistant steels with scales of 480D x 8.4 mm thick 92,1500×(b)HR3C,200× and 48. D x 10.16 mm thick, respectively. Chemical compositions as well as heat treatment conditions of their crostructure of T92 sample is presented in Fig. la, which base materials are listed in Table 1, which are in accor- displays a typical tempered lath martensitic microstruc- dance with the requirements of ASME SA-213 T92 and ture. Similarly, metallographic microstructure of HR3C TP310HCbN specifications. Etched in agent of picric sample was also obtained after being etched in the agent acid(2, 4, 6-trinitrophenol)1. 25 g, HCI 20 ml, ethanol of CuSO4 4 g, HCl 20 ml and ethanol 20 ml for 20 s 10 ml and H2O 10 ml for 40 s, the metallographic mi- As is shown in Fig. 1b, HR3C presents a fine-grained @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
84 Y. GONG et al. Table 1 Chemical compositions and heat treatment conditions of T92 and HR3C samples (wt%) Elements C Cr Mo V Nb Ni Mn P S Si N Al W B T92 Sample 0.11 8.76 0.36 0.21 0.059 0.25 0.46 0.016 0.002 0.39 0.044 0.01 1.63 0.0033 ASME 0.07–0.13 8.50–9.50 0.30–0.60 0.15–0.25 0.04–0.09 ≤0.40 0.30–0.60 ≤0.020 ≤0.010 ≤0.50 0.03–0.07 ≤0.04 1.50–2.00 0.001–0.006 SA-213 T92 HR3C 0.06 24.63 / / 0.49 20.29 1.24 0.012 0.001 0.39 0.24 / / / Sample ASME ≤0.10 23.00–27.00 / / 0.20–0.60 17.00–23.00 ≤2.00 ≤0.030 ≤0.030 ≤1.50 0.15–0.35 / / / SA-213 TP310HCbN Heat treatment conditions: T92: 1050 ◦C × 20 min (normalizing) + 760 ◦C × 60 min (tempering). HR3C: solution-treated at 1110 ◦C minimum. creep rupture performances of them and their welded joints, let alone the dissimilar steels welded joints between them. Therefore, a thorough assessment of the comprehensive properties, particularly the creep properties of T92/HR3C dissimilar steels welded joints, seems pretty urgent. In this paper, besides various conventional mechanical tests including tensile test, bending test and hardness survey, optical microscope (OM) was also applied to inspect the metallographic microstructures across the dissimilar steels welded joint between T92 and HR3C. Moreover, creep rupture test was particularly employed under different load stresses at 625 ◦C to investigate the creep features of the joints, whose fractograph was then observed by using scanning electron microscope (SEM) as well. Furthermore, the residual stress distribution of the welded joint after welding was calculated by using finite element method (FEM), which was a tentative approach to evaluate residual stress of the dissimilar steels welded joint between these two novel materials through the computational simulation method. Finally, based on the analysis results, not only the creep rupture performances and the degradation curves of T92/HR3C dissimilar steels welded joints were reported, but also the mechanism of voids initiating creep rupture was concretely discussed, which may have critical significance in both service-life prediction and future heat-resistant steels preparation for boiler components. EXPERIMENTAL Tested materials were nominal T92 and HR3C heatresistant steels with scales of 48O D × 8.4 mm thick and 48.26O D × 10.16 mm thick, respectively. Chemical compositions as well as heat treatment conditions of their base materials are listed in Table 1, which are in accordance with the requirements of ASME SA-213 T92 and TP310HCbN specifications.30 Etched in agent of picric acid (2, 4, 6-trinitrophenol) 1.25 g, HCl 20 ml, ethanol 10 ml and H2O 10 ml for 40 s, the metallographic miFig. 1 Metallographic microstructures of tested base materials (a) T92, 1500× (b) HR3C, 200×. crostructure of T92 sample is presented in Fig. 1a, which displays a typical tempered lath martensitic microstructure. Similarly, metallographic microstructure of HR3C sample was also obtained after being etched in the agent of CuSO4 4 g, HCl 20 ml and ethanol 20 ml for 20 s. As is shown in Fig. 1b, HR3C presents a fine-grained c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
CREEP PROPERTIES OF T92/HR3C WELDED JOINTS 85 Table 2 Chemical compositions of welding wire ERNiCr-3(wt%) P Si Cu Ni Ti ERNiCr-3 welding wire 00302901.300.0040.0010.040.0172.50.3120.0 Nb240 ASME SFA-5.14(AWS) ERNICI-3≤0.102.5-3.5≤3.0≤0.030≤0.015≤0.50≤0.50≥67.0≤0.7518.00-22.002.0-3.0 Table 3 Tensile test results of t92/HR3C dissimilar steels welded joints Tensile strength 50 Sample No (os, MPa) Rupture position 707 unit: m T92 base material Fig 2 Dimension of creep rupture test specimen T92 specification ≥620 HR3C specification 2655 austenitic microstructure with average grain size of about 7, which conforms with the requirement that the Table 4 Bending test results of T92/HR3C dissimilar steels grain size of TP310HCbN must be coarser than 7(in- welded joints cluding 7)in ASME specification. 30 The T92/HR3C dissimilar steels welded joints were Bending sty yle Sample No. Test condition welded by means of gas tungsten arc welding(GTAW) Face bending 1 D=4T,a=180° Qualified ith pure argon gas(Ar) as the shielding gas and Aws D=3T,a=50F ERNiCr-3(corresponding to INCONEL 82/182)as the Back bending D=4T, a= 180 Qualified welding wire, whose chemical compositions are listed in D=3T,a=50 Table 2. Subsequently, the welded joints were subjected to the post weld heat treatment (PWHT) at 760-770oc D denotes the bending diameter, 'T' denotes the material thickness; for 2 h to eliminate the residual stress ar denotes the bending angle. A variety of mechanical tests for the welded joints were then successively carried out. Tensile test, bending test RESULTS AND DISCUSSION and hardness survey were performed at room temperature Mechanical tests results according to the AsTME8-04, E290-97a(2004)and E92 82(2003)e2 standards, respectively. Also, metallographic As is clear from Table 3, the T92/HR3C dissimilar steels microstructure across the welded joint, especially in the welded joints exhibit qualified tensile strength, and the two heat-affected zones(HAz)and the weld seam, was in- T92 base material part is their weakest region under load spected under LEICA DMLM OM. In accordance with stresses.Table 4 reveals the sign that the welded joints also ASTME139-06 standard, the creep rupture test was con- present eligible toughness. In addition, no cracks were ducted at 625 oC under load stresses of 110, 120, 130, founded on the bended surfaces. Hardness survey results 140, 150, 160 and 180 MPa, respectively. Figure 2 shows are displayed in Fig 3, which indicates that the T92 base the round bar configuration of the creep rupture speci- material and weld seam are the two low-hardness parts men with the size of 10 mm diameter for base materials and 50 mm gage length for welded joint. After the creep Metallographic microstructure inspection rupture test, micro morphologies of the cross-sections of the ruptured samples were observed by using PHILIPS As is well known, a dissimilar steels welded joint is often KL30FEG SEM(Eindhoven, The Netherlands). approximately divided into five regions, i. e. base material Residual stress analysis of the welded joint after weld- A, HAZ of A, weld seam, HAZ of B and base material ing was carried out by means of finite element analysis B. In this paper, the metallographic microstructures of software ANSYS 10.0( Canonsburg Pennsylvania, USA). the five regions across the T92/HR3C dissimilar steels The FEM analysis result could clearly reflect the resid- welded joint were inspected under Om ual stress distribution of the T92/HR3C dissimilar steels Figure 4a presents the metallographic microstructure of welded joint in a convenient way, which may also theo- T92 base material after welding, in which no signs of dif- retically supplement the results of mechanical and creep ferential are detected when compared with that of ori inal material in Fig. la. However, in the HAZ of T92 @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
C R E E P P RO P E RTI E S O F T 9 2/H R 3 C W E LD ED JOINT S 85 Table 2 Chemical compositions of welding wire ERNiCr-3 (wt%) Elements C Mn Fe P S Si Cu Ni Ti Cr Nb+Ta ERNiCr-3 welding wire 0.030 2.90 1.30 0.004 0.001 0.04 0.01 72.5 0.31 20.0 Nb 2.40 ASME SFA-5.14 (AWS) ERNiCr-3 ≤0.10 2.5–3.5 ≤3.0 ≤0.030 ≤0.015 ≤0.50 ≤0.50 ≥67.0 ≤0.75 18.00–22.00 2.0–3.0 Fig. 2 Dimension of creep rupture test specimen. austenitic microstructure with average grain size of about 7, which conforms with the requirement that the grain size of TP310HCbN must be coarser than 7 (including 7) in ASME specification.30 The T92/HR3C dissimilar steels welded joints were welded by means of gas tungsten arc welding (GTAW) with pure argon gas (Ar) as the shielding gas and AWS ERNiCr-3 (corresponding to INCONEL 82/182) as the welding wire, whose chemical compositions are listed in Table 2.31 Subsequently, the welded joints were subjected to the post weld heat treatment (PWHT) at 760–770 ◦C for 2 h to eliminate the residual stress. A variety of mechanical tests for the welded joints were then successively carried out. Tensile test, bending test and hardness survey were performed at room temperature according to the ASTM E8-04, E290-97a(2004) and E92- 82(2003)e2 standards, respectively. Also, metallographic microstructure across the welded joint, especially in the two heat-affected zones (HAZ) and the weld seam, was inspected under LEICA DMLM OM. In accordance with ASTM E139-06 standard, the creep rupture test was conducted at 625 ◦C under load stresses of 110, 120, 130, 140, 150, 160 and 180 MPa, respectively. Figure 2 shows the round bar configuration of the creep rupture specimen with the size of 10 mm diameter for base materials and 50 mm gage length for welded joint. After the creep rupture test, micro morphologies of the cross-sections of the ruptured samples were observed by using PHILIPS XL30FEG SEM (Eindhoven, The Netherlands). Residual stress analysis of the welded joint after welding was carried out by means of finite element analysis software ANSYS 10.0 (Canonsburg Pennsylvania, USA). The FEM analysis result could clearly reflect the residual stress distribution of the T92/HR3C dissimilar steels welded joint in a convenient way, which may also theoretically supplement the results of mechanical and creep tests. Table 3 Tensile test results of T92/HR3C dissimilar steels welded joints Tensile strength Sample No. (σs, MPa) Rupture position 1 707 T92 base material 2 699 T92 base material T92 specification ≥620 / HR3C specification ≥655 / Table 4 Bending test results of T92/HR3C dissimilar steels welded joints Bending style Sample No. Test condition Result Face bending 1 D = 4T, α = 180◦ Qualified 2 D = 3T, α = 50◦ Back bending 1 D = 4T, α = 180◦ Qualified 2 D = 3T, α = 50◦ ‘D’ denotes the bending diameter; ‘T’ denotes the material thickness; ‘α’ denotes the bending angle. RESULTS AND DISCUSSION Mechanical tests results As is clear from Table 3, the T92/HR3C dissimilar steels welded joints exhibit qualified tensile strength, and the T92 base material part is their weakest region under load stresses. Table 4 reveals the sign that the welded joints also present eligible toughness. In addition, no cracks were founded on the bended surfaces. Hardness survey results are displayed in Fig. 3, which indicates that the T92 base material and weld seam are the two low-hardness parts. Metallographic microstructure inspection As is well known, a dissimilar steels welded joint is often approximately divided into five regions, i.e. base material A, HAZ of A, weld seam, HAZ of B and base material B. In this paper, the metallographic microstructures of the five regions across the T92/HR3C dissimilar steels welded joint were inspected under OM. Figure 4a presents the metallographic microstructure of T92 base material after welding, in which no signs of differential are detected when compared with that of original material in Fig. 1a. However, in the HAZ of T92, c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
Y. GoNG et al T92 T92 HAZ weld seam HR3C HAZ HR3C 270 它230 Fig, 3 Hardness distribution Distance(mm T92/HR3C dissimilar steels welded joints obvious sorbitic microstructure with coarsened laths load stresses is in accordance with the results in the lit could be observed in Fig. 4b. The width of the sorbite erature. 23 Together with the tensile test results, it can lath is nearly two times that of martensite lath, which may be concluded that the T92 base material part is actually result in an increase of hardness in HAZ of T92 as well the weakest region of the t92/HR3C dissimilar steels a decrease of toughness in this region simultaneously. welded joint under relatively higher stresses. Neverthe The mechanism can be explained that coarser laths may less, compared with the research data of Falat et al. 4 the block the growth of cracks under stresses, and eventu- rupture time of our T92/HR3C dissimilar steels welded ally lead to brittle rupture in this region Near the weld joints under 120 MPa at 625C actually even exceeds seam,a distinct boundary between HAZ of T92 and weld the counterpart value of P92/p92(P"denotes the word seam can be observed in Fig. 4c. Figure 4d shows the pipe, which contains the same chemical compositions of stripe-shaped austenitic microstructure of the weld seam, T, i.e. tube, but with a larger diameter and thickness) whose stripe width has already reached around 20 um. similar steels welded joint, 1174 h. Correspondingly, like Fig. 4c, Fig. 4e gives the boundary According to the classic theory about creep rupture test, between weld seam and HAZ of HR3C. Average grain a double logarithmic relationship between load stress o size of the austenite in HAZ of Hr3C is about 6(Fig. and rupture time t, i. e Igo versus lgt, can be expresse 40, and carbides as M23 C6 have precipitated at the grain in Eq.(1), in which the letters A and b both denote the boundaries. Compared with the average grain size of 7 material parameters of the tested samples. Accordingly HR3C may also lead to increase in hardness in this region. fitting result ofi plot of lgo versus lgt and the linear in HR3C base material, the coarsened grains in HAZ of Fig. 5 presents the Fig 4g is the metallographic microstructure of HR3C status too. Collectively, the schematic diagram as well gt=lgA_blgo base material, which shows no changes with its original (1) as the metallographic microstructures distribution across the T92/HR3C dissimilar steels welded joint is displayed hus, the threshold steam stress for service of the in Fig. 4h. T92/HR3C dissimilar steels welded joints exposed at 625C, which can be determined by extrapolation of the fitted line to 10 h, is 61.89 MPa. However, in practi Creep rupture test results al applications, some other unpredicted factors may also affect the creep strength of welded joints. Hence, safety Table 5 lists the creep rupture test results under load coefficient n is always adopted to modify the predicted stresses ranging from 110 to 180 MPa at 625C. This threshold stress obtained from linear extrapolation Con phenomenon that rupture positions vary with change in sequently, permitted stress [o] can be expressed as Eq (2) @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
86 Y. GONG et al. Fig. 3 Hardness distribution across the T92/HR3C dissimilar steels welded joints. obvious sorbitic microstructure with coarsened laths could be observed in Fig. 4b. The width of the sorbite lath is nearly two times that of martensite lath, which may result in an increase of hardness in HAZ of T92 as well as a decrease of toughness in this region simultaneously. The mechanism can be explained that coarser laths may block the growth of cracks under stresses, and eventually lead to brittle rupture in this region. Near the weld seam, a distinct boundary between HAZ of T92 and weld seam can be observed in Fig. 4c. Figure 4d shows the stripe-shaped austenitic microstructure of the weld seam, whose stripe width has already reached around 20 μm. Correspondingly, like Fig. 4c, Fig. 4e gives the boundary between weld seam and HAZ of HR3C. Average grain size of the austenite in HAZ of HR3C is about 6 (Fig. 4f), and carbides as M23C6 have precipitated at the grain boundaries. Compared with the average grain size of 7 in HR3C base material, the coarsened grains in HAZ of HR3C may also lead to increase in hardness in this region. Fig. 4g is the metallographic microstructure of HR3C base material, which shows no changes with its original status too. Collectively, the schematic diagram as well as the metallographic microstructures distribution across the T92/HR3C dissimilar steels welded joint is displayed in Fig. 4h. Creep rupture test results Table 5 lists the creep rupture test results under load stresses ranging from 110 to 180 MPa at 625 ◦C. This phenomenon that rupture positions vary with change in load stresses is in accordance with the results in the literature.23 Together with the tensile test results, it can be concluded that the T92 base material part is actually the weakest region of the T92/HR3C dissimilar steels welded joint under relatively higher stresses. Nevertheless, compared with the research data of Falat et al., 24 the rupture time of our T92/HR3C dissimilar steels welded joints under 120 MPa at 625 ◦C actually even exceeds the counterpart value of P92/P92 (‘P’ denotes the word ‘pipe’, which contains the same chemical compositions of ‘T’, i.e. ‘tube’, but with a larger diameter and thickness) similar steels welded joint, 1174 h. According to the classic theory about creep rupture test, a double logarithmic relationship between load stress σ and rupture time t, i.e. lgσ versus lgt, can be expressed in Eq. (1), in which the letters A and B both denote the material parameters of the tested samples. Accordingly, Fig. 5 presents the plot of lgσ versus lgt and the linear fitting result of it. lg t = lg A − B lg σ. (1) Thus, the threshold steam stress for service of the T92/HR3C dissimilar steels welded joints exposed at 625 ◦C, which can be determined by extrapolation of the fitted line to 105 h, is 61.89 MPa. However, in practical applications, some other unpredicted factors may also affect the creep strength of welded joints. Hence, safety coefficient n is always adopted to modify the predicted threshold stress obtained from linear extrapolation. Consequently, permitted stress [σ] can be expressed as Eq. (2), c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
CREEP PROPERTIES OF T92/HR3C WELDED JOINTS 87 (b) (c) 少了m austenite T92 HAZ HR3C HAZ 0 1cm 2 Fig. 4 Metallographic microstructures of the five different regions across the welded joint(a)T92 base material, (b) HAZ of T92, (c) boundary between HAZ of T92 and weld seam, (d)weld seam, (e) boundary between weld seam and HAZ of HR3C,(D) HAZ of HR3C, (g) HR3C base material, (h)schematic diagram of welded joint @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
C R E E P P RO P E RTI E S O F T 9 2/H R 3 C W E LD ED JOINT S 87 Fig. 4 Metallographic microstructures of the five different regions across the welded joint (a) T92 base material, (b) HAZ of T92, (c) boundary between HAZ of T92 and weld seam, (d) weld seam, (e) boundary between weld seam and HAZ of HR3C, (f) HAZ of HR3C, (g) HR3C base material, (h) schematic diagram of welded joint. c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
Table 5 Creep rupture test(625C)results of T92/HR3C dissimilar steels welded joints 625℃ Rupture No.(, MPa) Time(h) 180 3 T92 base material 1234567 160 T92 base material 150 T92 base material HAZ of T92 Weld seam 110 4182 Weld seam 1000 LMP=T(25+g(tr)×10 Fig. 6 Plot of stresses with LMP for welded joints. Table 6 LMP values and rupture times under stresses of 30, 35 and40 MPa at625°C Service stress (MPa) LMP value(x10-) Rupture time(h) 27.15 168060 19198 87631 fers from 19.5 to 36 according to a variety of references Fig 5 Double logarithmic plot of load stress versus rupture time involved.33,34 However, thanks to sufficient practical ap- plication experiences of T92 in the past decade, the value of C has been modified according to lots of firsthand and the value of n is generally ranging from 1. 2 to 1.65 data. Hence, a value of 25 is now commonly adopted for C to predict the service lives of T92 for better accordance tIv, in thi (2) rupture data for the T92/HR3C dissimilar steels welded In this case, we take n= 1.5. and the permitted joints can be plotted in terms of lgo versus Larson-Miller stress [ a] of T92/HR3C dissimil elded joint parameters(LMP)in Fig. 6, where the LMP is expressed can be determined from Eq (3),w value 41.26 Eq(5) MPa also exceeds the steam stress of USC conditions LMP=T(25+Igt). 6819=q 1541.26MPa. (3) Also, the Igo versus LMP curve displayed in Fig. 6 is then polynomial fitted for the convenience of service life pre In terms of service life prediction for high-temperature diction for the welded joints under different stresse components in boilers, the Larson-Miller equation, seen for USC boilers, service steam stresses commonly in Eq(4), is always applied to estimate the allowable from 30 to 40 MPa, whose LMP values can then be easily stresses under specific steam conditions read from Fig. 6. Table 6 lists the lmP values and thei P=T(C +Igt) corresponding rupture times under stresses of 30, 35 and (4) 40 MPa, respectively. It is obvious that rupture times un- where T is the absolute temperature in K, and t, is the der 30 and 35 MPa at 625 oC both exceed the recom- pture time in hours. Actually, C is a constant depend- mended service life of 10 h, even if the rupture time of ing on different materials. Owing to the fact that T92 40 MPa is long enough in practical applications (nearly base material is the weakest part of T92/HR3C dissimi- 10 years). Actually, the rupture time under 40 MPa at lar steels welded joint, the value of C can be determined 625C is still longer than that counterpart values of from T92 matrix material. As for T92, the value of C dif- P91/P22 dissimilar ferritic steels welded joint(less tha @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
88 Y. GONG et al. Table 5 Creep rupture test (625 ◦C) results of T92/HR3C dissimilar steels welded joints Sample Load stress Rupture Rupture No. (σ, MPa) Time (h) position 1 180 378 T92 base material 2 160 865 T92 base material 3 150 930 T92 base material 4 140 1640 T92 base material 5 130 2787 HAZ of T92 6 120 3164 Weld seam 7 110 4182 Weld seam Fig. 5 Double logarithmic plot of load stress versus rupture time for the welded joints. and the value of n is generally ranging from 1.2 to 1.65.22 [σ] = σ625 ◦C 105 n . (2) In this case, we take n = 1.5, and therefore the permitted stress [σ] of T92/HR3C dissimilar steels welded joints can be determined from Eq. (3), whose result value 41.26 MPa also exceeds the steam stress of USC conditions. [σ] = 61.89 1.5 = 41.26 MPa. (3) In terms of service life prediction for high-temperature components in boilers, the Larson–Miller equation, seen in Eq. (4), is always applied to estimate the allowable stresses under specific steam conditions:32 P = T(C + lg tr ), (4) where T is the absolute temperature in K, and tr is the rupture time in hours. Actually, C is a constant depending on different materials. Owing to the fact that T92 base material is the weakest part of T92/HR3C dissimilar steels welded joint, the value of C can be determined from T92 matrix material. As for T92, the value of C difFig. 6 Plot of stresses with LMP for welded joints. Table 6 LMP values and rupture times under stresses of 30, 35 and 40 MPa at 625 ◦C Service stress (MPa) LMP value (×10−3) Rupture time (h) 30 27.15 168060 35 27.01 119198 40 26.89 87631 fers from 19.5 to 36 according to a variety of references involved.33,34 However, thanks to sufficient practical application experiences of T92 in the past decade, the value of C has been modified according to lots of firsthand data. Hence, a value of 25 is now commonly adopted for C to predict the service lives of T92 for better accordance with the actual conditions. Consequently, in this case, the rupture data for the T92/HR3C dissimilar steels welded joints can be plotted in terms of lgσ versus Larson–Miller parameters (LMP) in Fig. 6, where the LMP is expressed as given in Eq. (5). LMP = T(25 + lg tr ). (5) Also, the lgσ versus LMP curve displayed in Fig. 6 is then polynomial fitted for the convenience of service life prediction for the welded joints under different stresses. As for USC boilers, service steam stresses commonly range from 30 to 40 MPa, whose LMP values can then be easily read from Fig. 6. Table 6 lists the LMP values and their corresponding rupture times under stresses of 30, 35 and 40 MPa, respectively. It is obvious that rupture times under 30 and 35 MPa at 625 ◦C both exceed the recommended service life of 105 h, even if the rupture time of 40 MPa is long enough in practical applications (nearly 10 years). Actually, the rupture time under 40 MPa at 625 ◦C is still longer than that counterpart values of P91/P22 dissimilar ferritic steels welded joint (less than c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
CREEP PROPERTIES OF T92/HR3C WELDED JOINTS 89 o「 Fig. 7 Macroscopic morphologies of ruptured welded joints under different load stresses(a)180, (b)130, (c)110 MPa. 19 000 h) and P91/P91 similar steels welded joint (less Figure 8 presents the micromorphologies of cross- than 32 000 h). 26 Relevant data of T92 welded joints have section of the ruptured sample under 180 MPa. It can be seen from Fig. 8a that the diameter of the cross-section is about 2.5 mm. Compared with its original value of 6.0 mm, the reduction in area y can be calculated as about fractograph observation of creep ruptured samples 80%. As is shown in Fig. 8. the cross-section is covered with densely distributed spherical creep voids, whose di According to the creep rupture test results, the rupture ameter ranges from 5 to 50 um. The presence of creep positions of the T92/HR3C dissimilar steels welded joints voids, whose nucleation mechanisms still need to be fur under different load stresses, mainly including three dif- ther identified lonly represents a good toughness ferent parts, seen in Table 5. Hence, fractograph of sam- of materials at high temperatures. In some region,three ples under three representative load stresses: 110, 130 and neighbouring voids with diameters of about 5 um have 0 Mpa, whose rupture positions were respectively weld coalesced into a larger wave-like one, whose axial width seam,HAZ of T92 and T92 base material, were then has already reached nearly 20 um, as can be seen in Fig.8c concretely inspected by means of SEM. Fracture posi- and d. This is typical evidence for ductile materials that tions of joints under 110, 130 30 MPa located in creep rupture is led by the coalescence of microscopic three different positions. Figure 7 was the macroscopic cracks which are also generated by the sub-coalescence of morphology of them three. Then, samples were cut from creep voids the rupture positions to observe the micromorphologies Compared with the fractograph of ruptured sample un- of their cross-sections under sem der 180 MPa, the fractograph under 130 MPa displays @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
C R E E P P RO P E RTI E S O F T 9 2/H R 3 C W E LD ED JOINT S 89 Fig. 7 Macroscopic morphologies of ruptured welded joints under different load stresses (a) 180, (b) 130, (c) 110 MPa. 19 000 h) and P91/P91 similar steels welded joint (less than 32 000 h).26 Relevant data of T92 welded joints have hardly been found. Fractograph observation of creep ruptured samples According to the creep rupture test results, the rupture positions of the T92/HR3C dissimilar steels welded joints under different load stresses, mainly including three different parts, seen in Table 5. Hence, fractograph of samples under three representative load stresses: 110, 130 and 180 Mpa, whose rupture positions were respectively weld seam, HAZ of T92 and T92 base material, were then concretely inspected by means of SEM. Fracture positions of joints under 110, 130 and 180 MPa located in three different positions. Figure 7 was the macroscopic morphology of them three. Then, samples were cut from the rupture positions to observe the micromorphologies of their cross-sections under SEM. Figure 8 presents the micromorphologies of crosssection of the ruptured sample under 180 MPa. It can be seen from Fig. 8a that the diameter of the cross-section is about 2.5 mm. Compared with its original value of 6.0 mm, the reduction in area ψ can be calculated as about 80%. As is shown in Fig. 8b, the cross-section is covered with densely distributed spherical creep voids, whose diameter ranges from 5 to 50 μm. The presence of creep voids, whose nucleation mechanisms still need to be further identified, commonly represents a good toughness of materials at high temperatures. In some region, three neighbouring voids with diameters of about 5 μm have coalesced into a larger wave-like one, whose axial width has already reached nearly 20 μm, as can be seen in Fig. 8c and d. This is typical evidence for ductile materials that creep rupture is led by the coalescence of microscopic cracks which are also generated by the sub-coalescence of creep voids. Compared with the fractograph of ruptured sample under 180 MPa, the fractograph under 130 MPa displays c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
b) 10o Fig 8 SEM of the cross-section of creep ruptured sample under 180 MPa(a)total morphology, (b)dimples and voids, (c),(d) three-neighbouring voids an obvious brittle rupture morphology, seen in Fig. 9a. can be determined among the voids generated, respec Lots of slim dissociation steps resulting from narrow sor- tively, under 110, 130 and 180 Mpa. On the one hand, bite laths can be found on the cross-section Meanwhile. as is discussed above. the wide and shallow voids with di creep voids can also be observed in Fig. 9b. However, the ameter of about 10 um(Fig. 10d)may be accounted for amount of voids which have grown in a wide and shal- the low toughness of the weld seam. On the other hand, low form is far less than that on the cross-section under the larger distribution density of the creep voids under 180 MPa. Although a small amount of dimples can be 110 MPa than that generated under 130 MPa may be ad observed in Fig. 9b as well, they are not the dominant fac- counted for the lower hardness of the weld seam. Thus, tors of rupture. This phenomenon may be attributed to the fractograph of the ruptured sample under 110 MPa the highest hardness and residual stress in HAZ of T92. presents an intermediate micromorphology between that Figure 10 displays the fractograph of creep ruptured under 180 and 130 Mpa, which indicates its intermediat sample under 110 MPa. An obvious macroscopic disso- properties of weld seam. ation step rather than macroscopic creep voids can be detected in fig. 10a. Meanwhile, it can be learned that the Finite element method results reduction in area y is less than 5%. Actually, magnified by 500 times, the zoom is filled with randomly distributed FEM is the most widely used computational simulation creep voids, seen in Fig. 10b. Furthermore, neighbouring method for its superiorities as convenience, effectiveness voids also have the potential to coalesce into larger ones, accuracy, etc, and is always applied in physical field anal as shown in Fig. 10c. However, at least three differentials yses including thermal field, force field, magnetic field, Including vo pth, void area a nd distribution density etc and their coupled fields. 3-37In this case, the residual @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
90 Y. GONG et al. Fig. 8 SEM of the cross-section of creep ruptured sample under 180 MPa (a) total morphology, (b) dimples and voids, (c), (d) three-neighbouring voids. an obvious brittle rupture morphology, seen in Fig. 9a. Lots of slim dissociation steps resulting from narrow sorbite laths can be found on the cross-section. Meanwhile, creep voids can also be observed in Fig. 9b. However, the amount of voids which have grown in a wide and shallow form is far less than that on the cross-section under 180 MPa. Although a small amount of dimples can be observed in Fig. 9b as well, they are not the dominant factors of rupture. This phenomenon may be attributed to the highest hardness and residual stress in HAZ of T92. Figure 10 displays the fractograph of creep ruptured sample under 110 MPa. An obvious macroscopic dissociation step rather than macroscopic creep voids can be detected in Fig. 10a. Meanwhile, it can be learned that the reduction in area ψ is less than 5%. Actually, magnified by 500 times, the zoom is filled with randomly distributed creep voids, seen in Fig. 10b. Furthermore, neighbouring voids also have the potential to coalesce into larger ones, as shown in Fig. 10c. However, at least three differentials including void depth, void area and distribution density can be determined among the voids generated, respectively, under 110, 130 and 180 Mpa. On the one hand, as is discussed above, the wide and shallow voids with diameter of about 10 μm (Fig. 10d) may be accounted for the low toughness of the weld seam. On the other hand, the larger distribution density of the creep voids under 110 MPa than that generated under 130 MPa may be accounted for the lower hardness of the weld seam. Thus, the fractograph of the ruptured sample under 110 MPa presents an intermediate micromorphology between that under 180 and 130 Mpa, which indicates its intermediate properties of weld seam. Finite element method results FEM is the most widely used computational simulation method for its superiorities as convenience, effectiveness, accuracy, etc., and is always applied in physical field analyses including thermal field, force field, magnetic field, etc. and their coupled fields.35–37 In this case, the residual c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
CREEP PROPERTIES OF T92/HR3C WELDED JOINTS 91 among which the shear m dulus is commonly regarded as one-tenth of the elastic modulus The results of the cor onal analy presente in Fig. 13, which shows the Von Mises equivalent stress distribution across the welded joint. It can be obviously observed from Fig. 13a that residual stress mainly ac cumulates at the two HAZ regions. The largest resid ual stress occurs at the HAZ of T92 region, seen in Fig b, which has already reached around 330 MPa. Mean while, the residual stress of the haz of hr3c is also up to around 250 MPa. This could be ascribed to the preferable ductility of austenite HR3C, as it is able to off- set stresses through plastic deformation. The FEM result verified the fact that residual stresses usually accumulate at HAZ of the undermatched part, i. e the ferrite HAZ, in ferrite/austenite dissimilar steels welded joints. A conclu sion can then be put forward that the selection of welding wires with similar strength of ferrites could effectively lieve the residual stresses of the joints after welding; in other words, it can increase the comprehensive strength of the joir Compreh In terms of creep voids, they are commonly generated un- der the interaction among high temperature, load stress and ageing time. Till date there are several classic con troversial theories of the emergence of creep voids. 9, #0 By Greenwood, Argon, #. Raj and Ashby, #4 tion positions of void nucleation were argued; by Hull and Fig9 SEM of the cross-section of creep ruptured sample under Rimmer, +5 Needleman# and Hancock, "the controlling 30 MPa(a)total morphology() shallow voids factors of void growth were debated; and by Stowell, + s Nicolaou and Semiatin 49,50 and Chokshi, 5I the coales cence procedures of voids were discussed. In this paper, stress of the T92/HR3C dissimilar steels welded joint af- based on their works, a four-stage nucleation to crack ter welding was calculated by the FEM software ANSYs mechanism of creep voids was put forward to clearly un- 10.0. derstand the mechanisms of creep voids generation for the width of the welded joint is not sufficiently large, the T92/HR3C dissimilar steels welded joints their effect on residual stress distribution is generally ne- glected. Thus, the three-dimensional (3D)thermal-stress 1 Nucleation coupled field analysis can be simplified as 2D axisymmet- In this stage, the creep voids with irregular shapes are ric problem. The 2D FEM model (after being meshed) generated from the effect of grain boundaries sliding is shown in Fig. 1la by using PLANE 13 2D coupled d/or grain matrix deformation under specific load field solid element. Also, birth-death element was applied In in the weld seam with 21 lavers to simulate the welding the voids commonly nucleate at the grain boundaries of procedure, seen in Fig. llb, of all the layers that were the weakest part or the part with largest residual stress of initially dead, the lowest layer was firstly activated when tested material. As for the t92/HR3C dissimilar steels being welded, then the rest ones would be activated layer welded joints, the rupture positions i. e the voids nucle by layer above the former ones sequentially. Displace ation positions varied under different creep condition ment in horizontal and vertical directions of the left and seen in Table 5. This may be accounted for the different the right borders of the welded joint set zero were the properties of the five regions across the joint. boundary conditions. The physical properties of T92 and 2 Growth HR3C base materials as well as ErNiCr-3 welding wire Creep voids continuously grow under constant temper used in calculation are listed in Fig. 12 and Table 7, 12.3 ature and load stress in this growth stage. However, the @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
C R E E P P RO P E RTI E S O F T 9 2/H R 3 C W E LD ED JOINT S 91 Fig. 9 SEM of the cross-section of creep ruptured sample under 130 MPa (a) total morphology (b) shallow voids. stress of the T92/HR3C dissimilar steels welded joint after welding was calculated by the FEM software ANSYS 10.0. As the width of the welded joint is not sufficiently large, their effect on residual stress distribution is generally neglected. Thus, the three-dimensional (3D) thermal-stress coupled field analysis can be simplified as 2D axisymmetric problem. The 2D FEM model (after being meshed) is shown in Fig. 11a by using PLANE 13 2D coupled field solid element. Also, birth–death element was applied in the weld seam with 21 layers to simulate the welding procedure, seen in Fig. 11b, of all the layers that were initially dead, the lowest layer was firstly activated when being welded, then the rest ones would be activated layer by layer above the former ones sequentially. Displacement in horizontal and vertical directions of the left and the right borders of the welded joint set zero were the boundary conditions. The physical properties of T92 and HR3C base materials as well as ERNiCr-3 welding wire used in calculation are listed in Fig. 12 and Table 7,12,38 among which the shear modulus is commonly regarded as one-tenth of the elastic modulus. The results of the computational analysis are presented in Fig. 13, which shows the Von Mises equivalent stress distribution across the welded joint. It can be obviously observed from Fig. 13a that residual stress mainly accumulates at the two HAZ regions. The largest residual stress occurs at the HAZ of T92 region, seen in Fig. 13b, which has already reached around 330 MPa. Meanwhile, the residual stress of the HAZ of HR3C is also up to around 250 MPa. This could be ascribed to the preferable ductility of austenite HR3C, as it is able to offset stresses through plastic deformation. The FEM result verified the fact that residual stresses usually accumulate at HAZ of the undermatched part, i.e. the ferrite HAZ, in ferrite/austenite dissimilar steels welded joints. A conclusion can then be put forward that the selection of welding wires with similar strength of ferrites could effectively relieve the residual stresses of the joints after welding; in other words, it can increase the comprehensive strength of the joints. Comprehensive analysis In terms of creep voids, they are commonly generated under the interaction among high temperature, load stress and ageing time. Till date, there are several classic controversial theories of the emergence of creep voids.39,40 By Greenwood,41 Argon,42,43 Raj and Ashby,44 the initiation positions of void nucleation were argued; by Hull and Rimmer,45 Needleman46 and Hancock,47 the controlling factors of void growth were debated; and by Stowell,48 Nicolaou and Semiatin,49,50 and Chokshi,51 the coalescence procedures of voids were discussed. In this paper, based on their works, a four-stage ‘nucleation to crack’ mechanism of creep voids was put forward to clearly understand the mechanisms of creep voids generation for the T92/HR3C dissimilar steels welded joints. 1 Nucleation In this stage, the creep voids with irregular shapes are generated from the effect of grain boundaries sliding and/or grain matrix deformation under specific load stress and temperature in creep process. Accordingly, the voids commonly nucleate at the grain boundaries of the weakest part or the part with largest residual stress of tested material. As for the T92/HR3C dissimilar steels welded joints, the rupture positions i.e. the voids nucleation positions varied under different creep conditions, seen in Table 5. This may be accounted for the different properties of the five regions across the joint. 2 Growth Creep voids continuously grow under constant temperature and load stress in this growth stage. However, the c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96
Y. GoNG et al (b) oK So Mo DEt ws Fig. 10 SEM of the cross-section of creep ruptured sample under 110 MPa(a)total morphology, (b)dimples and voids, (c) r-neighbouring voids, (d)magnification of void. Meshed FEM model of T92/HR3C steels welded joint (a)total model welded joint. ultimate void volumes may vary in a wide range of values is constrained in a wide and shallow form with only a according to different toughness of different materials small amount owing to the rigidness of matrix mat As for ductile materials like T92, the preferable tough rials. ' Meanwhile, in the process of voids growth, the ness ensures a good freedom for the voids to grow, which original voids with irregular shapes transform to uni- may lead to a great amount of deep voids. However, in formly spherical ones. This can be explained that voids terms of brittle materials like HAZ of T92, voids growth with irregular shapes possess higher surface-free energy @2010 Blackwell Publishing Ltd Fatigue Fract Engng Mater Struct 34, 83-96
92 Y. GONG et al. Fig. 10 SEM of the cross-section of creep ruptured sample under 110 MPa (a) total morphology, (b) dimples and voids, (c) four-neighbouring voids, (d) magnification of void. Fig. 11 Meshed FEM model of T92/HR3C dissimilar steels welded joint (a) total model and (b) welded joint. ultimate void volumes may vary in a wide range of values according to different toughness of different materials. As for ductile materials like T92, the preferable toughness ensures a good freedom for the voids to grow, which may lead to a great amount of deep voids. However, in terms of brittle materials like HAZ of T92, voids’ growth is constrained in a wide and shallow form with only a small amount owing to the rigidness of matrix materials.52 Meanwhile, in the process of voids growth, the original voids with irregular shapes transform to uniformly spherical ones. This can be explained that voids with irregular shapes possess higher surface-free energy, c 2010 Blackwell Publishing Ltd. Fatigue Fract Engng Mater Struct 34, 83–96