evaluation of weld residual stress near the cladding and · pdf filewelding current (a) 995...

50 ISSN-1883-9894/10 © 2010 – JSM and the authors. All rights reserved.

E-Journal of Advanced Maintenance Vol.2 (2010) 50-64Japan Society of Maintenology

Evaluation of Weld Residual Stress near the Cladding and J-weld in Reactor Pressure Vessel Head for the assessment of PWSCC Behavior Jinya KATSUYAMA1,*, Makoto UDAGAWA1, Hiroyuki NISHIKAWA1,†, Mitsuyuki NAKAMURA1,‡ and Kunio ONIZAWA1

1 Nuclear Safety Research Center, Japan Atomic Energy Agency, 2-4 Shirakata-shirane, Tokai-mura, Naka-gun, Ibaraki 319-1195, Japan

*Corresponding author, E-mail: [email protected] †Present address: Mizuho Information & Research Institute, Inc., 2-3 Nishiki-cho, Kanda, Chiyoda-ku, Tokyo 101-8443, Japan ‡Present address: Research Organization for Information Science & Technology, 2-4 Shirakata-shirane, Tokai-mura, Naka-gun, Ibaraki 319-1195, Japan

ABSTRACT Weld residual stress is one of the most important factors in order to assess the structural integrity of safety-related components such as reactor pressure vessel (RPV) in long-term operation of nuclear power plant since the residual stress significantly affects crack initiation and growth behaviors. The inner surface of the RPV made of low alloy steel is protected against corrosion by weld-overlay cladding of austenitic stainless steel. At the J-weld of the vessel head penetrations, Ni-based alloys are used for weld material. The residual stresses generated within the cladding, J-weld and base material were measured as-welded and after PWHT conditions using the deep-hole-drilling method. Thermal-elastic-plastic-creep analyses considering the phase transformation were also performed to evaluate the weld residual stress. By comparing analytical results with the measured ones, it was shown that there was roughly a good agreement of residual stress distribution within the cladding, J-weld and base metal. It was also suggested that taking the phase transformation during welding and PWHT into account was important to improve the accuracy of weld residual stress analysis. Using the residual stress distributions, fracture mechanics analyses for primary water stress corrosion cracking (PWSCC) have been performed using probabilistic fracture mechanics analysis code. Effects of the weld residual stress and scatter of PWSCC growth rate on the crack penetration were evaluated through some case studies. KEYWORDS

weld residual stress, reactor pressure vessel, PWSCC, weld-overlay cladding, J-groove welding, deep-hole-drilling method, finite element analysis, structure integrity, crack growth

ARTICLE INFORMATION

Article history: Received 20 May 2010 Accepted 21 July 2010

1. Introduction

In order to assess the structural integrity of safety-related components in long-term operated light water reactors (LWRs), weld residual stress is one of the most important factors since the residual stress distribution significantly affects the structural integrity related to crack initiation and growth behaviors. This is because that high tensile stresses caused by the residual stress and/or operation loads result in high stress intensity factor (SIF) calculated for structural integrity assessment when a crack is found. To evaluate the residual stress distribution is also necessary to apply and improve the repair techniques by welding. Reactor pressure vessel (RPV) is one of the most important components for safe long-term operation of nuclear power plants. The inner surface of the RPV, which is made of low alloy steel of ASME SA533B-1, is protected against corrosion by weld-overlay cladding of austenitic stainless steel. In such dissimilar metal welds, residual stress is produced due to the cladding process and the difference of the thermal expansion coefficients between austenitic cladding and ferritic base metal. The residual stress near the cladding of beltline region may have an influence on the integrity of RPV during pressurized thermal shock (PTS) events. In addition to this, a number of cracks due to primary water stress corrosion cracking (PWSCC) have been observed on vessel head

J. Katsuyama, et al./ Evaluation of Weld Residual Stress near the Cladding and J-weld in Reactor Pressure Vessel Head for the assessment of PWSCC Behavior

51

penetrations (VHPs) [1]. At the J-weld of the VHPs, Ni-based alloys are used for weld material. In the dissimilar metal welds, the initiation and propagation of PWSCC resulting in leakage of reactor coolant may strongly depend on the weld residual stress distribution. Therefore, it is very important to evaluate the weld residual stress with high accuracy for the structural integrity assessment of RPV.

The simulation on the weld residual stress is usually used to evaluate the residual stress distribution related to the structural integrity. Many numerical simulations based on finite element analysis (FEA) have been performed to evaluate the weld residual stress of weld-overlay cladding. Siegele and Brand have performed FEA simulations to evaluate the stress fields of cladded plates [2]. They indicated results that weld-overlay cladding generated tensile stress in the cladding and compressive stress in the base metal close to the cladding. In our previous study, FEA has been performed, which treats weld-overlay cladding and PWHT, to evaluate the residual stress distribution in and near the cladding [3-6]. Experimental measurements based on sectioning and deep-hole-drilling (DHD) techniques have also been conducted to confirm the accuracy of numerical simulations. The results showed that the stress distribution obtained from the FEA almost agreed with the experimental results, e.g., tensile residual stress reaching to yield stress was generated in the cladding due to the difference in the thermal expansions between cladding and base metals. However, there is a difference in residual stress distributions between results from Siegele and Brand and our results, particularly in base metal close to cladding, i.e. heat affected zone (HAZ). Weld residual stress distributions around the J-weld have been also evaluated in some papers based on FEA and inherent strain method [7-9]. Residual stress distributions have been shown that, for example, tensile stresses are generated at the inner surface of nozzle tube and around the J-weld. However, there are very few descriptions about both effects of the cladding and J-weld on the residual stress found in the previous papers. Since residual stress distribution in such regions may affect on structural integrity of vessel heads, the weld residual stress due to cladding as well as J-weld is to be evaluated with high accuracy.

It is well known that various conditions of residual stress, SCC growth rate and flow stress related to PWSCC have scattered and therefore affect the structural integrity of RPV. Since probabilistic fracture mechanics (PFM) analysis method can treat such scatter and uncertainties in the structural integrity assessment, it is a promising method for assessing the structural integrity related to PWSCC at a J-weld as an example. PFM analysis method is also expected to provide an advancement of maintenance such as optimizing inspection period from analysis on scatter of inspection accuracy. In JAEA, we have been developing some PFM analysis codes based on a Monte Carlo method focusing on degradation mechanisms for safety-related reactor components. The PFM analysis code for PWSCC is called as PASCAL-NP (PFM Analysis of Structural Components in Aging LWR – NiSCC / PWSCC) [10].

In the present study, a welding simulation method for three-dimensional cladded plate specimen based on FEA has been improved to obtain more reliable results of the weld residual stress distribution. For the improvement, the phase transformation in base metal as well as creep behavior was incorporated in the simulation of weld-overlay cladding and PWHT. The analytical results have been compared with the residual stress distributions obtained from DHD method to confirm the accuracy of the welding simulation. Weld residual stress near the J-weld and cladding for J-groove welded specimen with cladding has been then evaluated by using the improved three-dimensional welding simulation and DHD measurements. Applying the evaluated residual stress distributions to PASCAL-NP, some case studies have been also performed on crack growth and leakage behavior. 2. Experimental 2.1. Cladded and J-Groove Welded Specimens

Cladded specimens are fabricated using thick plates of JIS SQV2A (equivalent to ASME

SA533B-1; called SA533B-1) ferritic steel (500L x 500W x 125t mm) as shown in Fig. 1. Weld-overlay cladding was performed by a single layer electro-slag welding (ESW) method in this study. Welding conditions are listed in Table 1. In this case, the electrode of Type 309L stainless steel (SS) is used to make weld metal similar to be Type 304L SS. After ESW, PWHT of about 615oC for 6.0 h was conducted.


52

To simulate a J-weld of VHPs, J-groove welded specimens are also fabricated through machining, buttering, J-groove welding and PWHT using the cladded specimen. Weld materials of the alloy 82 of Ni-based alloy are used for buttering and J-groove welding. Tungsten inert gas (TIG) welding is used for buttering and J-groove welding. Schematic flow of fabricating of J-groove welded specimen, and top-view of the J-weld specimen including welding directions of over-lay cladding and J-groove welding are illustrated in Fig. 2. Total numbers of welding-passes for buttering and J-groove welding are 40 and 107, respectively. PWHT of about 615oC for 1.2 h and 11 h were conducted after buttering and J-groove welding, respectively.

Fig. 1 Overlay welded test plate.

Table 1 Welding conditions of ESW.

ESW Material of band electrode Type 309L SS

Width of band electrode (mm) 50 Welding current (A) 995

Arc voltage (V) 26.0 Welding speed (mm/s) 2.5

Heat input (kJ/mm) 10.3 Number of pass (-) 9

(i) Weld-overlay Cladding and PWHT

(ii) Machining of penetration hole (R40)and J-groove

(iii) Buttering and PWHT

(iv) Insert the nozzle tube, J-groove welding and PWHT

(iv) Machining of penetration hole (R50)

1

2

3

4

5

6

7

8

9

Unit : mm

Welding direction of cladding

Pass no.

500

500 (24º)(37º)

Welding direction of J-groove

Location 1

Location 2

(a) Fabricating process of J-weld specimen (b) Top-view of J-weld specimen

Fig. 2 Schematic image of fabricating process of J-groove welded specimen.


53

2.2. Residual stress measurements

Sectioning method based on beam thinning technique was performed to measure the through-thickness distribution of residual stress for cladded specimen in as-welded and PWHT conditions. Beam shape specimens for the measurements were cut out from the center of cladded specimen by avoiding overlap regions of welded beads. Detail of experimental procedure of residual stress measurement by sectioning method is described in previous paper [3].

Steep and complicated distribution in residual stresses would be obtained even after PWHT around dissimilar metal welds, such as weld-overlay cladding, due to the difference of the thermal expansions between base metal and weld metal [3, 4]. Therefore, the DHD measurements for cladded specimens were carried out to evaluate the through-thickness distributions of residual stress for cladded and J-groove welded specimens. The DHD technique is a unique method to obtain detail of the residual stress by 0.2 mm pitch in depth. Details of the method are found in a separate paper [11]. A measurement point for the cladded specimen is the center of cladded specimen where corresponds to the center of weld bead. Measurement points for the J-groove welded specimen are illustrated in Fig. 2. Location 1 was selected to evaluate the residual stress distribution within the J-weld. Location 2 was selected to evaluate the residual stress distributions through cladding at the centers of cladded beads. 3. Analysis 3.1. Finite Element Analysis (FEA)

To obtain weld residual stress distribution due to overlay cladding and J-groove welding analytically, thermal-elastic-plastic-creep analysis is performed using Abaqus, which is a commercially available FEA code [12]. Welding methods assumed in the analyses are ESW for the single layer weld-overlay cladding, and TIG welding for the buttering and J-groove welding. 3.1.1. FEA Model

Geometry of the model of cladded plate specimen for the welding simulation of weld-overlay cladding is shown in Fig. 3. Shapes of the plate specimen and welded beads of the model were defined based on fabricated conditions as described above. As shown in Fig. 4, a half symmetric model is used for welding simulation of J-groove welded specimen to save the calculation time of FEA. In addition to this, welding simulations of buttering and J-groove welding are conducted based on lumping method of the welded beads as shown in Fig. 4 (b) (Same color region is treated as a one bead.); e.g. the number of J-groove welding passes is decreased from 107 of actual welding to 19 of FEA. Elements used in thermal analyses and elastic-plastic-creep analyses are three dimensional solid elements of DC3D8 and C3D8, respectively. Overlap of weld metal is neglected to simplify the analysis in this study.

Fig. 3 FEA model for a cladded specimen.


54

Buttering

J-weld

RPV base metal

Nozzle tube

Cladding

14.8

32.2

3.6

40.355o

(a) Overview of FEA model (b) Close up of J-weld (Unit: mm)

Fig. 4 FEA model for J-groove welded specimen. 3.1.2. Material Properties

Each material property is temperature dependent [3-6, 13]. In order to evaluate residual stress in with higher accuracy, the phase transformation of base metal is taken into account based on the continuous cooling transformation (CCT) diagrams of SA533B-1. The CCT diagrams used in this study consist of three CCT diagrams with different achieving maximal temperatures of 900, 1200 and 1400oC [14]. The relationship between stress and strain in plastic region is defined using the kinematic hardening law. Yield stress of each phase such as Martensite or Bainite within HAZ was estimated from correlation between measured hardness distribution and phase fraction distribution obtained from FEA. Thermal expansion coefficients for phases of base metal such as Martensite, Bainite and Austenite were determined from our experimental measurements [15]. The material properties of base metal were defined by considering the phase fraction which varied during welding. Heat transfer and radiation are considered at each surface of the model using a coefficient of heat transfer of 10 W/(m2 K) and an emissivity of 0.7. Back stress and equivalent plastic strain are reset to zero above 1200 oC in the elastic-plastic analysis.

afar c

y

q(x,0,0)

x

bc'

c

Qafar c

y

q(x,0,0)

x

bc'

c

Q

Fig. 5 Wide double-ellipsoid model.

3.1.3. Analysis of Weld-Overlay Cladding

Thermal analysis is performed to simulate temperature distribution during weld-overlay cladding by ESW method. A movement of welding heat source is simulated by sequential heat inputs to elements aligning along the welding direction. A double-ellipsoid model is usually applied to simulate the behavior of arc weld [16]. However, band electrodes of 50 mm in width are used for weld-overlay cladding. Therefore, in this study, the model has been modified by inserting the width of a bead in the center of the double-ellipsoid model. This “wide double-ellipsoid model” is depicted in Fig. 5.


55

Sequentially coupled thermal-elastic-plastic analysis has been performed to obtain stress distributions. In the thermal-elastic plastic analysis, adhesion of weld metal is considered at the same time of pass of heat source. 3.1.4. Analysis of Buttering and J-groove Welding

Whole processes for fabrication of J-groove welded specimen indicated in Fig. 2 such as machining, buttering, PWHT after buttering, J-groove welding and PWHT are simulated by FEA. The simulation is performed after mapping the residual stress distribution produced by ESW cladding and PWHT (as described in 3.1.3 and 3.1.5, respectively) into the J-weld FEA model by using map solution technique. Machining of penetration hole and J-weld groove are simulated by removing the meshes belonging to those regions. Movements of heat source during TIG welding for buttering and J-groove welding are simulated based on the same methodology as described in 3.1.3 using a normal double-ellipsoid model as a moving heat source [16]. Welding parameters of the double-ellipsoid model are determined by adjustments that analytical peak temperatures during welding become comparable to experimentally measured ones. Through the analyses of buttering and J-groove welding, the phase transformation of base metal based on the CCT diagrams is taken into account as the same way as weld-overlay cladding. Contact conditions without friction are applied to the interface between the outer surface of nozzle tube and the inner surface of penetration hole. 3.1.5. Analysis of PWHT

Creep analyses are performed to simulate PWHT under condition of 615oC after each welding process of weld-overlay cladding, buttering and J-groove welding. The durations of PWHT for after weld-overlay cladding, buttering and J-groove welding are 6.5 h, 1.2 h and 11 h, respectively. Norton creep law is adopted as creep constitutive equation in the analyses. The following equations are used as creep constitutive equations for each material. 24.5171082.2 σε ××= −cr& (SA533B-1) 88.9281049.1 σε ××= −cr& (Type 309L SS) (1) 11.6221068.1 σε ××= −cr& (Alloy 82) Where crε& is the equivalent creep strain rate (1/s) and σ is the Mises equivalent stress (MPa). The constants in the equations have been taken from experiments. Martensite produced by each welding process can be fully tempered by PWHT [15]. Therefore, each mechanical property in Martensite is gradually replaced into Bainite during PWHT to simulate the phase transformation of tempered Martensite. 3.2. Crack Growth Analysis 3.2.1. Analysis Methods in PASCAL-NP code

The crack growth rate (CGR) of PWSCC is known as scattered as shown in Fig. 6 [17]. Therefore, the structural integrity assessment including crack growth prediction becomes over-conservative when the upper bound CGR data is used in the deterministic manner. In the proposed code case of the JSME Rules on Fitness-for-Service, the mean CGR curve is used for considering such conservatism in the assessment. In such a case, PFM analysis can be the most useful and rational method to properly assess the structural integrity of nuclear reactor components with considering the scatters of experimental data and field experiences [18].

We have, therefore, used a probabilistic fracture mechanics analysis code PASCAL-NP for crack growth analysis. PWSCC may initiate at the inner or outer surface of cylinder-type structure of Ni-based alloys. PWSCC may also initiate at the surface of Ni-based alloy weldment such as J-groove weld. The orientation of the crack is axial and/or circumferential depending on the stress distribution. PASCAL-NP has a function to calculate the SIF for such various types of crack related to PWSCC. Figure 7 indicates the analysis flow chart of PASCAL-NP based on the Monte Carlo method. Using the flow chart, the leakage and/or break probabilities are evaluated after PWSCC growth analysis considering the scatters and uncertainties in CGR and flow stress. Details of PASCAL-NP are described in a separate paper [13].


56

3.2.2. Analytical Conditions

In order to evaluate the effects of scatter in the CGR of PWSCC, PFM analysis is performed at the J-weld of a VHP. Total iteration number of simulations based on Monte Carlo method was specified as 1,000,000 in this study so as to be enough for the evaluation of PWSCC growth behavior with the scatter of CGR. Both vessel head thickness, T, and the distance from nozzle inner surface to the base metal of vessel head, WL, are set as typical dimensions as shown in Fig. 8. In this study, WL was set to 55.1 mm based on the fabricated J- weld specimen as was shown in Fig. 4. T was set to 183 mm from a dimension of actual vessel head thickness. Operating temperature and pressure are set to 318 ºC and 15.4 MPa, respectively [19]. The material and temperature dependencies on CGR are considered in the analysis. The mean CGR of Ni-based alloy weld metal is higher than that of base

1 10 1001E-8

1E-7

1E-6

Exp. (325oC) Ave. (325oC) +2σ (325oC) +1σ (325oC) -1σ (325oC) -2σ (325oC)

Cra

ck g

row

th ra

te, d

a/dt

(mm

/sec

)

Stress intensity factor, KI-9 (MPa m0.5)

Fig. 6 Crack growth rate against stress intensity factor for PWSCC in Ni-based welds [17].

(i) Setting the analytical conditions・Piping geometry・Crack type・Residual stress・Operating condition・Initial crack size ・Total sampling number , etc

(iv) Statistical outputs

(ii) Sampling random valuables・Crack growth rate

(log-normal distribution)・Flow stress of material

(weibull distribution)

(iii) Evaluation of crack growthbehavior and integrity assessment・Time history of crack size・Leak / Break judgment

End of sampling ?

Yes

No

Mon

te C

arlo

Met

hod

End

Start

Fig. 7 Analytical flow chart of PASCAL-NP [10].

O

J-weld

T

a

b

aL

Nozzle tube

t

WL

Ri

RPV base metal

Crack

Items (mm) Inner radius of nozzle, Ri 35.2

Nozzle thickness, t 14.8 Vessel head thickness, T 183

WL 55.1 aL 32.2

Crack depth, a 0.3 Crack half length, b 0.9

Fig. 8 Geometry of J-weld for PFM analysis.


57

metal [17, 20]. Furthermore, the scatter of CGR of weld metal is smaller than that of base metal [20]. In PFM analysis, these scatters of CGR are expressed by log-normal distribution. A SIF calculator incorporated into PASCAL-NP based on API [21] is applied to the crack growth analysis. Initial crack depth is set as the size that the SIF at the deepest point of crack exceeds slightly the threshold value, Kth, of 9.0 MPa m . The aspect ratio, that is length per depth, of initial semi-elliptic crack is assumed to be six which is a typical value for a postulated crack in the assessment of structural integrity. In this study, the depth and half length of initial semi-elliptic crack were defined as 0.3 mm and 0.9 mm, respectively. The J-weld is judged as the occurrence of leakage when the deepest point of crack reached the full thickness of J-groove weld (32.2 mm). 4. Results and Discussion 4.1. Residual Stress Produced by Weld-Overlay Cladding

In residual stress analysis considering phase transformation of base metal, the accuracy of thermal analysis is important since the transformation temperature and phase fraction strongly depend on cooling rate during welding. Shown in Fig. 9 (a) are temperature histories during weld-overlay cladding obtained from experiment and analysis. The shape of molten pool obtained from analysis is shown in Fig. 9 (b). Penetration depth measured by macroscopic observation was around 0.4 mm for the ESW-cladded specimen. By adjusting the heat input parameter Q and shape parameters a, b and c, the molten pool shape simulated for ESW was found to be in good agreement with experiments. Temperature history such as peak temperature and cooling rate also agreed well as shown in the figure.

0 5000 10000 15000

0

200

400

600

800

1000

865

Tem

pera

ture

(o C)

Time (s)

Experiment Analysis836

Penetration depth ~ 0.4mm

(a) Temperature history (b) Molten pool shape

Fig. 9 Temperature history by FEA and experiment measured at the same positions of 15mm from cladded surface, and molten pool shape from FEA. (Gray region corresponds to molten pool.)

Calculated residual stresses after ESW cladding are shown in Fig. 10. The longitudinal stress

which is parallel to the welding direction became tensile in the cladding and compressive just below the cladding, and tensile below the compressive area in the plate. Residual stress at the interface area between weld beads in base metal indicated the maximum value of tensile stress. These analysis results qualitatively correspond to an experimental one measured by R. Kume et al. [22]. It can be seen that tensile stress became high as welding pass increased. This is because tensile stress in the longitudinal direction generated by preceding welding was reduced by subsequent welding. This has also been agreed with the results reported by Siegele and Brand [2].

Figure 11 represents the through-thickness distributions of longitudinal residual stress at the center of ESW specimen obtained from FEA with and without considering phase transformation. An experimental result measured by sectioning method [3] is also shown in the figure. It is clear that the accuracy of FEA was significantly improved by taken the phase transformation into account. In particular, the phase transformation improved to simulate compressive residual stress within base metal close to the cladding. Such compressive stress could be occurred by the phase transformation


58

behavior. In the case of SA533B-1 steel, Martensitic and Bainitic phases with different yield stress levels and with different volumes precipitate from Austenite phase during cooling. Due to the volume changes from Austenite to Martensite and Bainite as well as internal constraint from the plate, compressive stress could be generated. In order to evaluate the residual stress with high accuracy, therefore, it is important to take the phase transformation into account. Similar investigation has been found in reference [23] by Y. Mikami et al. for a fillet welding T joint.

Weld sequenceWeld sequence

(a) Longitudinal


(b) Transverse

Fig. 10 Longitudinal and transverse residual stresses after cladding at the center of cladded specimen.

0 20 40 60 80 100 120 140-300

-200

-100

0

100

200

300

400

Long

itudi

nal r

esid

ual s

tress

(MPa

)

Distance from cladded surface (mm)

Sectioning FEA FEA (w/o Phase trans.)

Cladding

Fig. 11 Through-thickness distribution of weld residual stress after cladding for cladded specimen.

Figure 12 shows residual stress distribution at the center of welding line after PWHT condition

for ESW cladded specimen. Residual stress within base metal near the cladding was dramatically relaxed by PWHT although residual stress around the interface between the weld beads still remained tensile stress. In the cladding, high tensile stress almost equivalent to yield stress was generated due to the difference in thermal expansions between cladding and base metal.

The through-thickness distribution of longitudinal residual stress at the center of cladded specimen of ESW evaluated from FEA, sectioning and DHD techniques are plotted in Fig. 13. In the cladding, high tensile stress around 280 MPa was obtained from FEA and each experimental measurement. The residual stress distribution indicates very steep change across the fusion line from tensile stress in cladding to compressive stress in base metal when the phase transformation is considered. The DHD measurement also showed compressive stress around the same area. On the other hand, FEA results without considering the phase transformation provides no compressive stress

(MPa)

-300-200-100+0.0+100+200+300


59

in base metal near cladding. Therefore, the compressive residual stress in base metal near the cladding is due to the phase transformation. In addition to the improvement of welding simulation of weld-overlay cladding, the creep analysis has also simulated the phase transformation of tempered Martensite as mentioned in the section of PWHT analysis procedure. It is suggested from the results that taking the phase transformation into account is necessary to obtain highly accurate residual stress even after PWHT.


(a) Longitudinal


(b) Transverse

Fig. 12 Longitudinal and transverse residual stresses after PWHT at the center of cladded specimen.

0 20 40 60 80 100 120 140-200

-100

0

100

200

300

400

Long

itudi

nal r

esid

ual s

tress

(MPa

)

Distance from cladded surface (mm)

Sectioning DHD FEA FEA (w/o Phase trans.)

Cladding

Fig. 13 Through-thickness distribution of residual stress after PWHT for cladded specimen. 4.2. Residual Stress Produced by Buttering and J-groove Welding

Residual stress distributions in as-welded and after PWHT conditions for J-weld model are shown in Figs. 14 and 15, respectively. Hoop and radial directions correspond to longitudinal and transverse ones of J-groove welding since the figure is drawn based on the cylindrical coordinate system of the nozzle tube. Tensile radial stress in the cladding surface surrounding J-weld can be seen. The tensile stress area became larger after PWHT. On the other hand, very high tensile hoop stress at the surface region of J-weld was decreased after PWHT. Through-thickness distributions of radial and hoop residual stresses inside and near the J-weld are shown in Figs. 16 and 17, respectively, obtained from the FEA. Locations 1 and 2 correspond to the measuring points plotted in Fig. 2. Experimental results measured by DHD method are also plotted in these figures. It can be found that residual stress distributions evaluated by FEA and DHD method agree roughly, although some differences of residual stress values within the J-weld exists as shown in Figs. 16 (a) and 17 (a). For example, the

+400(MPa)

-200-100+0.0+100+200+300


60

hoop stress from FEA is ~200 MPa higher than DHD measurement in the J-weld. The differences can be due to the effect of lumping of welded bead. The use of relatively large element size near the J-weld may also cause the differences. Therefore, the causes of differences are studied further by improving FEA models and methods. The error in DHD measurements also has to be considered. Nevertheless, it is noted that residual stresses within the cladding and base metal in PWHT condition obtained from FEA agree very well with the experimental measurement as shown in Figs. 16 (b) and 17 (b). It is concluded that weld residual stress can be predicted with high accuracy by considering the fabricating process of cladding, buttering, J-groove welding and PWHT as well as phase transformation of base metal.

From the results shown in Figs. 14 to 17, tensile stress within the cladding, which had been generated by weld-overlay cladding and PWHT, remains through the fabricating process of J-groove welding and PWHT. Inside the J-weld, high tensile stress in the hoop direction was generated by J-groove welding. After PWHT, the results from DHD measurement showed that the tensile stress in both directions within the J-weld decreases to less than about 200 MPa. Relaxation of the tensile stress can be also shown by creep analysis although the degree of decrease in the tensile stress was less than that of DHD measurement. Although there are such differences in residual stress values between FEA and experiment, it is noted that tensile stress in J-weld remains even after PWHT in the same manner as tensile stress within the cladding. It is also interesting that higher tensile stress is seen in buttered region, where is drawn by orange in Figs. 16 (a) and 17 (a), as compared with that of J-weld. The cause of this difference has not been made clear yet. As shown in Figs. 16 (b) and 17 (b), both results from FEA and DHD measurement suggested that tensile stress values within the cladding are increased by conducting PWHT. Therefore, the PWHT for the relief of the residual stress due to J-groove welding may have an influence on the increase in tensile stress within the cladding around the J-weld.

(a) Radial stress (b) Hoop stress

Fig. 14 Residual stress distribution in as-welded condition of J-weld specimen with cladding.

(a) Radial stress (b) Hoop stress

Fig. 15 Residual stress distribution after PWHT condition of J-weld specimen with cladding.

-200-100+0.0+100+200+300+400+500

(MPa)

-200-100+0.0+100+200+300+400+500

(MPa)


61

0 20 40 60 80 100 120

-200

-100

0

100

200

300

400

500

R

adia

l res

idua

l stre

ss (M

Pa)

Through-thickness distance from welded surface (mm)

As weld (FEA) After PWHT (FEA) As weld (DHD) After PWHT (DHD)

Buttering

J-weld

0 20 40 60 80 100 120-200

-100

0

100

200

300

400

500

Rad

ial r

esid

ual s

tress

(MP

a)



Cladding

(a) Location 1 (J-weld) (b) Location 2 (Cladding)

Fig. 16 Comparison of radial residual stresses between as-welded and PWHT conditions.

0 20 40 60 80 100 120

-200

-100

0

100

200

300

400

500

Buttering

Hoo

p re

sidu

al s

tress

(MPa

)



J-weld

0 20 40 60 80 100 120-200

-100

0

100

200

300

400

500

H

oop

resi

dual

stre

ss (M

Pa)



Cladding

(a) Location 1 (J-weld) (b) Location 2 (Cladding)

Fig. 17 Comparison of hoop residual stresses between as-welded and PWHT conditions. 4.3. Crack Growth Analysis for PWSCC

Using the residual stress distribution within J-groove weld measured by DHD method, PWSCC growth behavior was calculated by using PASCAL-NP. The stresses used in the analysis correspond to the hoop stress since a crack was assumed to exist along radial direction. In this case, it is assumed that a crack initiated at the welded surface of J-weld propagates to the crevice region between vessel head base metal and the nozzle. We note that a fourth-order polynomial curve fitting for the SIF calculation based on API [21] was a good approximation of the hoop stress distribution within the J-groove weld. Figure 18 indicates the results of crack growth behavior and distribution of crack depth as a function of operation time calculated by (a) deterministic, (b) probabilistic and (c) probabilistic methods. It is obtained from Fig. 18 (a), for example, that CGR of average+1σ and average+2σ cases become around 3.4 and 1.9 times faster, respectively, than the average CGR case. Large scatter of crack depth is recognized clearly from probabilistic analysis result shown in Fig. 18 (b). This is due to the large scatter of CGR of PWSCC as indicated in Fig. 6. The average CGR becomes very slow after around 10 years of operation as shown in Fig. 18 (c). This is possibly explained from the facts indicated in Fig. 18 (a) that cracks which have higher CGR reach the depth of J-weld groove in early stage of operation. For example, in case of CGR of average+1σ, leakage occurs before ~4 years.


62

To evaluate the effect of residual stress within the J-weld on probability of leakage, comparison was carried out using residual stress distributions obtained from FEA and DHD measurement as shown in Fig. 17 (a). Since there is a possibility that the residual stress distribution has a scatter of beyond 100 MPa caused by uncertainties of welding parameters [24], the residual stress distributions were used for the case studies, i.e., FEA stress for a higher case, and DHD measurement stress for a lower case. Figure 19 shows (a) the cumulative probabilities of leakage and (b) the average value of SIF at the deepest point of crack tip as a function of operating time. The average SIF shown in Fig. 19 (b) was calculated as the arithmetic mean of SIF values of all cracks not having reached the full thickness of J-groove weld. As shown in Fig. 19 (a), the probability of leakage based on FEA stress becomes higher from the early stage than that of DHD measurement. Operating times when the probabilities of leakage become 50% for FEA and DHD measurement are 3.5 and 7.8 years, respectively. The higher probability of leakage from FEA is caused by higher average SIF value for FEA than that of DHD measurement as shown in Fig. 19 (b). After 1.5 years from operation start, for instance, SIF based on FEA stress reaches about three times higher than DHD measurement. The difference in SIF value is due to the difference in the hoop stress distribution in J-weld of around 200 MPa between FEA and DHD measurement as shown in Fig. 17 (a). Therefore, it is necessary to evaluate the residual stress distribution within the J-weld with high accuracy for structural integrity assessment related to PWSCC behavior.

At present, PASCAL-NP can consider uncertainties of CGR and flow stress of the materials of interest. In case of PWSCC, taking uncertainties of weld residual stress, inspection accuracy and initial crack sizes into account is necessary for the structural integrity assessment based on PFM analyses. Further improvement of PASCAL-NP is currently ongoing so as to be applied to any countermeasures for PWSCC.

0 5 10 15 20 25 300

5

10

15

20

25

30

Aver

age

of c

rack

dep

th (m

m)

Operating time (year)

Crack growth rate Average+2σ Average+1σ Average +1s Average-1σ Average-2σ

(a) Crack depth from deterministic analysis

0 5 10 15 20 25 30

0

5

10

15

20

25

30(%)


Cra

ck d

epth

(mm

)

0.010000.022270.049590.11040.24600.54771.2202.7166.04913.4730.00

0 5 10 15 20 25 30

0

5

10

15

20

25

30

Aver

age

of c

rack

dep

th (m

m)

Operating time (year) (b) Probabilistic density of crack depth (c) Average crack depth by PFM analysis Fig. 18 Crack growth behavior and distribution of crack depth as a function of operating time calculated

based on the stress distribution from DHD measurement.


63

0 5 10 15 20 25 30

0

20

40

60

80

100

C

umul

ativ

e pr

obab

ility

of le

akag

e (%

)


DHD FEA

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 5 10 15 20 25 300

20

40

60

80

100

SIF

(FE

A) /

SIF

(DH

D)

SIFFEA/SIFDHD

DHD FEA

Aver

age

SIF

at th

e de

epes

t poi

nt o

f cra

ck ti

p(M

Pa

m0.

5 )

Operating time (year) (a) Cumulative probability of leakage (b) Average SIF at the deepest point of crack tip

Fig. 19 Cumulative probability of leakage due to PWSCC and SIF at the deepest point as a function of operating time in J-weld calculated by PASCAL-NP.

5. Conclusion

Thermal-elastic-plastic-creep analyses considering the phase transformation of base metal have been performed based on three-dimensional finite element method to evaluate residual stress distribution produced by weld-overlay cladding, J-groove welding and PWHT. Using the evaluated residual stress, the effect of residual stress on the structural integrity of RPV related to PWSCC has been assessed by using PFM analysis code of PASCAL-NP. From these analytical results, the following conclusions are drawn:

(1) By taking the phase transformation during weld-overlay cladding and PWHT into account, it was shown that residual stress distribution from FEA agreed well with that of experimental measurement. FEA results without considering the phase transformation could not provide compressive stress in base metal near cladding. In order to evaluate the weld residual stress near the cladding with high accuracy, the phase transformation of materials during not only welding but also PWHT should be considered.

(2) To evaluate the residual stress in and near the J-weld, the consideration of the phase transformation also has an influence on the improvement in accuracy of analysis results. However, some differences in residual stresses between FEA and DHD measurement within J-weld were found. From a viewpoint of FEA, the differences are necessary to be minimized by improving the FEA methods.

(3) Within the cladding, tensile residual stress produced by weld-overlay cladding remains after the fabrication process of J-groove welding including the PWHT.

(4) The effect of scatter of CGR on PWSCC behavior was quantitatively evaluated based on the crack growth analyses using the residual stress distributions within J-weld and operating loads.

(5) Since the CGR of PWSCC has a large scatter, it is suggested that the residual stress should be evaluated with high accuracy for structural integrity assessment including crack growth analysis.

Acknowledgement This study was performed under the contract between Nuclear and Industrial Safety Agency of

the Ministry of Economy, Trade and Industry of Japan and the JAEA.

References [1] P. Riccardella, N. Cofie, A. Miessi, S. Tang and B.Templeton; “Probabilistic Fracture Mechanics Analysis

to Support Inspection Intervals for RPV Top Head Nozzles,” Proceeding of the Conference on Vessel Penetration Inspection, Crack Growth and Repair, (2003).


64

[2] D. Siegele and M. Brand; “Numerical Simulation of Residual Stresses due to Cladding Process,” Proceedings of 2007 ASME Pressure Vessels and Piping Division Conference, PVP2007-26586 (2007).

[3] M. Udagawa, J. Katsuyama, and K. Onizawa; “Residual Stress Analysis of Weld Overlay Cladding on the Structural Integrity of Reactor Pressure Vessel,” Japan Welding Society, Proceedings of symposium on welding structure (2006), (in Japanese).

[4] M. Udagawa, J. Katsuyama and K. Onizawa; “Effects of residual stress by weld overlay cladding and PWHT on the structural integrity of RPV during PTS,” Proceedings of ASME Pressure Vessels and Piping Division Conference, PVP2007-26556 (2007).

[5] H. Nishikawa, J. Katsuyama and K. Onizawa; “Effects of Weld-overlay Cladding on the Structural Integrity of Reactor Pressure Vessel during Pressurized Thermal Shock,” Quarterly Journal of the Japan Welding Society Vol. 27, pp. 245s-250s (2009).

[6] H. Nishikawa, J. Katsuyama, M. Udagawa, M. Nakamura and K. Onizawa; “Weld residual stress evaluation of reactor pressure vessel considering material property changes of heat-affected zone due to weld-overlay cladding,” submitted to Transactions of the Japan Society of Mechanical Engineers (in Japanese).

[7] Japan Nuclear Energy Safety Organization; “Report of Integrity Assessment of Flawed Components with Structural Discontinuity,” pp. 45-81, (2004), (in Japanese).

[8] D. Rudland, G. Wilkowski, Y.-Y. Wang and W.Norris; “Analysis of Weld residual Stresses and Circumferential Through-Wall Crack K-solutions for CRDM Nozzles,” NUREG/CP-0180, (2003).

[9] K. Nakacho, T. Ohta, N. Ogawa, N. Ma, H. Hamaguchi, M. Satou and M. Nayama; “Measurement of Welding Residual Stresses of Reactor Vessel by Inherent Strain Method - Measurement of Residual Stresses of Pipe-Plate Penetration Joint -,” Quarterly Journal of The Japan Welding Society, 25-4, pp.581-589, (2007), (in Japanese).

[10] M. Udagawa, J. Katsuyama and K. Onizawa; “Study on PWSCC Behaviors at Nickel-Based Alloy Welds Based on Weld Residual Stress Analysis and Probabilistic Fracture Mechanics,” Proceedings of 2010 ASME Pressure Vessels and Piping Division Conference, PVP2010-26028 (2010).

[11] E.K. Kingston, D. Stefanescu, A.H. Mahmoudi, C.E. Truman and D.J. Smith; “Novel applications of the deep-hole drilling technique for measuring through-thickness residual stress distributions,” Journal of ASTM International, April 2006, Vol. 3, No. 4 Paper ID JAI12568 (2006).

[12] Dassault Systemes Simulia Corp.; ABAQUS analysis user’s manual, version 6.9 (2009). [13] M. Udagawa, J. Katsuyama and K. Onizawa; “On the evaluation of PWSCC growth near J-groove welds

of vessel head penetration nozzle,” Proceedings of 8th international Welding Symposium, (2008). [14] J. Katsuyama, T. Tobita, Y. Nishiyama and K. Onizawa; “Mechanical and Microstructural Characterization

of Heat-Affected Zone Materials of Reactore Pressure Vessel,” Proceedings of 2010 ASME Pressure Vessels and Piping Division Conference, PVP2010-25561 (2010).

[15] J. Katsuyama, H. Nishikawa, M. Udagawa, M. Nakamura and K. Onizawa; “Assessments of Residual Stress Due to Weld-Overlay Cladding and Structural Integrity of Reactor Pressure Vessel,” Proceedings of 2010 ASME Pressure Vessels and Piping Division Conference, PVP2010-25541 (2010).

[16] J. Goldak, A. Chakravarti and M. Bibby; “A New Finite Element Model for Welding Heat Sources,” Metallurgical Transactions B, 15B, pp. 299-305 (1984).

[17] Japan Nuclear Energy Safety Organization (JNES); “Proposed curves for SCC growth rates of Ni-base alloys,” JNES-SS-0517, (2005) (in Japanese).

[18] D.L. Rudland, H. Xu, G. Wilkovski, P. Scott, N. Ghadiali and F. Brust; “Development of a New Generation Computer Code (PRO-LOCA) for the Prediction of Break Probabilities for Commercial Nuclear Power Plants Loss-Of-Coolant Accidents,” PVP2006-ICPVT-11-93802, (2006).

[19] D. Rudland, G. Wilkowski, Y.-Y. Wang and W.Norris; “Analysis of Weld residual Stresses and Circumferential Through-Wall Crack K-solutions for CRDM Nozzles,” NUREG/CP-0180, (2003).

[20] Japan Society of Mechanical Engineers (JSME); “Codes for Nuclear Power Generation Facilities - Rules on Fitness-for-Service for Nuclear Power Plants-,” JSME S NA1-2008, (2008) (in Japanese).

[21] American Petroleum Institute: Fitness-For-Service API 579-1/ASME FFS-1, (2007). [22] R. Kume, H. Okabayashi and M. Asano; “Mechanism of Underclad Cracking of Residual Strain and

Heat-Affected Zone Ductility,” Journal of Engineering Materials and technology, Vol. 98, pp. 348-366 (1976).

[23] Y. Mikami, Y. Morikage, M. Mochizuki and M. Toyoda; “Angular distortion of fillet welded T joint using low transformation temperature welding wire,” Science and Technology of Welding & Joining, Vol. 14, no. 2, pp. 97-105 (2009).

[24] J. Katsuyama, T. Tobita, H. Itoh and K. Onizawa; “Effect of Welding Conditions on Residual Stress and SCC Behavior at Butt-Welding Joints of Recirculation Pipes,” Proceedings of 2008 ASME Pressure Vessels and Piping Division Conference, PVP2008-61430 (2008).

evaluation of weld residual stress near the cladding and · pdf filewelding current (a) 995...

Documents