ML17059B507

Final Rept, Analysis of Nine Mile Point Unit 1 Shroud Weld V9 & Weld V10 Cracking.

ML17059B507
Person / Time
Site:	Nine Mile Point
Issue date:	04/30/1997
From:	Michelle Manahan, Stonesifer R External (Affiliation Not Assigned), MPM RESEARCH & CONSULTING
To:
Shared Package
ML17059B506	List: ML17059B505 ML17059B507 ML18040A255
References
MPM-497439, NUDOCS 9704250254
Download: ML17059B507 (260)

Text

-* r Analysis of Nine Mile Point Unit 1 Shroud Weld V9

/

and Weld V10 Cracking 9704250254 9704ii ADOCK 05000220 PDR PDR April, 1997 Copyright 1997 MFM Technologies, Inc.

All rights reserved

....SERVING CLIENT NEEDS Report No. MPM-497439 THROUGH ADVANCEDTECHNOLOGY

Executive Summar At the end of cycle 11, NMPC installed four tie rods in the annular region of the vessel between the shroud and vesse1 to mitigate cracking in the circumferential shroud welds. The tie rods are intended to provide support to the shroud so that it meets its intended function under normal operation and accident conditions with postulated 360 degree through wall cracking in the circumferential welds. In principle, the tie rods eliminate the need to inspect the circumferential welds. However, the shroud vertical welds are not protected by the tie rods and structural integrity of the vertical welds must be ensured through inspection and analysis.

At the end of the current refueling outage (RFO 14, cycle 12), significant cracking was observed in the NMP-1 shroud vertical and circumferential welds. The inspections were performed in accordance with the Reference [BWRVIP96) guidelines and include both enhanced visual (EVT) and automated ultrasonic (UT) examinations. The focus of the work reported here is on assessment ofthe weld V9 and V10 cracking. These welds are located between the H4 and HS circumferential welds and are exposed to neutron irradiation along their entire length. The peak V9/V10 axial fast (E > 1 Mev) neutron flux position is located 22.8 inches below the H4 circumferential weld. The V9/V10 cracking is almost exclusively on the OD side of the weld and is contained within the HAZ.

Surface examinations revealed short circumferential cracks within the nomirially 0.25 inch wide HAZ

'hich are, in many cases, connected to the long axial cracks. 'The axial cracks were determined by UT to be deepest near the H4 weld where the fast neutron flux is highest. The correlation between fast fluence and crack depth is illustrated in Figure ES-1.

I This study was conducted to gain an in-depth understanding of the cracking which has been observed in the V9 and V10 welds. An important issue related to the cracking is the fact that the cracking is predominantly on the OD side of the welds. As shown in Figure ES-2, this type of cracking would not be ~ed for as-welded stress profiles ofdouble V-groove welds because these profiles are tensile on both the ID and OD and these surface stresses are of sufficient magnitude to initiate stress corrosion cracking (SCC). Anotherprimary focus of the work has been to determine a safe allowable operating time until the next shroud inspection. These structural evaluation calculations have been compared with the results of calculations performed by GE [GE97] which are based on conservative BWRVIP criteria. As discussed later, the best estimate analyses demonstrate a large margin for 10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> of additional operation as proposed in [GE97J.

As shown in Figure ES-2, the V9 and V10 welding residual stresses are sensitive to weld heat input rates. This effect is a result of larger through wall temperature gradients for low heat input weld passes. The baseline case of Figure ES-2 is for the mid-range heat input of 51 kJfinch which was documented in the weld specification. Based on fracture mechanics calculations for shallow flaws, a stress of about 20 to 30 ksi is required to initiate IGSCC at 0.050 inch deep flaws. Higher stresses are generally required for circumferential flaws than axial flaws since the length of circumferential laws are limited to the 0.25 inch wide region of weld sensitized HAZ material (thus lowering the residual stress induced K, for a given flaw depth). Examination of the surface stress

magnitudes given in Figure ES-2 shows that the predominant OD cracking cannot be explained by the as-welded stress fields alone.

The shroud with its 1.5 inch wall thickness and 176 inch ID is a flexible structure which is difficultto weld. After welding, the shroud section between H4 and H5 would flex about 2 inches under its own weight when placed on its side. Such flexing can result in plastic deformation near the welds due to residual stress levels being at or near the material yield strength. Industry practice, plant records, and remnants of weld attachments on the ID of the NMP-I shroud all indicate that mechanical alignment procedures (i.e., jacking) were used during fabrication of the shroud to facilitate fit-up. The effects of diametral "squeezing" has been analyzed for the three weld heat inputs (baseline, 0.75 baseline, and 0.5 baseline). The diametral squeezing was found to result in stress Gelds which cause predominant OD cracking. The results of these analyses are summarized in Table ES-I.

Examination ofthe surface stresses indicates that predominant OD cracking would be maximized by a combination low heat input and a large diametral squeeze. It can be seen from Figure ES-2 that lower heat input lowers the ID surface axial stresses while raising the axial stress at the OD surface.

A diametral squeeze lowers both axial and hoop stresses at the ID surface to levels below that required to initiate a stress corrosion crack. In the case of the 0.5 baseline heat input, the dead weight ofthe shroud on its side after the last pass (welds at 6 and 12 o'lock) would be sufficient to produce a condition favoring predominant OD cracking (see "weld+2"squeeze +op" in Table ES-I).

Additional NDE crack profiles willbe needed in the future to determine the best estimate stress fields for welds V9 and V10.. However analyses.perfozmedjiave narromeNhe candidate. stress Gelds to a limited set of plausible solutions. The stress solutions for the 0.75 baseline heat input case, one of the plausible solutions, is shown in Figure ES-3.

Crack growth calculations have been performed using the GE fluence dependent crack growth model [BWROG94]. The GE model accounts for radiation enhanced segregation which increases electrochemical potentiokinematic reactivation (EPR). The model also includes radiation effects on electrochemical corrosion potential (ECP) and reduction of stresses due to irradiation enhanced creep. The GE model has been compared with crack growth data measured at NMP-I and the model predictions are in good agreement with the measurements. The results of the crack growth calculations for the V9 and VI0 welds are shown in Figure ES-4 for the 0.75 baseline heat input case.

propagation ~

The model predicts that the applied stress intensity (K, ) will be less than the threshold for SCC after about I. I inches ofcrack growth in the high fluence region. Since little data are available on IGSCC arrest thresholds in complex stress Gelds, it is not prudent to postulate arrest without NDE confirmation. Integrating the observed cracking backward in time using a best estimate crack growth model suggests earlier initiation for the higher flux regions of the shroud.

Typical crack initiation incubation times of 6 to 10 EFPY have been predicted for the analyzed residual stress profiles. For all analyzed cases, the OD cracks initiate under the axial stress (circumferential cracks) and propagate to a depth required to reach the SCC threshold of 8 to 10 ksiVin under the hoop stress Geld. After reaching this critical depth, the hoop stresses dominate the flaw growth behavior. The axial stress driven initiation at the surface is supported by the UT and visual examination results which show both short circumferential cracks and iong axial cracks. It should be recognized that OD surface initiation under hoop stresses is possible in localized regions

where hoop stress peaks occur.

Structural margin assessment calculations were performed for weld V9 because the NDE data showed the V9 cracking to be more limiting than that of V10. Although K, was observed to drop below Kiscc in some crack growth simulations, credit for crack arrest was not taken for structural evaluations ofweld V9. Linear elastic and elastic-plastic &acture mechanics (LEFM and EPFM) as weH as limit load calculations were performed using the bounding crack depth approximation shown in Figure ES-5. Since the NDE exaniinations indicate integrity of the H4 and HS welds in the vicinity of V9 and V10, credit was taken for the presence of the circumferential welds in the structural assessment calculations. The analyses show that structural integrity will be maintained for at least an additional 2 years of hot operation. Therefore, there is a large margin between the allowable operating time predicted using the BWRVIP criteria (10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> or about 1.2 years of hot operation to next inspection) and the results obtained using a conservative, best estimate model.

In summary, the analyses performed for cracks discovered in the NMP-1 shroud during the recent inspection have demonstrated that safe operation can be ensured for at least another cycle of operation without the need for inspection. Data &om future inspections of V9 and V10 can be used in conjunction with in-depth models, such as reported here, to benchmark and develop confidence in model predictive capability. The development of a benchmarked plant-specific model may be useful in the future for developing meaningful in service inspection schedules. As an option in future shroud NDE inspections, NMPC can obtain additional benefit from limited inspection of welds H4 and HS. Veri6cation of the integrity of these welds would result in increased margins of safety and longer inspection intervals for V9 and V10.

Weld V9 OD Fluence and Crack Depth Profiles (left side-of V9) 2.60000E+20 1.50 (U 2.40000E+20 5

1.25 Eo 2.20000E+20 i

0) 2.00000E+20 1.00 Z O O (D

1.80000E+20 <<s~Eii>> lh. i~il 1' ~ ~

0.75 ~

o '1.60000E+20 0.50 ~

1.40000E+20 0.25 1.20000E+20 0.0 0 10 20 30 40 50 60 70 80 90 Distance Along Vg.Measured from H4 (inches)

Figure ES-1 Correlation Between Fast Neutron Fluence and Crack Depth at NMP-I Vertical Weld V9

0 Heat Input Sensitivity Weld + Operating at 550 F 50 40 30 Vl 20 lh C>

L 10 l J

~1.5 baseline

~ ~ ~ 0

~

-f3- baseline 0.75 baseline

~0.5 baseline

-10

-20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.6 Distance from Inner Surface (in)

Heat Input Sensitivity WeId + Operating at 550 F 40 30 20 tO 10 th 0

CO CL o -10 O

x

-20

.i ~1.5 baseline i": ~baseline

-30

~0.75 baseline 40 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure ES-2 Axial and Hoop Stresses at Operating Temperature for the V9/V10 Welds as a Function of Heat Input During Welding VI

Table ES-I Surface Stress Summary for Several Weld Heat Input Cases Showing the Effect of Diametral Squcczc on the Strcsscs HAZ Surface Stress Summary (Baseline Weld Heat)

Hoop Stress (ksi) Axial Stress (ksi)

ID OD ID OD as welded 19.8 -3.7 39.9 42.6 welded + operating weld+ 4" squeeze+ op 16.6 90

-2.8

-1.9 29.1 195 '55 32.7 HAZ Surface Stress Summary (0.75 Baseline Heat Weld) as welded Hoop Stress-(ksi)

ID .'OD 0.2

'9A ID 37.1

'D Axial Stress (ksi) 43.8 welded + operating 23.6 0.0 26.0 33.6 weld+ 2" squeeze+ op 17.2 0.7 20.2 26.4 weld + 4" squeeze+ op 8.4 -OA 13.4 26.0 weld + 6" squeeze+ op -2.6 0.7 8.1 14.0 HAZ Surface Stress Summary (0.5 Baseline Heat Weld)

Hoop Stress (ksi) Axial Stress (ksi)

ID OD ID OD as welded 38.8 11.0 27.7 50.3 welded + operating 28.4 8.5 19.6 38.8 weld+ 2" squeeze+ op 15.3 7.1 13.8 31.7 weld + 4" squeeze+ op 62 5A 10A 28.7 weld + 6" squeeze+ op -5.0 5.2 7.0 19.4

V9N10 Weld HAZ Stresses 0.75 Baseline Heat input 60 50 40 N

30 O

20 X

10

~

I ~

~as welded weld plus operating without squeeze

'operating after deadweight squeeze

~operating after 4 inch diametral squeeze,

'operating after 6 inch diametral squeeze

-10 0.0 0.2 OA 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

V9/V10 Weld HAZ Stresses 0.75 Baseline Heat Input 40 30 20 I

I

'I Vl p

to .1P G.

0 0 -20

-30 40

~

~weld

~

plus operating without squeeze operating after deadweight squeeze operating after 4 inch diametral squeeze r

~operating after 6 inch diametral squeeze

-50 0.0 0.2 0.4 0.6 0.8 1.0 1.2 14 1.6 Distance from Inner Surface (in)

Figure ES-3 Axial and Hoop Stress Profiles for 0.75 Heat Input Case with Various Diametral Squeeze Deflections Applied

OD Cracking 0.75 Baseline Heat -4 in. Squeeze - Min Flux Region 1.6 i I

~ r L a,-axial i .

I l I stress driven 1.4

'l hoop stress driven, V9 near H5 weld 1.2 I ~ t3

't' a

1 i l r 'I * '

t I ~

~~ 0.8 I crack arrest (K < Kthres)

O I 0.6 .l ......g,, I I

1' O ~

0,4 0.2 1

I 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

OD Cracking 0.75 Baseline Heat -4 in. Squeeze- Peak Flux Region 1.6 l

'I 1.4 1.2 I

I j I I

r I ~

crack arrest (K < Kthres)

~~ 0.8 Cl 0.6 O - axial stress driven 0.4 hoop stress driven 0.2 t3 V9 deepest crack I I 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

Figure ES-4 OD Crack Propagation Results for 0.75'Baseline Heat Input Case Showing Axial Stress Driven Crack Initiation Followed by Hoop Stress Driven Radial Propagation IX

Weld V9 OD Flux and Crack Depth Profiles conservatively estimated crack depth at 2 EFPY of exposure after 1.5 cycle 12 4.80000E+11 1.4 1.3 v) 4.40000E+11 1.2

'E 4.00000E+11 1.1 Z O 1.0

'i I 0.9 C3 3.60000E+1 . L G)

X

~

• 'I I ~ ~

0.8 M'.7 bounding

~ 3.20000E+11 approximation at end of 06 5 O

cycle 12 0.5 G) 2.80000E+11 04 (0 z 0.3 measured at end 0.2 of cycle 12 0.1 2.40000E+11 0.0 0 10. 20 30 40 50 60 70 80 90 Distance Along Vg Measured from H4 (inches)

Figure ES-5 Measured V9 Cmck Profile Along with Bounding Approzimation at the End of Cyde 12 and at 2 Years of Hot Operation After Cyde 12

/ ~

Table of Contents Executive Summary.......

1.0 Introduction.........,.

2.0 Stress Field Characterization of V9 and V10 Welds . 3 2.1 Welding Simulations 3 2.1.1 2D Plane Strain Finite Element Model 4 2.1.2 Material Properties . 5 2.1.3 Welding Procedure .5 2.1.4 Welding Heat Transfer Simulations .6 2.1.5 Welding Stress and Deformation Simulations 7 2.1.6 Sensitivity Studies . 9 2.1.7 Welding Residual Stress Mechanisms 10 2.2 Operating Stresses .

2.3 Shop Loads ... 12 2A 3D Weld E6ects . 14 Model.....

2.4.1 The 3D Finite Element 2.4.2 Solutions for V9/VIO Weld Shrinkage 2.4.3 Radial Mismatch at Welds*V9'and'V10"..

2.4.4 Radial Mismatch at Welds H4 and HS Variation...... '

" --" 17 15 16 18 2.5 Stress Intensity Factor Calculations 20 2.5.1 Method . / 20 2.5.2 Comparison of Finite Element Stress Intensity Factors to Handbook Values 21 2.5.3 Stress Intensity Factor Solutions for the V9 and V10 Welds at Operating Conditions 22 3.0 Crack Initiation and Growth Models 68 3.1 Threshold Stress Intensity Factor (KQ 68 3.2 Comparison of Applied Stress Intensity (Kg to K~ 68 3.3 Crack Growth Calculation Model 70 3.4 Crack Growth Model Predictions 71 4.0 Neutron Flux Data ..... 81 5.0 Crack Growth Calculations .. 95 5.1 Implementation of the Plant-Specific Crack Growth Model 95 5.2 Crack Growth Simulations at the OD Surface of V9 ... 97 5.3 Predicted Crack Growth Behavior for Continued Operation . .... 100 5.4 Structural Assessment of V9 for Continued Operation 100 XI

6.0 Summary and Conclusions 7.0 References .

8.0 Nomenclature ..

~ i ~

~ ~ 'i >wc> t %<a t~

XII

1.0 Introduction The NMP-1 vertical welds V9 and V10 are located at azimuthal positions 10 degrees and 190 degrees, respectively, and are between circumferential welds FI4 (above) and H5 (below). In the vicinity of the V9 and V10 welds, the shroud inner diameter is 176 inches and the wall thickness is 1.5 inches (See Figure 1-1). 'During the non-destructive examination (NDE) of the shroud at the end of cycle 12, cracks were found in both vertical and circumferential welds in the shroud. After the initial discovery of cracks, the examination program'as expanded following the Reference

[BWRVIP96] guidelines. The'inspections included both enhanced visual (EVT) and automated ultrasonic (UT) examinations. Although cracking was discovered in many of the shroud welds, the focus of the work reported here is on assessment of the weld V9 and V10 cracking. NMPC requested MPM to perform an in-depth study of these welds to provide an explanation for the predominant OD cracking which was observed. This type of cracking would not be expected for as-welded stress profiles of double V-groove welds because these profiles generally have tensile stresses on both the ID and OD surfaces and these surface stresses are generally believed to be of sufficient magnitude to initiate stress corrosion cracking (SCC).

f Another primary focus of the work has been to determine the safe allowable operating time until the next shroud inspection using conservative, but accurate models of crack growth and fracture resistance. These structural "evaluation caIculations"have been'ompared with the results. of calculations performed by GE [GE97] which are based on conservative BWRVIP criteria. As discussed later in the report, the best estimate analyses demonstrate a large margin for the 10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> of continued operation proposed in [GE97]. I

/

R=94.5 4.0 6.0 R=93.0 39.25 2.0 18.5 H4 Weld R=88.0 1.5

.v. v ivan'<aW. iIIC (>UL.vtl4~

h

)'I Vv v v ~ L. lt vc '.I 158.12 v

/

HS Weld 68.0 8.75 Schematic Representation of the NMP-1 Shroud in the Vicinity of the V9 and V10 Welds

2.0'tress Field Characterization of V9 and

~

~

V10 Welds This section describes the calculations for characterizing the stress fields that affect IGSCC in the sensitized heat affected zones (HAZs) of the axial shroud welds known as V9 and V10. The V9 and V10 welds (V indicating a vertical weld in the finished shroud) join the two halves of the central mid cylinder. The two welds are 180 degrees apart. The material joined by these welds is type 304 stainless steel (SS) and the weld filler is type 308 SS. The weld was done in the horizontal position using the submerged arc process. At the time the V9 and V10 welds were made, the two half cylinders being joined by these welds were unattached to other portions of the shroud assembly.

The two horizontal welds, H4 and H5, joining the cylinder formed by V9 and V10 to the upper (H4) and lower (H5) shroud assemblies were the last two major welds in the shroud fabrication.

Since the applied loads for the shroud are quite small, the welding residual stresses are the most significant component of the stress fields that affect potential cracking in the V9 and V10 HAZs. The characterization of the welding residual stresses in this study was done using two types of simulations. The first type of simulation assumed plane strain behavior and used a two-dimensional (2D) elastic-plastic finite element model (FEM) to simulate the weld pass metal deposition sequence. This simulation provided baseline welding induced residual stresses.

However, in recognition of the. fact. that three-dimensional,(3D) features',of. the welding.,process (starts and stops, weld repairs, and fit-up variations, etc.) may have resulted in some perturbatioh of the idealized weld residual stress fields, a 3D linear elastic finite element model was used to determine the magnitudes and distributions of stresses associated with a series of weld shrinkage and fit-up assumptions. I

/

Weldin imulations The welding simulation methods that were used in this study were developed over the past twenty years beginning with NRC and Electric Power Research Institute (EPRI) supported work at the Battelle Memorial Institute and then continuing through direct industry support through consultation and analysis services provided by Computational Mechanics, Inc.. Because welding simulation involves modeling complex material and geometric behaviors such as high temperature plasticity, annealing, and-material deposition, welding simulation with common commercial finite element packages is difficult. As a result, the proprietary WELD3 [WEL3] software package was developed for doing welding simulations. Although using WELD3 software tends to simplify the process of doing welding simulations, there is more to welding simulation than using the correct software. Studies have shown that simplifications that are sometimes used by novice weld simulators lead to unrealistic residual stress predictions. Perhaps the most common and probably the worst mistake that one can make is to try to simulate a weld with numerous layers of weld passes as ifall of the weld metal is deposited at one time. This type of so-called weld simulation can lead to unrealistic residual stress magnitudes and distributions. Welding induces stresses through several simple mechanisms. Ifa welding simulation is approached as a means to understanding how these

0 mechanisms interact in a given weld geometry, for a given welding sequence, and for a given set of welding parameters; and due attention is given to studying the effects of assumptions made in the modeling process, then there is no more reason for a welding simulation to provide misleading predictions than any other finite element simulation.

The axial nature of the V9 and V10 welds makes it natural to assume plane strain behavior in the 2D finite element model. The small weld cross section c'ompared to the total shroud cross sectional area makes the plane strain assumption (i.e., zero axial strain) very reasonable.

Although stresses from 2D simulations tend to be representative of actual component stresses, the differences between a 2D weld simulation and the actual 3D welding process leads to the 2D model generally predicting smaller shrinkages than actually occur. The basic reason is that in the 2D model, the entire length of the weld is essentially being welded simultaneously. The relative size of the hot region in the 2D model is therefore much larger than in the actual case. This leads to the model thermally expanding during the time that heat is being input to the model to a larger extent than occurs in the real 3D geometry. In the real 3D geometry, cool material surrounds the weld region and therefore the initial thermal expansion of the weld puddle region is largely restrained. The contraction behavior during cooling is reasonably well represented by the 2D model because temperature gradients in the longitudinal weld direction for the cooling material tend to be small compared to gradients in the through thickness and transverse weld directions. However, since the thermal contraction,.is starting &om,a.more. thermally expanded geometry, than. in the real.3D case, the predicted net welding induced shrinkage is less than in the actual 3D geometry. Since the cooling process is mostly responsible for the creation of the residual stresses and is well represented by the 2D model, the welding residual stresses from the 2D simulations are generally good approximations to those in the 3D geometry.

WELD3 incorporates a special option for exploring the effect that the 2D assumption may have on the predicted shrinkages and stresses. The option attempts to introduce artificial 3D constraint on the weld region during the times when it is increasing in temperature. Past experience with the use of this option shows that it has a much more pronounced effect on predicted displacements and shrinkages than on predicted stresses [WELV]. The effect on the shrinkage predictions is geometry dependent. When the weld is on a relatively thin component that requires only a few weld passes, the predicted shrinkages can be nearly doubled by using this option. For thicker sections requiring more passes, the effect is generally less. This is because even in the 2D model, the cooler material tends to restrain thermal expansion of the weld region and therefore the 2D model tends to be a better approximation for simulating welds in thicker sections.

.1.1 2D lane train Finite Element Model The 2D plane strain model used in the welding simulations is shown in Figure 2-2. This model assumes symmetry about both the V9 and V10 weld center lines. It also assumes that interaction of the two welds is negligible. This latter assumption is possible due to the large compliance of the shell. This large compliance means that stresses induced at one weld by

I shrinkage at the opposite weld are very small. Since interaction of the welds is negligible, it is advantageous in the numerical simulation to assume that the two welds are made simultaneously.

This assumption leads to another plane of symmetry, which means that the finite element grid can be limited to one quarter of the shroud circumference. The section A-A is used for plotting the weld HAZ stresses.

2.1.2 aterial Pro erties The shroud base metal is type 304 SS. The weld metal is type 308 SS. The material properties used in this study are summarized in Tables 2-1 and 2-2. Table 2-1 summarizes the elastic and elastic-plastic mechanical properties. These properties are used as functions of temperature in the finite element simulations. The properties for type 304 SS were taken from Reference [Int]

except for the coefficient of thermal expansion which was taken from Reference [Aero]. The properties at the higher temperatures of these tables are often not reported and must be estimated.

The elastic and plastic moduli at 2100'F reflect the fact that the material cannot sustain much stress at temperatures in this range and are not based on measured values. The material properties for these very high temperatures are not critical to the calculation of residual stresses at temperatures of practical interest because the yield stress, and hence the stresses, are very low at these high temperatures.

i

~

/

The material properties.used;fordhe<type,3085S.weld, metal are assumed,to:be.the. same as for the type 304 SS base metal except for the yield stress. The yield stress of the weld metal is assumed to be 1.25 times the yield stress of the base metal. Although it is generally believed that the type 308 SS yield stress is above that of type 304 SS, yield stress data tends to be very inconsistent. The factor of 1.25 is believed to be a reasonable estimate and is not expected to have a large effect on the hoop residual stresses in the V9 and V10 HAZs. The coefficients of thermal expansion (CTE) for the two materials appear to be different in these tables. The difference, however, is in the reference temperature. Since the weld metal initially starts in a molten state, a reference temperature of 2100'F is used instead of 70'F. Both tables reflect the same material behavior. The heat transfer analysis used in this-study was based on an analytical solution for moving point heat sources. This solution does not permit temperature dependent properties to be used. The properties in Table 2-2 therefore represent average heat transfer properties over the range of temperatures experienced during welding.

2.1.3 Weldin Procedure The weld procedure for V9 and V10 was reasonably well documented in the P. F. Avery weld procedure sketches and welder records attached to [GE771]. The weld was a submerged arc weld. The weld groove was a symmetric (equal depth from inner and outer surfaces) double J type.

The sides of the grooves were 22.5 degrees and both root regions had a 0.25 in. radius. A minimum preheat temperature of 60'F and a maximum interpass temperature of 350'F was specified.

Welding was done in the "flat" position. The specified welding parameters were as follows:

~ass a . ~Am ~Vlt Travel Filler Dia.

1-12 240-325 25-31 10-20 ipm 5164" BALANCE 475-550 28-32 16-20 ipm 5/32" Although the welding sequence was not specified, the actual sequence could be discerned from the welder records. Passes 1-6 were root pass welds done at the inner surface. Passes 7-12 were root pass welds done at the outer surface. Weld pass 7, however, did riot immediately follow weld pass

6. Instead, the entire inner weld groove was filled before back chipping and welding the outer surface root passes. There is no record of the number of passes required to finish the inner and outer surface welds after the root passes were completed. However, recent inspections appear to indicate that the last weld layer of the V9 and V10 welds contained two passes.

.1.4 Weldin Heat Transfer imulation The WELD3 module called PLTTMP was used to perform the heat transfer calculations.

PLTTMP provides the finite element temperatures used in the thermal elastic-plastic stress calculations. PLTTMP uses an analytical solution to the problem of a constant velocity point heat source in a plate. In this model, the surfaces of the plate are treated as insulated. The PLTTMP calculations use the temperature independent conductivity and heat capacity values given in Table 2-2. r I

a A range of weld parameters were used in the weld pass thermal simulations. The mid range parameters (and thus mid range heat input) of the Avery weld specification is referred to as the baseline heat input case. This case resulted in a heat input rate of 32 kJ/in for the root passes and a heat input rate of 51 kJ/in for the fill passes. Since the analytical thermal model simulates the actual movement of the arc, and since the material properties were input in terms of BTU's and seconds these heat input rates were converted to time rates with the units of BTU:

~Pas e nut root passes 7.5 0.25 fillpasses 14.6 0.30 The submerged arc weld process is a very efficient process in the sense that nearly 100% of the electrical input energy gets into the weld as heat. A 100% efficiency was therefore assumed in this simulation. Heat input sensitivity analyses were run using 0.5, 0.75, and 1.5 times the baseline heat inputs given above.

Since the number of passes actually used to fill the weld grooves was not provided in the available weld documentation, PLTTMP was used to determine the minimum number of passes that could be reasonably used in the simulation for the heat input rates resulting from the mid range welding parameters. Since previous experience has shown that combining passes of a weld layer into one modeled pass provides reasonable stress predictions, and since there was no way to know the order in which passes of a weld layer were actually deposited, the procedure was to model passes

as though they extended across the entire weld groove width. PLTMP allows heat input to be spread over an entire weld layer by dividing the total heat input among several point sources distributed across the weld groove. Numerical experimentation indicated that the residual stresses based on the baseline thermal simulation were insensitive to the manner in which the heat input was distributed across the width of the weld layer.

With the heat input rates and travel speeds that were specified, it was found that each weld groove could be reasonably simulated as being completed with four weld layers. This was based on the fact that with this assumption the predicted temperatures for each pass (layer) results in the new pass material starting at a temperature that is well above the temperature at which the material can bear substantial stress. Ifmore passes were simulated, these high heat inputs would cause substantial portions of the previous pass to be heated to the temperature range in which stresses are totally relieved. This would not be expected to lead to substantially different residual stress predictions from those obtained using four passes (layers) on each side. However, weld shrinkages tend to be more affected by the number of weld passes than the stresses, and therefore predicted weld shrinkages could be smaller than actual shrinkages to the extent that the model used fewer passes than were actually applied. Figure 2-2 shows the manner in which the finite elements in the weld region were divided into a total of eight weld passes. The numbers in the weld elements indicate the order in which the weld passes were deposited in the simulation..

I I

A temperature history;for each:finite.element of the,2D plane strain model for each weld pass deposition was computed by PLTTMP. In this analysis, an initial temperature of 70'F was assumed and it was assumed that the weld region cools to 70'F before the next weld is deposited. There is nothing in the weld documentation that provides any information on these initial and interpass temperatures other than the minimum and maximum values of 60'F and 350'F. The results of this study are not believed to be sensitive to reasonable deviations from these assumed values but this was not verified.

The temperature history for each finite element is approximated in a piece-wise linear way by computing temperatures at specific times using PLTTMP and then allowing the WELD3 elastic-plastic stress calculation module (STRESS2D) to linearly interpolate betweerr these distributions for any other times that temperatures may be required. In this study, the temperature history for each pass was defined in terms of temperatures at five times. The first time was 1.5 seconds after the arc passes. The second, third, and fourth times were for when the new weld pass cooled to 1500'F, 1000'F, and 500'F. The final time was for when the entire weld region cooled to 70'F. Appendix A contains a number of temperature contour plots (isotherms) which illustrate these five temperature distributions. The distributions are referred to by the pass number plus one of the letters A, B, C, D, or E. Distribution 3A is the distribution for pass 3 at 1.5 seconds.

Distribution 3C is the pass 3 distribution when the weld has cooled to 1000'F, etc.

.1.5 Weldin tress and Deformation imulation Having the finite element temperature histories defined by PLTTMP, the elastic-plastic finite

element calculation for the welding simulation was done. The WELD3 module for this step in the analysis is STRESS2D. The temperature dependent properties of Table 2-1 are used in the STRESS2D calculations.

The final deformed shape predicted by the baseline welding simulation is shown in Figure 2-3. Because of the plane strain assumption, no axial shrinkage is predicted. It can be seen that a residual radial outward deflection of the weld region is predicted. The final outward deflection is predicted to be about 0.19 inches. This is substantially less than the 1.5 inch radial deflection that was predicted to exist after the completion of the ID welding. This radial deflection is not localized to the weld. When artificial 3D constraint was added to the model, the outward deflection was decreased to essentially zero (-0.02 in.).. Since the shrinkage behavior with the 3D constraint added to the model is generally believed to be more realistic, it appears that the symmetric weld groove pattern has the desirable effect of producing near zero weld induced bending type deflection of the shell.

Figure 2-4 shows the predicted weld induced axial and hoop residual stresses in the HAZs of the V9 or V10 weld at 70'F. These stresses resulted from using the baseline heat input defined above. Results both with, and without added artificial constraint are shown. The stresses are plotted as a function of distance from the inner shroud surface. The distance is the radial, distance and is not measured along the 22.5 degree path of the HAZ Rom which stresses were taken. The inside shroud surface is at the left side of the plots and the outside is at the right side. Since it is the hoop stress component that would drive axial cracks in the HAZ, these are the stresses of most interest.

However, the axial stresses, as can be seen, are substantially tensile at the inner and outer shroud surfaces and may play a significant role in IGSCC initiation.

/

The hoop stresses of Figure 2-4 show several interesting behaviors. First, the hoop stresses at the outer surface are substantially below the yield level that might be expected for the material in the last pass; and are actually predicted to be slightly compressive. Second, although the outer surface hoop stresses are not tensile, there are tensile hoop stresses just below the outer surface.

Third, although a significant portion of the mid wall tends to be in strong hoop compression, the inner surface has significantly tensile hoop stresses. The reasons (i.e., mechanisms) behind this hoop stress behavior can be discerned from the simulation and will be discussed below.

The distribution of hoop residual stress seen in Figure 2-4 is similar to the axial stress distributions that have been measured and predicted by welding simulations for circumferential welds in large diameter pipes with wall thicknesses greater than one inch [EP1743]. Figure 2-5, taken from [EP1743], shows both residual stress data and the results of a circumferential weld simulation for a 26 inch pipe with a 1.3 inch wall thickness. The hoop stresses (i.e., stresses acting transverse to the weld) from the simulations are in good agreement with the data, and both show a through wall distribution that is qualitatively similar to that predicted for the shroud V9 and V10 welds in this study. The key similarities are the low tensile or compressive OD stresses with a more highly tensile zone beneath the OD surface, a mid wall compressive zone, and highly tensile ID stresses. It can be seen that the hoop (circumferential) stresses are at tensile yield in the pipe but that

they are not tensile at all points tlirough the wall. This points out the fact that while the transverse weld direction residual stresses for the two geometries are similar, the large difference in the radius to thickness ratio (R/t) does lead to differences in behavior. It is also worth noting here that the pipe weld was a single sided weld. Reversing the weld sequence in the shroud so that the ID is welded last, would be expected to result in a much different axial distribution from that obtained above.

The data of Figure 2-5 represents two pipes. Similar data has been obtained and compiled for other pipes and has been presented in Appendix A of [NUREG0313]. Figure 2-6 is a reproduction of Figure 3 from this appendix. It can be seen that the data show a trend that is very similar to that of the hoop stress predictions for the V9 and V10 welds. The deeper penetration of the yield level ID stresses in the V9 and V10 welds is believed to be the result of the decreased resistance to bending that is offered in the present, large R/t, axial weld configuration (plane strain) as compared to the circumferential weld configuration (axisymmetry).

2..6 en. itivi tudies It was described previously that the WELD3 sofbvare offers an option which is intended to introduce into the 2D model some of the constraint that would be operative in the real 3D geometry.

Since this constraint is introduced without actually introducing restraining forces in the model, this constraint is referred to as artificial 3D constraint. Comparing the solutions in Figure 2-4, it can be seen that the effect on stresses of including the artificial constraint, is not large,.for this weld problem.

Due to this lack of significant effect; all.subsequent, analyses, were. done without:;artificial,3D constraint.

Figure 2-7 shows the radial shrinkage behavior from the welding simulation. Each point on the curves corresponds to one of the five times used in computing the finite element temperature histories. One curve is from the solution without artificial 3D constraint and the other is from the solution with artificial constraint. It can be seen that the addition of the constraint had only a small effect on the radial deflection behavior.

Figure 2-8 shows the effect of weld heat input magnitude on the predicted residual stresses.

It can be seen that the effect of the heat input is significant. The effect of heat input was more significant than was expected based on experience on circumferential welding on comparatively smaller R/t (i.e., pipe) geometries. The most significant effect of increasing the heat input is to decrease the hoop stress magnitudes. The increase in heat also significantly increases the mid wall axial stress magnitudes. The decrease in the hoop stresses with an increase in heat input can best be understood by noting first that the increase in heat leads to smaller through thickness temperature gradients, and second that the higher heat input and associated high temperature region tends to relieve the stresses created by the previous layer of passes.

The hoop stress levels predicted near the OD surface of the V9 and V10 heat affected zones for the baseline heat input case are relatively low. Although the addition of constraint via the WELD3 artificial constraint option did not significantly alter the predicted stresses, there was

concern that the model may be under predicting the OD hoop stress magnitudes due to some inherent shortcoming of the 2D model. A number of attempts at introducing a 3D type of constraint into the 2D model by means other than the WELD3 artificial constraint were made. None of these analyses resulted in significantly different stresses. A number of simulations were done in which the effect of modeling entire layers of passes (i.e., two passes) being deposited simultaneously were explored.

Again, it was found that the resulting stresses were largely unaffected. It was concluded that the heat input was the only model input that had a significant effect on the predicted residual stresses.

Z.1.7 Weldin Residual tress Mechanisms Probably the most interesting aspect of the predicted hoop residual stress distribution shown in Figure 2-4 is the tendency for low stresses at the outer surface. The two fundamental driving forces for welding residual stress and deformation are shrinkage and shrinkage. The first shrinkage is the shrinkage parallel to the welding direction (axial. for V9/V10) and the second shrinkage is shrinkage transverse to the welding direction (hoop for V9/V10). In axisymmetric geometries the circumferential shrinkage gives rise to yet a third shrinkage and that is radial shrinkage. For the V9 and V10 axial welds (planar welds rather than axisymmetric), the radial shrinkage effect is not present.

The transverse weld shrinkage gives. rise to hoop stresses. These stresses will vary through the thickness ifthe. wall thickness is large enough that significant through. wall.temperature gradients can exist. This is clearly the case for the shroud. The transverse shrinkage tends to result in tension in the newly deposited material and compression in the material beneath the newly deposited material. Basically the hotter, outer material wants to shrink during cooling but cannot shrink as much as it would like because the adjacent cooler material does not want to shrink as much as the hotter material. This puts the hotter material in tension and the cooler material in compression. This explains the tendency for compressive transverse weld stress in the mid wall region of the shroud.

Unless there is a some external restraint or variation in stress along the length of the weld, the transverse tension at the section must balance the transverse compression.

For circumferential welds (i.e., H4 and HS), the large yield level hoop stresses in the most recently deposited pass can be thought of as a band being shrunk onto the shroud at the weld region.

This band causes an inward deflection of the wall at the weld,. This inward deflection induces bending in the wall which affects the transverse weld stresses in the weld region. This bending that is induced during weld cooling tends to induce tensile axial stresses at the ID and compressive axial stresses at the OD. This behavior, when combined with the above transverse shrinkage behavior, therefore explains not only the reduced tension levels at the OD but also the large tension stresses at the ID.

For the axial welds V9 and V10, and for plate welds in general, the radial shrinkage induced bending mechanism is not operative. However, the transverse weld shrinkage can still lead to the creation of the bending type of deformation in the weld region that leads to increased ID tension and reduced OD tension. This bending results from the transverse shrinkage behavior not being 10

symmetric with respect to the mid wall. The nonsymmetry results from the hotter outer surface wanting to contract in the transverse direction more than the cooler inner surface. Since bending of this type is largely unrestrained (unlike the axisymmetric geometry), the effect of this bending can have a significant effect on the through thickness transverse weld stress distribution. The relatively deep penetration of the ID tensile hoop stresses predicted for the V9 and V10 welds is largely the effect of this transverse weld shrinkage induced bending. The decrease in OD transverse stress to, near zero levels can be understood by noting that the through thick'ness temperature gradients are not linear, and therefore cannot be accommodated by the linear strain field of the induced bending deformation without inducing residual stresses. In, effect, the cooling of the subsurface weld metal induces bending at a rate that effectively keeps shrinkage induced tensile stresses from building up at the OD.

The tendency for transverse and radial shrinkage induced bending to result in relatively low transverse stresses at the OD is seen in the large diameter pipe axial stress data of Figure 2-6. It is also apparent in the axial stress data from a horizontal shroud weld (H5) shown in Figure 2-9 [Pa96].

Although the data of this figure do not extend to the OD surface, the tendency for a decrease in tensile stresses as the OD surface is approached is apparent. The data also show the tendency for a mid wall compressive zone and a highly. tensile inner surface. While this data is consistent with the mechanisms that are described above and which are active in the V9 and V10 welds of this study, there are several differences in the geometry of the H5 weld of Figure 2-9 an'd the V9/V10 welds of this study that make their residual stresses quantitatively. different.* The first major difference is.the fact that the H5 weld is a circumferential weld. This means that the radial shrinkage mechanism is active, and that the hoop constraint of the axisymmetric geometry can resist bending due to transverse shrinkage. The second is that the data of Figure 2-9 are from a 2 inch thick Ishroud.

/

.2 eratin tre se Operating conditions for the V9/V10 region consist of a 6 psi outward pressure difference, an upward axial force due to the same 6 psi pressure acting on the shroud head, and a downward axial force due to the deadweight of the upper portion of the shroud. The 6 psi head load results in a net upward force of 163 kips. The deadweight load is downward with a magnitude of 130 kips.

The net force is therefore 33 kips upward. This force results in a net axial stress in the section above the central ring of 37 psi tension. The 6 psi outward pressure induces hoop stresses in the wall of about 350 psi. It can be seen therefore that the stresses due to the mechanical loads are quite small.

The most significant aspect of applying the operating conditions is the increase in temperature to 550'F. To allow axial thermal expansion in the plane strain model, the coefficient of thermal expansion was set to zero before increasing to operating temperature. Although this uniform temperature change induces no stresses due to differential thermal expansion, the decrease in yield stress does impact the predicted residual stresses. This is illustrated in Figure 2-10 for the baseline weld heat input analysis. The largest change in stress is seen in the axial stresses. Although 11

the hoop stresses at 70'F were already largely below the uniaxial yield at 550'F, the hoop stress magnitudes are also reduced by the heating to 550'F. This is believed to be primarily a Poisson ratio type effect driven by the axial stress reduction. Figure 2-11 summarizes the operating condition stress distributions from the four weld heat input levels that were simulated in this study. By comparing with Figure 2-8, it can be seen that the hoop stresses near the outer surface are decreased by going to operating conditions. The amount of the decrease is lagger for the lower heat input stress distributions.

Figures 2-12 and 2-13 contain contour plots of. the axial and hoop stresses at operating conditions based on the baseline heat input simulation. Figures 2-14 and 2-15 contain contour plots of the axial and hoop stresses at operating conditions based on the 0.5 baseline heat input simulation.

It can be seen from the axial stress contours that the through thickness HAZ stress plots that have been presented to this point would not change substantially by selecting a slightly different section through the HAZ. The locations that would be most affected by a shiA would be the OD surface points. Shifting further from the weld fusion line by about 0.1 in. would increase the OD surface stresses in the through thickness plots by about 5 ksi.

.3 ho oad The recent shroud inspection revealed:that-OD cracking at the V9 and V10 welds is much more prevalent than ID cracking. Furthermore, the OD cracking is not localized but extends over much of the weld length. It can be seen from Figure 2-10 that the baseline heat input case predicts that the ID hoop stresses are much greater than the OD hoop stresses and that the ID axial stresses are about the same as the OD axial stresses. With these predicted stresses, it is difficultto explain the relative lack of ID cracking. It was desirable to find a plausible explanation for this apparent discrepancy.

As noted above, a decrease in heat input was able to raise the OD hoop stress, but it also tended to raise the ID hoop stress. Therefore, the heat input effect was not sufficient to resolve the discrepancy. It appeared that a physical (i.e., non modeling related) factor was being over looked.

The required effect of the missing factor was that ID stresses be lowered relative to the OD st'resses.

While various 3D effects, to be explored below, were already selected for study, it was realized that these 3D effects would probably only be sufficient to explain local variations in cracking behavior.

The fact that most of the OD weld was cracked while virtually no ID cracking was observed meant that a more global effect was required.

One form of loading that was determined to have the potential for altering the welding residual stresses along the entire weld length was a diametral loading. Consider the cylinder lying on its side (axis parallel to the floor) such as when welding V9 and V10. Ifone weld is at the top and one is at the bottom, such as during welding, then a downward load at the top of the cylinder will tend to put the ID of both welds into tension. Ifthis added tension results in plasticity, then upon removal of the load, the ID hoop stresses will be less tensile than before the loading. At the OD, the induced hoop stresses would be compressive, and any plasticity would result in more tensile stresses 12

upon unloading. This is precisely the type of stress change that is needed to explain the predominately OD cracking behavior.

Using the 2D finite element model, the described diametral loading was simulated as being applied after the completion of welding. The results are shown in Figure 2-16 for the baseline weld heat input case. The plotted stresses are those existing afler the diametral loading has been removed.

It can be seen that the diametral loading significantly reduces the highly tensile axial stresses near the ID and OD surfaces. The relatively high ID tensile hoop stresses are also significantly reduced.

The OD hoop stresses are generally decreased but to a much smaller extent. For sufficiently high diametral loading, the point nearest the OD surface has a small increase in tension. The small decrease in the OD stresses is believed to be a Poisson effect due to the significant decrease in hoop stresses. Only the small increase in hoop tensile stress at the OD surface point, and for the largest diametral loads, is due to the OD hoop stresses exceeding the yield stress during the diametral loading such as described above.

The magnitude of the applied diametral loading is indicated in terms of the decrease of the shroud diameter along the line of the applied loading during the load application. To put these deflections into perspective, it is helpful to know that the deflection due to just the weight of the shroud is 1.6 inches. This deadweight loading induces a peak ID hoop stress of about 14 ksi.

Comparing the post diametral "squeeze" stresses at the ID and OD surfacey', it can be seen that the OD stresses tend to become more. conducive to.IGSCC.. than the ID. stresses after about a 4.to 6 inch diametral squeeze loading. This conclusion is based on the assumption of a threshold stress intensity factor and an assumed flaw depth that taken together require a surface stress of about 20 ksi to initiate IGSCC. Since the OD hoop stress is below this stress level, initiation at the OD is assumed to occur due to the axial stresses. Figure 2-17 shows the operating stress distributions for the baseline weld input followed by a 4 inch diametral squeeze.

Figures 2-18 and 2-19 summarize the diametral squeeze loading effect for the 0.75 and 0.5 baseline weld heat input cases. The effect of the diametral squeeze is similar to that for the baseline case except that the ID hoop stresses are more efficiently reduced by the diametral loading. This is because the ID axial stresses are closer to yield at the start of the diametral loading, and therefore the ID plastic deformation starts to occur immediately. The result is that a smaller diametral loading is necessary to reach the point at which IGSCC initiation at the ID becomes difficult. In fact, it can be seen that the deadweight loading is adequate for the 0.5 baseline case while a 2 to 4 inch squeeze is needed for the 0.75 baseline case.

Table 2-3 summarize the HAZ surface stresses from the analyses that consider the application of a diametral loading that is centered on the V9 ad V 10 welds. For the sake of discussion, assume that a surface stress of less than 20 ksi will not initiate IGSCC. For the baseline weld heat input case at operating conditions, both the ID and OD axial stresses are capable of initiating IGSCC. Based on the hoop stress distribution of Figure 2-17, it can be seen that once initiated by axial stresses, the ID subsurface hoop stresses will tend to drive cracking at a faster rate than the OD subsurface hoop stresses. This behavior is contrary to observed behavior. For the baseline case with a 4 inch 13

0 H

diametral squeeze, only the OD axial stresses are above the threshold level for IGSCC and thus all cracking would be at the OD. For the half baseline weld heat case, both ID surface stresses are below 20 ksi after only a 2 ksi diametral squeeze. This level of diametral loading is attained entirely from deadweight. Since the V9 and V10 welds are 180 degrees apart, the first of the two welds to be completed would have easily been subjected to this level of diametral loading during the welding of the opposite vertical weld. Whereas the difference in ID and OD axial stresses was only 6 ksi for the 4 inch squeeze, baseline weld case, the OD axial stress exceeds the ID value by 18 ksi after the deadweight (2 inch) diametral squeeze of the half baseline heat input case. Both ID stresses are much less than the assumed 20 ksi threshold for IGSCC initiation while the OD axial stress is much greater than 20 ksi. Increasing the diametral loading beyond the deadweight level, decreases ID and OD stress levels. With a 6 inch squeeze, all surface stresses are below 20 ksi. It therefore is concluded that a 2 to 4 inch squeeze is consistent with the tendency for predominately OD cracking.

2.4 3 Weld <

ects The residual stresses predicted by the 2D welding simulation are believed to be reasonable approximations of the average stress behavior for the V9 and V10 shroud welds. The 2D model cannot, however, account for 3D effects such as weld starting and stopping, localized weld repair operations, changes in welding parameters, or variations in the weld groove fit-up that required mechanical alignment forces to be applied before welding; and which lead to'esidual stresses upon their removal after welding.

The hoop stresses from the 2D model result in nearly a zero net hoop force across the weld HAZ. This near zero value of net force is due to the very compliant nature of the shroud. The degree of this compliance is demonstrated by applying uniformly distributed, opposite and equal, radial forces at the V9 and V10 weld grooves for one of the half cylindrical shell components. The radial load that induces a diametral change of 2 inches (about 15 Ib/in of shell length) leads to a peak bending stress at 90 degrees &om the load points of only about 3.5 ksi. The membrane stress at this peak bending stress location would be about 0.01 ksi. It can therefore be seen that the hoop direction shrinkage of the weld is practically unrestrained by the shell geometry. In the real 3D structure, the net hoop directed force across the entire V9 and V10 welds must be essentially zero, but at any given location along the welds it is possible that the average hoop stress may not be zero. Equilibrium only requires that every location with net hoop tension be offset by some location having net hoop compression.

The actual 3D residual stress state may be thought of as the 2D stresses from the plane strain simulation plus some 3D stress field representing the variations from the 2D plane strain baseline field. The purpose of the analyses of this section was to estimate the magnitudes and distributions of stresses that one might expect for the 3D component of the stress field.

Three types of 3D effects were considered in this study. The first effect was based on the assumption that the transverse weld shrinkage (i.e., hoop weld shrinkage) would not be the same at all points along the V9 and V10 welds. The second effect was based on the assumption that the weld

II grooves of the two half cylinders did not match perfectly in the radial direction without some form of alignment forces being applied before and during welding. The third effect was based on the assumption that the pre-welding alignment of the H4 and H5 weld grooves required the application of some form of alignment forces. The first effect is referred to as a transverse weld slirinkage variation. The second and third are referred to as radial mismatch at the V9/V10 welds or at the H4/H5 welds.

Since no data on actual transverse weld shrinkage variations or pre-weld radial mismatch was available, it was necessary to assume both the magnitude and the distribution of the assumed shrinkage and radial mismatch variations. To keep the analysis as simple as possible, the shrinkage and radial mismatch variation magnitudes were modeled as being sinusoidal with respect to distance along the welds. The results are expected to be useful for assessing the effects of less regular variations. Since the analyses of this study on 3D effects assumed linear elastic material behavior, the results are scalable.

Since it was clear that the stresses from these sinusoidal variations would depend on the frequency of the assumed variations, a series of analyses were performed. For the radial mismatch and transverse weld shrinkage at V9/V10, the lowest frequency resulted in a half cycle variation from the one end of the V9/V10 weld to the opposite end. The highest frequency resulted in three full cycles of radial mismatch or transverse weld shrinkage. For the radial niismatch at H4/H5, the lowest frequency resulted in two complete cycles of varia'tion around: the full shroud cir'cumferei1ce:.

The two cycle case corresponds to both upper and lower H4 weld prep regions being elliptical before welding and the major axis of the top being aligned with the minor axis of the bottom. In addition to the two cycle case, four, eight, and sixteen complete cycles of H4/H5 radial mismatch variation around the circumference were considered. It was found from these analyses, that the axial dimension of the shroud between H4 and H5 was sufficient to preclude interaction between the radial mismatch effects of the H4 and H5 welds.

2.4.1 The Finite lement

~

odel The 3D finite element grid used in this study was a 90 degree sector of the shroud. The axial length of the model was half of the 90.12 inch total cylinder length. The grid is shown in Figure 2-

20. The grid consists of 6840 eight-noded isoparametric elements and 10614 nodes. There are 57 planes of elements in the circumferential direction and 60 in the axial direction. The finite element program used for these 3D calculations is called ALT3D [ALT3D]. The ALT3D element formulation used in this calculation includes extra incompatible shape functions which allow the elements to accurately represent bending behavior with only a single element representing the entire wall thickness. Two elements were used through the wall in the model primarily to improve the quality of stress contour plots. A less refined version of the model was used for the radial mismatch analysis at V9/V10 but was replaced with the more refined grid of Figure 2-20 before the V9/V10 shrinkage variation analyses and the H4/H5 radial mismatch analyses. The less refined grid had 16 elements in the axial direction and 55 elements in the circumferential direction and is shown in Figure 2-21.

15

2.4.2 olutions for V /V10 Weld hrinka e Variation The assumed sinusoidal variation in transverse weld shrinkage was introduced to the 3D model as a circumferential displacement variation along the edge of the grid corresponding to the vertical weld centerline. While the circumferential displacements were forced to vary by the amount specified by the sinusoidal shrinkage variation, the assumed symmetry of behavior about the weld centerline required that the radial and axial displacements be unrestrained. The boundary conditions at 90'rom the weld plane represented a plane of symmetry. The boundary representing the mid section of the cylinder was also modeled as a plane of symmetry. The boundary conditions at the H4/HS boundary reflected symmetry or anti-symmetry depending on the number of shrinkage variation cycles being modeled. For the 0.5 and 1.5 cycle cases, anti-symmetry was modeled. For the other cases, a symmetry boundary was modeled.

The introduction of the weld shrinkage. variation induced bending stresses in addition to the more directly induced axial variation in hoop membrane stress. The bending shows up in a deformed shape plot of the V9/V10 cross-section as shown in Figure 2-22. The radial displacement magnitudes are exaggerated here for clarity but are, in fact about three times the magnitude of the applied circumferential displacements (which were 0.003 in. or half the assumed full maximum shrinkage variation magnitude of 0.006 in.). The length of the shrinkage variation (L) is defined as the distance over which the variation goes from zero to the peak value and back to zero (i.e., one half cycle) and is-illustrated in Figure 2-22 for the Z,cycle case. This length is intended to correlate. the sinusoidal variation stresses to stresses at a truly local variation of size L. A weld repair would be one possibility for a weld shrinkage variation. The length of the repair would be represented by L.

The hoop stress behavior at the V9/V10 weld is shown in Figure 2-23 for a representative portion of the 3 cycle solution. An axial stress bending inflection point lies at the middle of the densely contoured region. Just to the right of the inflection point is a small region in which hoop stresses are about -18 ksi at all point through the wall. Just to the left of the inflection point is a similar region where the hoop stresses are +18 ksi at all points through the wall. The through thickness gradients indicated by the contours in these small neighboring regions are not correct representations of the computed stresses. The plotting error is due to the large stress gradients relative to the grid refinement in the axial direction. Beyond these small regions of pure hoop membrane stress are regions of combined hoop bending and membrane stress where the stresses are on the order of half the peak membrane values. Figure 2-24 shows a section taken at right angles to the weld centerline plane at a position half way between the inflection points of Figure 2-23. It can be seen that the shrinkage variation induced stresses in the HAZ are not significantly reduced from those at the weld centerline but that the stresses do decay with distance from the weld.

The stress effects of the transverse shrinkage variation simulations are summarized in Figure 2-25. For this plot, a peak axial shrinkage variation magnitude of 0.006 in. is again assumed. Since the analysis is linear, these results scale with the assumed peak shrinkage. The 0.006 in. shrinkage assumed here is on the order of 10% of the total transverse weld shrinkage that was computed in the welding simulation. This value is believed to be a conservative estimate of the variation that could 16

0 reasonably be expected in the V9 and V10 weld for the lengths of variations being considered here (15 in. to 90 inch). However, no data is available to support this estimate. For the shorter variation lengths considered here, and for even shorter lengths than those which were considered, it seems that a 10% variation may be overly conservative. This is because it becomes more difficult to create a variation of a fixed magnitude over shorter and shorter lengths (as evidenced by the larger induced stresses).

There are two distinct stress phenomena occurring for the transverse shrinkage stresses as can be seen from Figure 2-23 and Figure 2-25. One results in relatively large localized membrane stresses that decrease as L is decreased. The other results in smaller but more wide. spread membrane plus bending stress behavior for which stress magnitudes increase as L is decreased. It seems likely that both types of behavior would be observed in the vicinity of a localized (i.e., non cyclic) shrinkage variation due a weld repair. It can be seen that the ID stress magnitudes for the second type of behavior are always larger than the OD stresses. It should be kept in mind that the peak tensile stresses shown in Figure 2-25 are accompanied by their compressive counterpart. For every tensile regio'n there is a similar compressive region at some other location along the weld.

2.4.3 Radial Mismatch at Weld V and V10 The assumed sinusoidal variation in radial mismatch was introduced to the 3D model as a radial displacement variation along the edge ofthe grid corresponding to the vertical weld centerline.

The radial displacements were forced to vary by the amount specified by the sinusoidal mismatch variation. Due to the assumed antisymmetry of behavior about the weld centerline, the axial displacements were forced to be zero and the hoop direction displacements were unrestrained. The boundary conditions at 90'rom the weld plane represented a plane of symmetry. The boundary representing the mid section of the cylinder was also modeled as a plane of symmetry. The boundary conditions at the H4/H5 boundary imposed symmetry or anti-symmetry depending on the number of shrinkage variation cycles being modeled. For the 0.5 cycle case, anti-symmetry was modeled. For the other cases, a symmetry boundary was modeled.

The form of the initial radial mismatch is illustrated in the deformed shape plot of Figure 2-

26. The left end corresponds to the mid section and the right to either the H4 or H5 weld. The peak radial mismatch at the V9/V10 weld is 0.1 inch. It should be noted that though the analysis calculates the stresses in going from the initial, perfect, undeformed geometry to the deformed shape depicted in this figure, these stresses are exactly the same except for a sign change as the stresses that would be produced in going from a geometry that is initially stress free in the shape of the deformed mesh in Figure 2-26 and that ends up in the perfectly cylindrical shape of the undeformed mesh.

Figure 2-27 shows hoop stress contours for two shroud sections for the 3 cycle radial mismatch behavior. The top section plot is representative of the cylinder mid section and all other sections where the hoop bending stresses peak. Allowing for a sign reversal, this section is representative of stress magnitudes at seven sections along the 90.12 inch weld. Due to the assumed antisymmetry at the weld centerline for this type of mismatch, the hoop bending stresses have an 17

0 l

inflection point at the weld centerline. The contour plots do not precisely show this behavior due to the large gradients relative to the grid. The hoop stresses in the HAZ are shown in the bottom section plot of Figure 2-27. It can be seen that these stresses are not substantially smaller than the peak bending stresses which occur at about two wall thickness from the weld centerline. For the cases with fewer cycles, the peak stresses occur farther from thesveld centerline, and thus the HAZ stresses are somewhat lower than the peak stresses. Figure 2-28 shows similar contour plots for the axial stresses. The axial stresses are also antisymmetric with respect to the weld centerline and are similar in distribution and magnitude to the hoop stresses.

Figure 2-29 summarizes, the. HAZ>oop andaxial stresses that result from an assumed maximum radial mismatch of 0.1 in. These plots show that the hoop and axial stresses are primarily bending stresses with the OD hoop stresses being slightly smaller than their ID counterparts and the OD axial stresses being larger than their ID counterparts. The General Electric (GE) specified tolerance for the shroud of 176 in. diameter +0.5 in. suggests that mismatches as large 0.5 in. might have occurred. While this is probably a reasonable and conservative value for the lower cycle (larger variation length, L) cases, it seems that as higher cycle cases are considered, a lower peak radial mismatch would be more reasonable. Again, there is no data to guide the selection of radial mismatch magnitudes or frequencies.-

2.4.4 Radial Mismatch at Weld>> 4 and 5 The assumed sinusoidal variation in radial mismatch was introduced to the 3D model as a radial displacement variation along the edge of the grid corresponding to the H4 or H5 weld centerline. Half of the total mismatch is applied to the modeled portion of the shroud. By applying half of the total assumed radial mismatch, it is inherently assumed that the stiffness of the upper portion of the shroud is equal to that of the modeled portion. This is an approximation, but it is believed to be a reasonable one considering the uncertainty of actual radial mismatch magnitudes.

The radial displacements at the H4/H5 weld boundary were forced to vary by the amount specified by the assumed sinusoidal mismatch variation. Due to the assumed antisymmetry of behavior about the H4/H5 weld centerline, the hoop directed displacements were forced to be zero and the axial displacements were unrestrained. The boundary conditions at the V9/V10 centerline and at 90'rom the V9/V10 weld centerline plane imposed symmetry. The boundary representing the mid section of the cylinder was also modeled as a plane of symmetry. It was found that the boundary condition at this section had no effect on the V9/V10 HAZ stresses because the H4/H5 mismatch induced stresses decayed to negligible values before the mid section was reached.

The form of the initial radial mismatch is illustrated by the deformed shape plot of the 8 cycle case shown in Figure 2-30. The assumed peak radial mismatch at the H4/H5 weld is 0.1 inch and occurs at the section depicted in this figure. The maximum displacements shown in the figure are half of the total mismatch since it-is assumed that. the radial-displacements are antisymmetric about the H4/H5 weld.

Figure 2-,31 shows contours of hoop stress for the 8 cycle case. The top plot shows the

stresses in the V9/V10 HAZ. It can be seen that the radial mismatch loading at the H4/H5 weld induces stresses only over the portion of the V9/V10 welds that are closest to the H4/H5 welds. The lower cycle cases have somewhat larger zones of influence, but even the 2 cycle case did not affect more than about a quarter of the vertical weld. It can be seen that near the H4/H5 weld, the vertical weld HAZ could have a significant tensile (or compressive) stress at all points through the thickness due to a radial mismatch at the horizontal weld.

Figure 2-32 shows contours of axial stress for the 8 cycle case. Whereas the hoop stress in the vertical weld HAZ was primarily a membrane stress, the axial stresses are primarily bending related.

Again it can be seen that the stresses decay quickly with distance from the horizontal weld.

Figure 2-33 summarizes the V9/V10 HAZ stresses due to assumed radial mismatch loading at the H4/H5 welds. The axial stresses are essentially pure bending stresses. While the plotted stresses indicate that the ID stresses are tensile and the OD stresses are compressive, stresses with the opposite sign are just as frequent. The hoop stresses have both a bending and a membrane component. The membrane component is fairly independent of the number of cycles (i.e., variation length, L) with a typical value of about 15 ksi. For the variation length considered, this membrane component is larger than the bending component thus leading to the possibility of through wall tensile hoop stresses in the vertical weld HAZ.

The General Electric (GE) specified tolerance for the shroud of 176 in. diameter +0.5.in.

suggests that mismatches as large 0.5 in. might have occurred. While this is probably a reasonable and conservative value for the lower cycle (larger variation length, L) cases, it seems that as higher cycle cases are considered, a lower peak radial mismatch would be more reasonable. Again, there is no data to guide the selection of radial mismatch magnitudes or frequencies.

There are three main conclusions that can be made based on the results of this radial mismatch analysis:

1. Radial mismatch at the H4/H5 welds can induce very large stresses. in the V9/V10 HAZ.

The hoop stress in the vertical weld HAZ due to a radial mismatch at the horizontal weld can be tensile at all points through the wall with an average value of about 15 ksi for an assumed 0.1 mismatch.

The effect of the horizontal weld radial mismatch on the vertical weld HAZ stresses decays rapidly with distance from the horizontal weld.

The results of the 3D analyses of this section are summarized in Table 2-4.

19

. ~s tress Intensi Factor alculation The prediction of stress corrosion cracking behavior at the V9 and V10 welds requires that the stress intensity factor be known as a function of crack depth for the operating steady state condition. Since a handbook solution applicable to the large R/t-ratio of the shroud was not found, it was judged necessary to compute stress intensity factors using. the finite element model.

2.5.1 Method The method used for computing stress intensity factors with the finite element model is based on the concept of energy release rate, the relationship between energy release rate and the stress intensity factor, and the crack closure integral method [IRW57] for computing crack energy release rates. The crack closure integral allows the energy release rate for a cracked body to be determined from the stress in the vicinity of the crack tip just prior to some small crack extension (b,a) and the crack opening displacements after this small extension. Basically, the energy release rate is equal to the work that must be done to close the crack over the distance b,a (thus restoring the crack to the state that existed before the increment in crack size) divided by the change in crack area. For a finite element model, this integral is most conveniently formulated in terms of the nodal forces at the coupled crack tip nodes prior to a crack-extension and the opening displacement between these nodes after they are uncoupled and the crack-is-allowed-to extend by one element. The work done to reattach the nodes is S/2-,where-f- is the node force before the nodes are released; and 6 is. the opening between the nodes after release. Since the energy release rate calculation involves an increment in crack growth, the resulting energy release rate and stress intensity factor are assumed to be most representative of some intermediate crack size during the crack size increment.

/

The crack closure integral method can be used to compute Mode I (K,) and Mode II (K>>)

stress intensity factors. The energy due to crack plane normal node forces and displacements leads to the Mode I energy release rate (Gi) and the energy due to the crack plane parallel node forces and displacements leads to the Mode II energy release rate (G,g. For plane strain conditions these energy release rates are related to K, and K>> by [TPI73]:

G =

(1-v E

)

K, 2

andG II

=

(1-v E

)

K>>

Since cracks tend to grow in the direction that maximizes the Mode I energy release rate (consequently minimizing the Mode II energy release rate), the Mode II energy release rate is often small compared to the Mode I value. Therefore, it is rare that Mode II energy release rates or stress intensity factors are used in crack growth models: =In the current analysis, Mode II related energy release rates were assumed to be negligible.

Cracking can occur in both HAZs of a weld. However, for the purposes of computing K 20

I V

V

it is generally assumed that only one HAZ has a crack. Ifa crack is assumed to exist in both HAZs, interaction between the two crack stress fields would lead to a smaller K, than for a single HAZ crack. Since the weld simulation model assumed weld centerline symmetry, growing a crack in the HAZ of that model would be, in effect, simulating a crack in each HAZ. Therefore, it was necessary to modify the 2D plane strain finite element model prior to doing the K, calculations. A new grid was generated (see Figure 2-34) which was very similar to that used for the welding simulations but which did not have the elements in the weld region distorted to simulate the weld groove shape or the weld crown. With this model, it was possible to simulate a single crack while still modeling only a quarter of the full shroud circumference. This was done by putting the crack at an assumed plane of symmetry. With the crack being modeled at an assumed plane of symmetry, the crack plane node displacements represent half of the total opening. Therefore, the node displacements had to be doubled (so as to represent the total crack opening displacement) prior to computing the energy release rate.

2.5.2 om ari on o finite lement tres Inten. i Factors t Handbook Values To check that the finite element stress intensity factor calculation procedure was correct and to determine its accuracy, the procedure was used to compute a solution to a problem that could be found in a handbook. Such a solution was easily obtained from the current model by simply removing the symmetry boundary condition from the section at 90 degrees from the crack plane.

With this end of the model free, the model is essentially the equivalent'of a single edge cracked plate (SECP). By assuming a uniform crack plane normal stress existed prior to cracking, the model produces the solution for a single edge cracked plate under a remote uniform tension. Since the loading in the current model is limited to the crack plane nodes, the curved nature of the finite element model had virtually no effect on the K, solution.

The current single edge cracked plate K, solution is compared in Figure 2-35 to a handbook solution [TPI73]. The top plot compares the handbook SECP solution to the current SECP and shroud solutions in terms of the normalized stress intensity factor (F) where F = K/(sigma*sqrt(pi*a))

The handbook and the current SECP solutions are nearly identical. The extra restraint provided by the symmetry boundary conditions at 90 degrees from the crack plane in the shroud significantly reduces K, from those in the SECP for crack depths greater than about 50%. The plot at the bottom of Figure 2-35 compares the same solutions but uses a slightly different normalization. With this different normalization, the singular part of the SECP solution as a/t approaches unity is removed.

It can be seen that the current calculation methods provide excellent agreement with the handbook SECP solution. There is every reason to expect that the shroud-K, solutions will have similar accuracy to that found for the SECP solution.

Figure 2-36 shows one last comparison of the SECP and shroud solutions. In this figure, the 21

normalization used to remove the singular behavior as a/t approaches unity is altered so that the shroud solution appears to be nonsingular rather than the SECP solution. The exponent on the (1-a/t) factor to achieve this behavior was 1.1. The fact that this value is less than the 1.5 value that removed the singular behavior from the SECP solution is reasonable. The normalization of K, solutions for geometries which do not result in bending of the remaining crack ligament (e.g., double edge cracked plate and center cracked plate ) as a/t approaches unity requires an exponent of 0.5.

Therefore, the 1.1 value of the shroud geometry puts it about midway between the two limiting conditions. This normalization provides a rational basis for computing K, at crack depths approaching a/t = 1. Such very deeply cracked solutions could not be reasonably computed with the finite element model since there would be two few finite elements representing the uncracked ligament.

2.5.3 tress Intensi Factor olutions for the V and V10 Welds at eratin onditions The stress intensity factor solutions for a crack growing in the HAZ of the V9/V10 welds under operating conditions are shown in Figure 2-37. These solutions include welding residual stresses plus the hoop stress due to the 6 psi pressure differential across the shroud, plus the effect of the 1050 psi pressure acting on the crack faces. This figure contains one plot for cracks growing from the ID and another plot for cracks growing from the OD. Solutions for the four weld heat input levels considered in this study are given. The solutions for crack depths greater than 1.3 inches are extrapolated from the finite element solution using the normalization procedure described above.

/

The K, solutions of Figures 2-37 suggest that through wall crack growth cannot occur for ID cracks since K, becomes negative once a crack reaches a depth of about 50%. However, the K, for the OD cracks remain positive and thus suggest that through wall crack growth cannot be ruled out.

It can be seen that the K, for small ID cracks tend to be larger than those for small OD cracks. This suggests that ID cracking is more likely than OD cracking.

Figure 2-38 gives Ki solutions for uniform 1 ksi hoop stress and 1 ksi bending stress distributions. The uniform stress case can be used to decompose the K, solutions for Figure 2-37 into residual and operating stress components. This decomposition will be necessary if creep relaxation of secondary loads is considered in the crack growth analysis portion of this study.

Figures 2-39 and 2-40 give the K, solutions for the 0.75 baseline and 0.5 baseline weld heat input cases. Various diametral loadings were applied and removed before finally adding the operating loads. With sufficient diametral loading, the ID surface stresses were reduced to the extent that hoop and axial stresses are smaller than those needed to initiate IGSCC at the ID surface.

Therefore, OD crack initiation is made more likely than ID crack initiation which is consistent with observed cracking. However, it can be seen from these figures that ifa crack does somehow initiate at the ID, there are still substantial K, to drive the crack. It can be seen that the diametral loading 22

does not alter the tendency for ID cracks to eventually arrest due to negative K, at about mid wall.

It can be seen that the diametral loading tends to reduce the K, for both ID and OD cracks. The application of the diametral loading did not alter the K, levels for OD cracks at a depth-of 1.0 to 1.2 inches. The K, level for cracks of this depth remain at about 3 ksi]in for the 0.5 baseline case and about 10 ksi)in for the 0.75 baseline case. If a threshold K~ for crack growth of 3 to 10 ksi)in is assumed, then crack growth would arrest at a depth of about 1 inch for both cases.

23

Table 2-1 Temperature Dependent Mechanical Propcrtics for Type 304 SS Base Metal and Type 308 SS Weld Metal MATE<RIAL1 SS 304 BASE Mean Thermal E<lastic Poissons Expansion Hardening Yield Temperature Modulus Ratio Coefficient ('F') Modulus Strength (psi) (70'F Reference (psi) (psi)

Temperature)

50. 28700000. 0.26 .0000086 539700. 36000.

300. 27100000. 0.28 .0000090 452300. 31100.

550. 25800000. 0.31 ~ 0000095 364800. 25900.

750. 24200000. 0.32 .0000098 296300. 22300.

1000. 22500000. 0.30 .0000102 217900. 18500.

1300. 20200000. 0.28 .0000105 139000. 14900.

1600. 16000000. 0.24 .0000108 79600. 10200.

2100. 10000. 0.22 .0000113 1000. 1000.

MATERIAL2 SS 308 WELD Mean Thermal Elastic Poissons Expansion Hardening Yield Temperature Modulus Ratio Coefficient ('F') Modulus Strength (psi) (2100'F Reference (psi) (psi)

Temperature)

50. 28700000. 0.26 .0000113 539700. 45000.

300. 27100000. 0.28 .0000116 452300. 38900.

550. 25800000. 0.31 .0000119 364800. 32400.

750. 24200000. 0.32 .0000121 296300. 27900.

1000. 22500000. 0.30 .0000122 217900. 23100.

1300. 20200000. 0.28 .0000125 139000. 18600.

1600. 16000000. 0.24 .0000128 79600. 12800.

2100. 10000. 0.22 .0000133 1000. 1000.

24

Table 2-2 Heat Transfer Material Properties for Stainless Steel Base and %'eld Metals Conductivity: 0.000266 BTU/in/sec/F Heat Capacity: 0.03S 1 BTU/in'/F 25

Table 2-3 V9 and VIO HAZ Surface Stress Summary for Weld Plus Operating Conditions Considered in this Study HAZ Surface Stress Summary (Baseline Weld Heat)

Hoop Stress (ksi) Axial Stress (ksi)

ID OD ID OD as welded 19.8 -3;7 39.9 42.6 welded + operating 16.6 -2.8 29.1 32.7 weld+ 4" squeeze+ op 9.0 -1.9 19.5 25.5 HAZ Surface Stress Summary (0.75 Baseline Heat Weld)

Hoop Stress (ksi) Axial Stress (ksi)

ID OD ID OD as welded 29A 0.2 37.1 43.8 welded + operating 23.6 0.0 26.0 33.6 weld + 2" squeeze+ op 17.2 0.7 20.2 26.4 /

weld+ 4" squeeze+ op 8.4 -0.4 13.4 26.0 weld+ 6" squeeze+ op -2.6 0.7 8.1 14.0 HAZ Surface Stress Summa (0.5 Baseline Heat Weld)

Hoop Stress (ksi) Axial Stress (ksi)

ID OD ID OD as welded 38.8 11.0 27.7 50.3 welded + operating 28.4 8.5 19.6 38.8 weld+ 2" squeeze+ op 15.3 7.1 'I3.8 31.7 weld+ 4" squeeze+ op 6.2 54 10.4 28.7 weld+ 6" squeeze+ op -5.0 5.2 7.0 19.4 26

Table2-4 V9 and V10 HAZ Stress Summary for the 39 Residual Stress Effects Considered in this Study Radial Mismatch at V9/1 0 (0.1 inch )

Stress (ksi)- . -.

ID OD cycles L (in) Axial Hoop Axial Hoop 3 15 10.1 -10.6 -11.7 10.1 2 23 4.5 -5.2 -5.5 4.9 1 45 1;1 -1.4 -1.6 1.4 05 90 0.2 -0.4 -0.4 0.4 V9/1 0 Weld Shrinkage Variation (0.006 inch)

Stress (ksi)

Most of Cross-Section Localized Behavior ID OD (membrane stress) cycles L(in) Axial Hoop Axial Hoop Axial Hoop 3 15 10.4 9.4 1.6 4.8 9.5 18.0 2 23 8.4 5.8 -0.2 1.0 10.5 20.0'1.0 1.5 30 6.1 3.8 -0.2 -0.4 22.0 1 45 3.6 2.0 0.2 -0.8 11.7 24.0 05 90 1.2 0.6 0.3 -0.5 12.0 25.0 Radial Mismatch at H4 (0.1 inch)

Stress (ksi)

ID OD cycles L (in) Axial Hoop Axial Hoop 2 139 0.8 15.1 -0.8 15.5 4 70 1.2 15.2 -1.2 15.6 8 35 2.5 14.0 -2.5 17.0 16 17 8.0 6.0 -8.0 25.0 27

A R= 88.0 Figure 2-1 The 2D Plane Strain Finite Element Grid 28

OD 22.5 '

6 6: 6; '." ~~ .,

2: 2 2

1 2

'6:6:6 6

6:6 6:6 5:5:5 5:5 61 1;1 SS 2:,2:2:2 I~

2

, 2 2

' ~ '"'~"

ID Simulation Figure 2-2 The Assumed Weld Passes and Welding Sequence for the V9/V10 Weld 29

Deformed Shape (xl0)

V9!V10 Weld Completed (70 Fj Figure 2-3 The Deformed Shape of the shroud Predicted by the WE<LD3 Simulation (Deformation Exaggerated by a Factor of 10), I 30

V9IV10 As Welded Stress at 70 F 50 45 40 ~

35 ro

~ro 30 lO 25 V) nr X

20 15 i

I

~

~

with constraint w/o constraint; 10 0

00 0.2 0.4 0.6 0.8 1.0 '.2 1.4 1.6 Distance from Inner Surface (in)

V9N10 As Welded'Stress:af 70 F 30 20 10 rh rn 0 rh sr L

V)

~ -10 0

0

-20 I

~ w/o constraint

-30 AO 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface'(In)

Figure 2-4 The Predicted Welding Residual Stresses in the Heat'Affected Zones of the V9 and VIO Welds at 70'F 31

Residuol Axial Stress, MPo Residuol Circumferenliol Stress, MPo

-300 -200 - IOO 0 !00 200 I.30 (33mm) -200 -IOO 0 100 200 300 400 0 Measured stress i heat (7(92 l.2 6 I 12 0 Measured stress (30.48 VI T'6.00 wu (30.48 Ih A (660 mm) 4l heat 834264 Predicted residuol stress (ZS.4)

Cl V

0 I.O (ZS.4)

V C

(:

distribution (20.32 Cfl 0.8 ~ II 0.8 C (2032 C C C H

0 Cl CP 0.6 O, CL

( I5.24) o. .6 o.

E' IS.24 0.4 CP (IO.I6) EJ 0.4 C

o 1 e s ' (IO.I 6) C O

(5.08) O O 0.2 0.2 (5.08)

-40 -20 0 20 40 -40 -ZO 0 20 40 60 Residuol Axial Stress,',ksi Residua) Circumterentiat Stress, ksi I

Figure 2-5 Predicted and Measured Through>vali Welding Residual Stresses in a 26 inch Diameter Pipe With a 1.3 inch Wall Thickness [EP1743]

32

rHSIDE WALL OUTS IDE WALL 50 o GE26 p GE26 (4 azin<ulhs)

< AHL 26 t2 ozimuIhsj 0 AHL 26 (IH"SKRYICK FROM KR8) o AHL20 p.

P pp' ay p

Ce 0

0 0 u p 0 0

p g80 p O

'i Op p

0 0,2 04 0.6 0.8 I.O Through-eaEl Distribution of Axial Residual Stress in Larga-Diameter Pipes (t o 3. in.)

Figure 2-6 Measured Throughwall Axial Welding Residual Stress Distributions From Various Pipes With Wall Thicknesses Greater Than One Inch 33

V9N10 Baseline Heat Input 2.0 1.5 C

o 1.0 E

V iu CL V)

~ 05 ru

'U 0.0 ~

~with without constraint artificial 3D constraint l I

I l

j 0 50 100 150 200 250 300 350 400 450 500 Solution Increment Figure 2-7 Predicted Axial Radial Displacement at Weld V9/V10 Both With and Without Use of the Artificial3D Constraint Option of WELD3 34

Il Heat Input Sensitivity As Welded Stress at 70 F 60 50 40 V) 30 lO Cl cs 20 I

~

~ 1.5 baseline baseline I

10.

~0.75 b I'0.5 baseline 0

-10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Heat Input Sensitivity As Welded Stress at 70 F 50 40 30 20 10 th Vl 0

V) o~ 10

-20

~1.5 baseline

-30 'baseline 40

-50 I

~

'0.75baseline 0.5 baseline

~

I 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-8 Predicted V9/Vlo HAZ Residual Stresses Due to Four Different Assumed Weld Heat Input Levels 35

Neutron Diffraction Residual Stress Data for a2 in. Thick H5 Weld 50 40 30 I

20

~ ~

~

ij 1

I I

I Vl I I I I r 10 lO

~ I CO 0 I I l I rr I

-10 li 1

t

-20

-30 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Normalized Distance Figure 2-9 Neutron Diffraction Axial Residual Stress Data for a 2 inch Thick H5 Shroud Weld 36

V9N10 Weld HAZ Stresses 50 45 40 35 Ol 30 lh Ol

~ 25 ro m 20 X

15 10

'as

~operating welded 0

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

.,V9N10 Weld AAZStresses 30 20 Fn 10 EO Vl 0

Co CL O

0 -10 ~

~as operating welded

-20

-30 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-10 Predicted V9/V10 HAZ Residual Stresses With Operating Loads and at an Operating Temperature of 550'F 37

Heat Input Sensitivity Weld + Operating at 550 F 50 .

40 30 Vl 20 I ~ v= ~ ~ ~

Ol L

10 EO l ~1.5 baseline i, ~baseline 0

~0.75 baseline

~0.5 baseline

-10

-20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Heat Input Sensitivity Weld+ Operating at 550 F 40 /

30 20 lh 10 M

0 Co CL o -10 xO

-20

-30

~

~1.5 baseline

~ baseline 0.75 baseline

~0.5 baseline 40 0.0 0.2 0.4 0.6 " 0.8 '.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-11 Predicted V9/Vlo HAZ Residual Stresses Due to Four Differen Assumed Weld Heat Input Levels with Operating Loads Applied and at Operating Temperature 38

Hoop Stre (ksi)

A -25 B -20 C -15 D -10 E -5 F 0 G 5 H 10 I l5 J 20 K 25 V9N10 Weld (baseline heat Input) + Operating (550 F)

Baseline%eld Heat Input and Operating Figure 2-I2 Hoop Stress Contours in the-V9 or VIQ Weld Region for Conditions 39

-(ksil A -10 8 -5 C 0 0 5 E I0 F 15 G 20 H 25 I 30 J 35 K 40 L 45 V9N10 Weld (baseline heat input) ~ Operating (550 Fj Weld Heat Input and Operating Figure 2-13 Axial Stress Contours in the V9 or V10 Weld Region for Baseline Conditions I

40

Hoop S (ksi)

A -35 B -30 C -25 0 -20 E -15 F -10 G -5 H 0 I 5 J 10 K 15 L 20 M 25 N 30 0 35 V9NI0 Weld (0.5 baseline heat Input) + Operating (550 F)

Case of Half the Baseline Weld Heat Figure 2-14 Hoop Stress Contours in the.U9 or U10 Weld Region for the Input and Operating Conditions 41

waafs S

(ksg A -10 B -5 C 0 0 5 E 10 F 15 G 20 H 25 I 30 J 35 K 40 I 45 ygNI0 Weld (0.5 baseline heat Input) + Operating (550 9 Weld Heat Figure 2-15 Axial Stress Contours in the-U9 or U10 Weld Region for the Case of Half the Baseline Input and Operating Conditions 42

Post Weld Diametral Loading Baseline Heat Input 50.

45 4Q 35 3Q th Ol oo m 20

)C 15 10 ~

~as welded after deadweight loading

~after 4 inch diametral squeeze

~affer 6 inch diametral squeeze

~after 8 inch diametral squeeze 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.6 Distance from Inner Surface (in)

Post Weld Diametral Loading Baseline Heat Input 30 20

'v) 10 CO CO 0

N Q.

O o 10

~as welded

~after deadweight loading

-20

-30

~

~aiter 4 inch diametral squeeze after 6 inch diametral squeeze

~after 8 in diametral squeeze 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-16 Baseline Heat Input HAZ Residual Stresses After the Application and Removal of Several Magnitudes of a Diametral Shroud Load Centered on the V9 and V10 Welds 43

V9N10 Baseline Weld HAZ Stresses 50 45 40 35 rh

~N 30 Vl g 25 CO m 20 15

~as welded 10 ~weld plus operating without squeeze

~weld plus 4 inch diametral squeeze

'~operating

, after 4 inch diametral squeeze 0

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

V9IV10 Baseline Weld HAZ Stresses 30 f ~

I 20

'g) 10 rh lO IcL.

0 O

0 -10

~as welded

~weld plus operating without squeeze

-20

~weld plus 4 inch diametral squeeze

-30

~ operating after 4 inch diametral squeeze 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from inner Surface (in)

Figure2-17 Baseline Heat Input HAZ Residual and Operating Stresses After the Application and Removal of a 4 Inch Diametral Load 44

0' V9/V10 Weld HAZ Stresses 0.75 Baseline Heat Input 60 50 40 lh y) 30 Vl Cl 20 C$

X 10

~as welded

-El weld plus operating without squeeze

~operating after deadweight squeeze

~operating after 4 inch diametral squeeze ',

~operating after 6 inch diametral squeeze '.

-10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.6 Distance from Inner Surface (in)

V9/V10 Weld HAZ Stresses'.76 Baseline Heat Input 40 30 20 10 vl 0 to -10 Q.

O 0 -20

-30

~as welded

~weltrptus operating without squeeze

~operating after deadweight squeeze

-40 ~operating after 4 inch diametral squeeze

~operating after 6 inch diametral squeeze

-50 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-1S Half Baseline Heat Input HAZ Residual and Operating Stresses After the Application and Removal of Several Magnitudes of Diametral Load 45

V9N10 Weld HAZ Stresses Oa5 Baseline Heat Input 50 40 30 to 20 ti0 ttg CO 10 C$

X

'as l e I

welded

-10 ~

~weld.pius operating without squeeze operating after deadweight squeeze

~operating after 4 inch diametral squeeze l

~operating atter 6 inch diametral squeeze I

-20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

V9/V10 Weld HAZ Stresses'.5 Baseline He~at In ut 50

~as welded I

40 ~weld plus operating without squeeze 30 ~

~operating after deadweight squeeze

-rr-operating after 4 inch diametral squeeze e pe raring airer 6 inch diam elrar squeeze f.

20 tO 10 to to 0

Co o~ 10 O

-20

-30 AO

-50 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Distance from Inner Surface (in)

Figure 2-19 Half Baseline Heat Input HAZ Residual and Operating Stresses After the Application and Removal of Several Magnitudes of Diametral Load 46

H4 or H5 weld centerline 45.06 cylinder mid section 90'rom V9/V10 V9 or V10 weld centerline Figure 2-20 The 3D Finite Element Grid Used for the Transverse Weld Shrinkage Variation and 84/H5 Radial Weld Mismatch Simulations 47

H4 or H5 weld centerline 45.06 90'rom V9/V10 cylinder mid section V9 or V10 weld centerline Figure 2-21 The 3D Finite Element Grid Used for the V9/V10 Radial Weld Mismatch Simulations 48

0 Mid H4/H5 section L

V9IIO 0.006 In. Weld Shrinkage Variation (2 cycle)

Detormed Shape (xl00)

Figure 2-22 Deformed Shape Plot Shoxving the Radial Displacements at the V9/V10 Weld for a 2 Cycle Sinusoidal Transverse Weld Shrinkage Variation 49

ltoop (ksi)

A -16 8 4 C 4 D -4 E -2 F - 0 G 2 H 4 I 6 J II K I6 n'.

V9IIO 0.006 In. Weld Shrinkage Variatlon P cycle)

Figure 2-23 Hoop Stress Contours at the V9/VIO Weld HAZ Due to a 3 Cycle Transverse Weld Shrinkage Variation 50

Iloop (ksi)

A -9 B -8 C -7 D

E -5 F -4 G -3

, H -2 I -I J 0

&I10 0.006 In. Weld Shrinkage yarlatlon (3 ~cle)

I'ig ure 2-24

- u Mid Hoop Stress Contours at the Shroud i Section ection Due to a 3 Cycle Tra cle Transverse Weld Shrinkage Variation at the V9 and V10 Welds 51

Weld V9/10 Transverse Shrinkage Variation (0.006 inch peak variation) 14 12 10 lh M

C) ~OD CO ~po & QDppcpp '.

X 6 p p N 4. I X:

C)

Ql

-2 10 20 30 40 50 60 70 80 90 Variation Length (in)

Weld V9/10 Transverse Shrinkage Variation (0.006 inch peak variation) 28 24 i p g 20 16 Po K

CL o

O 12 ~

~OD ID ~

8

~ID &ODlocal i CO Z 4 10 20 30 40 50 60 70 80 90 Variation Length (in)

Figure 2-25 Summary of V9/V10 Peak HAZ Stresses as a Function of Transverse Weld Shrinkage Variation Length 52

V9110 0.10 In. Radial Mismatch (3 cycle)

Deformed Shape (x10)

Figure 2-26 Deformed Shape Plot Shosving the Radial Displacements at the V9/V10 Weld for a 3 Cycle Sinusoidal Radial Weld Groove Mismatch at the V9 and V10 Welds 53

Hoop Stress A -15 8 -10 C -5 0 0 E 5 F 10 G 15 22 a to f.2 Figure 2-27 Hoop Stress Contours at the V9/V10 Weld HAZ and at the Shroud Mid Section Due to a 3 Cycle Radial Weld Groove Mismatch at the V9 and V10 Welds 54

Axial Stress

'ksQ A -15 B -10 C -5 D 0 E 5 F 10 G 15 Figure 2-28 Axial Stress Contours at the V9/V10 Weld HAZ and at the Shroud Mid Section Due to a 3 Cycle Radial Weld Groove Mismatch at the V9 and V10 Welds 55

Weld V9/10 Radial Fit-Up Induced Stresses (0.1 inch radial mismatch) 15.

10 Ol lO Cl L

CO EI$

N X 5 C)

~ID CFl i'~OD i

-10

-15 10 20 30 40 50 60 70 80 90 Variation Length (in)

Weld V9/10 Radial Fit-up Induced Stresses (0.1 inch radial mismatch) 15 10 5

CO C.

o 0 xO N

CO EA

.5 i ~i l~~oo 0

-10

-15 10 20 30 40 50 60 70 80 90 Variation Length (in)

Figure 2-29 Summary of V9/V10 Peak HAZ Stresses as a Function of V9/V10 Radial Weld Groove Mismatch 56

0 H4 Radial Mismatch (8 cycfe)

I Deformed Shape (xl0)

Figure 2-30 Deformed Shape Plot Showing the Radial Displacements at the H4/H5 Weld for an 8 Cycle Sinusoidal Radial Weld Groove Mismatch at the H4/H5 Weld 57

Hoop Stress Pcsg A 0 B 2 C 4 D 6 E 0 F 10 G 12 H 14 I 16 J 18 Hoop Stress A -'15 B -10 C -5 D 0 E 5 F 10 G 15 Figure 2-31 Hoop Stress Contours at the V9/Vlo Weld HAZ and at the 84/85 Weld Centerline Due to an 8 Cycle Radial Weld Groove Mismatch at the H4/H5 Weld 58

Axial stress A

8 -6 Q .4 D -2 E 0 F 2 G 4 H 6 I 8 Adal Stress A

8 -1 C 0 D l E 2 Figure 2-32 Axial Stress Contours at the V9/Vlo Weld HAZ and at the H4/85 Weld Centerline Due to an S Cycle Radial Weld Groove Mismatch at the H4/H5 Weld 59

Weld H4 Radial Fit-up Effect at V9/10 (0.1 inch radial mismatch) 10 6

e 4 V)

C$

N 2 x

O

-8

-10 0 20 40 60 80 100 120 140 Variation Length (in)

Weld H4 Radial Fit-up Effect at V9/10 (0.1 inch radial mismatch) 28 24 20

~~0 D 16 O.

O 0

12 N

x g 8

)

20 40 60 80 100 120 140 Variation Length (in)

Figure 2-33 Summary of V9/VIO Peak HAZ Stresses as a Function of H4/H5 Radial Weld Groove Mismatch Length 60

II,Iwt,~~g

e'C,J. ~

tt kgb t'Agp~

" 4g~

%A-t' fg C+C

~4~

Ct~~

4~,

II..,"I:....-".::,-.!i::.aaa55RRR%~~

it:"-II',-:-i',.I:Itiaa%1gSSSQ+

I

" it III I attui%$

..',"$ %5gg+~~

iWttStt101NIN1tt~gRR+

~

R ii081 5 t 1%1%%

5 S tt 1 8 S 5%%%%Sg 1WSkiN'FN1SSi 555gggg l ch I.p Ctt

+I 11 Itt ttl tt aa t

tti tt I

1 1

'I tt tt 11

~

I I t I 'I I I II I ' I

Normalized Stress Intensity Factor Uniform Stress (ID or OD) 18.O 16.0 ~

14.0 12.0 ld Current FEM shroud 10.0

-- 0 - -Current FEM SEC Handbook SEC 8.0 6.0 4.0 2.0 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 aft Normalized Stress Intensity Factor Uniform Stress (ID or OD) 1.15 1.10 El Current FEM shroud

-- 0 - -Current FEM SEC Handbook SEC 1.05 IA 1.00

~~ 0.95 LL 0.90 0.85 0.80 0.0 0.1 0.2 0.3 04 0.5 0.6 07 0.8 0.9 1.0

. a/t Figure 2-35 Comparison of Shroud Stress Intensity Factor Behavior to that of a Single Edge Cracked Plate Geometry and Comparison of the Current Solution with a Handbook Solution 62

Normalized Stress Intensity Factor Uniform stress (ID or OD) 3.00 2.80 2.60

~

Curreul FEM shroud 0 - -Current FEM SEC Handbook SEC 2AO 2.20 2.00 1.80 LL 1.60 1.40 1.20 1.00 I i

I i I 0.80 0.0 0.1 0.2 0.3 0.4 0.5 06 0.7 0.8 0.9 1.0 an Figure 2-36 A Special Normalization of the Shroud Stress Intensity Factor Solution that Makes it Possible to Estimate Values for Very Deep Cracks 63

OD Axial Crack Operating Conditions 60 M

50 40

~

~1.5 baseline weld heat baseline weld heat

~0.75 baseline weld heat ',

~0.5 baseline weld heat i 30 Vl 20 10 0

-10 0.0 0.1 0.2 0,3 OA 0,5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

ID Axial Crack  !

Operating Conditions 60 I I

50

~1.5 baseline weld heat baseline weld heat

~0.75 baseline weld heat 40 ~0.5 baseline weld heat 30 M

20 10 0

-10 0.0 0.1 0.2 0.3 04 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

Figure2-37 Stress Intensity Factor Solutions for V9/V10 HAZ Cracks at Operating

~ ~

Conditions for Four Assumed Weld Heat Input Levels 64

Axial Crack Under Uniform or Bending Stress (ID or OD) 40

-Et- uniform 1 ksi stress 35

~ ~1ksi bending stress  !

30 25 20 15 10 0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

Figure 2-38 Stress Intensity Factor Solutions for V9/VIO HAZ Cracks Due to a Uniform 1 ksi Hoop Stress or a 1 ksi Bending Stress Distribution 65

OD Axial Crack Stress Intensity Factor Behavior 0.75 Baseline Weld Heat Input 50 40

~

~weld weld + operating e 2 in. Iqneeze e nperadng I

4 in. squeeze+ operating wefd+

6 in. squeeze+ operating j 'weld+

30 j I I t I I 20 I

10 0

-10 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

ID Axial Crack Stress Intensity Factor Behavior 0.75 Baseline Weld Heat Input,'0 50 ~

~weld+ operating

~weld+ 2 in. squeeze+ operating i a

~weld+ 4 in. squeeze+ operating I

~weld+ 6 in. squeeze+ operating 30 20 I

~ I

-10 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

Figure2-39 Stress Intensity Factor Solutions for V9/V10 HAZ Cracks at Operating Conditions for the 0.75 Baseline Weld Heat Input Case with Several Levels of Diametral Loading 66

OD Axial Crack Stress Intensity Factor Behavior 0.5 Baseline Weld Heat Input 50 45 ~wetd+ operating

~weld+ 2 in. squeeze+ operating 40

~weld+ 4 in. squeeze+ operating "

35 ~weld+ 6 in. squeeze+ operating 30 rn 25

~ 20 s,

i J g 15 1

I e l

I 10 I

(

r 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 O.g 1.0 1.1 1.2 1.3 1.4 1.5 a (in}

ID Axial Crack Stress Intensity Factor Behavior 0.5 Baseline Weld Heat Input 50

~weld+ operating 40 ~weld+ 2 in. squeeze+ operating

~weld+ 4 in. squeeze+ operating

~ueid e g in. squeeze e egereiing 30 C

ol 20 z 10

-10 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.6 0.9 1.0 1.1 1.2 1.3 1.4 1.5 a (in)

Figure2-40 Stress Intensity Factor Solutions for V9/Vlo HAZ Cracks at Operating Conditions for the Half-Baseline Weld Heat Input Case with Several Levels of Diametral Loading 67

1 3.0 Crack Initiation and Growth Models A goal of this study was to explain the predominately OD cracking in the V9 and V10 welds.

Relatively few sites of ID cracking were observed. Where cracking was observed, the cracks were relatively shallow compared to the OD'cracking. Just as significant, however, is the fact that most of the ID surface did not show even signs of crack initiation. While the stress fields of Section 2 were being calculated, one factor that was considered in the selection of cases was whether they could lead to an explanation of the lack of ID cracking. In this section we describe an ad hoc criterion that was used in screening the surface stresses for initiation potential. The criterion basically involves comparing the stress intensity factor for a selected small crack depth to a SCC threshold stress intensity factor (K>><<).

3.1 Threshold tress Intensi actor There are numerous examples of K>>cc measurements in aggressive water environments in the open literature. An example is given in Figure 3-1 which shows K>><< to be in the 8-10 ksi v"in.

range. However, few measurements of K>>cc in typical BWR primary system water are available.

Reference [Ma94] reports on measurements of K>><< for furnace sensitized 304 SS (649'C for 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />) which was tested in oxygenated water (4 ppm) at temperatures ranging from 252'C to 288'C.

The conductivity ranged from 0.5 pS/cm to 1.3 pS/cm during these experiments. Experiments were conducted which made use of compact type specimens C(T) that had been modified to provide external surfaces that are isolated from the specimen, so that the current flowing between the crack and the external cathode could be readily monitored. This was done by coating the C(T)'specimens with baked-on polytetrafluoroethylene (PTFE) so that only the crack (after fatigue precracking) was exposed to the environment. Cathodes were then mounted on the sides of the C(T) specimen and the current was monitored, using a zero resistance ammeter (ZRA) of conventional design, as the stress intensity and water chemistry were varied. Typical results are shown in Figure 3-2 which suggests that K>><< is between 10 ksi Kin. and 20 ksi Kin.. Based on the literature data, and the data given in Figure 3-2, we have taken K>>cc to be 8 to 10 ksi v in.

3.2 m aris n of A lied tress Inten i K,) to~,sec For the purpose of comparing the applied stress intensity (Ki) to K>><< for shallow initial flaw depths, the solution for a long surface crack in a half space was used when considering initiation under hoop stresses (i.e., cracks aligned with the HAZ):

Ki = 1.12 alma where o = surface stress (ksi) a = flaw depth (in.)

68

t J

Assuming a threshold value of 10 ksi Kin, and a starter-crack depth of 0.050 in., a stress of 22.5 ksi is required to have the applied K, equal to Kscc. When considering crack initiation under axial stresses, the solution for a finite length, semi-elliptical crack in a half space was used due to the limiting width of the HAZ:

K,=F aftra where F is a function of the crack depth (a) to half length (c) ratio. Assuming a starter-crack depth of 0.050 in., and a HAZ width of 0.25 in., then c for the starter-crack is 0.125 in. And the alc ratio is 0.4. The value of F for this crack shape is about 0.94. For an applied K, of 10 ksi fin, we must have a surface stress of 26.8 ksi to initiate IGSCC. The use of the 0.050 in. clearly affects the stress levels that are assumed to initiate IGSCC. However, since the use of these computed initiation stress levels is only to compare the tendency for initiation at ID versus OD surfaces, it is believed that the values computed in this manner serve their purpose.

Using these initiation stress levels, it is possible to compare the potential for initiation at the ID and OD surfaces based on the HAZ stress levels summarized in Table 2-3 of this report. For the baseline weld heat input case of Table 2-3, it can be seen that the "weld+ operating" hoop stresses are below our initiation stress values computed above. However, the axial stresses are both above our initiation values. The OD axial stress of 32.7 is moderately larger than the ID value of 29.1, but this does not explain the observed lack of ID cracking. When a 4 in. diametral squeeze is app/ied and released before applying operating loads, the hoop stresses are further reduced and the diQerence between the ID and OD axial stresses becomes larger. With this magnitude of squeeze loading, the ID axial stress is well below our assumed initiation value of 26.8 ksi and the OD stress is just slightly below it. Ifa smaller diametral squeeze type loading had been applied, it appears that all HAZ surface stresses would be below their initiation values except the OD axial stress. It appears likely that the diametral squashing due to just the deadweight may have been sufficient for this condition to occur.

For the 0.75 baseline heat input case, the-ID and OD axial stresses are again above out initiation values for the "welded+ operating" condition. The ID hoop stress is above the initiation value while the OD hoop stress is below. While OD axial stress initiation is most likely, ID initiation due to hoop stress is also expected. It can agin be seen that the application of a diametral squeeze that induces tensile hoop directed plasticity at the ID of the weld results in a decrease in stress levels at the ID. Again, the deadweight induced diametral load (2" squeeze) is sufficient to reduce the ID stresses below are initiation values. The OD axial stress in this case is just about equal to the initiation value of 26.8 ksi. An increase to a 4 in. magnitude of diametral loading further decreases the ID stress levels without significantly lowering the OD value of axial stress. A further increase to 6 in. Of diametral squeeze reduces all surface stresses to values well below our initiation threshold values.

The surface stress behavior for the 0.5 baseline heat input case is similar to that of the baseline and 0.75 baseline cases. However, it can be seen that the decrease in weld heat input results 69

'9 in an even more marked difference between ID and OD surface stress values. For the 2 in. (I.e.,

deadweight) squeeze case, the OD axial stress (31.7 ksi) is not only well above our initiation threshold stress level, but it is more than twice the magnitude of either ID surface stress component.

It is concluded that it is likely that some form of diametral loading could easily have been applied to the V9/V10 shroud subassembly during or after welding of the V9 and V10 welds. The fact that a deadweight loading appears adequate to induce the ID surface initiation retardation effect makes this type of loading very plausible. The fact that the welds are 180 degrees apart, would seem to also add to the likelihood of this form of loading since the first completed weld would have been at the bottom (6 o'lock position) during welding of the OD side of the opposite weld (at the 12 o'lock position). The assumption of the ID welding being done at the 6 o'lock position and the OD welding being done at the 12 o'lock position is based on the use of the submerged arc weld process in which flux is held in the weld groove under the action of gravity.

3.3 rack rowth alculation odel In order to account for radiation effects on crack initiation and growth, consideration must be given to neutron damage in the steel and to gamma induced radiolytic decomposition of the water.

Irradiation assisted stress corrosion cracking (IASCC) is not an important concern in the non-sensitized 304 SS shroud base metal because the fluences through the en'd of cycle 12 have not reached threshold levels over significant azimuthal regions in the beltline. The peak base metal fluence at the end of cycle 12 is 6 x 10" n/cm', which is just at the threshold for IASCC to start to become a concern. However, most of the base plate regions are below the fluence level where one would expect to see detectable IASCC effects. This fact, coupled with the fact that almost all of the cracks observed in the NMP-1 shroud were contained within the HAZ, has led to the conclusion that IASCC is not at present a concern for the NMP-1 shroud.

However, General Electric (GE) has suggested [BWROG94] that there may be an effect on the sensitized material (such as the HAZ of V9/V10) at fast neutron fluences above -1 x 10" n/cm .

Through the end of cycle 12, the GE model implies a neutron induced increase in the electrochemical potentiokinematic reactivation (EPR) level of about 7 C/cm'. This effect on the sensitized grain boundaries, in addition to the acceleration of crack growth due to radiolytic decomposition of water (higher levels of 0, and H,O,), are accounted for in the GE neutron fluence dependent crack growth rate model.

The GE model, which has been adapted for the NMP-1 calculations, is fully described in P3WROG94]. The model includes the following effects

~ ~ radiation enhanced segregation which increases EPR

~ radiation effects on ECP

~ reduction of stresses due to irradiation enhanced creep 70

The functional form of the model is as follows:

dt

= 11.055 n 'tCK "]" (incheslhr)

where, K = ksi/in (adjusted for stress relaxation)

C=4.1x10"; fluence z 1.4x10'/cm'

= 1.14 x 10" ln(fluence) - 4.98 x 10" 1.4 x 10'/cm' fluence c 3.0 x 10" n/cm n = determined from GE fits to measured data and depends on EPR, ECP, and conductivity The EPR is determined from the following equation developed by GE:

EPR = 15+ 3.36 x 10" (fluence)"'C/cm')

In this model it is assumed that the NMP-1 shroud welds were initially sensitized to an EPR of 15 (C/cm'). Since the fluence effect on EPR is negligible below a fluence of 3.0 x 10" n/cm', the EPR was modeled as a fixed value of 15 below a fluence of 3.0 x 10" n/cm'.

With regard to stress relaxation due to irradiation enhanced creep, GE has proposed a model which has been fit to data and is only a function of fluence. Although most stress relaxation models which have been developed for non-nuclear applications show a stress dependence, the.GE correlation is plausible because of the high supersaturation of vacancies in a neutron field. The GE model was conservatively linearized into two linear segments with a threshold fluence of 1.0 x The linearization has been defined by the following table of data:

10'/cm'.

/

FastNeu r ue ce n/cm2 Fracti n e idual tre Remainin 1.0 x 10'~ 1.0 1.1 x 10" 0.83 1.0 x 10" 0.30 3.4 rack rowth del redictions As shown in Figures 3-3 and 3-4, plant-specific crack growth rate data for NMP-1 have been measured in the core and in the recirculation line. The location for the in-core measurement is shown in Figure 3-5. As summarized in Table 3-1, the in-core crack growth rate is -10 times higher than the recirculation line crack growth rate. The fast neutron flux (assumed to attenuate in a qualitatively similar fashion as for gamma radiation) attenuation (Table 3-1) of two orders of magnitude from the core to the shroud indicates that the crack growth rate at the H4 weld is lower than that in the core since there is less radiolysis at the H4 location. Therefore, the plant-specific crack growth rates were interpolated using electrochemical corrosion potential (ECP) data. The use of ECP as an interpolation parameter is reasonable since water radiolysis creates non-equilibrium concentrations of electroactive species (O~, H~, 8 O, g, OH, etc.). These species result in a potential drop across the metal-solution interface (referred to as a mixed potential). Thus, the ECP 71

is a useful measure of the corrosion potential and can be used to interpolate crack growth data in cases where the water chemistry is equivalent but the gamma field is different.

Reference [Be93] made use of available thermohydraulic data'and radiolysis data reported in Reference [EPRI89] to calculate corrosion potential maps for 8 BWRs. The results for NMP-1 are summarized in Table 3-1 and are plotted in Figure 3-6. These data were used to interpolate the NMP-1 plant-specific ECP measurements to obtain ECP data at the ID and OD surfaces of the H4 weld. The interpolated crack growth data were then obtained using the measured ECP data as the interpolation parameter. The results are summarized in Table 3-1. The GE model yields crack growth rates which are in reasonable agreement with the NMP-1 plant data.

A sample case was run for a constant K of 20 ksiv'in to illustrate the features of the model.

The ECP was assumed constant at 150 mV~ and the conductivity was also assumed constant at 0.1 NS/cm. The results of the calculation are shown in Figure 3-7. Radiation effects become significant after a threshold fluence of about 1 x 10" n/cm'. As shown in Figure 3-7, radiation enhanced segregation effects are significant at 7-8 x 10" n/cm . After a fluence of about 2 x 10'/cd, irradiation enhanced stress relaxation becomes dominant and the crack growth rate is significantly reduced as fluence increases.

72

Table 3-1 Measured and Calculated Crack Growth Rates for Nine Mile Point Unit 1 Fast Neutron Flux ECP (mV>>,) Measured Crack GE Crack Plant Location (n/cm'/sec) Growth Growth Model measured calculated Rate (in/hr) Prediction (in/hr)

In-Core 4.85 x 10" (see d) 1.42 x 10" (see a) 2.13 x 10 4 (see e)

Shroud Weld H4 8.97 x 10" 217 250'67'50'5'47 ID Shroud Weld H4 4.82 x 10" 211 OD Recirculation low 184 1.27 x 10'see a) 6.54 x 10 6 (see f)

Line I

a) Measured data reported in [ESEERCO88]

b) Reported in [Be93]

c) Interpolated using calculated ECP data d) The calculated flux is'in good agreement with the 3D MONICORE code computations performed using 1993 in-core measurements.

These calculations resulted in a flux of 4.77 x 10" n/cm'/sec at local power range monitor (LPRM) number 19.

e) Nominal conditions of: ECP = 200 mV~,; conductivity = 0.1 pS/cm; initial EPR = 15 C/cm', K = 25 ksiv'in.

f) Nominal conditions of: ECP = 75 mV,,; conductivity = 0.1 pS/cm; initial EPR = 15 C/cm; K = 25 ksi/in.

73

Type SG4 St n/ess S/ee/

OC8 Specimens to C0ns/on/ Lood Tas/

ZZR NoC/

Temp.=/OS C to

~ to" O

CL (0

(3 todl ~ Anne@ed Air Coo(ed fran l060'C 8 Sensitized 50Hrs ot650 C to 0 6 2O SO 40 5O 60 Stress 1tAmity, MN m Figure 3-1 Effect of Stress Intensity of Stress Corrosion Cracking (SCC) Velocity for Solution Annealed and Sensitized Type 304 SS in 22% NaCI Solution at 105 C (Sp>>j 74

Water Temperature ~ 638. 8 Pressure -" 4200 psfg = 40 ksiVin K>

2500 K, = 30 ksiVin 700 600 2000 Load 600 Z 1600 400 K, = 20 ksiV'in 300 0

> 1000 200 K, = 10 ksiVin 100 SOO

-1 00 0,00 0.60 5,00 1.60 2.00 2.60 3.00 3.60 7Jtna (hrs.).

I Figure 3-2 Load and ZRA Current Versus Time for Multiple Loading/Unloading Cycles to Various K, Levels 75

304 STAINLESS STEEL NOMINAL: 15 Ccm-2 (EPR) 10 7 NOMINAL: 25 Ksi ~in. (K)

'N-CORE NOMINAL: "HIGH PURITY" (MATERIAL)

ASSUME 0.1-0.3 pS cm

'(WATER CONDUCTIVITY)

SENSITIZED SENSOR I

)- 8

<10 K

IN-PIPE SENSITIZED Z -SENSOR 0

(3 0

CC 0

~10 9 O 0.3 pS cm 1 0.1 pS cm 1 10 10 600 400 200 0 + 200 CORROSION POTENTIAL (mVshe)

Figure 3-3 Observed vs. Theoretical Predicted Crack Growth Rates for Sensitized Stainless Steel Sensors [ESEERCOSS]

76

4.2

4. 1 4.8 N

5 3.9 8 CA 8

3.0 Q LJ bJ 3.? K G:

3.6 288 LLI 0

3.5 Measured Crack Cirowth Rate = 1.42 x 10 in Z hr LJ 188

3. '1 3.3 8 1888 2888 3888 4888 5888 6888 ?888 8888 9888 188BB HOURS

/

Figure 3-4 Sensitized Type 304 SS in NMP-1 Core tESEERCO88]

77

52 51 ECP 50 49 48 47 SCM 46 45 ~

44 43 LPRM LOCATION EACH LOCATION 42 41 R IRM LOCATION (8) 39 40 37-38 SAM LOCATION (4) 35-36 34

• SOURCE LOCATION (5)

~O 33 CONTROL ROD (12) 32 31-29-30 POISON CURTAINS (264)

REMOVED 27-28 O WRNM 25-26 23-24 NOTES:

21-22 19-20 1. ASSFMBLIES WITH 17-18 2. CO.ORDINATES ARE 16 15 3. CONTROL ROD 14 ~

13 12 a) IF CIRCLE 10 11 b) IF CIRCLE 8

9 7 c) IF A NUMBER 6

5 4

2 3

CONTROL ROD I

4 6 8 101214 1618 202224 26 28 30 32 3438 38 4042 444648 50 FUEL 1 I 3 I 5 I 7 I 9 l11113I15 ) 17 I )9 I 2) I 23I 25I 27l 29 I 3)/ 33135[37 f 39(4) I 43 I 4 5 I 47 I 49 l 5 Figure 3-5 Location of the NMP-1 In-Core Stress Corrosion Monitor (Designated SCM) [ESEERC088]

0.2

~

Mid Fuel Channel Core Bypass 0.0 Upper Plenum .

-0.2

~ Dotrncomer Inlet ElGSCC Recirculation Inlet

/

-OA Recirculation Outlet

-0.6

~ Bottom Plenum

-0 8 Feedwater Hydrogen Concentration (ppm)

Figure 3-6 Electrochemical Corrosion Potential as a Function of Feedwater Hydrogen Concentration for NMP-1 79

~ 0.000025

~o (0 0.000020' EPR effect signifi '-

0.000015 below'luence radl tion irradiation threshold effe ts enhanced sign icant

• 'nt stress I(3 0.000010 ~ ~ ~ ooNO relaxation dominant threshold-or stress relaxation O

gQ gQ gQ Fast Neutron Fluence (n/cm 2)

K = 20 ksi(in)"0.5 conductivity = 0.1 microS/cm ECP = 150 mVshe Figure 3-7 Crack Growth Prediction Using GE Crack Growth Model 0 ( for K=20 ksiv'in, ECP = 150 mV,,;.conductivity = 0.1 pS/cm) 80

4.0 Neutron Flux Data The radial (R)-angle (8) neutron transport data was taken from References [BCL84,BCLS4b].

Table 4-1 documents the peak azimuthal data which was extracted from Reference [BCL84b] and converted to fast flux. Figure 4-1 is a plot of the fluxes as a function of azimuthal angle and Figure 4-2 is a plot of the peak azimuthal fluences..

In order to account for axial peaking, the data in Reference [GE86] was normalized. In order for the Reference [GE86] data to'e compatible with the'Reference [BCLS4] data, the axial fluxes from Reference [GES6] were normalized at the capsule elevation: The c'enter of the surveillance capsule is located at elevation 2S1 5/16 inches (plant drawing 945E180). Since the bottom of the active fuel is at 209 5/16 inches and the active fuel height is.144 inches (plant drawing 104R859),

the surveillance capsule is located at the mid-height of the active fuel region which corresponds with the average fluxes for nodes 12 and 13 of Reference [GE86]." The locations of Reference [GE86]

nodes are given in Table 4-2 and the results from Reference [GES6] are given in Table 4-3. The normalization factor for the ID surface is 3.099x10 and for the I/4~T is 1.823x10.. It was assumed that the flux normalization at the vessel ID surface applies to the shroud. The normalized axial flux profile data are given in Table 4-4 and the data are plotted in Figure 4-3...I The azimuthal flux data and normalized GE axial flux profile were used to calculate the Wial flux profile at the ID and OD surfaces of V9 and V10. These data are provided in Tables 4-5 and 4-6. Plots of the V9/V10 fluxes and fluences are given in Figures 4-4 through 4-7. As a result of octal symmetry, the fluxes at V9 are identical to those at V10.

/

81

Table 4-1 Peak Azimuthal Flux and Fluence at thc End of Cycle 12 at Shroud ID Surface (E<xtracted from Reference [BCL84b])

~ I Azimuthal An le Peak Azimuthal Flux Peak Azimuthal Fluence (degrees) (n/cm2/sec) (n/cm2) 1.7 5.2E+11 2.8E+20 5.2. 5.9E+11 3.1E+20 8.7 7.2E+11 3.8E+20 10.0 7.6E+11 4.0E+20 12.2 8.3E+11 4.4E+20 15.7 I.OE+12 5.4E+20 19.1 I.OE+12 5.5E+20 22.4 7.7E+11 4.1E+20 25.0 6.3E+11 3.3E+20 273 5.0E+11 2.6E+20 29.2 4.1E+11 2.2E+20 30.9 3.4E+11 1.8E+20 32.7 2.8E+11 1.5E+20 35.0 2.4E+11 1.3E+20 37.7 2.0E+11 I.IE+20 40.6 1.8E+11 9.6E+19 43.5 1.8E+11 9.6E+19 82

Table 4-2 Reference [GE86j Locations of Nodes and Mid-nodes Node Elev. Top of Elev. Hid Node Inches- Node Inches 24 144.89 141.88 23 138.85 135. 83 22 132.82 129.80 21 127. 78 123.76 20 121. 75 117. 72 19 114. 71 111. 68 18 108. 67 105.65 17 102. 63 99. 61 16 96.60 93.57 15 90.55 87.50 14 84.52 81. 50 13 78.48 75.46 12 72.44 69.42 ll 66.41 63!39 10 60.37 57.35 9 54.33 51. 31 8 48.29 45.28 7 42.26 39.24 6 36.22 33.21 5 30. 19 27. 16 4 24 15

~ 21.13 3 18. 11 15. 09 2 12.07 9.06 1 6.03 3.02 0 0.00 83

Table 4-3 Comparisons Behveen Axial Fast Flux Calculations and Power by Node

[GAL<86]

( ppTp ppR pip-CYCLE CYCLE 7)

POWER FLUX (E'.0 el< u)

REL H20 NODAL CORNER DENSITY INSIDE WALL ]/4 T NODE AVE. BUNDLE.- ..-. (REG 2) --. ~ (INT 74) ( I NT 81) 24 0. 26S9 v.1=08 0.6]~5 i. 157E+C>9 6. 80 . E+<.>8 C> 4654 0.2168 0.6177 1. 996E~C>9 17 <<E+Ci9

<.>,6" <<5 v, 7<<2 0. 6 "42 2. 54CiE+<.>9 494E~<:.9 21 Ci. 7471 <.> ~ <<] ] 4 0. 6325 2. 91]E+v9 7 1 2EN<.>9 2 <.> <.>. 8446 <.> . =. 4 <.>.=. 0. 6426 <<. ]78E~C>9 868E+09 19 < > ~ 9 t >7 <>. <<641 0.6544 <<. <<78E+t.>9 986E+Ci9 18 C>. 9762 v..<<824 v.6682 <<.508E+09 C>6.:E ~09 17 1 ~ <.> <.> 7 0 ..=.9=.9 C>. 68)) 3. 528E+Ci9 0; 5E~C>9 16 1. V244 <.i. =956 0. 7<>2 << <<. 5C>9E+09 064E+v9 1 1. =99<.> C> ..<<880 <.>. 7'<<29 <<. 447E~C>9 Ci27~<.>9 14 1. 0 <<49 0 <<7=8 0. 746C> 3. <<08E+<>9 945E+C>9 1.0 78 0. <<6<.><.> <.i. 7717. 3. ]7 <<E~v9 866k~K>9 i=' 1. 04.=.<.> 0. <<486 <.>. 8<.>02 .026E+v9 l. 78C>E+09

1. 0=45 <>. <<4 1 .. 0. 8.<<16 2. 84vE~<.>9 l. 67]E+<.i9

<.> ]. <>4 <<<< <.>. <<<<74 0. 8655 2. 714E+v9 l. 596E+<.>9 9 1. <.>8.<<7 . <<5<.> Ci. 9<.>] << =. 625E+<.>9 /1. 54.:<<E+<.>9 8 1. 1477 <.>. 9378 2. 544 E+C>9 1. 496E+v9 7 1. 2 <<<.>8 0. 9725 2. 461E~C>9 l. 447E~v9 6 1. <<249 0. =-164 1 . t.><.><.>5 2 ..'7.<<EwV9 1 ~ <<97E+<'.>9 5 .1. 4128 C> ..<<<.> <<5 1 ~ C> 1 9<.> 2. 27<<E+09 l. <<<<SE+!.>9

1. 4588 0 '874 1 0 77~ 2. i60E+C>9 ~ 1. 27<.>E~C>9

1. <<9C>9 0.2678 r. 0=<<is 2. 0] ] E~<.>9 l. ]8:<<E~09

i. 1.:94 0.2:<<64 1. 0 <<51 1. 757E+v9 l. 0..4E+<.>9

0. 7018 C>. 16: l 1. C>.=76 1. ]59E~C>9 6. 8 1 ..E+<.>8

Table 4-4 Normalized Axial Flux Profile Height Above Normalized Flux Normalized Flux Bottom of Profile at Profile at Active Fuel Vessel ID 1/4 T (inches) Surface Position 3.02 0.373 0.373 9.06 0.566

'=':- " 0.567 15.09 0.648 0.648 21.13 0.696 0.696 27.16 0.733 0.733 33.21 0.765 0.766 39.24 0.793 0.793 45.28 0.82 0.82 51.31 0.846 0.846 57.35 0.875 0.875 63.39 0.916 0.916 69.42 0.976 0.976 75.46 1.023 1.023 81.50 1.067 1.066 87.50 1.112 1.1 1 1 93.57 1.132 1.132 99.61 1.138 1.138 105.65 -1.131 1.131 ~...

111.68 1.089 1.089 117.72 1.025 1.024 123.76 0.939 0.939 129.80 0.819 0.819 135.83 0.643 0.643 141.88 0.373 0.373 85

E I

Table 4-5 Axial Flux and Fluence at the End of Cycle 12 at the ID Surface of Shroud Welds V9/Vl0 Axial Elevation (inches Axial Flux Profile at ID Surface Axial Fluence Profile at ID above bottom of active of Welds V9 and V10 Surface of Welds V9 and V10 fuel) (n/cm2/sec) (n/cm2) 3.02 2.51E+11 1.32E+20 9.06 3.80E+11 2.01E+20 15.09 4.35E+11 2.30E+20 21.13 4.67E+11 2.47E+20 27.16 4.91E+11 2.60E+20 32.31 H5 5.09E+11 2.69E+20 33.21 5.13E+11 2.71E+20 39.24 5.32E+11 2.81E+20 45.28 5.50E+11 2.91E+20 51.31 5.67E+11 3.00E+20 57.35 5.87E+11 I 3.10E+20 63.39 6.14E+11 3.25E+20 69.42 6.54E+11 3.46E+20 75.46 6.86E+11 3.63E+20 81.50 7.15E+11 3.78E+20 87.50 7.45E+11 3.94E+20 93.57 7.59E+11 4.01E+20 99.61 7.63E+11 '.03E+20 105.65 7.58E+11 4.01E+20 111.68 7.30E+11 3.86E+20 117.72 6.87E+11 3.63E+20 122.43 H4 6.42K 11 3.39E+20 123.76 6.29E+11 3.33E+20 129.80 5.49E+11 2.90E+20 135.83 4.31E+11 2.28E+20 141.88 2.50E+11 1.32E+20 86

0 Table 4-6 Axial Flux and Fluence at the End of Cycle 12 at the OD Surface of Shroud Welds V9/VIO ta evation inc es ia ux ro i eat ia uence ro i e at above bottom of active Surface. of Welds.V9 and,V10 Surface of Welds VB and V10 fuel) (n/cm2/sec) (n/cm2) 3.02 1.36E+11 7.21E+19 9.06 2.07E+11 1.09E+20 15.09 2.37E+11 1.25E+20 21.13 2.54E+11 1.34E+20 27.16 2.67E+11 1.41E+20 32.31 H5 2.77E+11 1.47E+20 33.21 2.79E+11 1.48E+20 39.24 2.89E+11 1.53E+20 45.28 2.99E+11 1.58E+20 51.31 3.09E+11 1.63E+20 57.35 3.19E+11 1.69E+20 63.39 3.34E+11 '.77E+20 69.42 3.56E+11 1.88E+20 75.46 3.73E+11 1.97E+20 81.50 3.89E+11 2.06E+20 87.50 4.05E+11 2.14E+20 93.57 4.13E+11 2.18E+20 99.61 4.15E+11 2.19E+20 105.65 4.13E+11 '.18E+20 111.68 3.97E+11 2.10E+20 117.72 3.74E+11 1.98E+20 122.43 H4 3.49E+11 1.85E+20 123.76 3.42E+11 1.81E+20 129.80 2.99E+11 1.58E+20 135.83 2.35E+11 1.24E+20 141.88 1.36E+11 7.20E+19 87

Peak Azimuthal Flux at Shroud ID Surface

~ ( I ~

Q ~

~ y

'1.20000E+12 o 1.00000E+12

~ 8.00000E+11 oE 6.00000E+11 X

LL, 4.00000E+11 V9 and V1 0 equivalent O azimuthal position Z',

~~ 2.00000E+11

~ ~

0 5 10 15 20 25 30 35 40 45 Azimuthal Angle (degrees)

Figure 4-1 Plot of Peak Fast Neutron Flux at Shroud ID Surface as a Function of Azimuthal Angle 88

Peak Azimuthal Fluence at Shroud ID Surface 6.00000E+20 4 ~

P 5.00000E+20 ~ w II: ~ t o 4.00000E+20 oD 3.00000E+20 4 2.00000E+20 V9 and 0 V1 0 equivalent azimuthal position Z',

1.00000E+20 ~ ~

0 5 10 15 20 25 30 35 40 45 Azimuthal Angle (degrees)

Figure 4-2 Plot of Peak Fast Neutron Fluence at Shroud ID Surface as a Function of Azimuthal Angle 89

1.5 LL C

1.0 u

05 E

O Z

I/O T 0.0 ID Surface 0 24 72 96 120 144 Height Above Bottom of Active Fuel (inches)

Figure 4-3 Plot of Normalized Axial Flux Profile as a Function of Distance Above the Bottom of the Active Fuel 90

1 ID Axial Flux Profile for Welds V9 and V10

. ~ \~ \ Tg tP t \

~

+ 1, i 'i ~

1.00000E+12 I I

o 9.00000E+11 I I

I

~ 8.00000E+11 I I

I o 7.00000E+11 I I

x 6.00000E+11 I I

~ 5.00000E+11 I I

I 0 I I 4.00000 E+11 I I I

I

~ 3.00000E+11 I

I I

I weld- weld LL 2.00000E+11 '5 I

I I

'H4 I

I I 1.00000E+11 0 24 48 72 96 120 144 Height Above Bottom of Active Fuel (inches)

Figure 4-4 Plot'of Axial Fast Neutron Flux Profile at Shroud Weld V9/V10 ID Surface as a Function of Height Above Bottom of Active Fuel 91

ID Axial Fluence. Profile for Welds Vg and V10

~ E ~ i ' ~ 4 AI 4 ~ ~ I

• 5.00000E+20 I I I I I I I I E I I

~ 4.00000E+20 I I I I I I I I O I I

~ 3.00000E+20 I I I I

I I 0 I I

I I I I I z 2.00000E+20 I I

I I

I weld I weld

'H5 I I I I 1.00000E+20 0 24 48 72 96 'j20 'j44 Height Above. Bottom of Active Fuel (inches)

Figure 4-5 Plot of Axial Fast Neutron Fluence Profile at Shroud Weld V9/Vlo ID Surface as a Function of Height Above Bottom of Active Fuel 92

I~ 4 ~

OD Axial Flux Profile for Welds V9 and V10

<< ~ ~

"" uiuisiiienL 'l iCQuCC'I Sic' "" " ~

':." ~

1I 'K

'1.00000E+12 I I

\ \

~ ~ C ~ ' '

I o 9 00000E+11 I I I I

~ 8.00000E+11 I

I I

I I I I I o 7.00000E+11 I I

~

I I x 6.00000E+11 I I I I

~ 5.00000E+11 I

I I

I weld I IH4 0 weld I

4.00000E+11 ,

H5 I

I I

3.00000E+11 Lt 2.00000E+11 1.00000E+11 0 24 48 72 96 120 144 Height Above Bottom of Active Fuel (inches)

Figure 4-6 Plot of Axial Fast Neutron Flux Profile at Shroud Weld V9/V10 OD Surface as a Function of Height Above Bottom of Active Fuel 93

y f OD Axial Fluence Profile for Welds V9 and V10 5.00000E+20 I I I I I I I I E I I

~ 4.00000E+20 I I

I I

I I I I O I I I I

~ 3.00000E+20 I I

I I

I I I I 0 I I I I weld

'eld R 2.00000E+20 'HS I

I I

1.00000E+20 0 24 48 72 96 120 144 Height Above Bottom of Active Fuel (inches)

Figure 4-7 Plot of Axial Fast Neutron Fluence Profile at Shroud Weld V9/Vlo OD Surface as a Function of Height Above Bottom of Active Fuel 94

5.0 Crack Growth Calculations The axial stresses at the OD surface of the V9 and V10 HAZ due to welding and operating stresses were predicted by the finite element simulations to be large enough to initiate IGSCC. If a diametral loading is assumed after welding and before operation, the ID surface HAZ stresses can be left in a state for which IGSCC initiation is much slower than at the OD or perhaps circumvented entirely. The limited IGSCC that has been observed at the ID surfaces of V9 and V10, can be explained in terms of the existence of localized, high stress regions.

Lo'calized regions of high stress can occur as the result of fabrication related procedures, including weld repairs, and jacking to align parts before welding.'These localized stress inducing procedures were studied and their associated stres's distnbutions'and magnitudes calculated in Section 2.

This section describes the crack growth simulations that have been done using the plant specific crack growth model that was described in Section 3 and the stress intensity factors that were computed in this work and described in Section 2. The crack initiation process, that is, the process whereby grain boundaries are degraded first at the surface and then within a volume of material adjoining the surface, eventually leading to a systematic separation of grain boundaries that can be reasonably described as a crack, is not well understood. The time for a crack-like defect to form can be on the order of 2 to 10 years depending on the seve'rity of the stresses, the susceptibility of the material, and the severity of the corrosive environment. The model that is used in the current crack growth simulations does not attempt to predict the time that is required to incubate crack-like defects. Rather, the simulation is started from a given defect depth that can be arbitrarily selected but which more typically would correspond to the depth of a particular crack that has been sized by UT. The crack growth model is then used to essentiallyiintegrate crack growth backward in time to determine the time at which the crack initiated. In this way, the model leads to the determination of the incubation time for the selected crack. More importantly, however, the model can be used to integrate forward in time to predict future behavior for the crack of interest. This predictive capability for future crack growth behavior is essential as input to the engineering calculations that are required to demonstrate that structural safety margins will be maintained until the next time that UT inspections can be performed.

Throughout this section of the report, time is represented as effective full power years (EFPY).

The crack growth model is a function of fluence but the important issue related to water chemistry is hot operation time. Therefore, EFPY should be interpreted as hot operation time for cases where the reactor is not run at full power.

The equations comprising the crack growth model include crack growth law parameter dependencies of a sort that are not readily incorporated into general purpose, commercial,

&acture mechanics sofhvare. The coefficients of the growth law are both fluence and water chemistry (conductivity and ECP) dependent. Furthermore, the law assumes that irradiation induced creep effects relax residual stresses with time. Another complicating factor is that fluence

i is not constant through the wall thickness. The fluence at the ID is about 80% greater than at the OD. Due to these complications, it was decided to write a FORTRAN program to implement the crack growth model.

The approach of creating custom crack growth simulation sofhvare allowed several features of the V9 and V10 crack growth environment to be modeled that would have been difBcult or impossible with offthe shelf software. One was that the actual history of water chemistry could be incorporated in the calculations. The higher conductivities that existed during the early years of operation have a significant effect on predicted IGSCC rates. Figure 5-1 shows the measured mean conductivity history that was used by the sofbvare and shows how the recent low conductivity levels are assumed for future operation. While ECP could also be input as a time dependent parameter, insufficient data on its history led to the assumption that ECP is constant at 150 mV,H . Another feature of the custom sofbvare was that the fluence dependent portion ofthe crack growth model could be based on the actual fluence at the current through wall position of the crack tip as well as on the axial position along the weld. The software used plotting utilities to provide graphical output of key model behavior so as to make its use more efficient. Figure 5-2 shows a plot of crack tip fluence that was generated by the software. At any given point in the shroud, the fluence increases linearly with time. Since the crack tip is moving from the OD surface toward the ID surface in the simulation leading to this plot, the fluence is nonlinear with time. Since the EPR of the sensitized material is assumed by the model to increase with fluence, the EPR at the crack tip also increases in a nonlinear fashion due to crack growth.

This effect is shown in Figure 5-3. Figure 5-4 shows the crack growth model's irradiation induced creep effect. The residual stress component of K, (i.e., not including the operating component) was reduced by the fluence dependent factor that is plotted in Figure 5P. It can be seen that the creep related stress reduction is significant. The various titles on these ~lots indicate that the plots are for an OD crack assuming baseline weld residual stresses with no diametral squeeze effect and that the crack is at the position along the height of the weld where flux is maximum.

Two FORTRAN programs were actually written (circgrow and vgrow). One (circgrow) was for simulating growth of circumferential flaws and the other (vgrow) was for simulating growth of axially oriented flaws. The only portions of the programs that were significantly different were the stress intensity factor calculation routines. The axial crack program had the various K, versus crack depth curves computed in Section 2 directly encoded in the program. The circumferential crack program used a handbook solution for a semi-elliptical surface crack in a finite thickness plate [SIFH90]. This solution allows linear variation of stress with depth, and allows the crack depth to length aspect ratio (a/c) to be in the range 0 to I. A linear stress distribution is quite adequate for the current application since only the stresses over the first half inch of depth (the range of analyzed circumferential crack depths) needed to be fit. The coefficients of this linear stress distribution were inputs to the circumferential crack growth soRware. The handbook solution for the semi-elliptical crack is given as a polynomial function of a/c and crack depth to thickness (a/t) ratio and thus was easily programmed. As described above, the length of the circumferential cracks is limited by the width of the weld sensitized region 96

(assumed to be 0.25 in. wide). This limitation on length resulted in a/c reaching a value of unity (semi-circular crack) when the crack depth reached 0.125 in. For circumferential crack growth beyond a depth of 0.125 in., the crack must "tunnel" (i.e., crack depth becomes greater than half the surface crack length). The handbook solution did not cover this type of cracking and therefore an assumption had to be made. The crack depth dependence of K, for a tunneling crack is reduced since crack opening at the deepest point on the crack &ont is limited by the short surface length of the crack. At some depth the value of K, becomes largely independent of the crack depth dimension and depends primarily on the crack length dimension. It was therefore assumed in the simulations that K, for tunneling cracks was limited in magnitude to that which occurred when the crack was semi-circular (0.125 in. depth).

The transition Rom a circumferential crack orientation to an axial orientation may require some amount of time that is not represented in the model For cases in which numerous

~

circumferential flaws exist within the extent of the eventual axial crack, the transition will be much facilitated. In any case, the transition is best viewed as a continuous process rather than as a sudden change in crack orientation. Initially, axial stresses will dominate cracking and grain boundary attack will be oriented to reflect this. As the hoop stresses increase with depth, the grain boundary attack willbecome less oriented to the axial stresses. Eventually, the greater e6ect of the hoop stresses will tend to orient the attack along axial planes. This transition will occur near the idealized crack tip location and therefore will not be initiallyvisible &om the surface. However, once the cracking is dominated by hoop stresses, the axially oriented cracking will tend to propagate axially oriented cracks back toward the OD surface thus eventually leading to obviously axial cracks at the surface. It is based on this cracking scenario that cracks in the early and transition phase are expected to have less axially oriented cracks visible at the surface than deeper cracks for which cracking is dominated by hoop stresses.

/

It was described above that the crack growth model does not include a model for predicting the time for IGSCC to produce a crack which is large enough for K, to exceed Kiscc.

Rather, this crack incubation time was calculated by taking a measured crack depth at a specific location in the shroud, and integrating backward in time until the crack is at its threshold depth.

Then the crack model is integrated forward in time until arrest or through wall growth is predicted. Therefore, by inputting a single crack depth corresponding to the current crack depth at a specific location in the shroud, a complete cracking history can be computed for that crack.

This predicted cracking history is dependent on the assumed through wall residual stress distribution that is used in the modeling. By using a variety of assumed residual stress distributions, it is possible to illustrate the sensitivity of the predicted cracking behavior to the assumed residual stresses.

It has been stated that a backward integration in time is used to determine the early cracking behavior. This terminology is used because it clearly conveys the intent of the procedure. It should be noted, however, that even through such a backward integration algorithm 97

k is possible with the currently used crack growth model, the current sofhvare crack growth algorithm is actually for a forward integration in time. The early portions of the crack growth history were determined by selecting crack depths and time which when integrated forward in time led to the desired crack depth at the time of inspection. This selection wa's done on a trial .

and error basis. The initial crack depths for the circumferential crack model were selected so that K~ would be slightly exceeded. This initial crack depth ranged &om 0.035 in. to 0.050 in. If, the selected crack depth and time data point used to anchor the calculation was in the hoop stress dominated growth regime (axial crack), then the initial conditions for the circumferential crack growth model were selected to give a continuous and smooth transition to the circumferential crack growth model behavior. For the axial crack growth portion of the model, initial crack depth and time were again selected on a trial and error basis so as to have the crack growth history pass through the anchoring data point or to provide a continuous and smooth transition from the previously computed axial stress dominated crack growth behavior.

Figure 5-5 shows the results of crack growth modeling for two OD surface cracks. The top plot shows the growth history of a crack that was in the lowest flux region of the V9 weld (near the HS weld). The bottom plot shows the history for a V9 crack that was in the highest flux region of the V9 weld. The time for the plotted data corresponds to the current time in EFPY.

Since the wall thickness is 1.5 in., a crack depth of 1.5 in. indicates a through wall crack. The crack growth modeling of this figure assumed the residual stress distributions from the baseline weld simulation without any diametral squeeze loading before operation. The computed crack history is indicated by first a dotted line and then a solid line. The dotted line is the portion of the history computed with the circumferential crack model. The solid line was computed with the axial crack model. The slight mismatch at the intersection of the lines is due to the trial and error approach used to select initial conditions for each phase of the calculation. The lower ends of both portions of the computed curve indicates the depth at which K, first exceeded K~. A remarkable feature of these computed histories is that the two portions of the history tend to have similar slopes at the transition point. This was not expected. However, there is a reason for this behavior. The explanation is that the K, of the circumferential crack never substantially exceeds Kiscc. This is due to the limiting effect of the HAZ width for the tunneling circumferential crack.

Typically, K, did not exceed 13 ksi(in during the circumferential growth stage. Further, this peak value tended to occur before the transition to hoop dominated behavior due to the creep effects included in the model. Therefore, at the transition time, both models were using K, values that were near K~ in magnitude.

The crack growth behavior of Figure 5-5 indicates that the low flux region crack initiated growth after about 12 EFPY. The deeper crack at the highest flux region was predicted to initiate after about 6 EFPY. Although selecting other actually observed crack depths would lead to different predicted initiation times &om these values, the fact that the crack depths have been shown to correlate with flux leads to the conclusion that initiation times will also correlate with flux. The residual stresses that were used in the crack growth modeling of this figure lead to the model predicting through wall crack growth. The deepest crack (high flux) is predicted to grow through wall by the end of 17 EFPY. The shallower crack (low flux) is predicted to grow 98

through wall after about 23 EFPY.

Figure 5-6 shows the effect on the crack growth behavior of assuming a different residual stress state. For this figure, the residual stress state is that due to using half of the baseline weld heat input. It can be seen that the hoop stress driven portion of the crack growth history leads to very rapid growth rates in the mid depth region (0.4 to 0.8 in.) but then predicts a rapid deceleration in growth rate as the depth approaches 1.0 in. At about 1.0 in., K, is predicted to drop below K . The horizontal lines at this depth indicate the remaining crack growth behavior If ifthe crack arrests. cracking does not arrest due to dropping below g K~ then crack growth proceeds at a much reduced rate until a depth of about 1.2 in. At that time the crack growth accelerates rapidly and through wall growth is imminent. The reduced slope portion of the crack g

growth curve between 1.0 and 1.2 in. is due to being close to zero. Ifit had dropped below zero, this would indicate compressive stresses are acting to close the crack tip. Such a condition would be a more reliable indication of crack arrest behavior than having K, drop below K~.

Because the deep crack data point of the bottom plot is above the slow growth portion of the crack growth curve, the predicted incubation time for the crack is fairly sensitive to the crack depth data point. The left most curve, which actually passes through the data point, gives a relatively short incubation time of about 3 EFPY. The history that was computed so that K, dropped below K~ just before the current outage leads to a predicted incubation time of about 13 years. These two incubation times are near the extreme ends of the range of incubations times that are believed to be reasonable. Based on this assessment of incubation times, it seems that the actual cracking behavior is probably between the two extreme cases that are plotted.

Figure 5-7 shows the effect on crack growth behavior of adding a 2 in. diametral squeeze after welding and before operation. The 0.5 baseline heat input is still being used, so the results of this figure can be compared to Figure 5-6 to see the effect of the squeeze load. Basically, it can be seen that the predicted crack growth behavior for these OD cracks is not significantly different due to the squeeze load. Since this squeeze loading is believed to be suf6cient to prevent wide spread IGSCC initiation at the ID surface, it can be concluded that the squeeze loading effect is much more significant at the ID than at the OD, Figure 5-8 shows the effect on crack growth behavior of assuming the residual stresses from the 0.75 baseline weld heat input case with a 4 in. diametral squeeze after welding and before operation. It can be seen that the predicted cracking behavior is, as would be expected, somewhere between that of the baseline heat case (Figure 5.5) and the 0.5 baseline heat cases (Figure 5-6 and 5-7). The tendency for possible arrest at a depth of about 1.15 in. is similar to that of the 0.5 baseline case but with a slightly deeper arrest depth. The same bounding approach described for the deep crack case of the 0.5 baseline plots leads to a much narrower range of behavior due to the time associated with the potential arrest behavior being smaller. The shortened plateau of these curves is due to K, not approaching 0 as closely as for the 0.5 baseline case. Comparing the crack growth histories of Figure 5-8 to those for the baseline case of Figure 5-5, it can be seen that the crack histories are very similar. For the top plots, at a given crack depth, the times for the two curves are generally within 1 EFPY of being the same. For the 99

bottom plots, the times for equal depths on the two curves are within 1 to 2 EFPY of being the

~

~ ~ ~

same. The principal difFerences are that the baseline case tends to be smoother and predicts through wall cracking behavior.

To show that the shroud satisfies structural safety margins during continued operation, it is necessary to predict the crack growth behavior during the continued operation. Clearly, this crack growth prediction should be a conservative prediction (i.e., more crack growth predicted than actually occurs) since otherwise actual margins could be smaller than calculated. It is, however, useful to also do structural margin evaluations based on best estimated crack growth behavior so that the sensitivity and conservatism of the structural assessment can be better understood. With this desire for being realistic, but nevertheless conservative, the baseline weld heat residual stress distribution was selected for predicting crack growth behavior during the next period of operation. The cracking behavior &om this case is very representative of the other residual stresses cases which have been run but has the conservative feature of not predicting a significant slowing of crack growth rate (or arrest) near a depth of 1.15 in.

Weld V9 is the most severely cracked of the two welds, and thus is the case selected for structural margin calculations. The V9 UT based crack depth profile was 'conservatively simplified for structural analysis as illustrated in Figure 5-9. The axial dimensions of the two cracked regions were conservatively increased to reflect UT uncertainty and 2 EFPY of axial crack growth. The crack depths after two EFP Y of operation were predicted based on the baseline heat input case as shown in Figure 5-10. The deepest cracks for the V10 weld were predicted to grow to 1.16 in. The deepest portion of the V9 crack &ont was predicted to grow through wall very soon after operation commences. The portion of the crack front with depths of 0.5 in. or less was predicted to have a maximum depth of 0.82 in. after two EFPY of operation.

Linear elastic &acture mechanics (LEFM), elastic-plastic &acture mechanics (EPFM), and limit load analyses were. conducted to assess the margins for the V9 weld. Since the V10 cracks were predicted to not grow through wall for the 2 EFPY period being considered, V9 was clearly the limiting case. The limiting load case, based on [GE97] is the faulted condition which results in a b,p across the shroud wall of 22 psi. Using the 1.5 safety factor appropriate to this load case, the h,p used in the structural assessment calculations is 33 psi. For the 88 inch inner radius and 1.5 inch wall thickness of the shroud, this b,p results in a membrane stress of 1.94 ksi.

An LEFM handbook solution was used to compute K, for the through crack [TPI85]. The solution is for a finite length axial through crack in an infinite length cylinder. The infinite length cylinder solution is appropriate since the current inspection has shown that the H4 and H5 welds are not significantly degraded by IGSCC in the vicinity of the V9 and V10 welds. The computed value of Ki for the 41.84 inch through wall crack was 35.1 ksi)in. This is about one fourth of the 100

allowable value (Kg of 150 ksi)in [GE97]. To determine the applicability of the LEFM handbook solution to this crack con6guration', an EPFM handbook solution P)FH91] was used to compute J for this con6guration. This J was then converted to an equivalent K,. It was found that this value of K, agreed with the LEFM computed value to within a few percent. This con6rms that the use of LEFM is appropriate to this crack geometry and load level.

After the deepest portion of the crack Pont grows through wall, the effective applied K, at the end that adjoins the part through crack will tend to be larger than that computed above using the handbook solution. This is because the net section of the 0.68 inch remaining ligament is already at a higher average stress than the nominal membrane stress. This is due to the reduction in wall thickness due to the part through crack. The nominal average stress in the remaining part through crack ligament will be increased by the ratio of the initial wall thickness to the remaining ligament thickness (1.5/0.68 = 2.21). To get the effective K, we then scale up the originally computed value by the 2.21 factor. This leads to total effective K, at the intersection of the part through and through cracks of 77.6 ksi(in. This is still about half of the allowable value and thus we conclude that the remaining 0.68 inch thick ligament will not failure due to overload.

A limit load analysis was also done. It was conservatively assumed that the entire pressure load on the V9 weld must be born by the remaining V9 cross section. This is conservative because the H4 and HS welds and the material beyond these welds has not been signi6cantly degraded by IGSCC and thus are capable of sharing the V9 hoop load. The total pressure load that must be born by the weld is 262 kips. The allowable stress (flow stress) is 3Sm or 50.7 ksi

[GE97]. The rermmung V9 cross section after two EFP Y is 40.8 square inches. The required area (262/50.7) is 5.2 square inches. It can be seen that the predicted remaining cross section is about 7.8 times that required by the limit load criterion.

/

101

Assumed Conductivity History 10.00

'mean (uS/cm)

~+2 sigma (uS/cm);

-X-2 sigma (uS/cm) 1.00 O

CO O

'U 0

O 0.10 0.01 10 15 20 25 30 35 EFPY Figure 5-1 Conductivity History for NMP-1 Vsed in the Crack Growth Simulation 102

Crack Tip Pluence History 50.0 45.0 40.0 35.0 ~~ . ~ .) ~ ~ ~ ~ I 30.0 25 0 NJ 20.0 g 150 . ~ .)"

~

""""(-"

~ I (EPPES) 10.0 5.0 ' ~ ~

0

.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 time Figure 5-2 Crack Tip Fluence History from a Crack Growth Simulation 103

Crack Tip EPR. M.story 20.0 19,5 19.0 I

18.5 ~~ ) - ~

18.0 ~ ~ ~

17.5 B

17.0 16.5 ~ ~~ ~~ ) ~ ~ ~~ ~~ + ~ ~~ )~' "~""

16.0 8

I 15.0 20.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 time (BFFY)

Figure 5-3 Crack Tip EPR History from a Crack Growth Simulation

~ i- 5 ~$ e9 l ~ II Creep. Stress RelaxatiomPactor History 1.00 90 80 ~~ ~~~~~ J ~ A ~~ ~ ~> ~ ~

.70 4

0

.60 Qr

.40 0

,30

.20 ~

t'0 00 0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 tixnc (EFP~

Figure 5-4 Creep Stress Relaxation Factor from a Crack Growth Simulation 105

I OD Cracking Baseline Heat- Min Flux Region 1.6 1.4

- - - - - -axial stress driven 1.2 t hoop eireee driven R 1

0 V9 near H5 weld

~~ 0.8 R

e O R 0.6 no<<he Weevehonvv ad ++4~ r e vwa~*ee e ~. ~ ~ ~

O 0.4 0.2 R' j

I e

e 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

OD Cracking Baseline Heat- Peak Flux Region 1.6 1.4 It R 1.2 1

a

~~ 0.8 R

'i O I R

I 0.6 I n.

e

- - - - - -axial stress driven O

0.4 hoop stress drrven e

0 V9 deepest crack 0.2 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

Figure 5-5 Predicted Crack Growth History for OD Cracking Based on the Baseline Weld Heat Residual Stress Simulation

OD Cracking 0.5 Baseline Heat- Min Flux Region 1.6 1.4 1.2

- - - - - -axial stress driven 0

hoop stress driven VQ near H5 weld 1

I I f I i I ' I i l l

~ 0.8 crack arrest (K ( Kthres)

Cl

~o 0.6 'sfs I'hl lfrrkl l r 'l':VI2 nilrooh vsC trQI"CCClll. r.~L.<< ~ ".'I--

O ~

0.4

~ ~ I~ t.s

'.2 s

0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

OD Cracking 0.5 Baseline Heat- Peak Flux Region 1.6 I I f t I

I 1.4 1.2 I I I 1

~~ 0.8 l crack arrest (K ( Kthres) )

O ooo 06 O

0.4 - - -- ~ -axial stress driven hoop stress driven 0.2 D V9 deepest crack 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

Figure 5-6 Predicted Crack Growth History for OD Cracking Based on the 0.5 Baseline Weld Heat Residual Stress Simulation 107

OD Cracking 0.5 Baseline Heat 2 in. Squeeze - Min Flux Region 1.6 1.4 - - - - - -axial stress driven hoop stress driven 1.2 . Ct V9 near H5 weld e

r 1

I

~~ 0.8 D crack arres t(K ( Kthree)t 0.6 O

0.4 e I

~

l I r

I 0.2

~I 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

OD Cracking /

0.6 Baseline Heat-2 in. Squeeze- Peak Flux Region 1.6 l

l I

1.4 1.2 i

I 1

I r

~~ 0.8 D .. crack arrest (K ( Kthres) 0.6 L I O i i 0.4 . ~

--- ~ - -axial stress driven I

I r hoop stress driven 0.2 0 V9 deepest crack 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

Figure 5-7 Predicted Crack Growth History for OD Cracking Based on the 0.5Baseline Weld Heat Residual Stress Simulation with a 2 Inch Diametral Squeeze 108

OD Cracking 0.75 Baseline Heat -4 in. Squeeze - INin Flux Region 1.6 I  !

I I 1.4 .. ~ . axial stress driven hoop stress driven 1.2 D V9 near H5 weld 1

~~ 0.8 f I V I j crack arrest,(K < Kthres).

O V I'

0.6 f

~ r 'I g II ~

O I 0.4 I I

, t II 0.2 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time (EFPY)

OD Cracking /

0.75 Baseline Heat -4 in. Squeeze- Peak Flux Region 1.6 l l I

f I 1.4 1.2 1 f r

lcrack arrest (K < Kthres)

~~ 0.8 O

0.6

! I I

O - - - - - -axial stress driven 0.4 hoop stress driven 0.2 0 V9 deepest crack 0

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 Time(EFPY)

'igure 5-8 Predicted Crack Growth History for OD Cracking Based on the 0.75Baseline Weld Heat Residual Stress Simulation with a 4 Inch Diametral Squeeze 109

Weld VB OD Flux and Crack Depth Profiles conservatively estimated crack depth at 2 EFPY of exposure after 1.5

'cycle 12

4. 80000E+11 1.4 1.3 (o 4.40000E+11 1.2 1

1.1

'E 4.00000E+11 1.0 X

3.60000E+11 0.9 0.8 C3 M

bounding 07 ~

c- 3.20000E+11 approximation 0.6 5 at end of O cycle 12 0.5 ~

2.80000E+11 04

~ 0.3 measured at end 0.2 of cycle 12 0.1 2.40000 E+11 0.0 0 10 20 30 40 50 60 70 80 90 Distance Along Vg Measured from H4 (inches)

Figure 5-9 Comparison of UT Based Crack Depth Profile for V9 and the Conservative Approximation Used for Structural Margin Calculations 110

Two Yearn of Predicted OD Crack Growth Baseline Heat 1.5 t t  !

14 1.3 El- Vg high flux

%- V9and V10lovvhux 1.2 W V10 high flux

-- -... axial stress driven hoop stress driven

." ""a i

5...

r 0.8 0.7

~ 0.6 0.5 0.4 02 02 0.1 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Time (EFPY)

Figure 5-10 Predicted Crack Growth Behavior for Use in the Structural Margin Calculations for Weld V9 111

Analyses have been performed to gain an in-depth understanding of the cracking which has been observed during the non-destructive examination of the NMP-1 shroud at the end of cycle 12. These analyses have been performed to explain why cracking along V9 and V10 is predominantly OD with very little ID cracking. This type of cracking behavior would not be expected based on as-welded stress proflles for double V-groove welds. Another primary focus of the work has been to perform conservative, but accurate (non-bounding), calculations to determine the safe. operating time until the next shroud inspection. These calculations have been performed so that they can be compared with results of GE calculations which are based on recently developed BWRVIP criteria.

The performed calculations and analyses have led to the following conclusions pertaining to cracking at NMP-1 shroud welds V9 and V10:

Stress analyses have shown that fabrication practices control whether cracking in vertical:welds is. predominantly OD or both ID and OD. In the case of welds V9 and V10, a combination of low heat input and a diametral squeeze produce the most plausible stress Geld which results in predominantly OD cracking.

The stress fields show that OD IGSCC cracks initiate mainly under axial stress.

For the first few tenths of an inch of propagation, the axial stresses continue to dominate the crack growth behavior; however, aAer a depth of 0.3 to 0.5 inches, the crack growth behavior becomes dominated by the hoop stresses. These observations have been supported by UT and visual examination results which show both short circumferential cracks and long axial cracks. Also supportive is the observation that regions of more shallow cracking, tend to be more circumferential in nature. Initiation of cracks at the OD surface under hoop stresses is possible in localized regions where hoop stress peaks occur.

The cracking mechanism is IGSCC of the HAZ of the welds with radiation sects playing a role after an incubation dose of about 1 x 10" n/cm~. Visual and UT examinations have confirmed that the cracking in the vicinity of the V9 and V10 welds is contained within the HAZ.

Through the end of cycle 12, the shroud base metal fluences have not reached IASCC threshold over significant portions of the material. Therefore, IASCC in the plates is not a concern at present.

Plots of fluence and UT measured crack depth show a correlation between crack depth and fluence. Integrating the observed cracking backward in time using a best estimate crack growth model suggests earlier initiation for the higher flux 112

regions of the shroud. Typical incubation times of 6 EFPY have been predicted for the analyzed residual stress profile.

Crack growth simulations using the most plausible stress fields (those which lead to predominant OD cracking) have shown that deep flaws (-1.2 inches deep) may arrest as a result of the applied stress intensity factor falling below the IGSCC threshold level (K~- 10 ksiVin). Since little data are available on IGSCC arrest thresholds in complex stress fields, future NDE should focus on accurate crack depth profiles of V9.

Under the assumption that crack arrest does not occur, a structural evaluation of V9 was performed. Based on the NDE results, credit for the integrity of H4 and H5 in the vicinity of V9 was taken. LEFM, EPFM, and limit load calculations show that structural integrity will be maintained for at least an additional 2 years of hot operation. Therefore, there is a large margin between the allowable operating time predicted using the BWRVIP criteria (10,600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> or about 1.2 years of hot operation to next inspection) and the results obtained using a conservative, best estimate model.

7.0 References

[Aero] Aerospace Metals Handbook, 1991 Edition.

[ALT3D] ALT3D, Version 2.0, Sofbvare for the Analysis of Cracks in Complex 3D Geometries Using the Finite Element Alternating Method, Computational Mechanics, Inc., 1994.

[BCL84] Battelle Report BCL-585-84-6, "Examination, Testing, and Evaluation of Irradiated Pressure Vessel Surveillance Specimens 6om the Nine Mile Point Nuclear Power Station", July 18, 1984

[BCL84b] Computer Output Listing used to prepare Battelle Report BCL-585-84-6, "DOT Run for Nine Mile Point 300 Capsule", 5/16/84

[Be93] A. Bertuch, et al., "Modelling the Corrosion Behaviors of the Heat Transport Circuits of Light Water Nuclear Reactors", Sixth International Symposium on Environmental Degradation in Nuclear Power Systems, August, 1993.

1

[BWROG94] M. L. Herrera, et. al., "BWR Core Shroud Evaluation", prepared by GE for BWR Owners Group, GE-NE-523-148-1193, April, 1994.

[BWRVIP96] BWRVIP Document EPRI TR-105747, "Guidelines for Reinspection of Core Shrouds (BWRVIP-07)", February, 1996 /

[DFH91] (in three volumes), prepared for Novetech Corp. And Electric Power Research Institute, by Akram Zahoor, Research Project 1757-69, EPRI, Palo Alto, CA, 1991.

[EP1743] "Effect of Weld Parameters on Residual Stresses in BWR Piping Systems",'eport Prepared by Battelle Columbus Laboratories, EPRI NP-1743-, March, 1981.

[EPRI89] C.P. Ruiz, et al., "Modelling Hydrogen Water Chemistry for BWR Applications",

EPRI Report NP-6386, 1989.

[ESEERC 088] GE Nuclear Energy, "In-Reactor Stress Corrosion Monitor Prototype Development, Installation, and Operation", Final Report to ESEERCO, Research Report E85-37, February, 1988).

[GE771] Shroud Fabrication and Operational History Data, Nine Mile Point 1, GE Nuclear Energy Report, GENE 771-28-0494, June, 1994.

114

[GE86] GE Report No. MDE 236-1085, DRF No. A00-02518, Revision 1, "Fast Neutron Axial Pressure Vessel Calculations for Kine Mile Point Unit No. 1", January 21, 19&6

[GE97] GE Report GE-NE-B13-01869-043, Rev. 0,"Assessment of the Vertical Weld Cracking on the NMP1 Shroud, April, 1997

[GENE523] B.J. Branlund and B.M. Gordon, "Structural Evaluation and Justi6cation of the Kine Mile Point 1 Core Shroud for Continued Operation", GE Nuclear Energy Report No. GENE-523-A161-1094.

[Int] "Mechanical and Physical Properties of the Austenitic Chromium-Nickel Stainless Steels at Elevated Temperatures",Section I, Bulletin B, Chromium Nickel Stainless Steel Data, The International Nickel Company, Inc.

[IRW57] G.R. Irwin, "Analysis of Stresses and Strains Near the End of a Crack Traversing a Plate", Journal of Applied Mechanics, 24, pp. 361-364, 1957.

[Ma94] M.P. Manahan, Sr., D.D. Macdonald, A.J. Peterson, Jr., "Determination of the Fate of the Current in the Stress-Corrosion Cracking of Sensitized Type 304 SS in High Temperature Aqueous Systems", Corrosion Science, October, 1994.

0313] "Technical Report on Material Selection and Processing Guidelines for BWR Coolant Pressure Boundary Piping", U.S. Nuclear Regulatory Commission, NUIT-0313, Rev. 2, January, 1988.

[Pa96] Payzant, E.A., Spooner, S., Zhu, X., Hubbard, C.R., Rosinski, S.T., and Dowicki, J., "Experimental Determination of Residual Stress by Neutron DiFraction in a Boiling Water Reactor Core Shroud", ASME PVP Conference, Montreal, Canada, July 21-26, 1996, PVP-Vol. 322, NDE-Vol. 15, pp. 55-61.

[Sp81] M. Speidel, Metallurgical Transactions, Vol. 12A, p. 779 (1981).

[SIFH90] Stress Intensity Factors Handbook, Y. Murakami, Reprinted with correction 1990

[TPI85] H. Tada, P.C. Paris, G.R. Irwin, "The Stress Analysis of Cracks Handbook", Paris Productions, Inc. (Del Research Corporation), St. Louis, Missouti, 1985

[WEL3] , "WELD3 Version 2.3", Computer Sofbvare for Incremental, Thermal, Elastic-Plastic Finite Element Analysis, Computational Mechanics, Inc., 1996.

115

8.0 Nomenclature e azimuthal angle 2D Two-Dimensional 3D Three-Dimensional Crack Length BWR Boiling Water Reactor BWROG BWR Owners Group C(T) Compact Type Specimen CTE Coef5cient of Thermal Expansion DCB Double Cantilever Beam DOS Degree of Sensitization DOT Discrete Ordinates Transport ECP Electrochemical Corrosion Potential ~

EFPY Effective Full Power Years EPFM elastic-plastic &acture mechanics EOL end-of -license EPR Electrochemical Potentiokinematic Reactivation EPRI Electric Power Research Institute I

EVT enhanced visual examination FEM Finite Element Method G, Mode I Energy Release Rate Ga Mode II Energy Release Rate GE General Electric GPUNC GPU Nuclear Corporation H4/H5 Shroud Girth Weld Joining the Upper and Central Mid Cylinders HAZ Heat Affected Zone IASCC Irradiation Assisted Stress Corrosion Cracking ID Inner Diameter IGSCC Intergranular Stress Corrosion Cracking K, Mode I Stress Intensity Factor Ka Mode II Stress Intensity Factor Km+ SCC Threshold Stress Intensity Factor L Surface Length 117

PVELV] "%ELD3 VeriBcation Manual, Version 2.3", Computational Mechanics, Inc.,

1996.

116

LEFM Linear Elastic Fracture Mechanics LPRM Local Power Range Monitor MPM MPM Technologies, Inc.

NDE non-destructive examination NMP-1 Nine Mile Point Unit 1 NMPC Niag'ara Mohawk Power Corporation NRC Nuclear Regulatory Commission OD Outer Diameter PTFE PolytetraQuoro ethylene QA quality assurance R radius from the core center RPV reactor pressure vessel R/t Radius to Thickness Ratio SCC Stress Corrosion Cracking SEC Single Edge Crack SS Stainless Steel UT automated ultrasonic V9/V10 NMP-1 vertical shroud weld located between H4 and H5 ZRA Zero Resistance Ammeter 118

'