Professional Documents
Culture Documents
Materials Reliability Program: Finite-Element Model Validation For Dissimilar Metal Butt-Welds (MRP-316 Revision 1) : Volumes 1 and 2
Materials Reliability Program: Finite-Element Model Validation For Dissimilar Metal Butt-Welds (MRP-316 Revision 1) : Volumes 1 and 2
Materials Reliability Program: Finite-Element Model Validation For Dissimilar Metal Butt-Welds (MRP-316 Revision 1) : Volumes 1 and 2
3002005498
Final Report, September 2015
Acknowledgments
The following organization, under contract to the Electric Power
Research Institute (EPRI), prepared this report:
Dominion Engineering, Inc.
12100 Sunrise Valley Drive, Suite 220
Reston, VA 20191
Principal Investigator
J. Broussard
This report describes research sponsored by EPRI.
iii
Abstract
The residual stresses imparted by the welding process are a principal
factor in the process of primary water stress corrosion cracking
(PWSCC) of Alloy 82/182 nickel-alloy dissimilar metal (DM)
piping butt welds in pressurized water reactors (PWRs). Analytical
models are frequently used to simulate the welding process in order
to predict the residual stress distribution in the weld and base
material as an input to crack growth calculations. The crack growth
calculations, in turn, have demonstrated a high sensitivity to the
welding residual stress distribution inputs. As part of the industrys
proactive approach to addressing materials degradation, a multiyear
project has been conducted to validate the analytical models used to
perform welding residual stress analysis against measured residual
stresses.
This report documents the analytical modeling and measurement
work performed over the course of the project. The report was
originally published in 2011 in a single volume bearing EPRI
product ID 1022861. It was revised in 2015 to add a second volume
that reports the results of a subsequent work scope covering
additional validation topics. The original 2011 report has now been
designated as Volume 1 but has otherwise remained unchanged, with
the exception of correction of some errata in one subset of
measurements (for details, see the Introduction to Volume 1).
Keywords
Finite-element analysis (FEA) modeling
Primary water stress corrosion cracking (PWSCC)
Materials Reliability Program (MRP)
Weld residual stress
Table of Contents
Section 1: Introduction ........................................ 1-1
Background ............................................................... 1-1
Validation Program Plan ............................................. 1-2
Approach ............................................................ 1-2
Project Phases ...................................................... 1-2
Residual Stress Measurements ................................ 1-3
Report ....................................................................... 1-3
Revision 1 ................................................................. 1-4
Section 2: Residual Stress Measurement Methods 2-1
Residual Stress Measurement Plan ................................ 2-1
Residual Stress Technique Background ......................... 2-2
Neutron Diffraction ............................................... 2-2
Deep Hole Drilling ................................................ 2-4
X-Ray Diffraction ................................................... 2-4
Hole Drilling and Ring Core ................................... 2-5
Contour Method ................................................... 2-6
Section 3: Phase 1: Plate and Cylinder Specimens3-1
Specimen Design and Fabrication ................................ 3-1
Plate Specimens (Phase 1A) ................................... 3-1
Cylinder Specimens (Phase 1B) .............................. 3-5
In Process Measurements ..................................... 3-10
Residual Stress Measurements.................................... 3-18
Surface Measurement Results ............................... 3-19
Through-Wall Stress Measurements ....................... 3-24
Welding Residual Stress Modeling ............................. 3-28
Model Geometry ................................................ 3-28
Material Properties.............................................. 3-29
Heat Input Model ................................................ 3-33
Model Stress Results and Measurement Comparison 3-38
Conclusions ............................................................. 3-45
vii
ix
List of Figures
Figure 1-1 Representative PWR DMW Configuration ........... 1-4
Figure 3-1 Phase 1A Plate Dimensions (inch units) ............... 3-3
Figure 3-2 Phase 1A Plate Arrangement ............................. 3-4
Figure 3-3 Phase 1A Plate Assembly .................................. 3-4
Figure 3-4 Phase 1B Cylinder Specimens ........................... 3-6
Figure 3-5 Phase 1B Typical Cylinder and Weld Prep
Dimensions (inch units)................................................ 3-6
Figure 3-6 Phase 1B Cylinder C-5 Repair Zone Dimensions
(inch units) ................................................................. 3-9
Figure 3-7 Phase 1A Plate Specimen Thermocouple
Locations ................................................................. 3-11
Figure 3-8 Phase 1B Cylinder Specimen Thermocouple
Locations ................................................................. 3-12
Figure 3-9 Plate P-3 Pass 1 Thermocouple History.............. 3-13
Figure 3-10 Cylinder C-1 Pass 3 Thermocouple History ...... 3-13
Figure 3-11 Plate P-3 Weld Bead Profile Measurements (inch
units) ...................................................................... 3-14
Figure 3-12 Plate P-4 Weld Bead Profile Measurements (inch
units) ...................................................................... 3-15
Figure 3-13 Plate P-6 Weld Bead Profile Measurements (inch
units) ...................................................................... 3-16
Figure 3-14 Cylinder C-3 Weld Bead Profile Measurements
(inch units) Representative of All Cylinder Welds ....... 3-17
Figure 3-15 Plate P-3 Surface Stress Measurements
Comparison ............................................................ 3-22
Figure 3-16 Plate P-4 Surface Stress Measurements
Comparison ............................................................ 3-23
xi
xiv
List of Tables
Table 3-1 Phase 1A Plate Specimen Summary and Weld
Parameters ................................................................ 3-5
Table 3-2 Phase 1B Cylinder Specimen Summary................ 3-7
Table 3-3 Phase 1B Cylinder Specimen Weld Parameters .... 3-7
Table 3-4 Phase 1A Residual Stress Measurement
Summary................................................................. 3-18
Table 3-5 Phase 1B Residual Stress Measurement
Summary................................................................. 3-19
Table 4-1 Phase 2 Surge Nozzle Mockup Weld
Parameters ................................................................ 4-3
xv
Section 1: Introduction
This introductory section defines the purpose and scope of this study, and
outlines the approach used in this program.
Background
The residual stresses imparted by the welding process are a principal factor in the
process of primary water stress corrosion cracking (PWSCC) of Alloy 82/182
nickel-alloy weld materials used for piping butt welds in pressurized water
reactors (PWRs). These materials are used at numerous butt weld locations
within the primary loop of PWRs, typically in places where carbon or low alloy
steel components are joined to stainless steel ones; e.g., the butt weld joining the
low alloy steel reactor pressure vessel (RPV) nozzle to stainless steel piping.
Because they are frequently used to join dissimilar metals (i.e., carbon/low alloy
steel and stainless steel), these welds are often referred to dissimilar metal welds,
or DMWs. A DMW geometry with many elements common to these welds is
shown in Figure 1-1; however, it is noted that a wide variety of sizes and
configurations are present within PWRs.
Numerical methods by finite element analysis (FEA) have been used for a
number of years to predict the residual stress distribution in DMWs [1, 2, 3, 4]
for the purpose of performing crack growth predictions or other degradation
evaluations. Within the U.S. nuclear industry, analyses have been performed by a
variety of organizations in support of both the nuclear industry and the NRC. It
is noted that the calculation of welding residual stresses involves highly nonlinear analyses with a number of simplifying assumptions applied. Each
organization has developed individual techniques for performing these
calculations.
Welding residual stress calculations using FEA modeling were explored in detail
as part of work sponsored by EPRI investigating the effect of flaw shape
evolution on leak before break in pressurizer nozzle DMWs [5]; the NRC also
performed confirmatory analysis work on this issue [6]. The flaw growth
calculations and overall conclusions were found to demonstrate a relatively high
sensitivity to the welding residual stress distribution input to the flaw growth
calculations. Comparisons between the welding residual stresses calculated by
different participating organizations indicated some agreement in trends, but the
studies were found to lack residual stress measurements with which to compare
the calculations [7].
1-1
In order to address the uncertainties associated with modeling results and the
lack of stress measurements for prototypical pressurized water reactor (PWR)
nozzle DMW configurations, a welding residual stress model validation program
was proposed that would:
Report
The project phases are discussed in detail in separate sections of this report. Each
project phase section summarizes the results obtained from the different
measurement techniques, the modeling performed, the results of modeling and
measurements comparisons, and the lessons learned from the project phase. A
summary and conclusions section at the end describes the overall results of the
program, and examines the successes and challenges experienced by the multiyear effort.
1-3
Revision 1
This report was originally published in 2011. It was revised in 2015 to add a
second volume that reports the results of a subsequent work scope covering
additional validation topics. The original 2011 report has been renamed as
Volume 1 and has otherwise remained unchanged with one exception. A subset
of the Deep-Hole Drilling (DHD) measurements reported in Section 4 was
identified after publication as having been mislabeled. One set of hoop and axial
stresses were reversed for the measurements performed after the stainless steel
weld, leading to incorrect figures. These figures have been corrected.
Nozzle
Nozzle Butter
Dissimilar Metal Weld
SS Safe-end
SS Safe-end to
SS pipe weld
SS Piping
Figure 1-1
Representative PWR DMW Configuration
1-4
The following through-wall stress measurement techniques were selected for use
in this program:
Contour Method
Ring-Core Method
Slitting
n = 2d hkl sin B
where,
n = arbitrary positive integer
= neutron DeBroglie wavelength ()
Eq. 2-1
hkl =
d hkl d 0
d0
Eq. 2-2
where,
ij
=
Ehkl hkl
hkl
11hkl + 22hkl + 33hkl )
(
ij +
(1 + hkl )
(1 hkl )
Eq. 2-3
where,
hkl
hkl
= strain measured in three orthogonal directions
11hkl , 22
, 33
The diffraction elastic constants, Ehkl and vhkl, are specific to a particular set of
lattice planes (hkl). Because hkl does not typically vary significantly from a value
of 0.3, values were used from literature for both the stainless steel base metal and
the nickel alloy weld metal. However, variation is more significant for Ehkl and
these values need to be experimentally determined, which is included in the
measurement plan.
It is straightforward that the strained lattice spacing values, dhkl, are found by
measuring the weld specimens themselves. However, the strain-free lattice
spacings, d0, require that a sample be prepared that has been mechanically relaxed
by either slicing teeth into the material or sectioning it into small cubes.
Local metallurgical effects caused by the welding process result in large variations
in d0 in and around the weld region. Therefore, it was necessary to collect
samples for the d0 measurements at a cross section equivalent to where strain
2-3
A small diameter reference hole is gun-drilled through the bushes and the
component.
In-plane strains are generated from the change in diameter of the reference hole
due to trepanning and the normal strain component is developed from the
capacitance gauges. These strains in turn can be used to derive residual stresses.
The technique has reduced accuracy near the surfaces of the component because
the gun-drilled reference hole allows some relaxation close to the surface, which
is restrained from occurring in the bulk [11]. An incremental version of this
method has been developed in order to improve measurements taken in regions
near yielding where plasticity effects can lead to significant errors if the
incremental method is not used [12].
X-Ray Diffraction
Residual stress measurement with X-ray diffraction (XRD) is based upon the
same principles as neutron diffraction. However, X-rays penetrate into a sample
on the order of tens of microns as opposed to the centimeters possible with
neutrons. The limited penetration means that X-ray diffraction is effectively a
2-4
surface measurement technique, and since the stress normal to the free surface is
always zero, measurements only need to be performed in the directions parallel
and perpendicular to the weld seam.
Since X-ray diffraction measures a biaxial stress state, a strain-free reference
sample is not required to obtain the strain-free lattice parameter, d0. Instead a
value for d0 is obtained through the sin2 method, where a series of
measurements are made from different side tilts.
The microstructure of weld metal poses a challenge for X-ray diffraction because
texture and large-grain sizes can adversely affect the measurement results. The
presence of texture causes the d vs. sin2 plots to have an oscillatory curve
rather than be strictly linear. Large grains generate spotty rather than smooth
diffraction cones, which can cause problems in detecting the diffracted X-rays
[13].
Hole Drilling and Ring Core
Both the hole-drilling and ring-core techniques are based upon the residual stress
measurement principle of mechanical relaxation. For both measurement
techniques, a three-element strain gauge rosette is attached to the surface of the
component, material is removed in the vicinity of the gauges, the relieved strains
are measured, and the residual stress is computed from the strain data. As the
names imply, a hole is drilled through the center of a rosette in hole-drilling and
a ring is trepanned around a rosette in the ring-core method. In both instances,
the stress calculation assumes that the relaxation occurs elastically and because
strain is measured in three directions, the principal stress magnitudes and
orientation can be identified.
By drilling or trepanning in steps, the techniques provide residual stress profiles
in the near surface of the component. The ASTM standard allows a hole drilled
to a depth of 40% of the mean diameter of the strain gauge circle [14]. Although
no specific standard exists for the ring-core, measurements typically extend to a
depth of 30% of the outer diameter of the strain gauge rosette.
It has been identified that hole-drilling can only be accurately applied to
materials when the largest principal stress is lower than 50% to 70% of the
material yield strength [15]. This is because the hole concentrates the local
biaxial stress state by a factor ranging from two to four, depending on the
magnitude and orientation of the principal stresses. Therefore, if a region of high
stress is being mechanically relaxed, plastic deformation can occur around the
hole where the strain gauges are mounted. Because the hole-drilling technique,
like many other residual stress measurement techniques, is based on elastic
relaxation, induced plastic deformation will result in erroneous measured stresses.
Although welds typically do generate high regions of residual stress, there is a
significant amount of strain hardening that occurs in the material. Therefore
elevated stress levels may still be capable of being validly measured with holedrilling. The diameter of the ring-core gauges are about three times larger than
2-5
2-6
3-2
Figure 3-1
Phase 1A Plate Dimensions (inch units)
3-3
Fasteners
(Gr. 8 Alloy Steel)
Fixture Backing Plate
(Alloy 6061 T651
Aluminum)
Disc Springs
(Carbon Steel)
Figure 3-2
Phase 1A Plate Arrangement
Multi-pass Weld
Hex Nut
Flat Washer
Hex Nut
Flat Washer
Belleville
Washers
Belleville
Washers
Fixture Plate
Clamp
Fixture Plate
Clamp
Plate Weld Specimen
Backing
Plate
Backing
Plate
Flat Washer
Flat Washer
Krytox Lubricant
X-2 Heat Sink Compound
Bolt
Figure 3-3
Phase 1A Plate Assembly
3-4
Bolt
Phase 1A Plates
Table 3-1
Phase 1A Plate Specimen Summary and Weld Parameters
ID
Variable
Tested
No.
Passes
Current
Voltage
Travel
Speed
(in/min)
P-3
Plate Base
Case
11
275/2
25
11.5
P-4
Decrease
Travel Speed
275/2
25
P-5
Increase
Amperage
and Wire
Feed Rate
P-6
Decrease
Amperage
and Wire
Feed Rate
23
Remaining
Passes
6.0
76
96
11.5
3.5
76
96
375/3
25
11.8
6.0
136
136
175/1
25
10.8
6.0
39
39
Buttering
(Alloy 82)
Base Material
(304L SS)
Safe-End
(304L SS)
Girth Weld
(Alloy 82)
Girth Weld
(Alloy 82)
Base Material
(304L SS)
Girth Weld
(Alloy 82)
Girth Weld
(E308L SS)
Buttering
(Alloy 82)
Base Material
(304L SS)
Base Material
(Carbon Steel)
Figure 3-4
Phase 1B Cylinder Specimens
Figure 3-5
Phase 1B Typical Cylinder and Weld Prep Dimensions (inch units)
3-6
Table 3-2 summarizes the cylinder specimens developed as part of this validation
project. The weld parameters used for each of the weld specimens are shown in
Table 3-3.
Phase 1B Cylinders
Table 3-2
Phase 1B Cylinder Specimen Summary
ID
Specimen Description
No.
Passes
Butter
PWHT
C-1
C-3
Yes
C-4
No
C-5
Yes
SS-ER308L-SS ButterCS-A82SS
SS-A82-SS
Table 3-3
Phase 1B Cylinder Specimen Weld Parameters
Weld Passes
Current
(A)
Voltage
Range (V)
Travel Speed
Range
(in/min)
Wire Feed
Speed
(in/min)
210/160
9.0-9.8
5.7-6.2
22 2
210/160
9.0-9.8
5.7-6.1
52.5 5
250/220
9.4-9.8
5.5-5.9
100 5
4 and up
350/300
10.5-11.5
5.5-5.9
96 5
190/150
9.4-10.0
5.7-6.2
25 2
210/160
9.0-9.8
5.7-6.1
57 6
250/220
9.4-9.8
5.5-5.9
80 5
4 and up
300/270
10.5-11.5
5.5-5.9
96 5
190/150
9.4-10.0
5.7-6.2
25 2
190/150
9.0-9.8
5.7-6.1
47 5
250/220
9.4-9.8
5.5-5.9
75 5
4 and up
300/270
10.5-11.5
5.5-5.9
96 5
3-7
Cylinder C-5 was the most complex cylinder specimen, including many aspects
of a plant DM weld, although simplified into a cylindrical shape. As shown in
Figure 3-4, cylinder C-5 was fabricated with two girth welds: 1) a DM weld
between a buttered carbon steel nozzle and a short stainless steel cylinder
(representing a nozzle safe end), and 2) a stainless steel weld between the safe
end ring and a longer stainless steel cylinder (representing the plant piping). Prior
to the stainless steel weld, a repair weld was also fabricated in cylinder C-5 in
order to investigate its effect upon residual stress. The repair fabrication followed
the same progression that would occur in plant welds. A 90 long, 75% throughwall repair cavity was machined into the Alloy 82 weld from the OD using a
slitter blade with a 3 inch diameter. This left the ends of the repair groove
tapered to the OD surface with a slight curvature. A 62 arc at full-depth with a
0.12 inch land between the bottom of the groove and the ID of the weld
specimen is seen in Figure 3-6. The repair weld was axially placed at the interface
between the Alloy 82 weld and the Alloy 82 buttering, where a lack of fusion
defect could occur. The repair weld cavity was then filled using a manual GTAW
procedure with Alloy 82 since the tapering geometry of the repair groove made
automated welding difficult.
3-8
Figure 3-6
Phase 1B Cylinder C-5 Repair Zone Dimensions (inch units)
3-9
In Process Measurements
In order to characterize the welds and develop accurate finite element models, a
series of measurements were taken for every weld pass that was deposited in the
Phase 1 specimens.
A total of seven thermocouples were spot-welded onto each plate weld specimen
in order to measure temperatures caused by welding at various locations on the
specimen. Five thermocouples (labeled as 3 through 7) were attached on the
topside of the plate and were arranged in an L pattern as shown in Figure 3-7 to
record temperatures both along the groove and perpendicular to it. The
remaining two thermocouples (labeled as 1 and 2) were attached to the underside
of the plate immediately adjacent to one another for redundancy. Thermocouples
were also placed on the inner and outer surface of the cylinders during buttering
as well as girth welding. The thermocouples were placed in a fashion similar to
the plate specimens as is indicated by Figure 3-8. Complete temperature histories
for the plate welds and the cylinder girth welds are shown in Appendix A, an
example of the temperatures from plate P-3 (pass 1) and cylinder C-1 (pass 3) are
shown in Figure 3-9 and Figure 3-10, respectively.
Measurements were made of the bead profile for each pass using a laser
profilometry system developed at EWI for this purpose. The system uses a line
laser to record the depth of measurement from the detector to the sample.
Measurements were made at the mid-length of the weld for the plate specimens
and 180 from the center of the start/stop region for the cylinder specimens. This
allowed measurements to not only be in the region of fully-developed stress away
from the starts and stops but also be taken at the same location as thermocouple
measurements. Laser profilometry data for plates P-3, P-4, and P-6, as well as
for cylinder C-3 are presented in Figure 3-11 through Figure 3-14. The bead
profiles for plate P-5 are very similar to those for P-4, and are not shown. The
cylinder weld bead profiles are all nearly identical, and the laser profilometry data
for cylinder C-3 were the best of the set.
Weld shrinkage was measured after every weld pass by measuring the distance
between a series of punch marks that were pressed into each plate and each
cylinder. The measurements were taken with calipers equipped with center-line
gauges from two sets of punchmarks: a set that spanned the weld seam and
another set that measured from the edge of the plate fixture or cylinder end to
one of the punchmarks previously used. The first series measured the relative
shrinkage across the weld seam and the second allowed the relative measurements
to be made into absolute measurements. This way the measurements would be
able to indicate if the plate was translating as well as shrinking in the fixture or if
one side of the cylinder was shrinking more than another. In addition to
shrinkage, distortion measurements between the bottom of the plate specimens
and the backing plate were made using feeler gauges after the weld specimen had
cooled to room temperature.
3-10
Weld
Weld Start
Start
7
6
12 3 4 5
2"
Topside
Underside
2"
" "
3/16"
Weld Stop
Weld Stop
Figure 3-7
Phase 1A Plate Specimen Thermocouple Locations
3-11
6 78
12 3
9
4
5 10
67 8
123
9
10
5
Figure 3-8
Phase 1B Cylinder Specimen Thermocouple Locations
3-12
Temperature (F)
1400
TC 1
TC 2
1200
TC 3
1000
TC 4
TC 5
800
TC 6
TC 7
600
400
200
0
0
50000
100000
150000
Time (ms)
200000
250000
Figure 3-9
Plate P-3 Pass 1 Thermocouple History
C-1 (W-1) Pass 3 Temperatures
2400
2200
2000
1800
TC 1
1600
TC 2
Temperature (F)
TC 3
1400
TC 4
1200
TC 5
1000
TC 6
TC 7
800
TC 8
600
400
200
0
0
50000
100000
150000
Time (ms)
Figure 3-10
Cylinder C-1 Pass 3 Thermocouple History
3-13
200000
250000
0.2
Pass 1
0.1
Pass 2
Pass 3
0.0
Pass 4
Pass 5
-0.1
Pass 6
Pass 7
-0.2
Pass 8
Pass 9
-0.3
Pass 10
Pass 11
-0.4
-0.5
-0.8
-0.6
-0.4
-0.2
Figure 3-11
Plate P-3 Weld Bead Profile Measurements (inch units)
3-14
0.0
0.2
0.4
0.6
0.8
Pass 0
0.1
Pass 1
Pass 2
0.0
Pass 3
-0.1
Pass 4
-0.2
Pass 5
-0.3
Pass 6
Pass 7
-0.4
-0.5
-0.8
-0.6
-0.4
-0.2
Figure 3-12
Plate P-4 Weld Bead Profile Measurements (inch units)
3-15
0.0
0.2
0.4
0.6
0.8
Pass 0
Pass 1
Pass 2
Pass 3
Pass 4
Pass 5
Pass 6
Pass 7
Pass 10
Pass 11
Pass 12
Pass 13
Pass 14
Pass 15
Pass 16
Pass 17
Pass 18
Pass 19
Pass 20
Pass 21
Pass 22
Pass 23
0.2
0.1
0.0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.8
-0.6
-0.4
-0.2
Figure 3-13
Plate P-6 Weld Bead Profile Measurements (inch units)
3-16
0.0
0.2
0.4
0.6
0.8
0.000
-20.000
-15.000
-10.000
-5.000
0.000
5.000
10.000
15.000
20.000
Pass 0
Pass 1
X (mm)
-2.000
Pass 2
Pass 3
-4.000
Pass 4
Pass 5
-6.000
Pass 6
Pass 7
-8.000
Y (mm)
-10.000
-12.000
Figure 3-14
Cylinder C-3 Weld Bead Profile Measurements (inch units) Representative of All Cylinder Welds
3-17
Vendor
Location
Directions
Measured
Neutron
Diffraction:
Basic
Measurements
ORNL
45 Point Grid in 7
Lines on Crosssection Plane
Longitudinal
Transverse
Normal
ORNL
7 Depths along
Weld Centerline
2 Depths in Base
Metal
6 Directions
Neutron
Diffraction:
Longitudinal
Traverse
ORNL
8 Longitudinal
Positions,
3 Depth in WM & 3
Depths in BM
Longitudinal
Transverse
Normal
Contour
Hill
Eng.
1 Longitudinal
Measurement Slice
1 Transverse
Measurement Slice
Longitudinal
Transverse
TEC
7 Surface Points
Across Weld
On Topside of
Specimen
Longitudinal
Transverse
LTI
7 Surface Points
Across Weld
On Topside of
Specimen
Longitudinal
Transverse
In-plane
Shear
Ring-Core
LTI
2 Longitudinal
Positions,
Both at Weld
Centerline
on Topside of
Specimen
Longitudinal
Transverse
In-plane
Shear
Slitting
Hill
Eng.
4 Transverse
Measurement Slots
Transverse
Neutron
Diffraction:
Full Strain Tensor
X-ray
Diffraction
Surface Hole
Drilling
3-18
Measured Specimens
P-3
P-4
P-5
P-6
Table 3-5
Phase 1B Residual Stress Measurement Summary
RS
Measurement
Method
Vendor
Location
Directions
Measured
Neutron
Diffraction:
Basic
Measurements
ORNL
80 Point Grid
on Cross-section
Plane
Deep Hole
Drilling
Contour
Measured Specimens
C-1
C-3
C-4
C-5
C-5
R
Hoop
Axial
Radial
VEQTER
1 Hole through
Centerline of
Weld
Hoop
Axial
In-plane
Shear
Hill Eng.
2 Longitudinal
(Hoop) Meas.
Slices
2 Transverse
(Axial) Meas.
Slices
Hoop
Axial
with stresses along the weld axis (longitudinal) and transverse to the weld axis
presented for the two measurements performed in each plate. The measured
stress values stabilize at about 0.03 to 0.05 inch (0.5 to 1.0 mm) below the surface
of the weld. The measurement results show the near-surfae longitudinal stress to
be higher than the transverse stress, with longitudinal stresses between 200 and
300 MPa and transverse stresses between 0 and 100 MPa.
The ring core measurement results demonstrate significantly greater stability in
measurement as a function of depth than the hole drilling measurements.
Additionally, the results for the two independent measurement locations are
more consistent with each other than XRD and hole-drilling. Both the stability
and repeatability of the measurements reinforce confidence in this measurement
technique. The only drawback to this technique is the large gauge volume, which
encompassed a significant amount of the weld top surface. The ring core
technique provides an estimate of the average stress over the in-plane dimensions
of the removed core (about 8 mm or 0.3 inch). The size of the averaged region is
therefore close to the width of the entire weld cross section. This fact also makes
the ring core results more difficult to compare to the hole drilling and XRD
measurements, since these techniques measure stresses over a smaller area: the
XRD aperture was 1.5 mm wide and hole drilling measures the average stress
over the 1.5 mm hole diameter.
Slitting
The slitting method was used to measure the transverse residual stress at the
center of the weld groove for plate P-5 [18]. Measurements were taken at three
locations equally spaced along the length of the weld; the results are provided in
Appendix A. The slitting measurements also demonstrate stability in
measurement as a function of depth, like the ring core measurements do. The
first measurement location was found to have noticeably different results than the
other two measurement locations (which are nearly identical). This is likely a
result of the measurement being taken closer to a slot cut into the plate, which
had been used to make comb samples for the neutron diffraction measurements.
Measurement Comparison
The surface stress results for the different techniques used in this study may be
compared at the weld center of plate specimens P-3 and P-4, where there is
overlap among the various techniques. The results are presented in Figure 3-15
and Figure 3-16. Given the small scale of the measurement depths, the 0.005inch electropolishing performed prior to XRD is reflected in the location of the
XRD measurement results. The hole drilling was performed at a different
longitudinal position along the weld, and therefore starts off slightly higher than
the XRD.
In general, the results do not demonstrate significant consistency or agreement
among different techniques. Based on this investigation, additional research
appears to be necessary in order to develop reliable surface stress measurement
techniques for weld regions.
3-21
500
300
400
200
300
200
100
0
X-ray Diffraction
Surface Hole Drilling
Ring-Core
-100
-200
-2.00
-1.50
-1.00
-0.50
0.00
0.50
Figure 3-15
Plate P-3 Surface Stress Measurements Comparison
3-22
1.00
100
-100
-200
-300
-2.00
X-ray Diffraction
Surface Hole Drilling
Ring-Core
-1.50
-1.00
-0.50
0.00
0.50
1.00
500
200
400
100
300
200
100
0
X-ray Diffraction
Ring-Core
Surface Hole Drilling
-100
-200
-2.00
-1.50
-1.00
-0.50
0.00
0.50
Figure 3-16
Plate P-4 Surface Stress Measurements Comparison
3-23
1.00
-100
-200
X-ray Diffraction
Ring-Core
-300
Slitting
-1.50
-1.00
-0.50
0.00
0.50
1.00
Contour Transverse
- Plate P-4
ND Locations
- all plate specimens
- longitudinal, transverse, normal directions
Figure 3-17
Summary of Through Wall Stress Measurements for Plate Specimens (P-4 Shown)
3-24
Contour Hoop
- full cross section
- Cylinder C-3
Contour Axial
- Cylinder C-3
Contour Axial
- Cylinder C-3
OD
ID
ND Locations
- all cylinder specimens
- hoop, axial, radial directions
Figure 3-18
Summary of Through Wall Stress Measurements for Cylinder Specimens (C-3
Shown)
cylinder and the transverse stress plane in the plate all have a region of selfconsistent stress results. This expected result demonstrates the good repeatability
of the contour method measurements.
Neutron Diffraction
The most comprehensive set of measurements for the Phase 1 specimens was
performed using neutron diffraction. An extensive grid of measurement points
was established for all plate and cylinder specimens. The complete results of the
neutron diffraction measurements are provided in Appendix A. The neutron
diffraction data plots for the plates are arranged in pairs (e.g., lines 1 and 7, lines
2 and 6, etc.). Each line pair shown is symmetric about the weld cross section
center line (line 4). The cylinder data are shown on a line by line basis. It is noted
that the cylinder data are reported from the ID to the OD, in contrast to the
other measurement techniques.
The neutron diffraction data for the plates generally show longitudinal stresses in
excess of the transverse stresses. Not all locations investigated were found to yield
consistent stresses through the measured section; in some cases the stresses varied
significantly from point to point through the wall. When consistent results are
reported, the stress trend tends to be a general decrease in stress from the top of
the plate to the bottom of the plate.
The neutron diffraction cylinder results generally show hoop stresses in excess of
the axial stresses. Like the plate data, some locations investigated were found to
have stress data that varied significantly from point to point through the wall,
and other locations that are reported to have zero stress in any orientation
through a large portion of the wall. The results for cylinder C-1 through the weld
sections show stress trends consistent with the other measurement techniques,
with compressive stresses at the OD rising to tensile stresses at the ID region.
These results are more consistent with the DHD data trend, indicating a plateau
in stress about 25% through the wall from the OD side.
An important technical detail identified in the process of developing the residual
stress measurements using neutron diffraction is that positional accuracy is highly
important when there are high gradients of strain or large changes of chemistry.
Both the stainless steel and the A82/182 weld metal are face center cubic and
form solid solutions with varying d-spacings. The difference of the stress free dspacings for A182 and 316L stainless steel is approximately 4000 ppm. In order
to avoid potential d-spacing errors, it is necessary to measure the d-spacing values
at each neutron diffraction location in the stressed and stress-free conditions.
This necessitates cutting the specimen in order to remove the measured location
and then making a comb or other stress relieved sample with which to measure
the d-zero (stress-free condition) spacing.
When performing the neutron diffraction measurements for the Phase 1
specimens, the stress-free d-zero spacings were measured from a comb sample
taken from a portion of the weld cross section. However, the cross section used to
make the comb sample was not the precise cross section where the original d 3-27
spacings were measured; therefore, the d-zero spacing values are taken at a
different spatial location. Given the sensitivity to position error caused by solid
solution composition, this is an unquantified source of error in the Phase 1
neutron diffraction measurements.
Measurement Comparison
Comparison among the various through-wall stress measurement techniques are
performed in conjunction with comparisons to calculated stress values. These
comparisons are performed and discussed later in this section.
Welding Residual Stress Modeling
The Phase 1 portion of the program, given its research and development focus,
was used as a test bed for application of welding residual stress analysis
techniques. The extensive set of measurements taken during the weld process
permitted model development and benchmarking against the measured data.
This section summarizes the key areas where the Phase 1 specimens were used to
investigate FEA models.
Model Geometry
FEA models of the plate and cylinder specimens were developed from the overall
fabrication drawings and the laser profilometry measurements performed after
every weld pass. The weld bead surfaces were generated by picking key points
from the laser profilometry data plots, then fitting splines between the key points
to make the rounded bead profiles.
The points at the edge of the bead profile required adjustment for them to match
the initial model geometry since the weld cavity closes (and shrinks inward in the
case of the cylinders). An example of this behavior is shown in Figure 3-14,
where the edge of the weld cavity closes inwards for the later weld passes. This
approximation is endemic to currently identified engineering scale welding
simulation analysis techniques. Standard FEA analysis packages require that
every bead profile be defined at the start of the model. Therefore, the necessity of
approximating bead shapes to fit the initial weldment geometry inherently limits
the accuracy that can be gained from defining bead shapes based on laser
profilometry or etched cross sections. The degree of limitation will depend on the
change in shape of the cavity during the welding process.
Etched weld cross sections were also prepared for all of the weld specimens.
These cross sections were not considered as reliable as the laser profilometry
measurements for use in developing the bead surfaces. Comparison of laser
profilometry and etched weld bead cross sections, as shown in Figure 3-19,
demonstrates the difficulty in using an etched cross section to determine the true
weld bead profile. Due to melting caused by later weld passes, there is substantial
uncertainty in the actual weld surface for all but the last weld passes.
3-28
Figure 3-19
Comparison of Laser Profilometry and Etched Weld Cross Section (Plate P-3)
Material Properties
Material properties for the Alloy 82 weld metal and stainless steel base metal
were taken from previously developed welding residual stress analysis models.
Both thermal (specific heat and conductivity) and mechanical (elastic modulus
and coefficient of thermal expansion) properties at a wide range of temperatures
are required for the welding analysis model. The properties used in this analysis
have been developed over time from a variety of inputs, including the ASME
Boiler and Pressure Vessel Code [22] and Inconel product literature [23]. Plots
of the mechanical material properties as a function of temperature are shown in
Figure 3-21, and plots of the thermal material properties are shown in Figure
3-22.
Because the weld material and a portion of the base material are heated to
elevated temperatures, the very low yield strengths at these temperatures result in
the material experiencing plastic deformation. Therefore, material hardening
inputs are required for the weld simulation. Unless otherwise specified, the
models analyzed in this study are assumed to behave in an elastic perfectly plastic
fashion, i.e., they experience no additional material hardening once they reach a
defined elastic limit. This assumption effectively limits the expansion of the yield
surface in reversing plasticity; overexpansion of the yield surface can lead to
unrealistic stress results at room temperature. The elastic limit as a function of
temperature for the materials in the model is based on an average of the yield and
tensile strength at temperature. A plot of the elastic limit as a function of
temperature for the materials used in this analysis is presented in Figure 3-20.
3-29
The strain hardening law, as well as the inputs supplied for the hardening law,
form a significant input to a welding residual stress model. While DEI did not
specifically perform analyses using hardening rules such as kinematic hardening
or isotropic hardening, the results from other modeles using these hardening
rules are included as comparison.
Alloy 82/182
SS Base Metal
80
70
60
50
40
30
20
10
0
0
500
1000
1500
2000
Temperature, F
Figure 3-20
DEI Welding Residual Stress Model Elastic Limit (Zero Hardening After Yielding)
3-30
2500
Alloy 82/182
Stainless Steel
3.50E+07
3.00E+07
2.50E+07
2.00E+07
1.50E+07
1.00E+07
5.00E+06
0.00E+00
0
500
1000
1500
2000
2500
3000
3500
4000
Temperature, F
Alloy 82/182
Stainless Steel
1.20E-05
1.00E-05
8.00E-06
6.00E-06
4.00E-06
2.00E-06
0.00E+00
0
500
1000
1500
2000
2500
Temperature, F
Figure 3-21
DEI Welding Residual Stress Model Mechanical Properties
3-31
3000
3500
4000
Alloy 82/182
Stainless Steel
25.0
20.0
15.0
10.0
5.0
0.0
0
500
1000
1500
2000
2500
3000
3500
4000
Temperature, F
Alloy 82/182
Stainless Steel
0.250
0.200
0.150
0.100
0.050
0.000
0
500
1000
1500
2000
2500
Temperature, F
Figure 3-22
DEI Welding Residual Stress Model Thermal Properties
3-32
3000
3500
4000
Static heat source: A specified set of weld elements are loaded at a given time
with a volumetric heat flux. In many cases, the heat flux is based on the total
energy input to the weld pass, requiring knowledge (or assumption) of the
electrical power, the weld efficiency, and the time required to complete the
weld. In a static heat source, the volumetric heat flux operates on a fixed set
of elements over a fixed time period. Therefore, the heat source does not
move in the longitudinal direction of the weld as in the case of a threedimensional model. By definition, all two-dimensional heat source models
are static models.
Moving heat source: Like a static heat source, a moving heat source loads the
model using a volumetric heat flux. However, instead of operating on a static
set of elements, the heat source volume changes at every time step, based on
the travel speed of the welding torch. This is the most complex analysis
model to use. If done correctly, it can also be the most accurate, since it
captures the three-dimensional nature of the heat transfer and the weld
solidification process with time.
A static heat source model was primarily investigated in this project since most
analysis work tends to be done with two-dimensional models. A static heat
source model derived from the three-dimensional Goldak case [20] is presented
in [21] by Rudland, et al. In this model, the total weld volume and the total weld
power are related to each other. A uniform power density is applied to the entire
3-33
weld volume and the power density diminishes with time. The rate of power
decrease is equal to the normal distribution curve used along a single ellipsoid
axis in the Goldak equation. In this way, the power density reduction in time is
similar to that experienced at a single point with a Goldak ellipsoid passing
through it.
The model presented in Reference [21] was slightly simplified, and the terms
arranged to be in the following form:
2
q=
3t2
Ke a
K
=
a=
E V A
Aw
Eq. 3-1
L
S
where
q
scaling coefficient
VxA
Aw
The model is, therefore, a decreasing exponential function with time. The
characteristic length, L, and the scaling coefficient, E, are arbitrary values that
influence the size and shape of the applied power versus time curve. Integrating
this applied power with respect to time also allows calculation of the total energy
density applied to the weld cross section in the thermal model. When multiplied
by the weld cross section, the total applied energy (per unit length) is calculated
which can be compared to the welding process input energy per unit length. This
comparison will yield the true process efficiency assumed for the thermal analysis.
The definite integral for the exponential function describing q in Equation 3-1 is
a known value. When this function is integrated from time equals zero to
infinity, multiplied by the weld cross section, then divided by the weld process
energy per unit length, the resulting effective process efficiency is equal to
0.5xExL (divided by a unit depth). Therefore, as an example, using values of 1.0
3-34
for both E and L will result in a total applied energy that is consistent with a
process efficiency of 0.5.
Model Application and Comparison to Measurements
As an application example, the static heat source model described above was
applied to two-dimensional models of plate specimen P-3 and of cylinder
specimen C-1 using the weld parameters and weld bead geometry information
recorded for these specimens.
The plate P-3 weld passes were input using a total weld process efficiency of
0.72, with a characteristic length of 1.0 inch and a scaling coefficient of 1.44.
This combination was found to yield peak weld pool temperatures of
approximately 3,000F to 3,500F, as well as provide a satisfactory match to the
thermocouple measurements directly below the weld groove. The results at all
thermocouple locations for the first weld pass on plate P-3 are shown in Figure
3-23. As shown in Figure 3-23, the peak temperatures match well at a number of
thermocouple locations. The slope after the peak temperature does not match
well; this difference is believed to be related to the conductive heat transfer
between the plate and the fixture, which was not accounted for in the model.
Subsequent passes were also shown to match well on peak temperature, as shown
in Figure 3-24, which is a similar plot for the fifth weld pass on P-3.
Similar analysis methodologies were explored for the C-1 cylinder specimen. The
static heat source model described in Equation 3-1 was also employed for the
cylinder models. Instead of an axisymmetric thermal model, a two-dimensional
planar thermal model was used (structural analyses were still performed using an
axisymmetric analysis). This change was made to maintain a unit depth for the
thermal analysis, rather than the entire circumferential extent of the material. By
doing so, the time scales for the weld passes match the thermocouple
measurements better. A total weld process efficiency of 0.95 to 1.0 was input for
the cylinder model, with a characteristic length of 1.9 to 2.0 inches and a scaling
coefficient of 1.0. The results at all thermocouple locations for the first weld pass
on cylinder C-1 are shown in Figure 3-25, and the fourth weld pass is shown in
Figure 3-26. As shown in these figures, the peak temperatures match well at a
number of thermocouple locations, and the slope after the peak temperature
matches equally well.
3-35
2,000
TC 1 (F)
1,800
TC 2 (F)
1,600
TC 3 (F)
TC 4 (F)
Temperature (F)
1,400
TC 5 (F)
1,200
FEA TC 1 (F)
FEA TC 3 (F)
1,000
FEA TC 4 (F)
800
FEA TC 5 (F)
600
400
200
0
0
20
40
60
80
100
120
Time (s)
Figure 3-23
Plate P-3 Weld Pass 1 Thermocouple Measurements and FEA Results
2,000
TC 1 (F)
1,800
TC 2 (F)
1,600
TC 3 (F)
TC 4 (F)
Temperature (F)
1,400
TC 5 (F)
1,200
FEA TC 1 (F)
FEA TC 3 (F)
1,000
FEA TC 4 (F)
800
FEA TC 5 (F)
600
400
200
0
0
20
40
60
Time (s)
80
100
Figure 3-24
Plate P-3 Weld Pass 5 Thermocouple Measurements and FEA Results
3-36
120
1800
1600
1400
1200
TC 1
FEA TC 1
Temperature (F)
1000
TC 2
FEA TC 2
800
TC 3
600
TC 6
FEA TC 3
FEA TC 6
TC 7
400
FEA TC 7
TC 8
200
FEA TC 8
0
0
50
100
Time (s)
150
200
250
Figure 3-25
Cylinder C-1 Weld Pass 1 Thermocouple Measurements and FEA Results
1800
1600
1400
1200
TC 1
FEA TC
TC 2
FEA TC
TC 3
FEA TC
TC 6
FEA TC
TC 7
FEA TC
TC 8
FEA TC
Temperature (F)
1000
800
600
400
200
0
0
50
100
Time (s)
150
200
250
Figure 3-26
Cylinder C-1 Weld Pass 4 Thermocouple Measurements and FEA Results
3-37
1
2
3
6
7
8
All models simulate the thermal aspects of the welding process by applying
power generation as a function of time to the weld bead cross section. Each
model uses a different approach for the input power function and for the
total amount of input energy. Review of the models indicates that each apply,
to the first order, the same amount of total input energy, and the energy is
consistent with the input energy of the weld process.
Model A uses an elastic-perfectly plastic material hardening law that sets the
plastic point equal to the material flow stress at temperatures (based on asdeposited material). Models B and C use a standard isotropic hardening law,
with the weld material stress-strain data based on annealed material. Model
D uses a bilinear kinematic hardening law with stress-strain data based on
annealed material.
Model A was simulated using ANSYS finite element analysis software. The
remaining models were performed using the ABAQUS FEA package.
3-39
FEA Model A
FEA Model B
FEA Model C
FEA Model D
Contour, Facility A
Contour, Facility C
600.0
500.0
400.0
300.0
200.0
100.0
0.0
-100.0
-200.0
-4.00
-2.00
0.00
2.00
4.00
6.00
8.00
10.00
12.00
3-40
14.00
16.00
FEA Model A
FEA Model B
FEA Model C
FEA Model D
Contour, Facility A
200.0
100.0
0.0
-100.0
-200.0
-300.0
-400.0
-4.00
-2.00
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
FEA Model B
FEA Model C
Contour - A
DHD / iDHD
600.0
500.0
400.0
300.0
200.0
100.0
0.0
-100.0
-200.0
0.00
2.00
4.00
6.00
8.00
10.00
3-43
12.00
14.00
400.0
FEA Model A
Contour - A
FEA Model C
Neutron Diff, Facility B
FEA Model B
DHD / iDHD
300.0
200.0
100.0
0.0
-100.0
-200.0
-300.0
-400.0
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
Examining Figure 3-29 (hoop stress), it is observed that the two mechanical
relaxation methods (DHD and contour) as well as the model results generally
agree in the trend through the weld; i.e., low stress at the OD increasing to
higher stress at the ID. While the DHD and contour data are nearly identical in
the first 25% of the wall thickness (from the OD), the two measurement
techniques differ in trend past this point, with DHD predicting a flat stress
distribution and contour predicting an increasing stress. The neutron diffraction
results agree with neither the model results nor the other measurement data in
trend or magnitude for hoop stress. As in the case for the plate P-4 longitudinal
stresses, the results for the models using an isotropic hardening rule (Models B
and C) as well as the model using elastic-perfectly plastic hardening (Model A)
overpredict hoop stress relative to the measurement data. However, the amount
of hoop stress overprediction is generally lower than in the case of the plate P-4
results, particularly towards the OD surface. It is also noted that the trend
through the weld for FEA Models A and C agree more with the contour method
measurement data than the DHD data.
Examining Figure 3-30 (axial stress), it is observed that the two mechanical
relaxation methods do not agree in the trend through the weld; the contour
method results predict an increase from compressive to tensile from OD to ID,
3-44
whereas the DHD results indicate a flat trend through the weld. The neutron
diffraction results, overall, predict a flat overall trend through the weld, but with
significant variation in stress progressing from point to point. Similar to the hoop
stresses, the axial stress model results agree with the trend from the contour
method data. It is notable that FEA Models A and C have better agreement than
Models B and C, even though Models B and C both use isotropic hardening.
The large differences between Model B and Models A and C are likely again due
to modeler assumptions regarding bead size and heat input.
Conclusions
The Phase 1 specimens proved to be a valuable aspect of the overall validation
program. By starting with small samples that were easily shipped and handled, a
greater number of measurement techniques were used in this program phase. The
full set of measurements made for the specimens allow for detailed comparison
with the modeling results, in particular the thermocouple comparisons to the
thermal model results.
The analysis model results had good agreement with the measurement data.
Some modeling and measurement differences are likely due to the small
specimens used in Phase 1. Because of the small number of weld beads in these
specimens relative to a typical PWR DM weld cross section, each additional bead
has a more significant effect on the predicted stress distribution. This, in turn,
magnifies the impact of modeling assumptions that do not play as large a role in
thicker cross section geometries. The smaller cross section in the Phase 1b
cylinders also led to significant changes in the weld groove cross section with
each weld bead; this effect is also less severe in the larger cross section welds.
3-45
4-1
DMW with
fill-in weld
F316L Safe End
SS Weld
Buttering
TP 316 SS Pipe
14-in Sch 160
The mockup was fabricated in the following four steps. The carbon steel nozzle
was buttered with 137 passes of Alloy 82. After heat treating (post-weld heat
treat to 1,100-1,200 F for three hours then air cool) and machining the butter,
40 passes of Alloy 82 were deposited to make up the main DM weld. The root of
the main weld was then machined and 27 passes were deposited with Alloy 82 to
make up the 360 degree fill-in weld at the ID of the component. Each weld pass
was a full circumferential ring; i.e., each weld pass stopped at its starting point.
Additionally, the weld passes were indexed from pass to pass; i.e., the start/stop
point of a weld pass was offset rotationally from the start/stop point of the
preceding weld pass (two inches for the V-groove weld and 60 for the fill-in
weld). After completion of the V-groove and fill-in welds, residual stress
measurements were made on the DM weld. The residual stress measurements
were followed by the TP308 stainless steel safe-end to pipe weld, with a second
set of residual stress measurements made in the DM weld to investigate the effect
from the safe-end to pipe weld. For the main DM weld and fill-in weld, laser
profilometry measurements were made to map the contour of each weld pass.
The welding parameters used for the two Alloy 82 welds and the stainless steel
weld performed for the mockup are given in Table 4-1. A combined laser
4-2
profilometry plot of the 40 passes used for the V-groove portion of the DM weld
is shown in Figure 4-2, and a similar plot of the 27 passes used for the ID fill-in
weld buildup is shown in Figure 4-3; these figures are taken from NRC
presentations on the round robin [31]. As shown in these figures, the weld prep
geometry changed as the initial weld passes were being deposited. Therefore, the
representations of the bead geometry may not fully reflect the size of the actual
weld bead (e.g., bead number 1 in Figure 4-2).
The temperature during welding was recorded as a function of time using
thermocouples placed on the ID and OD of the component. Six thermocouples
were placed on the top center location during the butter welding (three
thermocouples each on the ID and OD). When the main DM and fill-in welds
were performed, the same thermocouple location could not be used. Therefore,
the locations of the ID thermocouples were shifted axially.
SS Weld
Fill-In Weld
Main DM Weld
Table 4-1
Phase 2 Surge Nozzle Mockup Weld Parameters
Weld
Passes
Current
(A)
Voltage
Range (V)
Travel
Speed
Range
(in/min)
Wire Feed
Speed
(in/min)
125/75
9.8
6.0
20
150/100
9.8
6.0
30
220
10.7
6.0
60
4-5
220
9.6
6.0
85
6-35
260/220
10.2
5.75-6.125
90
36-40
220
9.6
6.125
85
All
200
15.5
6.0
20
90
9.2
6.0
110
9.2
6.0
3-9
115
26
6.0
SMAW
10-27
147
26
6.0
SMAW
28
115
26
6.0
SMAW
4-3
Figure 4-2
Phase 2 V-Groove Weld Laser Profilometry Measurements
Figure 4-3
Phase 2 Fill-In Weld Laser Profilometry Measurements
4-4
Figure 4-4
Phase 2 Residual Stress Measurement Locations
4-5
500
400
300
Stress, MPa
200
100
iDHD #1
iDHD #2
-100
-200
-300
-400
0
10
15
20
25
30
35
40
45
Figure 4-5
Phase 2 DHD/iDHD Axial Stress Measurements Prior to SS Weld
600
500
Stress, MPa
400
300
iDHD #1
iDHD #2
200
100
0
0
10
15
20
25
30
35
40
Figure 4-6
Phase 2 DHD/iDHD Hoop Stress Measurements Prior to SS Weld
4-6
45
600
500
400
Stress, MPa
300
200
iDHD #1
100
iDHD #2
0
-100
-200
-300
0
10
20
30
Distance from ID, mm
40
50
Figure 4-7
Phase 2 DHD/iDHD Axial Stress Measurements After SS Weld
500
400
Stress, MPa
300
200
iDHD #1
100
iDHD #2
0
-100
-200
0
10
20
30
Distance from ID, mm
40
Figure 4-8
Phase 2 DHD/iDHD Hoop Stress Measurements After SS Weld
4-7
50
4-8
37
38
36
35
40
39
34
33
32
30
31
27
29
28
26
25
24
23
21
18
22
19
20
16
17
14
15
12
13
10
11
5
2
4
3
Figure 4-9
DEI Phase 2 Model DM Weld V-Groove Sequence
1
3
5
13
8
14
9
15
4
6
7
10
16
11
17
Figure 4-10
DEI Phase 2 Model DM Weld Repair and Fill-In Weld Sequence
4-9
12
18
19
Material Properties
The material properties used for the Problem 1a solution were identical to those
detailed in Section 3 for the Phase 1 models. Thermal and linear mechanical
properties were taken from a variety of available sources, and plastic deformation
was simulated using an elastic-perfectly plastic hardening definition. In Problem
1c, the material properties provided by the NRC were used, including strain
hardening data for the different materials. When stress-strain data were input as
a nonlinear property in Problem 1c, the isotropic strain hardening law was
specified. In Problem 2, the material property set used in Problem 1a was
specified.
Heat Input Model
Since the model was two dimensional, a static heat source heat input model was,
by definition, used in the analysis. The power input routine described in Section
3 was used for the welding simulation in all problem sets. The voltage and
current recorded for the different passes was input, and a characteristic length of
2.5 inches was used for the Problem 1a set. After reviewing the thermocouple
data, a shorter characteristic length was used for the Problem 1b and 1c data sets,
reducing the applied energy for the weld beads. The scaling coefficient for the
power input routine was allowed to vary from bead to bead based on obtaining a
desirable peak temperature of 3,000F to 3,500F in the weld pool region.
Model Stress Results and Measurement Comparison
The results of the analysis performed by DEI are compared to the through-wall
stress measurement results in Figure 4-11 and Figure 4-12 for before the stainless
steel weld, and in Figure 4-13 and Figure 4-14 for after the stainless steel weld.
In Figure 4-11 and Figure 4-12 (before the stainless steel weld), two sets of
analysis results are presented, those from Problem 1a and those from Problem 1b.
As noted previously, the Problem 1b model was performed after reviewing the
thermocouple measurements for the mockup, and the input weld energy in the
model was reduced from the Problem 1a model. One set of results is presented
for after the stainless steel weld; the single Problem 2 analysis result is consistent
with the thermal inputs and material properties used for Problem 1a. The
Problem 1a inputs are used for the Problem 2 model comparison in order to
provide a full set of comparisons to the measurement data using a single
consistent set of model inputs.
The analysis results presented in Figure 4-11 and Figure 4-12 demonstrate good
agreement with the measured stresses through the DM weld prior to the stainless
steel weld. The axial stress distribution in Figure 4-11 has a trend that is
consistent with the measurement results, but does not capture the ID region
trend of a relatively flat stress distribution. The location and magnitude of the
distribution minimum is matched well, and the OD region trends are also
matched. The Problem 1b reduction in thermal energy has little effect on the
axial stress results from the model except for the last 10% of the wall thickness at
the OD region, where the stress trend is corrected relative to the Problem 1a
4-10
result. The hoop stress distribution in Figure 4-12 also has good agreement with
the trends of the measurement data. The model tends to overpredict the
measured hoop stresses by a modest and consistent offset through the entire cross
section. This result is consistent with the results obtained using similar model
inputs for the Phase 1 analyses described in Section 3. Very little difference is
observed between the hoop stress results from the Problem 1a and Problem 1b
models. The most significant difference is a shift of about two to three
millimeters (about 5% of the wall thickness) towards the ID for the location of
minimum stress.
iDHD #1
iDHD #2
600
500
400
Stress, MPa
300
200
100
0
-100
-200
-300
-400
0
10
15
20
25
30
35
40
45
50
Figure 4-11
DEI Phase 2 Model and Measurement Axial Stress Comparison Prior to SS Weld
4-11
iDHD #1
iDHD #2
700
600
Stress, MPa
500
400
300
200
100
0
0
10
15
20
25
30
35
40
45
50
Figure 4-12
DEI Phase 2 Model and Measurement Hoop Stress Comparison Prior to SS Weld
iDHD #1
iDHD #2
Model (Problem 2)
600
500
400
Stress, MPa
300
200
100
0
-100
-200
-300
0
10
15
20
25
30
Distance from ID, mm
35
40
45
50
Figure 4-13
DEI Phase 2 Model and Measurement Axial Stress Comparison After SS Weld
4-12
iDHD #1
iDHD #2
Model (Problem 2)
500
400
Stress, MPa
300
200
100
0
-100
-200
0
10
15
20
25
30
Distance from ID, mm
35
40
45
50
Figure 4-14
DEI Phase 2 Model and Measurement Hoop Stress Comparison After SS Weld
The analysis results presented in Figure 4-13 and Figure 4-14 also demonstrate
reasonable agreement with the measured stresses through the DM weld
following completion of the stainless steel weld. The model results agree well
with the trends of the measured data, including the location and magnitude of
changes in stress before and after the SS weld. The stresses in the DM weld
change linearly with radial position after the SS weld (which is consistent with
the SS weld causing a far-field, elastic stress at the DM weld). The hoop stress
change is nearly linear with radial position, being about -400 MPa at the ID and
-70 MPa at the OD. The axial stress change is also linear, being about -500 MPa
at the ID and +400 MP at the OD. Similar to the comparison from before the
stainless steel weld, the model hoop stresses tend to overpredict the measurement
results, with the exception of the last 25% of the wall thickness at the OD. The
model does tend to underpredict the measured stresses in the axial direction
towards the ID region of the model. Comparing Figure 4-11 to Figure 4-13, it
appears that the difference in axial stress between the model and the
measurements is roughly the same amount in the same location. Therefore, the
model is accurately predicting the change in stress due to the stainless steel weld,
and the differences are due to the initial DM weld differences identified in
Figure 4-11.
International Round Robin Stress Results
The collected hoop and axial stress results before and after the stainless steel weld
obtained from the modelers participating in the international round robin are
presented in Figure 4-15 through Figure 4-18. These results compare data from
4-13
the modeling Problem 1a submissions (Figure 4-15 and Figure 4-16), where each
modelers best estimate for material properties and thermal inputs were used for
the case of the DM weld alone (V-groove plus fill-in weld). Also compared are
the data from the modeling Problem 2 submissions (Figure 4-17 and Figure
4-18), which include the effects of the stainless steel weld. Also shown in the
figures are the DHD through wall measurement results, the average of the
collected set of stress distributions, and lines depicting three standard deviations
on the modeling stress distributions (assuming that the model results fit a normal
distribution at each point sampled). Additional analysis of the data is provided in
Figure 4-19 through Figure 4-22, which show the results from Problem 1a and
Problem 2 with the average results data for the models using isotropic hardening
and the average results data for models using kinematic hardening represented.
These figures show little difference, on average, for axial stress results from
models using the two different hardening rules. However, a significant difference
between the two average data sets is observed for the hoop stress results. In the
case of the hoop stress results prior to the stainless steel weld, the measured data
tends to fall between the isotropic hardening models average and the kinematic
hardening models average.
600
500
400
Stress (MPa)
300
200
100
0
-100
-200
-300
-400
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 4-15
International Round Robin Axial Stress Comparison Prior to SS Weld
4-14
A - MIXED
B - KIN
C - ISO
C - KIN
D - KIN
E - ISO
E - MIXED
E - KIN
F - ISO
G - ISO
H - ISO
I - ISO
I - KIN
J - ISO
K - KIN 3D
L - KIN
iDHD #1
iDHD #2
Average
800
700
600
Stress (MPa)
500
400
300
200
100
0
-100
-200
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
A - MIXED
B - KIN
C - ISO
C - KIN
D - KIN
E - ISO
E - MIXED
E - KIN
F - ISO
G - ISO
H - ISO
I - ISO
I - KIN
J - ISO
K - KIN 3D
L - KIN
iDHD #1
iDHD #2
Average
Figure 4-16
International Round Robin Hoop Stress Comparison Prior to SS Weld
600
500
B - ISO
400
C - ISO
B - KIN
C - KIN
300
D - KIN
200
E - ISO
100
E - KIN
E - MIXED
F - ISO
G - ISO
H - ISO
-100
I - ISO
-200
I - KIN
-300
iDHD #1
-400
J - ISO
iDHD #2
0
0.1
0.2
0.3
0.4
0.5
0.6
Distance from ID (x/t)
0.7
0.8
0.9
Figure 4-17
International Round Robin Axial Stress Comparison After SS Weld
4-15
FEA Average
600
500
B - ISO
400
C - ISO
B - KIN
C - KIN
300
D - KIN
200
E - ISO
100
E - KIN
E - MIXED
F - ISO
G - ISO
H - ISO
-100
I - ISO
I - KIN
-200
J - ISO
-300
-400
iDHD #1
iDHD #2
0
0.1
0.2
0.3
0.4
0.5
0.6
Distance from ID (x/t)
0.7
0.8
0.9
FEA Average
Figure 4-18
International Round Robin Hoop Stress Comparison After SS Weld
600
500
400
Isotropic Avg
Stress (MPa)
300
200
Kinematic
Avg
100
iDHD #1
iDHD #2
-100
FEA Average
-200
-300
-400
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 4-19
International Round Robin Average Axial Stress Comparison Prior to SS Weld
4-16
800
700
600
Isotropic Avg
Stress (MPa)
500
400
Kinematic Avg
300
iDHD #1
200
iDHD #2
100
0
FEA Average
-100
-200
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Figure 4-20
International Round Robin Average Hoop Stress Comparison Prior to SS Weld
600
500
400
300
Isotropic Avg
200
Kinematic Avg
100
iDHD #1
iDHD #2
-100
FEA Average
-200
-300
-400
0.1
0.2
0.3
0.4
0.5
0.6
Distance from ID (x/t)
0.7
0.8
0.9
Figure 4-21
International Round Robin Average Axial Stress Comparison After SS Weld
4-17
600
500
400
300
Isotropic Avg
200
Kinematic Avg
100
iDHD #1
iDHD #2
-100
FEA Average
-200
-300
-400
0.1
0.2
0.3
0.4
0.5
0.6
Distance from ID (x/t)
0.7
0.8
0.9
Figure 4-22
International Round Robin Average Hoop Stress Comparison After SS Weld
4-18
Figure 5-1
Phase 3 Nozzles As Initially Identified
5-1
Nozzle Configuration
Since the nozzles were taken from a cancelled plant, drawings were not readily
available for these nozzle components. Instead, the nozzles were measured to
obtain the relevant dimensions. Additionally, in order to obtain configuration
information that is normally available for in service nozzles and their DM welds,
Nozzle B was sacrificed for mechanical testing and to etch the cross section
revealing the through-thickness dimensions.
Figure 5-2 shows the overall dimensions measured for the nozzles. The ID
surface of Nozzles C and D are shown in Figure 5-3; as shown in this figure, a
repair region of unknown depth is present at the ID of Nozzle C. Figure 5-4
shows the etched cross section of the sacrificed nozzle with the various
components identified. The same cross section was dimensioned using the scale
in the photograph, as shown in Figure 5-5. The etched cross section also
provides key information regarding the size and number of weld beads in the
DM weld cross section. However, as noted in Section 3 of this report, etched
cross sections do not always reveal accurate bead size and shape information. The
etched cross section does accurately reflect the amount of heat energy input into
the model based on the depth of melting observed. Additionally, based on the
etched cross section, it appears that the DM weld was fabricated from ID to OD
in a single progression, with no indication of an inside surface backweld or fill-in
weld. The narrow gap at the ID surface tends to indicate that the original weld
land was at the as-found ID location, rather than at a smaller ID that was
machined away.
Prior to performing residual stress measurements, some alterations were made to
Nozzles C and D. Both nozzles had material removed from the rough cut
pressurizer shell section in order to facilitate shipping and measurement. The
effect of a stainless steel weld was investigated for Nozzle D; a weld between the
safe end and a section of pipe was performed prior to measuring the DM weld
stresses. The final, as-measured configuration of the two nozzles is shown in
Figure 5-6.
5-2
Figure 5-2
Phase 3 Nozzles Overall Configuration with Dimensions
Nozzle C
Figure 5-3
Phase 3 Nozzles Polished ID Viewed from Safe End
5-3
Nozzle D
A82/182
Weld
Low Alloy Steel
SS Safe-End
SS Cladding
Figure 5-4
Phase 3 Nozzle B Etched Cross Section
5-4
0.88 in
1.30 in
0.80 in
Figure 5-5
Phase 3 Nozzle B Cross Section Scaled Dimensions
5-5
1.35 in
0.13 in
1.39 in
0.12 in
Nozzle D
Nozzle C
Figure 5-6
Phase 3 Nozzles As Measured Configuration
5-6
Materials Characterization
As noted previously, nozzle B was selected to be sacrificed for materials testing
and characterization. Materials testing samples were removed from the nozzle
base material, the weld material, and from the safe end base material as shown in
Figure 5-7. These samples were used to perform tensile tests and to perform
optical emission spectrographic chemical analyses [25]. The chemical analyses
determined that the nozzle material met the requirements of SA-508 Grade 2,
the weld material met the requirements of Inconel Alloy 182, and the safe end
material met the requirements of Type 316 stainless steel. The tensile test results
were found to be consistent with typical properties for these materials; the
measurement data are reported in Appendix A.
Metallurgical
sample
A
0
3.00
0.50
270
90
Tensile test blank
180
Section AA
Figure 5-7
Phase 3 Nozzle Materials Characterization Sample Locations
5-7
axial stress at multiple circumferential locations was found to be very similar. The
Nozzle C axial results show similar agreement if the 105 position, which is
towards the edge of the measured plane, is excluded.
1, 2 = Contour Method Hoop Stress
A82/182
Weld
SS
Safe-End
iDHD/DHD 1&2
SS Cladding
Figure 5-8
Phase 3 Nozzle C Contour and DHD Measurement Locations
SS
Safe-End
iDHD/DHD 2
A82/182
Weld
iDHD/DHD 1
SS Cladding
Figure 5-9
Phase 3 Nozzle D Contour and DHD Measurement Locations
5-9
Neutron Diffraction
Neutron diffraction measurements were performed on a circumferential segment
of Nozzle D that remained after performing contour method measurements [29].
Strain measurements made during the parting out of the section were used to
determine the full residual stress profile. As noted previously, the measurements
were performed at a series of eleven through-wall lines spaced about 0.25-inch
apart. A gauge volume cross section of 5 mm by 5 mm was used, resulting in six
measurement points through the wall thickness at each measurement line.
Complete neutron diffraction measurement results are included in Appendix A.
Welding Residual Stress Modeling
Similar to the Phase 1 residual stress modeling described in Section 3, welding
residual stress models of the Phase 3 nozzles were developed and analyzed by
EPRI and NRC contributors to the validation project. Each modeler had access
to the full set of characterization information taken from Nozzle B. The results
presented are based on each modelers best effort at an initial prediction of the
welding residual stress distribution. Additional details about the models used in
this comparison are as follows.
All models simulate the thermal aspects of the welding process by applying
power generation as a function of time to the weld bead cross section. Each
model uses a different approach for the input power function and for the
total amount of input energy. Review of the models indicates that each apply,
to the first order, the same amount of total input energy. While the weld
procedure used for these nozzles is unknown, the energy for the FEA models
is consistent with the input energy typical for a shielded metal arc weld
(SMAW) process using Alloy 182 weld metal.
Model #1 uses an elastic-perfectly plastic material hardening law that sets the
plastic point equal to the material flow stress at temperatures. Model #2 and
#3 use a standard isotropic hardening law, with the weld material stressstrain data based on annealed material. As noted above, Model #4 uses the
same strain hardening data as Model #3, but with a specified kinematic
hardening law.
5-10
Model #1 was simulated using ANSYS finite element analysis software. The
remaining models were performed using the ABAQUS FEA package.
The welding residual stress calculations do not show any difference in through
wall stress distribution (hoop or axial stress) before or after the stainless steel
weld. Therefore, only one set of model results is presented. The neutron
diffraction results were also found to differ substantially from the rest of the
measurement and model data sets, and were excluded from the comparison. The
DHD measurement data from the nominal (unrepaired) region of Nozzle C were
initially compared to the contour measurements from Nozzle C and Nozzle D
and found to disagree considerably in both trend and magnitude with all contour
measurement planes. It is noted that one of the hoop contour method
measurement planes for Nozzle C is only 45 from the nominal region DHD
measurement location. If the Nozzle C nominal region DHD data were a result
of the repair, it would be expected to be reflected in the contour method
measurements as well. However, all the contour method results across both
nozzles are consistent with each other and are generally consistent with the
Nozzle D DHD data trends. For these reasons, the DHD measurement data for
the nominal region in Nozzle C were considered a potential outlier and were
excluded from the measurement results comparison.
Through wall stress distributions at the weld centerline are presented in Figure
5-10 for hoop stresses and in Figure 5-11 for axial stresses. Also shown in the
figures are the stress distributions obtained from empirical models based on
analysis and testing of butt welds in stainless steel piping [30], labeled as ASME
>1". The results shown in the figures demonstrate a degree of consistency
between the three models. It is notable that significant differences in assumptions
regarding bead size and shape were observed among the three models, which
likely accounts for some of the results differences.
Overall, the trends of the measurement results match well to all four analysis
models for both hoop and axial stresses, and reflect previous trends identified in
Phase 1. The elastic perfectly plastic and isotropic hardening analysis models
tend to overpredict hoop stress (particularly that measured using contour
method), and the kinematic hardening model result fall between the two
measurement data method results. The measurement results are in contrast to the
results from the Phase 1 cylinders and plates, where the contour method
measurements for hoop stress were generally higher than the DHD results.
Therefore, there is some uncertainty in the degree of conservatism on hoop stress
exhibited by these models. Axial stresses are likewise reasonable versus the
measurement data. Model #1 and Model #4 tend to agree with the lower data
reported from the contour measurements, whereas Models #2 and #3 tend to
agree with the higher results reported from the deep hole drilling.
5-11
FEA Model #1
DHD Nozzle D
FEA Model #2
Contour Nozzle C
FEA Model #3
Contour Nozzle D
FEA Model #4
ASME Code > 1"
700
600
500
Stress, Mpa
400
300
200
100
0
-100
-200
-300
10
15
20
25
Distance from ID, mm
30
35
40
Figure 5-10
Phase 3 Nozzle Weld Centerline Hoop Stress Model and Measurement
Comparison
FEA Model #1
DHD Nozzle D
FEA Model #2
Contour Nozzle C
FEA Model #3
Contour Nozzle D
FEA Model #4
ASME Code > 1"
400
300
Stress, Mpa
200
100
0
-100
-200
-300
-400
10
15
20
25
Distance from ID, mm
30
Figure 5-11
Phase 3 Nozzle Weld Centerline Axial Stress Model and Measurement
Comparison
5-12
35
40
Both the analysis and measured data are consistent with a weld fabricated from
the ID to the OD and without a backweld or fill-in weld. All measurements and
models predict low stresses in the hoop and axial directions at the ID surface of
the weld. It is further noted that the empirical model results based on stainless
steel piping butt welds do not match well to either the measurement or analysis
data, which demonstrates that nozzle-to-safe end welds do tend to result in
different stresses than pipe-to-pipe butt welds.
Conclusions
The Phase 3 nozzles represent the condition of least knowledge for the analyses
performed in the various project phases. Despite lacking substantial information
about the fabrication of the nozzles, reasonable agreement was obtained between
the measurement data and a range of analysis models. The results show that, in
the absence of repairs or other ID modifications, the stress trend for a V-groove
DM weld is compressive at the ID region in both hoop and axial stresses. The
Phase 3 nozzles also demonstrate that not all safe end to pipe welds will affect
the DM weld stress distribution; the stainless steel weld in this case was too far
away (i.e., the safe end was too long) and the weld section was too thin to impact
the thicker DM weld.
5-13
relatively thin weld cross sections only permitted a fraction of the number of weld
beads typically used to complete a dissimilar metal weldment.
Producing the Phase 1 specimens under the laboratory conditions provided a
wealth of data useful for development of the FEA models, including laser
profilometry of each weld bead pass and multiple thermocouple measurements
during welding. This data gathering was repeated for the Phase 2 fabricated
mockup. Sacrificing one of the three Phase 3 nozzles to characterize the materials
and develop an etched weld cross section also provided a significant amount of
useful data about the welding processes used for plant components during initial
plant construction.
Measurement Techniques
Another success of this validation program has been the development of residual
stress measurement data for the broad variety of welded specimens discussed
above. It is expected that the data produced by this project will continue to be
used in additional model comparison and validation efforts for a number of years.
In particular, the use of multiple measurement techniques at comparable
locations allowed for an initial view into the potential spread of measured surface
stresses and through-wall residual stress distributions.
A broad variety of surface stress measurement techniques were used for the Phase
1 specimens, but the data produced by these techniques was not found to
demonstrate significant consistency or agreement among different techniques.
Based on this investigation to date, additional research appears to be necessary in
order to develop reliable surface stress measurement techniques for weld regions.
Neither mechanical relaxation techniques for diffraction based techniques were
found to yield reliable surface stress measurement data.
A similarly broad variety of through-wall stress measurement techniques were
used to characterize the Phase 1 cylinder specimens as well as the Phase 3
nozzles, with some success and some challenges. The measurement technique
relied upon for the bulk of the Phase 1 investigation was neutron diffraction,
where metallurgical composition and other complications introduced significant
challenges to this technique. Even with great care in measurement process, the
neutron diffraction technique produced a relatively limited set of reliable stress
information. Significant strides have been made during this project for using
neutron diffraction to measure residual stresses in DM weldments, and it should
not be ruled out for future measurement efforts. A greater amount of success was
found for the mechanical relaxation techniques used to measure through-wall
stress distributions, contour method and deep hole drilling. While each
measurement technique individually produced data that tended to be selfconsistent and repeatable, the agreement between the two results was less than
originally expected. It is noted that while these techniques are well established,
this project represented a small sample set of measurements. It is likely that with
additional consideration, further improvements can be made in the agreement
between the through-wall measurement techniques.
6-2
than originally anticipated, even when efforts were made to ensure the modelers
were using the same inputs. Further understanding of the cause of the spread
among analysis results from independent modelers is essential to improving
confidence in any one analysis result.
Validation
Validation of finite element analysis models used for welding residual stress
predictions was the central goal of this project. The work performed for this
research effort demonstrated that reasonable agreement in trend and magnitude
exists between measurement techniques and FEA models. However, the
challenges associated with differences between measurement results of different
techniques and the significant differences in modelers results were not resolved
within the timeline of this project. Although no specific acceptable level of
comparison difference between analysis and measurement was quantified, the
work performed for this project improved the understanding and confidence in
welding residual stress analysis models by quantifying the potential variation in
model results relative to the potential variation in measured stresses. Prior to this
project, neither of these concepts had been substantially quantified.
6-4
Section 7: References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10. D.J. Smith, P.J. Bouchard, and D. George. Measurement and prediction of
residual stresses in thick-section steel welds. Journal of Strain Analysis vol.
35. (2000): 287-305.
11. D. Stefanescu, C.E. Truman, and D.J. Smith. An integrated approach for
measuring near-surface and subsurface residual stress in engineering
components. Journal of Strain Analysis for Engineering Design vol. 39.
(2004): 483 497.
12. F. Hosseinzadeh, A.H. Mahmoudi, C.E. Truman, and D.J. Smith.
Prediction and Measurement of Through Thickness Residual Stresses in
Large Quenched Components. Proceedings of the World Congress on
Engineering 2009 Vol II WCE 2009, July 1-3, 2009, London, U.K.. WCE
(2009).
13. B.D. Cullity and S.R. Stock. Elements of X-ray Diffraction. 3rd ed. Upper
Saddle River, New Jersey: Prentice Hall, 2001.
14. Standard Test Method for Determining Residual Stresses by the HoleDrilling Strain-Gage Method. ASTM E 837-99. West Conshohocken,
PA: ASTM, 1999.
15. J.P. Nobre, et al. Local Stress-Ratio Criterion for Incremental Hole
Drilling Measurements of Shot-Peening Stresses. Journal of Engineering
Materials and Technology vol. 128. (2006): 193-201.
16. M. Prime et al. Laser Surface-Contouring and Spline Data-Smoothing for
Residual Stress Measurement. Experimental Mechanics vol. 44 (2004): 176184.
17. Determination of the Principal Subsurface Residual Stress Distributions by the
Ring Core Method and the Hole-Drilling Method in One 304L Stainless Steel
Plate, Lamda Research, Inc. Reports 1523-15267 (Plate P-3), 1523-15223
(Plate P-4), and 1523-15348 (Plate P-6).
18. Residual Stress Measurements on Welded Plate and Welded Cylinder Specimens,
Hill Engineering, LLC. Report No. HE050710, July 30, 2010.
19. Deep-Hole Drilling Residual Stress Measurement of EPRIs Phase 1B Cylinders,
VEQTER Ltd Report No. R09-015b, June 8, 2010.
20. J. Goldak, et al, A New Finite Element Model for Welding Heat Sources,
Metallurgical Transactions, Vol. 15B, June 1984: 299-305.
21. D. Rudland, et al, Comparison of Welding Residual Stress Solutions for
Control Rod Drive Mechanism Nozzles, PVP2007-26045, Proceedings of
the 2006 ASME Pressure Vessels and Piping Division Conference, 2007.
22. ASME Boiler and Pressure Vessel Code, Section II, Part D: Properties,
1992 Revision.
23. Inconel Alloy 600, Special Metals Corporation Publication No. SMC 027,
September 2002.
7-2
33. Materials Reliability Program: Welding Residual Stress Dissimilar Metal ButtWeld Finite Element Modeling Handbook (MRP-317). EPRI, Palo Alto, CA:
2011. 1022862.
7-3
A-1
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
Temperature (F)
1400
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
200000
250000
50000
Time (ms)
150000
200000
250000
Time (ms)
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
200000
250000
Time (ms)
50000
100000
150000
Time (ms)
A-2
200000
250000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
Temperature (F)
1400
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
200000
250000
50000
150000
200000
250000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
50000
100000
150000
200000
250000
0
0
50000
100000
150000
Time (ms)
Time (ms)
A-3
200000
250000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
50000
100000
150000
200000
250000
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
50000
100000
150000
50000
100000
150000
Time (ms)
2000
Temperature (F)
1400
Temperature (F)
Temperature (F)
1400
200000
250000
Time (ms)
A-4
200000
250000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
2000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
100000
200000
300000
100000
Time (ms)
300000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
1400
Temperature (F)
1400
Temperature (F)
200000
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
100000
200000
300000
Time (ms)
100000
200000
Time (ms)
A-5
300000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
100000
200000
300000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
100000
200000
100000
200000
Time (ms)
Time (ms)
Temperature (F)
1400
Temperature (F)
Temperature (F)
1400
300000
Time (ms)
A-6
300000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
Temperature (F)
1400
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
50000
100000
150000
Time (ms)
200000
250000
50000
200000
250000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
150000
Time (ms)
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
0
0
50000
100000
150000
Time (ms)
200000
250000
A-7
50000
100000
150000
Time (ms)
200000
250000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
50000
100000
150000
Time (ms)
200000
250000
Temperature (F)
1400
Temperature (F)
Temperature (F)
1400
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
50000
100000
150000
Time (ms)
200000
250000
A-8
50000
100000
150000
Time (ms)
200000
250000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
2000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
150000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
Time (ms)
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
0
0
50000
100000
150000
Time (ms)
50000
100000
Time (ms)
A-9
150000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
Temperature (F)
1400
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
150000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
100000
Time (ms)
Time (ms)
A-10
150000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
2000
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
50000
100000
150000
50000
Time (ms)
150000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
100000
Time (ms)
Time (ms)
A-11
150000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
Temperature (F)
1400
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
150000
2000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
Time (ms)
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
50000
100000
150000
Time (ms)
50000
100000
Time (ms)
A-12
150000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
2000
1000
800
600
400
400
200
200
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
50000
100000
150000
50000
Time (ms)
150000
2000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
1400
Temperature (F)
1400
Temperature (F)
100000
Time (ms)
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
50000
100000
Time (ms)
Time (ms)
A-13
150000
1800
1800
1600
1600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
1200
1000
800
600
400
200
0
0
50000
100000
50000
100000
Time (ms)
Time (ms)
Temperature (F)
1400
Temperature (F)
1400
Temperature (F)
2000
150000
Time (ms)
A-14
150000
A-15
2400
2400
2200
2200
2000
2000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
Temperature (F)
1600
1400
1200
1000
800
600
1800
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
1600
Temperature (F)
1800
1400
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
Time (ms)
200000
250000
50000
150000
200000
250000
Time (ms)
2400
2200
2200
2000
2000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
1600
1400
1200
1000
800
600
1800
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
1600
Temperature (F)
1800
Temperature (F)
100000
1400
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
Time (ms)
200000
250000
A-16
50000
100000
150000
Time (ms)
200000
250000
2400
2200
2200
2000
2000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
Temperature (F)
1600
1400
1200
1000
800
600
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
1600
1400
1200
1000
800
600
400
400
200
200
0
0
0
50000
100000
150000
Time (ms)
200000
250000
TC 1
TC 2
TC 3
TC 4
TC 5
TC 6
TC 7
TC 8
TC 9
TC 10
1600
1400
1200
1000
800
600
400
200
0
0
50000
100000
150000
200000
50000
100000
150000
Time (ms)
2400
Temperature (F)
1800
Temperature (F)
1800
250000
Time (ms)
A-17
200000
250000
1000
P-3 Longitudinal
800
P-3 Transverse
600
400
200
0
-200
-400
-600
-800
-1000
-0.6
-0.4
-0.2
0.2
0.4
0.6
1000
P-4 Longitudinal
800
P-4 Transverse
600
400
200
0
-200
-400
-600
-800
-1000
-0.6
-0.4
-0.2
0.2
A-18
0.4
0.6
1000
800
600
400
200
0
-200
-400
-600
-800
-1000
-0.6
P-5 Longitudinal
P-5 Transverse
-0.4
-0.2
0.2
0.4
0.6
1000
P-6 Longitudinal
800
P-6 Transverse
600
400
200
0
-200
-400
-600
-800
-1000
-0.6
-0.4
-0.2
0.2
A-19
0.4
0.6
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0.030
0.035
0.040
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
A-20
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0.030
0.035
0.040
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
A-21
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0.030
0.035
0.040
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
A-22
1600
1400
1200
1000
800
600
400
200
0
-200
-400
0.000
0.005
0.010
0.015
0.020
0.025
A-23
0.030
0.035
0.040
150
100
50
0
-50
-100
Longitudinal (RC #1)
Longitudinal (RC #2)
Transverse (RC #1)
Transverse (RC #2)
-150
-200
-250
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Depth (in)
150
100
50
0
-50
-100
Longitudinal (RC #1)
Longitudinal (RC #2)
Transverse (RC #1)
Transverse (RC #2)
-150
-200
-250
0.01
0.02
0.03
0.04
0.05
Depth (in)
A-24
0.06
0.07
0.08
0.09
200
100
0
Contour (mid-length)
Slitting (location 1)
Slitting (location 2)
Slitting (location 3)
-100
-200
-2
10
A-25
12
14
A-26
Cylinder C-1
A-27
Cylinder C-3
A-28
A-29
A-30
A-31
400
300
200
-8 mm from center
-4 mm from center
Center of groove
+4 mm from center
+8 mm from center
100
Plate P-5
Longitudinal stress
Contour method
-100
-2
10
A-32
12
14
300
Plate P-5
Transverse stress
mid-length
200
100
-100
-200
-2
10
A-33
12
14
600
Cylinder C-3
Hoop stress
Contour method
70-deg plane
400
200
0
Center of weld
1.775 mm towards
4.775 mm towards
1.775 mm towards
5.325 mm towards
-200
SS
SS
CS
CS
(Line 5)
(Line 4)
(Line 6)
(Line 7)
CS
-400
-2
10
12
A-34
14
304L SS
600
Cylinder C-3
Hoop stress
Contour method
270-deg plane
400
200
0
Center of weld
1.775 mm towards
4.775 mm towards
1.775 mm towards
5.325 mm towards
-200
SS
SS
CS
CS
(Line 5)
(Line 4)
(Line 6)
(Line 7)
CS
-400
-2
10
12
A-35
14
304L SS
400
200
0
Center of weld (270-deg)
Center of weld (70-deg)
-200
-400
-2
10
12
A-36
14
600
Edge of weld
400
Cylinder C-3
Axial stress
Contour method
Center of weld
200
-200
-400
-2
10
12
A-37
14
A-38
A-39
400
400
300
300
200
200
100
Longitudinal
-100
Transverse
-200
Normal
Stress (MPa)
Stress (MPa)
100
Transverse
-200
Normal
-300
-300
-400
-400
-500
Longitudinal
-100
-500
0
10
12
14
Depth (mm)
12
14
300
200
200
0
Longitudinal
-100
Transverse
-200
Normal
Stress (MPa)
100
100
Stress (MPa)
10
400
400
-400
-400
0
10
12
Transverse
-200
-300
-500
Longitudinal
-100
-300
Normal
-500
14
Depth (mm)
10
12
14
Depth (mm)
400
400
300
300
200
200
Longitudinal
-100
Transverse
-200
Normal
Stress (mm)
100
100
Stress (MPa)
8
Depth (mm)
Longitudinal
-100
Transverse
-200
-300
-300
-400
-400
Normal
-500
-500
2
10
12
14
Depth (mm)
Depth (mm)
Stress (MPa)
100
0
Longitudin
al
Transverse
-100
-200
Normal
-300
-400
-500
0
Depth (mm)
A-40
10
12
14
10
12
14
400
300
300
200
200
100
Longitudinal
-100
Transverse
-200
Normal
Stress (MPa)
Stress (MPa)
100
Longitudinal
-100
Transverse
-200
-300
-300
-400
-400
-500
-500
Normal
-600
-600
0
6
8
Depth (mm)
10
12
14
16
400
300
300
200
200
10
12
14
16
Longitudinal
-100
Transverse
-200
Normal
Stress (MPa)
100
-300
Longitudinal
-100
Transverse
-200
Normal
-300
-400
-400
-500
-500
-600
0
10
12
14
-600
16
Depth (mm)
10
12
14
16
Depth (mm)
400
300
300
200
200
Longitudinal
0
-100
Transverse
-200
Normal
-300
Stress (MPa)
100
100
Longitudinal
-100
Transverse
-200
Normal
-300
-400
-400
-500
-500
-600
-600
0
8
10
Depth (mm)
12
14
16
8
10
Depth (mm)
Stress (MPa)
100
Stress (MPa)
Depth (mm)
Longitudinal
0
-100
Transverse
-200
Normal
-300
-400
-500
-600
0
A-41
6
8
Depth (mm)
10
12
14
16
12
14
16
600
500
500
400
400
300
Stress (MPa)
200
Longitudinal
100
0
Transverse
-100
Normal
-200
Stress (MPa)
300
200
Longitudinal
100
0
Transverse
-100
Normal
-200
-300
-300
-400
-400
-500
-500
0
6
8
10
Depth (mm)
12
14
16
600
500
500
400
400
300
300
Longitudinal
100
0
Transverse
-100
Normal
-200
Stress (MPa)
600
200
200
100
Longitudinal
Transverse
-100
Normal
-200
-300
-300
-400
-400
-500
-500
0
10
12
16
14
10
15
Depth (mm)
Depth (mm)
500
500
400
400
300
300
200
100
Longitudinal
Transverse
-100
Normal
-200
Stress (MPa)
600
200
100
Longitudinal
Transverse
-100
Normal
-200
-300
-300
-400
-400
-500
-500
0
10
12
14
16
8
Depth (mm)
Depth (mm)
Stress (MPa)
15
Stress (MPa)
10
Depth (mm)
200
Longitudinal
100
0
Transverse
-100
Normal
-200
-300
-400
-500
0
A-42
8
10
Depth (mm)
12
14
16
10
12
14
16
C-1
A-43
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
Stress (Mpa)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.00
2.00
4.00
6.00
Depth From ID (mm)
A-44
8.00
10.00
12.00
Stress (Mpa)
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
Stress (Mpa)
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
Stress (Mpa)
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
Depth From ID (mm)
A-45
8.0
10.0
12.0
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
400
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
6
Depth From ID (mm)
A-46
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
Stress (MPA)
Axial
Radial
Hoop
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
6
Depth From ID (mm)
A-47
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
10
12
Stress (MPA)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0
6
Depth From ID (mm)
A-48
10
12
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
A-49
8.0
10.0
12.0
600
400
Stress (Mpa)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
8.0
10.0
12.0
600
400
Stress (Mpa)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
A-50
8.0
10.0
12.0
Stress (Mpa)
200
0
-200
Axial
-400
Radial
Hoop
-600
-800
-1000
0.0
2.0
4.0
6.0
8.0
Depth From ID (mm)
10.0
12.0
Stress (Mpa)
200
0
-200
Axial
-400
Radial
-600
Hoop
-800
-1000
0.0
2.0
4.0
6.0
A-51
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
Depth From ID (mm)
A-52
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
8.0
10.0
12.0
0
-200
-400
Axial
-600
Radial
-800
Hoop
-1000
-1200
0.0
2.0
4.0
6.0
Depth From ID (mm)
A-53
8.0
10.0
12.0
Sample Location
0.2% Offset
Yield
Strength
(ksi)
49.5
81.5
63
74
46.5
81.5
61
72
60.5
101.0
43
54
62.0
100.0
45
52
63.0
86.0
32
71
62.0
85.0
31
71
Ultimate
Strength
(ksi)
Elongation
(%)
Reduction
of Area
(%)
A-54
A-55
A-56
A-57
A-58
A-59
A-60
A-61
A-62
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
A-63
25
30
35
6.16 mm WM - Nozzle
800
600
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
-6.23 mm WM - Nozzle
800
600
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
A-64
25
30
35
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
Stress (Mpa)
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
25
30
35
400
200
0
Axial
-200
Radial
-400
Hoop
-600
-800
-1000
0
10
15
20
A-65
25
30
35
3002005498
Final Report, September 2015
THE TECHNICAL CONTENTS OF THIS PRODUCT WERE NOT PREPARED IN ACCORDANCE WITH THE EPRI
QUALITY PROGRAM MANUAL THAT FULFILLS THE REQUIREMENTS OF 10 CFR 50, APPENDIX B. THIS
PRODUCT IS NOT SUBJECT TO THE REQUIREMENTS OF 10 CFR PART 21
NOTE
For further information about EPRI, call the EPRI Customer Assistance Center at 800.313.3774 or
e-mail askepri@epri.com.
Electric Power Research Institute, EPRI, and TOGETHERSHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.
Copyright 2015 Electric Power Research Institute, Inc. All rights reserved.
Acknowledgments
The following organization, under contract to the Electric Power
Research Institute (EPRI), prepared this report:
Dominion Engineering, Inc.
12100 Sunrise Valley Drive, Suite 220
Reston, VA 20191
Principal Investigators
J. E. Broussard
The following organizations, also under contract to EPRI, performed
significant work in support of this project and contributed to the
preparation of the report:
Department of Mechanical and Aerospace Engineering, University
of California, Davis
One Shields Avenue
Davis, CA 95616
Principal Investigators
M. R. Hill, M. D. Olson, M. N. Tran
Oak Ridge National Laboratory
One Bethel Valley Rd.
Oak Ridge, Tennessee 37831-6095
Principal Investigator
Z. Feng
D. Qiao
X. Yu
J. Chen
This publication is a corporate
document that should be cited in the
literature in the following manner:
iii
Abstract
The residual stresses imparted by the welding process are a principal
factor in the process of primary water stress corrosion cracking
(PWSCC) of Alloy 82/182 nickel-alloy dissimilar metal (DM)
piping butt welds in pressurized water reactors (PWRs). Analytical
models are frequently used to simulate the welding process in order
to predict the residual stress distribution in the weld and base
material as an input to crack growth calculations. The crack growth
calculations, in turn, have demonstrated a high sensitivity to the
welding residual stress distribution inputs. As part of the industrys
proactive approach to addressing materials degradation, a multiyear
project has been conducted to validate the analytical models used to
perform welding residual stress analysis against measured residual
stresses.
This report documents the analytical modeling and measurement
work performed over the course of the project. The report was
originally published in 2011 in a single volume bearing EPRI
product ID 1022861. It was revised in 2015 to add a second volume
that reports the results of a subsequent work scope covering
additional validation topics. The original 2011 report has now been
designated as Volume 1 but has otherwise remained unchanged, with
the exception of correction of some errata in one subset of
measurements (for details, see the Introduction to Volume 1).
Keywords
Finite-element analysis (FEA) modeling
Primary water stress corrosion cracking (PWSCC)
Materials Reliability Program (MRP)
Weld residual stress
Table of Contents
Section 1: Introduction ........................................ 1-1
Background ............................................................... 1-1
Approach.................................................................. 1-2
Section 2: xLPR Welding Residual Stress Inputs ... 2-1
Weld Geometry and Fabrication Sequence ................... 2-1
Steam Generator DM Weld ................................... 2-2
Reactor Pressure Vessel DM Weld .......................... 2-3
Reactor Coolant Pump Inlet DM Weld ..................... 2-4
Stress Profile Development........................................... 2-5
WRS Calculations and Post Processing .................... 2-5
Uncertainty Cases ................................................. 2-6
Analysis Results .......................................................... 2-7
Steam Generator DM Weld ................................... 2-8
RPV Outlet Nozzle DM Weld ............................... 2-12
RCP Inlet Nozzle DM Weld ................................. 2-12
Discussion ............................................................... 2-18
Conclusions ............................................................. 2-19
Section 3: Phase 2b (2014) International Round
Robin on Welding Residual Stress
Analysis ............................................. 3-1
Problem Statement...................................................... 3-2
Mockup Design and Fabrication ............................. 3-2
Model Guidance .................................................. 3-5
Participant Questionnaire ...................................... 3-6
Residual Stress Measurement ....................................... 3-7
Modeler Results........................................................ 3-11
Discussion ............................................................... 3-19
Conclusions ............................................................. 3-19
Section 4: Development of Residual Stress
Measurement Techniques .................... 4-1
Introduction ............................................................... 4-1
Contour Method Precision and Uncertainty Studies ........ 4-2
vii
viii
ix
List of Figures
Figure 2-1 xLPR Steam Generator Nozzle DM Weld
Configuration ............................................................ 2-3
Figure 2-2 xLPR RPV Outlet Nozzle DM Weld Configuration 2-4
Figure 2-3 xLPR RCP Inlet Nozzle DM Weld Configuration ... 2-5
Figure 2-4 Steam Generator Nozzle Base Case Analysis
Results ...................................................................... 2-9
Figure 2-5 Steam Generator Nozzle 15% Repair Analysis
Results .................................................................... 2-10
Figure 2-6 Steam Generator Nozzle 50% Repair Analysis
Results .................................................................... 2-11
Figure 2-7 RPV Nozzle Base Case Analysis Results ............ 2-13
Figure 2-8 RPV Nozzle 15% Repair Analysis Results .......... 2-14
Figure 2-9 RPV Nozzle 50% Repair Analysis Results .......... 2-15
Figure 2-10 Reactor Coolant Pump Base Case Analysis
Results .................................................................... 2-16
Figure 2-11 Reactor Coolant Pump 15% Repair Analysis
Results .................................................................... 2-17
Figure 2-12 Reactor Coolant Pump 50% Repair Analysis
Results .................................................................... 2-18
Figure 3-1 International Round Robin Mockup Configuration
(Phase 2b) ................................................................. 3-3
Figure 3-2 Compiled Laser Profilometry for Mockup DM
Weld Passes .............................................................. 3-3
Figure 3-3 Compiled Laser Profilometry for Mockup Back
Weld Passes .............................................................. 3-4
Figure 3-4 Compiled Laser Profilometry for Stainless Steel
Weld Passes .............................................................. 3-4
Figure 3-5 DHD Measurement Raw Data Hoop Stress ........ 3-8
xi
Figure 3-6 DHD Measurement Raw Data Axial Stress ......... 3-8
Figure 3-7 Contour Method Longitudinal Slice Hoop Stress
Data ......................................................................... 3-9
Figure 3-8 Contour Method Axial Slice Axial Stress Data
Contour Plot .............................................................. 3-9
Figure 3-9 Hoop Stress Comparison DHD Measurements
and Mid Weld Contour Path ..................................... 3-10
Figure 3-10 Axial Stress Comparison DHD Measurements
and Average of Contour Paths ................................... 3-10
Figure 3-11 Phase 2b Round Robin Modeler Results Prior to
SS Weld Hoop Stress............................................. 3-14
Figure 3-12 Phase 2b Round Robin Modeler Results Prior to
SS Weld Axial Stress ............................................. 3-15
Figure 3-13 Phase 2b Round Robin Modeler Results
Following SS Weld Hoop Stress .............................. 3-16
Figure 3-14 Phase 2b Round Robin Modeler Results
Following SS Weld Axial Stress .............................. 3-17
Figure 3-15 Phase 2b Round Robin Modeler Results
Change in Hoop Stress due to SS Weld ..................... 3-18
Figure 3-16 Phase 2b Round Robin Modeler Results
Change in Axial Stress due to SS Weld ...................... 3-18
Figure 4-1 Diagram of Bueckners superposition as applied
for the contour method, with color-scale giving xx,
positive in red and negative in blue ............................. 4-2
Figure 4-2 Dimensioned aluminum bar with contour planes:
normalized dimension shown, where W=76.2 mm ........ 4-3
Figure 4-3 Measured stress in the stress relieved bar for (a)
plane 1, (b) plane 2A, (c) plane 2B, (d) plane 3A, and
plane 3B ................................................................... 4-5
Figure 4-4 Mean (a) and repeatability standard deviation
(b) of measured stresses for the stress relieved bar ......... 4-5
Figure 4-5 Measured stress in the quenched bar for (a)
plane 1, (b) plane 2A, (c) plane 2B, (d) plane 3A, and
plane 3B ................................................................... 4-6
Figure 4-6 Mean (a) and repeatability standard deviation
(b) of measured stresses for the quenched bar ............... 4-6
xii
xiii
xiv
xvi
xvii
List of Tables
Table 3-1 Phase 2b Round Robin Problem Statement
Mockup Weld Parameters ........................................... 3-5
Table 5-1 Summary of Dynamic Strain Hardening Test
Matrix....................................................................... 5-6
Table 5-2 Material Constants Temperature Dependent Strain
Hardening Model .................................................... 5-17
Table 6-1 Axial Position and Operating Time for Circ Flaw
to Reach 50% TW, based on Minimum Time Evaluation
and Maximum Average Axial Stress Evaluation ........... 6-17
Table 6-2 Axial Position and Operating Time for Axial Flaw
to Reach 75% TW, based on Minimum Time Evaluation
and Maximum Average Hoop Stress Evaluation .......... 6-17
Table 6-3 Comparison of Average RMS Difference Relative
to Measurement ....................................................... 6-18
Table 6-4 Mean of Measurement Data at DM Weld
Centerline, Phase 2b Mockup .................................... 6-21
xix
Section 1: Introduction
Background
The residual stresses imparted by the welding process are a principal factor in the
process of primary water stress corrosion cracking (PWSCC) of Alloy 82/182
nickel-alloy weld materials used for piping butt welds in pressurized water
reactors (PWRs). These materials are used at numerous butt weld locations
within the primary loop of PWRs, typically in places where carbon or low alloy
steel components are joined to stainless steel ones; e.g., the butt weld joining the
low alloy steel reactor pressure vessel (RPV) nozzle to stainless steel piping.
Because they are frequently used to join dissimilar metals (i.e., carbon/low alloy
steel and stainless steel), these welds are often referred to dissimilar metal welds,
or DM welds.
Numerical methods by finite element analysis (FEA) have been used for a
number of years to predict the residual stress distribution in DM welds for the
purpose of performing crack growth predictions or other degradation evaluations.
These flaw growth calculations are typically sensitive to the welding residual
stress (WRS) distribution used as an input. Since the calculation of welding
residual stresses involves highly non-linear analyses with a number of simplifying
assumptions, research efforts have focused on understanding precision and
accuracy of such calculations.
EPRI and the NRC have completed a previous joint research project into the
calculation of welding residual stresses by numerical analysis as well as the
measurement of welding residual stresses [1]. Tasks during this project
investigated welding residual stress modeling predictions for a variety of
fabricated mockups and canceled plant components and compared the calculated
stresses to those measured using multiple techniques. The results were used to
quantify the potential variations in welding residual stress resulting from
differences between: (1) modeler to modeler, (2) modeler to measurement, and
(3) measurement to measurement. The project resulted in a greater
understanding of the potential sources for differences between modelers, with the
assumptions and inputs used for material hardening playing a significant role [2].
Improving and reducing the variability among modelers and in relation to
measured values was a desired goal to improve confidence in the predictions of
crack growth that use WRS inputs.
1-1
Approach
A multifaceted effort was undertaken to extend the cooperative EPRI-NRC
research program into welding residual stress. The following investigations were
pursued.
The mechanical behavior of material that has been plastically strained then
exposed to short durations of elevated temperature was experimentally
investigated. The results were used to develop a new model that incorporates
the reduced strain hardening effects caused by the temperature exposure.
1-2
The nozzle is buttered and post weld heat treated, and the buttering is
machined to form one side of the weld prep for the DM weld.
The buttered nozzle is welded to the safe end using Alloy 52 weld material.
The safe end is a forged ring that is larger than the finished weld geometry.
The remainder of the safe end is machined to form the weld prep for the
stainless steel weld to the plant piping.
The safe end is welded to plant piping using stainless steel weld material.
Illustrations of the model geometry in the nozzle to safe end weld configuration
and in the stainless steel weld final configuration are shown in Figure 2-1.
2-2
DM weld to
safe end ring
SS Weld after
machining
Figure 2-1
xLPR Steam Generator Nozzle DM Weld Configuration
The nozzle is buttered and post weld heat treated, and the buttering is
machined to form one side of the weld prep for the DM weld.
The buttered nozzle is welded to the safe end using Alloy 82/182 weld
material. The safe end is a forged ring that is larger than the finished weld
geometry.
The ID and the OD of the weld is machined to the finished weld dimension.
The remainder of the safe end is machined to form the weld prep for the
stainless steel weld to the plant piping.
The safe end is welded to plant piping using stainless steel weld material.
2-3
Illustrations of the model geometry in the nozzle to safe end weld configuration
and in the stainless steel weld final configuration are shown in Figure 2-2.
DM weld to
safe end ring
SS Weld after
machining
Figure 2-2
xLPR RPV Outlet Nozzle DM Weld Configuration
The pipe spool end is buttered and post weld heat treated, and the buttering
is machined to form one side of the weld prep for the DM weld.
The buttered pipe spool end is welded to a 33-inch long safe end piece using
Alloy 82/182 weld material. The safe end is the same diameter as the carbon
steel pipe. A backing ring is used for the weld.
The backing ring is machined off of the DM weld. The ID of the weld is
then back clad. Specific details about the back clad are not known; it is
assumed that the back clad acts as a back weld, with 15% of the weld
thickness at the ID removed and rewelded.
The safe end is cut back significantly and machined to form the weld prep for
the stainless steel weld to a prefabricated transition piece attaching to the
RCP inlet nozzle.
The safe end is welded to the transition piece using stainless steel weld
material.
The transition piece is then welded to the RCP inlet nozzle. This step occurs
far from the DM weld, and is therefore not simulated.
Illustrations of the model geometry in the nozzle to safe end weld configuration
and in the stainless steel weld final configuration are shown in Figure 2-3.
SS weld to SS
transition piece
Figure 2-3
xLPR RCP Inlet Nozzle DM Weld Configuration
therefore affects the reversing strain experienced with progressive weld bead
deposition.
In order to investigate this effect, analyses are performed with the heat input at
all weld beads increased by 10% and decreased by 10%. It is believed that
performing the analysis with all beads shifted to one side of the variability will
bound the effect of the variability. The two sensitivity cases are performed for the
baseline RPV model geometry and for the 50% repair RPV model geometry.
Material Strength Variability
The material properties defined for use by the WRS group included a set of stress
versus strain data that are considered nominal for the materials included in the
DM weld geometries (carbon / low-alloy steel nozzle, nickel-based alloy weld
metal, and stainless steel safe end and weld metal). The material properties
include stress versus strain data input at temperatures ranging from room
temperature up to near melting.
The stress versus strain data used for the material was anticipated to have a
strong impact on the predicted residual stress. In order to explore the likely range
of potential effects, an assessment was made of the potential ranges of strengths
that may be found for the different materials. The xLPR materials database
included information for the distribution of ultimate strength of the different
materials. The 3rd and 97th percentiles of this distribution were compared to the
mean of the distribution to calculate an effective scaling parameter for upper and
lower bounds of material strength. This scaling parameter was then applied as a
uniform offset to the stress versus stain data at each temperature, yielding an
effect 3rd and 97th percentile to the data sets.
Welding residual stress analysis cases were performed for RPV nozzle geometry,
in both the base case configuration and in the 50% repair configuration. A 3rd
percentile material strength case was performed with all materials set to their
lower range stress versus strain values, and a 97th percentile material strength case
was performed with all materials set to their upper range stress versus strain
values.
Analysis Results
A summary of the results generated by the xLPR modelers are presented in this
section. The results from each modeler are the averaged values that are calculated
from the isotropic and kinematic hardening models. All stress profiles were taken
at operating temperature, but without operating pressure or other force loads.
Residual stress profiles were evaluated through the weld cross section at three
axial locations: one at the geometric center of the weld and one at each side of
the weld (adjacent to the butter and adjacent to the safe end). Based on
preliminary evaluation, the weld centerline profile was evaluated as being
sufficiently representative of the other two profile locations.
2-7
The defined input WRS values for the xLPR code were generated from these
analysis results in the form of hoop and axial stresses at 26 evenly distributed
points through the weld thickness. The baseline values for a given configuration
are the average of the results from the multiple modelers which analyzed that
configuration. Three configurations were considered for each weld geometry: a
base case configuration (no repair), a 15% repair configuration , and a 50% repair
configuration.
Steam Generator DM Weld
The steam generator DM weld analysis hoop and axial stress results for the base
case configuration are presented in Figure 2-4. The 15% repair results are
presented in Figure 2-5, and the 50% repair results are presented in Figure 2-6.
The excellent agreement between the modelers is notable in all three sets of
results, especially for the base case configuration of the steam generator weld.
One possibility for this excellent agreement is the narrow gap nature of the weld,
which focuses the response of the material into the thicker cross section rather
than across the face of the weld. The repair cases show greater variability among
the modeler results, especially in the hoop stress results. The greater variation is
likely a result of not specifying to the modelers any details of the repair to be
considered, except for the specified repair depth.
2-8
300
EMC2
NRC
DEI
Average
200
100
0
0.00
0.20
0.40
0.60
0.80
1.00
-100
-200
-300
500
EMC2
NRC
DEI
Average
400
300
200
100
0
0.00
-100
0.20
0.40
0.60
0.80
Figure 2-4
Steam Generator Nozzle Base Case Analysis Results
2-9
1.00
300
200
100
0
0.00
0.20
0.40
0.60
0.80
1.00
EMC2
-100
DEI
Average
-200
-300
500
400
300
200
EMC2
DEI
Average
100
0
0.00
-100
0.20
0.40
0.60
0.80
Figure 2-5
Steam Generator Nozzle 15% Repair Analysis Results
2-10
1.00
400
EMC2
300
NRC
Average
200
100
0
0.00
0.20
0.40
0.60
0.80
1.00
-100
-200
-300
500
400
300
200
EMC2
NRC
Average
100
0
0.00
-100
0.20
0.40
0.60
0.80
Figure 2-6
Steam Generator Nozzle 50% Repair Analysis Results
2-11
1.00
2-12
400
EMC2
NRC
DEI
Average
300
200
100
0
-100
0.00
-200
0.20
0.40
0.60
0.80
1.00
-300
500
EMC2
NRC
DEI
Average
400
300
200
100
0
0.00
-100
0.20
0.40
0.60
0.80
Figure 2-7
RPV Nozzle Base Case Analysis Results
2-13
1.00
400
EMC2
300
NRC
Average
200
100
0
-100
0.00
0.20
0.40
0.60
0.80
1.00
-200
-300
-400
500
EMC2
NRC
Average
400
300
200
100
0
0.00
-100
0.20
0.40
0.60
0.80
Figure 2-8
RPV Nozzle 15% Repair Analysis Results
2-14
1.00
400
EMC2
300
DEI
Average
200
100
0
-100
0.00
0.20
0.40
0.60
0.80
1.00
-200
-300
-400
400
EMC2
DEI
Average
300
200
100
0
0.00
0.20
0.40
0.60
0.80
-100
-200
Figure 2-9
RPV Nozzle 50% Repair Analysis Results
2-15
1.00
Figure 2-10
Reactor Coolant Pump Base Case Analysis Results
2-16
Figure 2-11
Reactor Coolant Pump 15% Repair Analysis Results
2-17
Figure 2-12
Reactor Coolant Pump 50% Repair Analysis Results
Discussion
The generally good agreement between the modelers observed in the xLPR
WRS inputs analysis work stands in contrast to the larger variability observed in
the international round robin studies on welding residual stress (see Section 4 of
[1] and Section 3 of this report) . At the same time, the xLPR analyses were
performed using a much smaller sample size, which severely limits the confidence
in the estimates of mean and standard deviation from the sets of analyses. In the
xLPR code, the modeler uncertainty will be based on both the international
round robin results and on the WRS analysis work described in this section.
2-18
Conclusions
The WRS profiles presented in this section are developed from three welding
geometries: (1) a steam generator (SG) nozzle-to-pipe weld, (2) a reactor
pressure vessel (RPV) nozzle-to-pipe weld, and (3) a reactor coolant pump
(RCP) nozzle-to-pipe weld. Weld geometry information for the three cases was
obtained from two nuclear steam supply system (NSSS) vendors. The xLPR
WRS Subgroup chose the welding geometries based upon actual piping systems
already approved for LBB in the U.S.
Development of the WRS inputs for probabilistic calculations involves
characterization of both the best-estimate WRS profile and uncertainty about the
mean. Uncertainty in WRS inputs may be estimated from a number of sources.
For example:
2-19
3-1
Problem Statement
A detailed problem statement was developed for the Phase 2b round robin study,
with the intention of specifying details that may have led to differences in
modeler results during the Phase 2a study. Information and guidance included in
the problem statement are summarized in the following sections. The full
problem statement is included in Appendix A of this report.
Mockup Design and Fabrication
The overall mockup design is similar to the previous Phase 2a mockup; the
geometry represents a pressurizer surge nozzle of a type with an Alloy 82/182
weld buildup on the ID of the DM weld region. This weld buildup was used to
seat the thermal liner for the nozzle. Because the ID weld buildup is performed
after completion of the V-groove DM weld, it presents a complex stress profile
through the weld thickness.
The mockup configuration is shown in Figure 3-1. It consists of a carbon steel
nozzle, an Alloy 182 butter layer on the nozzle, an Alloy 182 weld between the
buttered nozzle and a stainless steel safe end, and a stainless steel weld between
the safe end and a moderate length of piping. After buttering the nozzle, a
single-sided V-groove weld was completed between the nozzle and the safe end.
The ID portion of the V-groove weld was then machined out, and the back side
fill in weld was completed at the nozzle ID. The safe end was then machined to
make the stainless steel weld prep, and the stainless steel weld was completed
between the safe end and the piping segment.
Detailed dimensional information about the weld configuration was supplied in
the problem statement; it is noted that the amount of information provided in
the Phase 2b problem statement was similar to that provided for the Phase 2a
round robin. Hand-drawn bead maps of the different welds were supplied along
with spreadsheet data of laser profilometry measurements made after each weld
pass. Figure 3-2 shows the combined laser profilometry data for the 24 passes
used to make the DM weld, and Figure 3-3 shows the data for the 15 passes used
to make the back side fill-in weld. The combined laser profilometry for the 27
passes used to make the stainless steel closure weld are shown in Figure 3-4. The
welding parameters used to generate the welds were also provided in the problem
statement; they are reported in Table 3-1.
3-2
DMW with
fill-in weld
F316L Safe End
SS Weld
Buttering
TP 316 SS Pipe
14-in Sch 160
10.0
5.0
0.0
-5.0
-10.0
Safe End
Side
Nozzle
Side
-15.0
-20.0
-25.0
-30.0
-20.0
-10.0
0.0
10.0
Axial Dimension [mm]
Figure 3-2
Compiled Laser Profilometry for Mockup DM Weld Passes
3-3
20.0
30.0
6.0
4.0
2.0
0.0
-2.0
Nozzle
Side
Safe End
Side
-4.0
-6.0
-8.0
-50.0
-40.0
-30.0
-20.0
-10.0
0.0
Axial Dimension [mm]
10.0
20.0
Figure 3-3
Compiled Laser Profilometry for Mockup Back Weld Passes
15.0
10.0
5.0
0.0
Safe End
Side
-5.0
-10.0
-15.0
-20.0
SS Pipe
Side
-25.0
-30.0
-35.0
-50.0
-40.0
-10.0
-30.0
-20.0
Axial Dimension [mm]
Figure 3-4
Compiled Laser Profilometry for Stainless Steel Weld Passes
3-4
0.0
10.0
Table 3-1
Phase 2b Round Robin Problem Statement Mockup Weld Parameters
Weld
Current
Voltage
Travel Speed
DM Weld
130 A
25 V
2.3 mm/s
Back Weld
130 A
25 V
2.3 mm/s
Stainless Steel
Weld
125 A
26 V
1.7 mm/s
Model Guidance
The Phase 2a round robin problem statement did not include any restrictions or
guidance on how to generate and analyze the models; all modelers were
encouraged to use their own best practices for the work. In order to explore the
impact of modeling guidance on modeler variability, the Phase 2b round robin
problem statement included more explicit guidance on certain aspects of
modeling. As noted previously, the modeling guidance used in the Phase 2b
problem statement was based on: (1) investigations published in EPRIs WRS
modeling handbook [5] and (2) the xLPR WRS inputs group modeling
framework. The Phase 2b problem statement included guidance on the following
topics.
Material Hardening Law
There are two simple approaches commonly used to simulate the hardening
experienced by metals undergoing cyclic straining: isotropic hardening and
kinematic hardening. More nuanced approaches to material hardening are also
possible, but they are used less frequently due to the larger amount of stressstrain information required as input. Additional discussion of material hardening
and WRS analysis may be found in Section 3 of MRP-317.
In the Phase 2a round robin, modelers were permitted to use their own approach
to material hardening, with all approaches considered equally valid potential
results. The Phase 2a results were found to show that the scatter in results was
driven to some degree by choice of hardening law. The models with isotropic
hardening tended to predict larger stress magnitudes (both positive and negative)
than the models that used kinematic hardening.
The Phase 2b problem statement, in contrast, required modelers to perform the
analysis twice: once using isotropic hardening and once using kinematic
hardening. In addition, the problem statement provided input files defining the
isotropic and kinematic hardening values in both ANSYS and ABAQUS
command formats, in order to ensure uniformity of material hardening among
the different analysts.
3-5
FEA model setup, including number of weld beads and weld sequence
Thermal model details, including information about the heat applied during
the weld passes
3-6
3-7
DHD 22
500
DHD 112
DHD 202
DHD 292
400
300
200
100
0
-100
-200
-300
0
10
20
30
Distance from Weld OD (mm)
40
Figure 3-5
DHD Measurement Raw Data Hoop Stress
DHD 22
500
DHD 112
DHD 202
DHD 292
400
300
200
100
0
-100
-200
-300
0
10
20
30
Distance from Weld OD (mm)
Figure 3-6
DHD Measurement Raw Data Axial Stress
3-8
40
500
400
300
200
100
0
-100
-200
mid weld -5 mm
-300
mid weld
-400
mid weld +5 mm
mid weld +10 mm
-500
0
10
15
20
25
30
35
40
Figure 3-7
Contour Method Longitudinal Slice Hoop Stress Data
500
400
300
200
100
0
-100
25 deg
-200
35 deg
-300
45 deg
-400
55 deg
65 deg
-500
0
10
15
20
25
Figure 3-8
Contour Method Axial Slice Axial Stress Data Contour Plot
3-9
30
35
40
DHD 22
500
DHD 112
DHD 202
DHD 292
400
300
200
100
0
-100
-200
-300
0
10
20
30
Distance from Weld ID (mm)
40
Figure 3-9
Hoop Stress Comparison DHD Measurements and Mid Weld Contour Path
DHD 22
DHD 112
DHD 202
DHD 292
Average Contour
500
400
300
200
100
0
-100
-200
-300
0
10
20
30
Distance from Weld ID (mm)
Figure 3-10
Axial Stress Comparison DHD Measurements and Average of Contour Paths
3-10
40
Comparison of the measured axial residual stress values shows the contour results
are about 100 MPa lower than the DHD values from the ID surface to 35%
percent of the wall. Then, from 35% to 65% of the wall (from the ID), when the
DHD results pass through a local minimum, the contour method results are
about 100 MPa higher than the DHD values. Finally, from 65% of the wall to
the OD, the measured DHD axial stress values increase more rapidly to their
maximum value and hold, whereas the contour method values reach the same
maximum value, but over a longer distance.
Comparison of the measured hoop residual stress values shows the DHD values
are about 100 MPa higher than the contour method values at the ID. The
contour method measurements increase rapidly, and two measurement
techniques agree well from 5% to 20% through wall (from the ID). From 20%
through wall to the OD, the contour method values show a more uniform hoop
stress distribution, whereas the DHD values show greater local maxima at 30%
and 75% through wall and a greater local minimum at 50% through wall.
Towards the OD of the weld, the DHD values range from 100 MPa to 150 MPa
higher than the contour method values.
Modeler Results
As noted previously, the modelers were requested to submit results along a path
through the center of the V-groove portion of the DM weld, and which forms a
straight line from the machined ID of the fill-in weld to the machined OD of the
V-groove weld. The path data were to be taken at 26 points through the wall:
one point at the ID, one point at the OD, and 24 evenly-distributed points in
between. Axial and hoop stress data were requested both prior to and after
completion of the stainless steel closure weld from the isotropic and kinematic
hardening models.
A total of 10 WRS finite element analysis submissions were received. The final
list of participants in the Phase 2b study included experienced modelers from
throughout the world, as follows:
ANSTO / UC Davis
AREVA, Inc.
Battelle
Dominion Engineering, Inc. (DEI)
Engineering Mechanics Corporation of Columbus (EMC2)
Inspecta Technology AB
KEPCO E&C, Inc.
U.S. Nuclear Regulatory Commission
Oak Ridge National Laboratory
Structural Integrity Associates (SIA)
There was some variation in the submissions: one participant used the Phase 2a
geometry, two participants used slightly altered structural boundary conditions,
one participant included results from 2D and 3D models (the latter using
3-11
The one exception is for the hoop stress values from 40% to 60% through wall,
where the modeler data tends to agree better with the DHD measurements.
An additional evaluation of the modeler data was performed following on from
the observation that more dispersion was found in the results after the stainless
steel weld. The effect of the stainless steel weld was calculated using the
hardening average results data set, and the stress at each point through the wall
from the pre stainless steel weld was subtracted from the post stainless steel
weld value. The results of the calculation for each modelers data set are
presented in Figure 3-15 and Figure 3-16, for hoop and axial stress, respectively.
The figures show that the stainless steel weld is calculated to impart a generally
linear stress distribution to the DM weld stress profile, as noted previously.
Additionally, the figures demonstrate that there is a substantial variation in
modeler prediction of this effect, especially for two modelers. Finally, the
modeler that predicts a significantly lower impact of the stainless steel weld is the
modeler with results that are higher than other modeler results in the calculated
through-wall stress distributions, and likewise for the modeler that predicts a
larger impact of the stainless steel weld. Therefore, the larger dispersion observed
in the post stainless steel weld analysis results is significantly affected by the
calculated impact of the stainless steel weld on the stresses at the DM weld
centerline.
3-13
ISO
KIN
AVG
Figure 3-11
Phase 2b Round Robin Modeler Results Prior to SS Weld Hoop Stress
3-14
ISO
KIN
AVG
Figure 3-12
Phase 2b Round Robin Modeler Results Prior to SS Weld Axial Stress
3-15
ISO
KIN
AVG
Figure 3-13
Phase 2b Round Robin Modeler Results Following SS Weld Hoop Stress
3-16
ISO
KIN
AVG
Figure 3-14
Phase 2b Round Robin Modeler Results Following SS Weld Axial Stress
3-17
Figure 3-15
Phase 2b Round Robin Modeler Results Change in Hoop Stress due to SS Weld
Figure 3-16
Phase 2b Round Robin Modeler Results Change in Axial Stress due to SS Weld
3-18
Discussion
Given the factors that were controlled in the problem statement, there remains a
relatively limited set of analysis considerations that can lead to modeler
variability, including:
Conclusions
The second international round robin on welding residual stress analysis has
provided significant additional data establishing the variability among modelers
trying to solve the same defined problem. Quantitative analysis of the data
generated from the program is performed in Section 5 of this report, which
considers validation approaches. Qualitative observations of the data generated by
the program indicate that substantial dispersion exists among the results of the
many modelers that submitted results, despite a well-controlled problem
statement including material property data.
3-19
additional stress component directions in that same cut plane section using
mechanical release techniques. This capability currently exists using neutron
diffraction (ND) measurement; however, additional techniques are also desirable
for cases where ND is not practical. In order to provide this capability, a
substantial amount of research work has been performed to establish techniques
where a series of slitting measurements are made on a released cross section,
enabling measurement of the orthogonal stress direction (e.g., axial stress
components taken from a hoop stress contour section). This biaxial stress
mapping methodology provides an alternative to ND in evaluating two stress
components at a single cross section.
Figure 4-1
Diagram of Bueckners superposition as applied for the contour method, with colorscale giving xx, positive in red and negative in blue
/3
2W
W
Figure 4-2
Dimensioned aluminum bar with contour planes: normalized dimension shown,
where W=76.2 mm
4-3
The average, repeatability standard deviation, and repeatability limit, were found
for each measurement as a function of in-plane position. The repeatability
standard deviation was calculated as
(, ) =
(, )
=1 (, )
Eq. 4-1
Results
Measured residual stress plots for the stress relieved bar are displayed in Figure
4-3. The results show low magnitude residual stresses around 20 MPa. The
mean and repeatability standard deviation plots for the stress relieved bar are
shown in Figure 4-4. The repeatability standard deviation is small in the interior,
under 5 MPa, and somewhat larger within about 2 mm of the upper and lower
boundaries of the contour plane, reaching a maximum of 14 MPa at the left side
of the upper edge. The repeatability limit is proportional to the repeatability
standard deviation, and therefore follows the trends seen in Figure 4-4b, being
less than 14 MPa in the interior and being larger near the upper and lower edges
and having a maximum of 39 MPa. The absolute maximum deviation of stress
from the mean is 16 MPa, which occurs at the same location as the maximum
standard deviation.
Measured residual stress plots for the quenched bar are displayed in Figure 4-5.
The results show much larger residual stresses than for the stress relieved bar,
with magnitudes of about 150 MPa. The features of the stress distribution agree
with previous measurements of stress in quenched aluminum bar performed by
Robinson, et al. [10, 11]. The mean and repeatability standard deviation for the
quenched bar are shown in Figure 4-6. The repeatability standard deviation is
small in the interior, around 5 to 10 MPa, and somewhat larger within 5 mm of
the edges, having a maximum of 20 MPa at the upper left corner. The
repeatability limit follows the same trend seen in the repeatability standard
deviation (Figure 4-6), being 14 to 28 MPa in the interior with a maximum of 55
MPa. The absolute maximum deviation of stress from the mean is 31 MPa in the
quenched bar, which occurs at the edges of the 76.2 mm width.
4-4
(a)
(b)
(c)
(d)
(e)
Figure 4-3
Measured stress in the stress relieved bar for (a) plane 1, (b) plane 2A, (c) plane
2B, (d) plane 3A, and plane 3B
(a)
(b)
Figure 4-4
Mean (a) and repeatability standard deviation (b) of measured stresses for the
stress relieved bar
4-5
(a)
(b)
(c)
(d)
(e)
Figure 4-5
Measured stress in the quenched bar for (a) plane 1, (b) plane 2A, (c) plane 2B,
(d) plane 3A, and plane 3B
(a)
(b)
Figure 4-6
Mean (a) and repeatability standard deviation (b) of measured stresses for the
quenched bar
4-6
Discussion
The repeatability data for the two measurement sets suggest a similar level of
precision for measurements in the stress relieved and quenched bars: 5-10 MPa
over most of the measurement plane and 20 MPa near the plane boundaries.
This indicates that the repeatability of the contour method may remain constant
over the range of stress magnitude addressed in this study. For high magnitude
residual stresses, as in the quenched bar, the repeatability standard deviation
found here is small relative to the stress magnitude.
Previous studies of residual stress measurement repeatability give useful context
to the results presented above. Published repeatability studies were available for
x-ray diffraction [12], slitting [13], and incremental hole drilling [14, 15]. The
level of repeatability for the contour method found in the simple aluminum bar
geometry is in reasonable agreement with that found in these other residual stress
measurement repeatability studies.
Welded Plate Mockup Repeatability Study
A welded plate mockup was also used to evaluate the repeatability of the contour
method for residual stress measurements. The geometry of the fabricated plate
mockup is shown in Figure 4-7. The plate is 25.4 mm (1 in) thick and 152.4 mm
(6 in) wide, with a 6.35 mm (0.25 in) deep groove, to be filled with a multi-pass
weld. The plate length was 1.22 m (48 in), intended to provide a long,
continuous weld that could be sectioned into several smaller coupons for
subsequent evaluation. The weld bead map used for the groove weld is also
shown in Figure 4-7.
17.3
70
25.4
12.7
6.35
R3.175
152.4
8
5 6
2 3
7
4
1
Figure 4-7
Dimensioned diagram of welded plate specimen cross section
The plate specimen was restrained during fabrication by tack welding the plate to
a large, stiff I-beam. The I-beam had a height and width of 203.2 mm (8 in),
4-7
web thickness of 10.8 mm (0.44 in), and flange thickness of 11.2 mm (0.43 in).
The tack welds were 9.5 mm (3/8 in) fillet welds, with complete welds along the
152.4 mm (6 in) ends of the plate, and seven equally spaced tack welds along the
1.22 m (48 in) edges, each 50.8 mm (2 in) long with a center-to-center pitch of
143.9 mm (5.667 in). After welding was complete, the tack welds were ground
off, so that the plate was free from the I-beam, which simplified residual stress
measurements.
Measurements
Contour method measurements were made at five planes along the plate length
to characterize the residual stress field in the plate, and simultaneously to cut it
into a set of nearly identical smaller samples. The first contour measurement was
at the center of the plate (plane 1 in Figure 4-8), followed by measurements at
the center of the remaining half plates (planes 2A, 2B), and finally by two
measurements at the center of two quarter plates (planes 3A, 3B). Analysis of
results from the five measurements provides information on the consistency of
the residual stress along the plate length, and assuming the stress is uniform with
length, an estimate of contour method measurement precision.
2B
3C
1
.5
190
3B
25
95.
2A
25
95.
3A
25
95.
25
95.
4
25.
15
2 .4
.0
762
25
95.
25
95.
Figure 4-8
Measurement plane locations, contour planes in red, neutron diffraction plane (3C)
in green, dimensions in mm
Results
The results of the contour measurements can be seen in Figure 4-9. The results
show high tensile stress in the weld region, with a maximum near 450 MPa,
giving way to nearby low magnitude compressive stresses (around -100 MPa),
which is typical of weld residual stress for similar stainless steel weld
specimens [16, 17, 18]. However, the results also show an area of larger
compressive stress (about -250 MPa) toward the transverse edges, which was not
4-8
expected, and may have been introduced during plate manufacture, prior to
welding. Planes 3A and 3B are near the mid-length of the intermittent tack weld
locations, while planes 1 and 2A fell between tack welds, and the results show
large tensile stresses near the bottom of the transverse edges at planes 3A and 3B
that are absent at planes 1 and 2A. The difference in stress state appears mostly
limited to an area within one plate thickness (25.4 mm) of the transverse edges.
(a)
(b)
(c)
(d)
Figure 4-9
Contour measurement results at plane (a) 1, (b) 2B, (c) 3A, and (d) 3B
stress toward the transverse edges. The repeatability standard deviation is less
than 20 MPa over a large portion of the cross-section, but is near 30 MPa in the
weld. A repeatability study assumes that measurements are made on articles with
the same stress field. However, near the transverse edges, the stress fields are not
identical because some measurements cut through tack welds were and others did
not. Therefore, the repeatability standard deviation within one plate thickness of
the transverse edges is omitted.
(a)
(b)
Figure 4-10
(a) Mean stress and (b) repeatability standard deviation of the four contour
measurements
Discussion
The level of contour method repeatability found here is consistent those found
for the quenched aluminum bars of rectangular cross-section. The results of the
aluminum bar study had repeatability standard deviation between 5 and 10 MPa
about 5 mm away from part boundary and up to 20 MPa at the part boundary. In
the welded plate specimen, the repeatability standard deviation is somewhat
higher, but still relatively small away from the boundary, being under 20 MPa
over most of the cross-section, and higher in the weld metal, with repeatability
standard deviation of about 30 MPa. The higher repeatability standard deviation
away from the boundary is consistent with the higher elastic modulus of the
stainless steel relative to the aluminum in the prior study. But, because welding is
generally more stochastic than quenching, and observed measurement precision is
limited by true variations of the unknown residual stress, the higher repeatability
standard deviation in the weld might be expected. The overall level of precision
for the contour method is in line with the precision of the most robust residual
stress measurement techniques.
To validate the measured residual stress, a weld simulation was performed [19]
and a complementary measurement was made. Complementary neutron
4-10
(a)
(b)
Figure 4-11
Line plots comparing the mean measured residual stress (Mechanical), with
repeatability standard deviation shown as error bars, the weld simulation output
(FE), and neutron diffraction measurements (ND) along the (a) horizontal at y = 17
mm, and (b) vertical at the weld center (x = 0)
4-11
Summary
The precision of the contour method has been established using repeatability
data for low stress (aluminum bar), moderate stress (quenched aluminum bar)
and elevated stress (welded specimen) conditions. In all three specimens, the
repeatability standard deviation was found to be roughly the same, with values of
30 MPa or less depending on the location. For typical welded specimens, this
level of precision is a low percentage of the total stress values.
Biaxial Residual Stress Mapping
Many methods exist for measuring residual stress and all provide a limited
portion of the stress tensor and have different limitations. For example, large
samples or samples with difficult microstructure (e.g., texture, large grains, etc.)
are difficult to measure with diffraction techniques [22]. A standard contour
method measurement provides the single component of residual stress in the
direction normal to the cut plane over the two-dimensional cut plane section.
However, it is also desirable to have the capability to measure additional stress
component directions in that same cut plane section using mechanical release
techniques. Multiple, progressively more complex test specimens were used to
develop and validate the techniques which can be used to generate a twodimensional map of the biaxial stress state (e.g., hoop and axial stress for a DM
weld) on a given cut plane.
The basic technique to develop the biaxial map is as follows:
Additional thin sections of the material are then made parallel to the contour
cut plane. The effect of cutting the thin sections on the stress state is
measured throughout the process. The multiple thin sections are assumed to
have identical stress distributions within them.
The stresses measured in the thin slices by slitting are combined with the
measured stress effects resulting from cutting out the slices to calculate the
total residual stress in the transverse direction.
Figure 4-12
Dimensioned diagram of the measurement article with the location of measurement
planes (W=77.8 mm, H=51.2 mm, and L=304.8 mm).
The longitudinal stress map as measured by contour method and the transverse
stress map as measured by multiple slitting measurements are shown in Figure
4-13. Both the longitudinal and transverse results are at the Biaxial Plane
shown in Figure 4-12. Figure 4-13b also indicates the locations where the slitting
profiles were measured.
4-13
(a)
(b)
Figure 4-13
Biaxial mapping plane measured stress results for (a) longitudinal stress and (b)
transverse stress
The comparison of the transverse stress from the biaxial map and the
confirmation measurements is shown in Figure 4-14. Since the stress in the
confirmation measurements is nominally constant away from the edges, the data
are averaged over x = 50 mm to 100 mm to better represent the underlying stress
field. The standard deviation is reported in the error bars in Figure 4-14. The
results show that the measurements agree well at all three intersecting planes.
The comparison at x = 38.9 mm has the largest disagreement of the three, with a
maximum difference at the top edge of 25 MPa. However, at most points, the
error bars from the two different measurement techniques are close to one
another so differences in technique are not statistically different. Overall, there is
excellent agreement between the two methods, validating the biaxial mapping
approach.
4-14
(a)
(b)
(c)
Figure 4-14
Line plots comparing the biaxial measurement of the transverse stress
(Superposition) and the confirmation measurement (Contour) at x = (a) 38.9
mm, (b) 19.9 mm, and (c) 58.9 mm
(a)
(b)
Figure 4-15
(a) Measured longitudinal stress using the contour method and (b) longitudinal
stress from computational weld model
(a)
(b)
Figure 4-16
(a) Measured total transverse stress using the slitting mapping technique and (b)
transverse stress from computational weld model
Line plots are displayed in Figure 4-17 comparing the transverse stress from: (1)
the biaxial stress mapping, (2) the FEA welding residual stress model, and (3) the
neutron diffraction measurement. The figure shows reasonable agreement
between all three methods. For the line plot traversing the x-direction at y = 17
mm (Figure 4-17a), there is excellent agreement between the mechanical
mapping and the weld model, and reasonable agreement with the neutron
diffraction results, except at x = 0 where the neutron result is an outlier. For the
line plot traversing the y-direction at weld center (Figure 4-17b), there is good
4-16
agreement between the mechanical mapping and the weld model in the weld (y >
12 mm), but a significant difference at the bottom of the plate where the
measured stress transitions very quickly from compression (-200 MPa at y = 5
mm) to high tension (400 MPa at y = 2 mm). The neutron diffraction results
agree fairly well with the other techniques along the weld center, but with results
at y = 17.5 and 22.5 mm being outliers.
The high tensile transverse stress at the bottom of the plate found with
mechanical biaxial mapping was not predicted by the computational weld model.
To further investigate transverse stress near the back face, additional x-ray
diffraction measurements were made in a removed slice identical to the slices
used for slitting measurements. The measured transverse stress from the
mechanical biaxial mapping, neutron diffraction, and x-ray diffraction are
nominally consistent, as shown in Figure 4-17b. This suggests that the FE model
might not be capturing some local conditions at the plate back face that were
present during welding, which resulted in a localized model discrepancy.
(a)
(b)
Figure 4-17
Line plots of the transverse stress found with biaxial mapping (Mechanical), finite
element weld simulation (FE), and neutron diffraction (ND) along the (a) horizontal
direction at y = 17 mm and (b) with x-ray diffraction (XRD) along the vertical at the
weld center (x = 0).
4-17
Figure 4-18
Phase 3 nozzle geometry; DM weld location is the 1.35-inch wide zone in the
middle of the picture
Figure 4-19
Phase 3 nozzle cross section with measurement coordinate system
4-18
The measurement was broken into a series of section cuts and measurements to
find the original state of stress in the nozzle. An outline of the section cuts and
measurements can be seen in Figure 4-20. The primary reason for the
measurement scheme developed here is to have the nozzle in a configuration
where axial stresses are readily accessible, which in this case corresponds to thin
slices. The sectioning steps are shown in Figure 4-20 (steps) and Figure 4-21
(cutting planes), and focus on determination of residual stress at a plane of
interest, noted in Figure 4-20 and labeled plane 1 in Figure 4-21. Axial stresses
in the thin slices made using cuts S1, S2, and S3 (configuration D) were found
using slitting as described previously. In order to obtain stress over the weld
region, several slitting measurements were performed on each slice (Figure 4-22).
Slitting measurement on a given plane used a single strain gage applied to the
OD of the nozzle slice at the slitting plane.
Figure 4-20
Phase 3 nozzle measurement outline
Figure 4-21
Phase 3 nozzle cutting planes
4-19
Figure 4-22
Phase 3 nozzle slitting measurement planes, identified by slice
Axial
FE
Measured
Hoop
Figure 4-23
Phase 3 nozzle hoop and axial stresses, as measured (top) and from finite element
weld simulation (bottom)
previously described for the Phase 3 nozzle. The cut planes used for the Phase 2a
mockup are shown in Figure 4-24. The slitting measurement locations used to
measure axial stress for the Phase 2a mockup are shown in Figure 4-25.
Figure 4-24
Phase 2a mockup cutting planes
Figure 4-25
Phase 2a mockup slitting measurement planes, identified by slice
The total measured residual stresses for the Phase 2a mockup DM weld are
compared to results from a welding simulation in Figure 4-26. The measurement
data and model output for hoop stress show significant differences in magnitude,
but have similar spatial features, both showing tensile stress near the OD and ID
and low magnitude stresses at the mid-thickness. For the axial stress, the
measurement data and model output have striking agreement.
The model output and measurement data for residual stress in the stainless steel
weld can be likewise compared in Figure 4-27. The trends in measurementmodel comparison noted for the DM weld apply equally to the stainless steel
weld. The spatial distribution of the hoop stress is in reasonable agreement, but
there is a significant difference in magnitude, and there is striking agreement for
the axial stress.
4-21
Axial
FE
Measured
Hoop
Figure 4-26
Phase 2a DM weld hoop and axial stresses, as measured (top) and from finite
element weld simulation (bottom)
Axial
FE
Measured
Hoop
Figure 4-27
Phase 2a SS weld hoop and axial stresses, as measured (top) and from finite
element weld simulation (bottom)
4-22
Figure 4-28
Phase 2b DM weld hoop and axial stresses, as measured (top) and from finite
element weld simulation (bottom)
Summary
A summary of the feasibility and applicability studies for a novel measurement
technique that can provide a biaxial map of residual stresses on a single cut plane
has been provided. Based on comparison to other measurement methods as well
as comparison to finite element welding simulation, this technique has
demonstrated that it produces reasonable and comparable results.
4-23
Conclusions
The precision of the contour method is better than 10 MPa, as described by the
repeatability standard deviation of measurements in two aluminum bars, one that
was stress relieved with low residual stresses (20 MPa) and one that was
quenched with relatively high residual stresses (150 MPa). Contour
measurements were somewhat less precise in a welded plate, being better than
30 MPa at most spatial locations. These levels of precision are similar to those
stated for other residual stress measurement techniques.
A new biaxial mapping residual stress measurement method was applied to three
different assets with DM welds typical of those in pressurized water reactors. The
measured results exhibit similar trends, with hoop stress being tensile toward the
OD and compressive toward the ID and axial stress being compressive at the ID
and tensile at the OD. The degree of agreement between model outputs and
measured biaxial residual stress maps is generally excellent for axial stress. Model
outputs for hoop stress show higher residual stress than found by measurements,
but the spatial distribution features of the model output and measurement data
are similar.
4-24
Initial investigation of Case 1 specimens and Case 2 specimens indicated that the
temperature at which the strain is induced has a limited effect on the subsequent
material recovery. Therefore, the majority of the testing was focused on Case 1
conditions.
5-3
Figure 5-1
Experimental Study Case Summary
Mechanical Testing
In this work, the following materials relevant to dissimilar metal welds in nuclear
reactor piping system were studied: Alloy 82, Alloy 52, Alloy 600, and 304L
stainless steel. Cylindrical test specimens were machined from wrought rods and
annealed at 1100 C in vacuum to remove any cold work in the material and to
establish uniform grain size. The sample dimensions are shown in Figure 5-2.
The gauge length of the specimen is 70 mm, and the diameter is 7mm.
Figure 5-2
Mechanical Test Specimen Geometry
The mechanical testing process for the Case 1 experimental study followed the
following general steps:
1. Initial pre-strain
The annealed specimens were pre-strained under uniaxial tensile at room
temperature to plastic strain values of 0.10 and 0.20 and then unloaded.
2. Specimen heating
The strained specimens were heat-treated in the GleebleTM thermal-mechanical
simulation system (see Figure 5-3). The specimens were heated rapidly (constant
rate of 200 C/s) to minimize any thermal effects on the material during the
heating process. After reaching the target temperature, the specimens were held
at that condition for a short time period ranging from 1 to 600 seconds. During
5-4
heating, the axial mechanical loads on the samples were controlled to be close to
zero, allowing free thermal expansion of the specimens. After the prescribed
short isothermal high temperature exposure, the samples were cooled down to
room temperature using compressed helium at a rate of about 100C/s. Helium
was chosen as the cooling medium instead of water because if the cooling rate is
too high, the temperature gradient from the center to the surface of the sample is
high enough to induce additional plastic deformation during cooling. An elevated
cooling rate is necessary to suppress the annealing of material during cooling
[25], [26].
3. Post-heating strain loading
After thermal treatment, the specimens were subject to a second monotonic
strain cycle to establish the new strain hardening behavior following heating.
Figure 5-3
Experimental Setup for High Temperature Mechanical Testing
While the specimen is heated, it has a non-uniform temperature along its axis.
The highest temperature is at the axial center of the specimen; the temperature at
this location is also used as the controlled target temperature. The specimen is
progressively cooler from the center out towards the ends, since the ends are in
contact with the water-cooled copper grips. This non-uniform temperature
distribution results in a variation of heat-treatment from center to ends and,
therefore, a gradient in mechanical properties, especially yield strength.
In order to correct for this non-uniform mechanical property distribution
following heating, a digital image correlation (DIC) technique was used to map
the strain distribution on the sample during the subsequent tensile test. During
the tensile test, the sample was painted with thin white background and random
black speckles. The DIC system took the pictures of sample during loading at
frame rate of 10Hz. The strain field along the length of the specimen can then be
calculated from a series of sequential images. An example of the DIC process is
shown in Figure 5-4. In this section, only the strain history at the center of
sample is extracted to plot stress strain curve with load from load frame.
5-5
Figure 5-4
Local Mechanical Property Measurement Using DIC
Pre-strain
level
Temperature
Range (C)
Hold Time
(sec)
Type 304L
10%
680-1000
1-300
Type 304L
20%
600-1150
1-600
Alloy 600
20%
680-1000
1-150
Alloy 52
20%
680-1100
1-150
Alloy 82
20%
680-1000
1-150
Alloy 82
10%
680-1000
1-150
5-6
Figure 5-5
Hardening Reduction due to High Temperature Exposure
For the fully annealed Reference case material, the initial yield stress Y0 (or the
initial yield surface) is 220 MPa. If the material is pre-strained to 17% plastic
strain, the flow stress is at Yp, which is 580 MPa. In plasticity theory, this means
the active yield surface1 is expanded to 580PMa. This significant increase in the
flow stress means that stainless steel experiences a significant strain hardening. If
the material unloads to zero stress and tensile load is applied again, plastic
deformation will occur again when the stress in the material reaches Yp.
However, as demonstrated by the red Case 1 curve, if the material is subject to
a high temperature isothermal excursion after plastic deformation even for 6
seconds at 900C the new yield point upon reloading at room temperature is
reduced to a value lower than Yp, referred to as Yp.
This reduction in strain hardening due to a short-time high temperature
excursion is quantified using a strain hardening reduction factor, R, which is
defined as shown below:
Eq. 5-1
5-7
Among all the materials studied so far, Alloy 600 has the highest R at a given
temperature and time, followed by Type 304L stainless steel, and Alloy 82.
Alloy 52 has lowest R. This means the Alloy 600 has the fastest recovery and
recrystallization kinetics (reduction in strain hardening), and Alloy 52 is the
most sluggish alloy. At 900 C, Alloy 600 reduces its strain hardening by
about 40% after 1 second, and 55% after 6 seconds. In comparison, Alloy 52
reduces only 10% and 22% respectively after 1 and 6 seconds. In summary, R
values are: Alloy 600 > SS304L > Alloy 82 > Alloy 52
5-8
Figure 5-6
Strain hardening reduction factor for Type 304L: Case 1, 20% nominal plastic
strain prior to isothermal heat treatment
Figure 5-7
Strain hardening reduction factor for Type 304L: Case 1, 10% nominal plastic
strain prior to isothermal heat treatment
5-9
Figure 5-8
Strain hardening reduction factor for Alloy 600: Case 1, 20% nominal plastic
strain prior to isothermal heat treatment
Figure 5-9
Strain hardening reduction factor for Alloy 52: Case 1, 20% nominal plastic strain
prior to isothermal heat treatment
5-10
Figure 5-10
Strain hardening reduction factor for Alloy 82: Case 1, 20% nominal plastic strain
prior to isothermal heat treatment
Figure 5-11
Strain hardening reduction factor for Alloy 82: Case 1, 10% nominal plastic strain
prior to isothermal heat treatment
5-11
Figure 5-12 presents an additional view of the effects of temperature and time on
R for the case of Type 304L stainless steel with an initial room temperature
plastic strain of 20% (nominal). As shown in this figure, the reduction in strain
hardening is generally negligible (less than 5 to 10%) for temperatures below 600
C, for the short time durations generally experienced during welding. At 825 C,
R increases from about 20% to about 40% as the isothermal time increases to 300
seconds. At 1000 C, R rapidly increases from about 60% to over 90% within a
few seconds. At 1100 C, over 90% of the original strain hardening was wiped
out within 1 second. This means that the flow stress nearly restores or reduces to
the initial yield stress of the annealed material of 220MPa.
Figure 5-12
Strain hardening reduction factor for Type 304L stainless steel as function of time
and temperature: Case 1, 20% nominal plastic strain prior to isothermal heat
treatment
The effects of material types on the dynamic strain hardening are clearly
distinguished with the use of LMP in Figure 5-13. For example, Alloy 600 has a
higher R value among the 4 materials tested, for a given LMP value.
Additionally, for a given material, the correlation between R and LMP can be
divided into two distinctive regimes in Figure 5-13. At low LMP (<27 for Alloy
600 and <29 for the rest alloy) representing low temperature and/or short time,
there appeared to be a linear correlation, suggesting a single mechanism/process
with a constant thermal activation energy in operation. It is reasonable to assume
that the process in this regime is predominately dislocation recovery. The slope of
the curve in this region is related to the rate of recovery kinetics. It is also noted
that the R vs LMP relationship for different materials are largely in parallel to
each other. At high LMP values, the relation becomes nonlinear and the
reduction in strain hardening appeared to be accelerated, indicating that
recrystallization started to operate in addition to recovery.
Figure 5-13
R values vs Larson-Miller parameter for the 4 materials studied in this work
Microstructural Analysis
The microstructures of Type 304L were analyzed with an optical microscope at
different time points within the experimental test steps. They are presented in
Figure 5-14. Figure 5-14a shows the well recrystallized grain microstructure after
the initial 1100C annealing heat treatment. The average grain size is about 40
um. In Figure 5-14b, highly deformed grains and extensive slip bands are present
after the initial 20% plastic pre-strain at room temperature. After performing an
isothermal heat treatment of the pre-strained material at 825C for 1 second, the
deformed grain morphology does not change much, although some of the slip
bands have disappeared, as shown in Figure 5-14d. The microstructure of a
different pre-strained specimen after isothermal heat treatment at 1100 C for 1s
5-13
is shown in Figure 5-14c, where both large grain and small recrystallized grains
can be found.
(a)
(b)
(c)
(d)
Figure 5-14
Microstructure of 304L: (a) Fully annealed prior to test; (b) after initial 20% prestrain; (c) isothermally treated at 1000C for 1 second; (d) isothermally treated at
825C for 1 second
Eq. 5-2
5-14
where is the time required to reach a given value of strain hardening reduction
factor R, Q is the activation energy of the thermally activated process, Rg the
universal gas constant, T the absolute temperature, and A the material constant.
In general, it is reasonable to assume the change in flow stress due to dynamic
strain hardening is a function of temperature and the flow stress,
= 0 ()
Then
= ()
0
Integrating over the time in which the flow stresses reduces from the Yp to Yp
(Figure 5-5):
We note that
= ()
0
0
0 0 = ()( 0 )
0 0 =
This yields
or
Therefore
and
0
0
= (1 ) and = 0
(1 ) = ()
1
=
() =
(1 )
() =
(1 ) =
This means that there should be a linear relation between ln( ln(1 )) and 1/T for the given isothermal annealing time:
( (1 )) =
5-15
Figure 5-15 depicts the ln( ln(1 )) versus -1/T relationship from
experimentally measured values of R for Type 304L stainless steel at different
temperatures and times. The linear correlation previously implied clearly exists.
Figure 5-15
Relation between R and isothermal temperature: Type 304L, Case 1 condition
or
(1 ) =
= 1
Eq. 5-3
Eq. 5-4
The above equation is considered applicable for the dislocation recovery process,
with a cutoff temperature of 1100 C. Above this cutoff temperature, the
reduction in strain hardening is considered to be complete (i.e. R = 1).
The material constants (activation energy Q and material constant C) are
obtained from the linear regression analysis of the experimental data, adjusted to
provide a good fit with the experimental data in the time range of 6-30 seconds.
The material constants for the four materials investigated in the study are
provided in Table 5-2 below.
5-16
Table 5-2
Material Constants Temperature Dependent Strain Hardening Model
Materials
Q (kJ/mol)
Type 304L
84.0
8.33
Alloy 600
52.1
4.43
Alloy 52
65.2
5.72
Alloy 82
61.0
6.02
Figure 5-16 compares the strain hardening reduction factor R calculated using
Equation 5-4 at different temperature versus the actual experimental data for
Type 304L stainless steel for the Case 1 study at 20% initial pre-strain. Figure
5-16 demonstrates that, despite of being time independent, the temperature
dependent model captures the basic behavior of the reduction in strain hardening
due to recovery and recrystallization.
Figure 5-16
Comparison of temperature dependent dynamic strain hardening model and
experimental data for Type 304L: Dashed lines are the strain hardening model
User Subroutine
The hardening reduction material behavior investigated and described here is not
a standard hardening selection that can be applied directly from a generalpurpose FEA modeling tool such as ANSYS or ABAQUS. Instead, specific
user-defined plasticity rules must be established in order to define the reduction
in strain hardening calculated using the simplified model described previously.
The modeling work described in this section makes use of ABAQUS; however,
ANSYS also has the capability to generate user-defined material plasticity rules.
The user-defined plasticity used for the FEA model implementation is defined in
conjunction with an isotropic hardening rule. The monotonic stress-strain curve
used as a basis of the user-defined plasticity rules is assumed to be the same as
that of the as-annealed material. As shown in Figure 5-5, this is a reasonable
assumption to make given that the slope of the red and blue stress strain curves
are similar. The extent of the strain hardening reduction for a material at a given
temperature is implemented by subtracting an equivalent amount of strain such
that the material has the desired reduced yield strength.
This approach is described in Figure 5-17, which depicts on the left a stressstrain curve in uniaxial tensile testing and depicts on the right the yield surface in
principal stress space. As shown in Figure 5-17(a), the material yield surface (Y)
expands as the plastic strain (p) accumulates; as the material is plastically
deformed by p, the material yield surface expands to Yp. The user-defined
plasticity rule allows the materials yield surface to shrink at a given temperature
due to the reduction in strain hardening caused by dislocation annihilation. In
the example, the yield surface shrinks to the value Yp, and this can be
equivalently treated as the material recovering its plastic strain from p to p', as
shown in Figure 5-17(b). It is noted if the extent of softening is sufficiently large,
the material would be completely annealed, i.e., all the prior hardening history is
removed and the yield surface is reset back to Y0.
5-18
Figure 5-17
Change in yield surface due to (a) monotonic strain hardening, and (b) dynamic
strain hardening
during the ramp up and down in temperature. A total of five temperature cycles
were imposed on the middle rod to include the effects of repeated reversing
thermal stress cycles.
Figure 5-18
Schematic of three-bar frame analysis problem
The results of the analyses are summarized in Figure 5-19, which provides results
for stress from both material property cases. As shown in Figure 5-19(b), for the
case with isotropic hardening, the plastic strain continuously accumulates over
the multiple thermal cycles. As a result, the residual stress in the middle bar
increases from 362 MPa at the end of the first cycle to 551 MPa at the end of the
fifth cycle, as shown in Figure 5-19(a). However, the reduced hardening model
predicts a very different plastic strain and stress evolution. As shown in Figure
5-19(b), the modified plastic strain in the reduced hardening model increases
initially as the temperature rises. However, it then starts to decrease when the
temperature rises above 1100K due to extensive reduction of strain hardening. As
the middle bar cools to lower temperatures, the modified plastic strain begins to
increase again. The same history repeats over the multiple thermal cycles. The
calculated plastic strain and resulting stress after the fifth cycle using the reduced
hardening model is much lower than that calculated using isotropic hardening.
As shown in Figure 5-19(a), the residual stress calculated using the isotropic
hardening model is 551 MPa, while the result using the reduced hardening
model is 344 MPa.
5-20
5-21
Figure 5-20
Phase 1 Cylinder Analysis Model
Figure 5-21 shows the calculated peak temperature distribution from the thermal
analysis. The gray region is predicted fusion zone in which peak temperature is
higher than 1400 C. Dashed line L1 in Figure 5-21 shows the actual fusion line
of the mock-up weld traced from a macro graph of weld cross section. The
calculated fusion zone was found to compare reasonably well with the actual
weld.
Dashed line L2 in Figure 5-21 shows the boundary of peak temperature that is
higher than 1100 C, and line L3 represents the 750C isothermal line.
According to the experiments described previously, any dislocations and plastic
deformation in the material between L1 and L2 accumulated during heating and
before cooling down to 1100 C will be nearly fully annealed. For material
between L2 and L3, only partial annealing would take place. However, this
partially annealed area is fairly wide compared to the fusion zone and the size of
each welding pass, suggesting the likelihood of considerable influence on the
5-22
plastic deformation behavior, the magnitude of flow stress, and the resultant final
residual stress.
In addition to the overall peak temperature plot in Figure 5-21, it is also noted
that each weld pass has its own set of L2 and L3 regions that influence the strain
hardening behavior of the overall cross section. As an example, the temperature
profile for weld pass 5 is shown in Figure 5-22.
Figure 5-21
Phase 1 Cylinder Overall Peak Temperature Plot
Figure 5-22
Phase 1 Cylinder Weld Pass 5 Peak Temperature Plot
Figure 5-23 compares the predicted equivalent plastic strain distribution in the
welds for conventional isotropic hardening and for the reduced hardening model.
As expected, the equivalent plastic strain with the reduced hardening model is
much lower than that of the isotropic strain hardening model.
5-23
(a)
(b)
Figure 5-23
Equivalent plastic strain distribution for (a) reduced hardening model and (b)
conventional isotropic hardening model
Figure 5-24 and Figure 5-25 compare the hoop and axial stress results,
respectively, among the three hardening rule cases considered. These figures
demonstrate considerable differences in the magnitude and distribution of weld
residual stress associated with different strain hardening rules.
Figure 5-24
Phase 1 cylinder hoop residual stress using different strain hardening rules
5-24
Figure 5-25
Phase 1 cylinder axial residual stress using different strain hardening rules
5-25
Figure 5-26
Through-wall hoop stress at the DM weld centerline for the Phase 1 cylinder
Figure 5-27
Through-wall axial stress at the DM weld centerline for the Phase 1 cylinder
Figure 5-28
International round robin Phase 2b mockup model geometry
The final residual stress distribution for the three different material hardening
cases considered are shown in Figure 5-29 and for hoop and axial stress,
respectively.
Figure 5-29
Phase 2b round robin hoop stress results using (a) the reduced hardening model,
(b) conventional isotropic hardening, and (c) conventional kinematic hardening
5-27
Figure 5-30
Phase 2b round robin axial stress results using (a) the reduced hardening model,
(b) conventional isotropic hardening, and (c) conventional kinematic hardening
The residual stress results at the centerline of the DM weld using the different
hardening rules considered are shown in Figure 5-31 and Figure 5-32 for hoop
and axial stress, respectively. The results are compared against the deep-hole
drilling measurement results. As shown in these figures, the reduced hardening
model tends to fall between the conventional isotropic and kinematic hardening
results. In many parts of the through-wall stress profile, the reduced hardening
model matches the measured results better than either of the two conventional
model results.
Figure 5-31
Phase 2b round robin hoop stress results at DM weld centerline
5-28
Figure 5-32
Phase 2b round robin axial stress results at DM weld centerline
Conclusions
The studies performed to develop an improved strain hardening model for
welding residual stress analysis have identified a significant potential
improvement to WRS modeling analysis. Experimental studies have identified a
notable effect of short exposures to elevated temperature on the strain hardening
material behavior. A new strain hardening model was developed to account for
this effect, and welding residual stress analyses which incorporate the model were
found to predict results which compare favorably to measured residual stresses.
5-29
6-1
Figure 6-1
Comparison of Phase 2a and Phase 2b Output Mean and Measurement Mean
computing uniform (constant) and bending (linear) section forces. Stress due to
the uniform section force (force divided by area) and the bending section force at
the inner diameter (moment divided by section modulus,) are then computed,
and normalized by the yield strength of Alloy 182.
Crack Tip Stress Intensity Factor (K)
Flaw assessment at welds, used to quantify the rate of potential SCC crack
growth, requires calculation of the SIF due to WRS and applied loads. The
applied loads include membrane stress, bending stress, and crack face pressure.
SIF calculations in this paper are based on the weight function method,
idealizing the nozzle as a pipe with geometry ri/ro = 0.8, and using closed form
weight functions from earlier work by Wu and Carlsson [33] and by Glinka [34].
There are typically three types of inner-diameter flaws considered in the analysis:
a complete internal circumferential flaw, an internal circumferential semielliptical flaw, and an internal axial semi-elliptical flaw. Because the K value for a
complete circumferential flaw can be similar to that at the deepest point of a
circumferential semi-elliptical flaw, only two flaw types are presented here, a
complete circumferential flaw and a semi-elliptical axial flaw. For the semielliptical axial flaw, we limit our analysis to the K at the deepest point and
consider a flaw shape with a constant aspect ratio c/a = 2. It is noted that the K
values reported for the semi-elliptical axial flaw case assume the residual hoop
stress profile at the center of the DM weld is present at all axial positions along
the crack face. This commonly used approach is consistent with the elevated
hoop stress distribution generally found on the weld cross section.
Because the circumferential flaw is driven by axial stress and the axial flaw is
driven by hoop stress, analysis of those two flaw types is considered sufficient for
validation. It should be pointed out that assuming a constant aspect ratio is not
typical of field assessments, but is useful here as a simplified basis for judging the
quality of WRS model outputs.
Crack Growth Time
Given K values, SCC crack growth time can be computed, and provides a metric
that can be judged relative to plant operational experience. The crack growth rate
is determined from the SIF using methods described in EPRI report MRP-115
[35], which is then integrated numerically to determine crack growth time. The
initial flaw is assumed to be of size a/t = 0.1, where a is the crack depth and t is
the thickness. Crack growth rate is determined at a set of crack size increments
a/t = 0.001, to a maximum size amax/t = 0.6 or 0.8, depending on the validity
limit of the respective weight function. A representative operating temperature of
343 C (650 F) [36] was used in the crack growth calculations. It should be
noted that in practice flaws would not be permitted to remain in service beyond
0.75t.
6-3
Benchmarking Results
Results are presented here for the four validation metrics described above using
the data from the Phase 2a and Phase 2b round robin studies. Due to differences
in time between performing the benchmarking studies for the Phase 2a and
Phase 2b data, some minor differences in approach are taken in the two
evaluations, as follows:
Different data were selected as the benchmark for the two studies. For Phase
2a, the benchmark is the mean of the modeler results; whereas the
benchmark for the Phase 2b results is the mean of the measurement data.
The Phase 2a round robin did not specify which hardening model to use;
therefore, each individual hardening result is treated as a potential solution.
In contrast, Phase 2b specified that all modelers perform the analysis twice,
once with isotropic and once with kinematic hardening, with the intent of
averaging the two sets of results.
6-4
Figure 6-2
Phase 2a Normalized RMS Difference Between Model Output Stress and the Mean
of All Model Outputs Benchmark
6-5
Figure 6-3
Phase 2b Normalized RMS Difference Between Model Output Stress and the
Measurement Mean Benchmark
6-6
Figure 6-4
Phase 2a Normalized Stress due to Net Section Force from Axial and Hoop
Stresses
The normalized stress at the weld ID due to bending moments from axial and
hoop stresses for the WRS data are compared in Figure 6-5. The bending from
axial stress is significantly larger than that from hoop stress. One analysis shows a
normalized bending moment from axial stress exceeding the nominal yield
strength of the material. Both moments are a result of the stainless steel weld
that is adjacent to the DM weld.
6-7
Figure 6-5
Phase 2a Normalized Stress due to Net Section Bending Moments from Axial and
Hoop Stress
6-8
Figure 6-6
Phase 2b Normalized Stress due to Net Section Force from Axial and Hoop
Stresses
Figure 6-7
Phase 2b Normalized Stress due to Net Section Bending Moments from Axial and
Hoop Stress
6-9
Figure 6-8
Phase 2a Crack Tip SIF (K) for Complete Internal Circumferential Flaw with
ri/ro = 0.8
Figure 6-9 provides results for the internal axial semi-elliptical flaw with a
cylinder geometry ri/ro = 0.8 and a crack aspect ratio c/a = 2. For the axial flaw,
the residual stress contributions to the SIF are positive for all model outputs and
measurements, which is consistent with the positive near-ID hoop residual stress
[1]. Also, the SIF from applied loads is smaller for the axial flaw than it was for
the circumferential flaw.
6-10
Figure 6-9
Phase 2a Crack Tip SIF (K) for Axial Surface Flaw with ri/ro = 0.8, c/a = 2 at
Crack Depth
6-11
Figure 6-10
Phase 2b Crack Tip SIF (K) for Complete Internal Circumferential Flaw with
ri/ro = 0.8
Figure 6-11
Phase 2b Crack Tip SIF (K) for Axial Surface Flaw with ri/ro = 0.8, c/a = 2 at
Crack Depth
circumferential flaw, no crack growth is predicted for WRS data from iDHD #1,
contour, and 4 out of 14 model outputs.
Figure 6-12
Phase 2a Crack Growth Time for Complete Internal Circumferential Flaw with
ri/ro = 0.8
The crack growth time for the deepest point of the internal axial semi-elliptical
flaw with aspect ratio c/a = 2 is shown in Figure 6-13. As demonstrated in Figure
6-9, there are no negative values of SIF for any of the cases. Additionally, the
SIFs are significantly higher for the axial flaw than for the circumferential flaws,
with is consistent with the hoop stress being large and mostly tensile through
wall. As a result, the flaw grows deeper and faster, in some cases reaching a/t =
0.8 in less than 80 months. Throughout the entire 720 months, SIF and crack
size computed from the mean model output are significantly greater than those
derived from the measurement data.
6-13
Figure 6-13
Phase 2a Crack Growth Time for Axial Surface Flaw with ri/ro = 0.8, c/a = 2 at
Crack Depth
6-14
Figure 6-14
Phase 2b Crack Growth Time for Complete Internal Circumferential Flaw with
ri/ro = 0.8
Figure 6-15
Phase 2b Crack Growth Time for Axial Surface Flaw with ri/ro = 0.8, c/a = 2 at
Crack Depth
analyst to select a conservative portion of the overall cross section rather than
focus on a single arbitrary location.
A computational analysis process was performed where the through-wall stress
information at an incrementally varying axial location on the cross section was
selected and evaluated for crack growth time. Then, for each model, the location
relative to the weld centerline where this minimum was found and the minimum
crack growth time was reported. A second computational analysis process was
also performed, where the axial location of maximum integrated average stress in
the first 25% of the wall thickness was calculated. This location was then also
used to calculate crack growth time and compared to the minimum previously
identified. Along with the model submissions, measurement data were included
in the computational analysis. The biaxial mapping measurement technique
(contour for hoop stress and multiple slitting measurements for axial stress) was
used to generate stress data along the mockup cross section. The deep hole
drilling measurements taken at the weld centerline were not considered in this
evaluation.
The results of this evaluation are reported in Table 6-1 for axial stress
(circumferential flaw case) and Table 6-2 for hoop stress (axial flaw case). In both
tables, the position shown is the distance from the weld centerline, with positive
values indicating towards the safe end side and negative values indicating towards
the nozzle side. The axial stress results (circumferential flaw case) in Table 6-1
are inconclusive, since the general stress condition along the inner portion of the
entire DM weld region was compressive. Therefore, nearly all modeler
submissions as well as the measurement data result in infinite crack growth times
due to negative K values. The higher hoop stress results (axial flaw case) in Table
6-2 provide much more information for consideration. As shown in Table 6-2,
there is generally good correlation in between the location which results in a
minimum time for a flaw to reach 75% through-wall and the location of the
maximum integrated average stress at the inner 25% of the wall. Furthermore,
comparing the results in Table 6-2 to the crack growth time data in Figure 6-15,
the dispersion in the calculated time to reach 75% through-wall is significantly
reduced when the location of maximum crack growth is considered rather than
an arbitrary location like the DM weld centerline.
6-16
Table 6-1
Axial Position and Operating Time for Circ Flaw to Reach 50% TW, based on
Minimum Time Evaluation and Maximum Average Axial Stress Evaluation
Location of
Minimum Crack
Growth Time
Case
Position
(mm)
Time
(mo)
Position
(mm)
Avg
Stress
(MPa)
Time
(mo)
A5
n/a
inf
-0.21
-118.65
inf
B1
+0.03
1861
-0.25
-19.74
inf
C1
n/a
inf
-0.25
-139.04
inf
D1
n/a
inf
-0.38
-88.07
inf
E1
n/a
inf
-0.63
-97.92
inf
G1
n/a
inf
-0.43
-28.72
inf
H1
n/a
inf
-0.52
-38.51
inf
I1
n/a
inf
+0.01
-173.48
inf
J1
-0.18
301
-0.06
51.83
319
Measured
n/a
inf
-0.28
-120.34
inf
Table 6-2
Axial Position and Operating Time for Axial Flaw to Reach 75% TW, based on
Minimum Time Evaluation and Maximum Average Hoop Stress Evaluation
Location of
Minimum Crack
Growth Time
Case
Position
(mm)
Time
(mo)
Position
(mm)
Avg
Stress
(MPa)
Time
(mo)
A5
-0.31
382
-0.34
102.10
434
B1
+0.21
175
+0.21
190.11
176
C1
-0.30
222
-0.33
155.39
224
D1
-0.33
145
-0.36
236.47
152
E1
-0.61
264
-0.71
334.96
617
G1
-0.29
154
-0.70
336.93
230
H1
-0.32
164
-0.64
299.87
226
I1
-0.23
347
-0.23
90.45
347
J1
-0.15
75
-0.15
388.97
75
Measured
-0.48
494
-0.50
64.89
499
6-17
Discussion
Although the RMS difference does not indicate whether a model has WRS
higher or lower than a benchmark, it provides a simple way to quantify
dispersion. The average RMS difference from the measurement mean for Phase
2a and Phase 2b are shown in Table 6-3 as a function of hardening rule, where
each row reports results for each hardening model, and the values listed are
specific to that hardening rule. For example, the row marked ISO reports the
average RMS difference of all Phase 2a or Phase 2b isotropic model outputs
relative to the Phase 2a or Phase 2b measurement mean, the row marked KIN
reports the average RMS difference of all kinematic model outputs relative to the
measurement mean, and so on. In the table, AVG refers to the average of
kinematic and isotropic models, and ALL refers to all of the isotropic and
kinematic results considered together. Average hardening could not be assessed
for Phase 2a because only three participants submitted results for both isotropic
and kinematic hardening.
For axial WRS, the average of isotropic and kinematic hardening used in Phase
2b (AVG) has the lowest average RMS difference between model and
measurement. For hoop WRS, the 5 models using kinematic hardening (KIN) in
Phase 2a and the average hardening in Phase 2b (AVG) show the lowest
difference between model and measurement. When not controlling for hardening
rule (ALL), there was a similar level of difference between models and
measurement in Phase 2a and Phase 2b.
Table 6-3
Comparison of Average RMS Difference Relative to Measurement
Axial
Hoop
Phase 2a
Phase 2b
Phase 2a
Phase 2b
ISO
28%
29%
44%
37%
KIN
27%
32%
27%
42%
AVG
--
21%
--
31%
ALL
28%
30%
38%
40%
The following are comparison points for Phase 2a and Phase 2b relative to other
validation metrics discussed above:
1. Stress due to Section Forces
Uniform section force from axial WRS were near zero in both studies
(see Figure 6-4 for Phase 2a and Figure 6-6 for Phase 2b), which is
consistent with mechanical equilibrium.
Stress due to uniform section force from hoop WRS is mostly positive in
the Phase 2b (see Figure 6-6) and was lower than for Phase 2a (see
Figure 6-4). There was more dispersion of the uniform section forces in
Phase 2a than in Phase 2b.
6-18
Stress due to bending section forces from axial and hoop WRS exhibit
similar dispersion in both studies, where the models are biased above the
measurement mean for hoop (the dashed red line in Figure 6-7) and not
strongly biased relative to the measurement mean for axial (the dashed
blue line in Figure 6-7).
For the axial semi-elliptical flaw, the level of agreement between the
output mean and measurement mean is quite good in Phase 2b (see
Figure 6-11), with even dispersion on both sides of the measurement
mean (the solid red line). In Phase 2a (see Figure 6-10), there is a clear
separation between the output mean and measurement mean, with all of
the model results falling above the measurement mean.
For the axial semi-elliptical flaw, a separation between output mean and
measurement mean can be seen in both studies (see Figure 6-13 for
Phase 2a and Figure 6-15 for Phase 2b). In the Phase 2b, the dispersion
among models is distributed evenly relative to the measurement mean. In
the Phase 2a, most dispersion is toward faster crack growth relative to
the measurement mean.
A significant factor in the dispersion of crack growth time for both Phase
2a and Phase 2b is the low K values at the ID surface for both flaw
orientations. Small differences in the K value at the ID can lead to very
large differences in calculated crack growth time under these conditions.
6-19
Validation Recommendations
The results of the validation approaches discussed in this section demonstrate the
importance of the accuracy of the welding residual stress input to crack growth
calculations. Given this importance, there is a benefit to demonstrating that
methodology used to generate a set of WRS inputs has been benchmarked
against known solutions. Therefore, it is recommended that a performance
demonstration process be used to validate models as part of the analysis process,
with the following requirements:
The recommended benchmark for the model validation is the average of the
residual stress measurements. Two independent measurement techniques
were used for the residual stress measurements, including multiple
independent measurements using the DHD technique. The measurements
were performed independent of the WRS analyses, so they are blind
measurement values. They were also performed after developing experience
on the similar Phase 2a round robin mockup, so they represent a matured
approach to this specific geometry and material condition.
6-20
Hoop
(MPa)
-222.8
-90.6
7.0
57.1
96.9
132.2
155.9
172.4
149.3
107.0
53.6
5.5
-23.1
-44.3
-42.5
-11.1
49.1
113.3
172.6
222.6
241.6
247.0
234.7
213.9
176.6
-22.3
6-21
Axial
(MPa)
-267.1
-221.0
-159.6
-110.6
-65.2
-38.0
-30.3
-45.2
-87.8
-133.0
-176.1
-206.0
-216.7
-217.7
-196.4
-147.3
-62.2
37.9
162.8
258.5
314.2
351.7
368.1
372.2
358.7
222.0
7-1
Section 8: References
1. EPRI, "Materials Reliability Program: Finite-Element Model Validation for
Dissimilar Metal Butt-Welds (MRP-316), Electric Power Research
Institute, Palo Alto, CA, MRP-316, 2011.
2. Weld Residual Stress Finite Element Analysis Validation: Part 1 Data
Development Effort (NUREG-2162)
3. H. J. Rathbun, L. F. Fredette, P. M. Scott, A. A. Csontos, and D. L.
Rudland, "NRC Welding Residual Stress Validation Program International
Round Robin Program and Findings," Proceedings of the ASME 2011
Pressure Vessels & Piping Division Conference, July 17-21, 2011, Baltimore,
Maryland, USA.
4. M. R. Hill, M. N. Tran, and J. E. Broussard, "Validation approaches for
weld residual stress simulation," Proceedings of the ASME 2014 Pressure
Vessels & Piping Division Conference, Anaheim, CA, USA, 2014.
5. EPRI MRP-317, "Materials Reliability Program: Welding Residual Stress
Dissimilar Metal Butt-Weld Finite Element Modeling Handbook (MRP317)," Report no. 1022862, Sect. 4, Electric Power Rearch Institute, Palo
Alto, CA: 2011.
6. M. B. Prime, "Cross-Sectional Mapping of Residual Stresses by Measuring
the Surface Contour after a Cut," Journal of Engineering Materials and
Technology, vol. 123, pp. 162-168, 2001.
7. ASTM, "Standard Practice for Use of the Terms Precision and Bias in
ASTM Test Methods", E177, ASTM International, West Conshohocken,
PA, 2010.
8. M. B. Prime and A. L. Kastengren, "The Contour Method Cutting
Assumption: Error Minimization and Correction", 507, Proceedings of the
SEM Annual Conference & Exposition on Experimental and Applied
Mechanics, Indianapolis, Indiana USA, 2010.
9. H. W. Coleman and W. G. Steele, "Experimentation, Validation, and
Uncertainty Analysis for Engineers", Hoboken, New Jersey, John Wiley &
Sons, Inc., 2009.
10. J. S. Robinson, D. A. Tanner, C. E. Truman, A. M. Paradowska, and R. C.
Wimpory, "The influence of quench sensitivity on residual stresses in the
aluminium alloys 7010 and 7075", Materials Characterization, vol. 65, pp.
73-85, 2012.
8-1
8-4
A-1
A-2
A-3
A-4
A-5
A-6
A-7
A-8
A-9
A-10
A-11
A-12
A-13
A-14
A-15
A-16
A-17
A-18
A-19
A-20
A-21
A-22
A-23
Program:
Pressurized Water Reactor Materials Reliability Program (MRP)
2015 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power
Research Institute, EPRI, and TOGETHER...SHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.
3002005498