-
DEVELOPMENT OF CRACK GROWTH RATE DISPOSITION CURVES FOR PRIMARY
WATER STRESS CORROSION CRACKING (PWSCC)
OF ALLOY 82, 182, AND 132 WELDMENTS
G. A. White1, N. S. Nordmann1, J. Hickling2, C. D.
Harrington31Dominion Engineering, Inc., 11730 Plaza America Drive,
Suite 310, Reston, VA 20190
2EPRI, 3412 Hillview Avenue, Palo Alto, CA 94304 3TXU Energy,
P.O. Box 1002, Glen Rose, TX 76043
Keywords: Nickel-base weld metals, Alloy 82, Alloy 132, Alloy
182, Primary Water Stress Corrosion Cracking, Crack growth rate
Abstract
Nickel-based austenitic alloys, including wrought Alloy 600 and
Alloy 82/182/132 weld metals, are used extensively in pressurized
water reactor (PWR) applications. In 2003, the authors reported the
results of work sponsored by the EPRI Materials Reliability Program
(MRP) to develop a crack growth rate (CGR) disposition curve for
primary water stress corrosion cracking (PWSCC) of thick-section
Alloy 600 material. This deterministic CGR equation has been
adopted by Section XI of the ASME Boiler & Pressure Vessel Code
for continued-service evaluation of PWSCC flaws detected (or
postulated to exist) in PWR reactor vessel upper head nozzles,
including control rod drive mechanism (CRDM) nozzles. Following
observations of cracking in primary circuit welds with high
residual stresses and in some J-groove welds attaching CRDM nozzles
to the reactor vessel upper head, the need for a similar equation
for Alloy 82/182/132 weldments was identified. A preliminary MRP
CGR curve for Alloy 182 material was published in 2000, but this
was based on a fairly limited experimental database and simplifying
assumptions. Weld metals are by definition as-cast structures and,
as such, are much more inhomogeneous than wrought materials. The
scatter introduced by the inhomogeneous nature of weld metals
necessitated the development of a more sophisticated approach.
Analogous to the procedure that resulted in the CGR equation for
Alloy 600 wrought material, an international panel of PWSCC experts
supported the MRP in its development of a deterministic CGR
equation for Alloy 82/182/132 weldments. After reviewing the key
metallurgical aspects of Alloys 82, 182, and 132, the data and
methods used to develop the CGR equation for such weldments are
described. The laboratory testing techniques that have been used to
generate CGR data for these weld metals in simulated PWR primary
water environments were analyzed. Appropriate screening procedures
were developed and applied to produce the final MRP database before
using an agreed data reduction methodology to derive separate CGR
curves as a function of temperature and stress intensity factor KI
for these weld metals, including consideration of the effects of
dendrite orientation. For stress intensity factors greater than 20
MPam, the new CGR curve for Alloy 182/132 weld metal is nearly
parallel to, and about four times higher than, the previously
reported curve for Alloy 600 wrought material. Comparisons are made
with other laboratory data not used in derivation of the new MRP
lines, with the limited field data available from repeat
non-destructive examination inspections of a cracked primary
circuit butt weld at the Ringhals PWR in Sweden, and with the CGR
disposition curves that have been proposed by other workers.
Finally, an example is provided of the way in which the curves
can be applied to the assessment of further growth through PWSCC of
piping butt weld flaws that might be detected in service.
I. Introduction
Nickel-based austenitic alloys, including wrought Alloy 600 and
weld metals Alloy 82, 182, and 132, are used extensively in
pressurized water reactor (PWR) applications. These materials offer
a useful combination of good mechanical properties and fracture
toughness, compatibility with other vessel or piping materials, and
corrosion resistance. However, recent incidents of primary water
stress corrosion cracking (PWSCC) of Alloy 600 components other
than steam generator tubes in the primary circuits of PWRs [1] have
highlighted the need for a qualified equation for crack growth
rates (CGRs) to evaluate flaws found by in-service inspection. This
requirement was fulfilled for the wrought Alloy 600 base material,
after much deliberation involving an international panel of PWSCC
experts, by the issuance in 2002 of the Materials Reliability
Program (MRP) MRP-55 report [2], whose main contents were later
published as Reference [3]. The disposition curve established in
that work has since been incorporated into the ASME Section XI Code
for flaw evaluation [4]. A similar requirement has also been
identified for Alloy 82/182/132 weldments following observations of
cracking in primary circuit welds with high residual stresses and
in some J-groove welds attaching control rod drive mechanism (CRDM)
and bottom mounted instrumentation (BMI) nozzles to the reactor
upper head [1,5]. A preliminary MRP CGR curve for Alloy 182
material was published as a proprietary report in 2000 [6] and
later made public [7], but this was based on a fairly limited
experimental database and an assumption that the results could be
described by application of a simple multiplication factor to the
then current Scott model for base metal, which had been derived
from field data on thin-walled steam generator tubing [8]. Weld
metals are by definition as-cast structures and, as such, are much
more inhomogeneous than wrought materials. The scatter introduced
by the inhomogeneous nature on a microscopic scale of weld metals
makes the simple multiplication factor approach not suitable for
extensive use, and necessitated the development of a more
sophisticated methodology. The approach taken here to deriving a
more appropriate model for nickel-based weld metals was analogous
to that used to establish the MRP-55 curve for thick-walled wrought
Alloy 600, namely detailed consideration and screening by the MRP
PWSCC Expert Panel of all available laboratory data from relevant
CGR
Proceedings of the 12th International Conference onEnvironmental
Degradation of Materials in Nuclear Power System Water Reactors
Edited by T.R. Allen, P.J. King, and L. Nelson TMS (The
Minerals, Metals & Materials Society), 2005
511
-
tests in simulated PWR water and statistical derivation of
best-fitcurves, taking into account, as far as possible, the
particular natureof PWSCC in weld material and its possible
influence on theexperimental results that had been obtained. The
present articlestarts by describing key metallurgical aspects of
Alloys 82, 182, and 132. It continues with a description of the
laboratory testingtechniques that have been used and details the
screeningprocedures that were applied to produce the final MRP
database.After setting out the data reduction methodology used to
deriveseparate CGR curves as a function of the stress intensity
factor KIfor nickel-based weld metals, comparisons are made with
otherlaboratory data not used in derivation of the new CGR lines,
thelimited field data available from repeat
non-destructiveexamination (NDE) inspections of a cracked primary
circuit buttweld at the Ringhals PWR in Sweden, and with the
CGRdisposition curves that have been proposed by other workers.
Finally, an example is provided of the way in which the curvescan
be applied to the assessment of further growth of cracks byPWSCC
which might be detected in service. Report MRP-115 [9]presents the
full methodology and results of the MRP study ofCGRs for PWSCC of
Alloy 82/182/132 weld metals.
II. Background on Metallurgical Aspects of Nickel-BasedWeld
Materials and Their Effects on Crack Growth Rates
The incentive for covering metallurgical aspects is that
highvariability is observed in the measured CGRs of Alloys 82,
182,and 132, and it is therefore important to understand
howmetallurgical factors contribute to this variability. However,
while discussion of these factors provides useful
backgroundinformation, firm correlations between metallurgical
factors andCGR are not available except for differences in CGR
between Alloys 182/132 and Alloy 82. For this reason, the CGRs
fordifferent weld alloys need to be addressed on a statistical
basis,and not by correlation with specific metallurgical
features.
II.A Macrostructural and Microstructural Features of
Nickel-Based Weld Metals
The CGR in Alloy 82/182/132 is strongly affected by
themicrostructure of the weld and by the orientation of the crack
growth with respect to the microstructure. For this reason, it is
important to develop an understanding of the main features ofAlloy
82/182/132 microstructure.
Weld metal forms by solidification from a molten state,which
leads to the formation of dendrites growing in the directionof the
heat flow, i.e., perpendicular to the solid material on whichthe
weld is deposited. Most welds are made with multiple passes.The
grain structure of dendrites in subsequent passes is
normallyrelated to that of previous passes as a result of epitaxy,
i.e., by thetendency of a crystal forming on a substrate to have
the same structural orientation as the substrate. This results in
the dendritespersisting through several or many weld passes.
Themacrostructure of a typical weld is shown in Figure 1 [9].
Theweld shown in Figure 1 was made with Alloy 82H (Alloy 82H is the
same as Alloy 82 but with C content of 0.030.10wt% instead of
0.10wt% max.). Features to note regarding Figure 1 include: The
weld was made with over 30 weld passes. There is a strong pattern
of columnar grains formed by
dendrites, and the pattern persists through many weld
passes.
The dendrites tend to be perpendicular to the base material
atthe weld-base material interface, and tend to become
vertical(root to crown direction) as the weld thickness
increases.The dendrites are mainly vertical in the central region
of theweld.
Crown
Root
1 mm
Fig. 1. Transverse Section of Alloy 82H Weld ShowingColumnar
Grain Structure [9]
The typical weld structure is shown schematically in Figure 2
[10], which reflects growth of the dendrites perpendicular to
thegroove wall and intersecting near the middle of the weld,
wherethey form a high energy solidification grain boundary. This
situation is somewhat different than that shown in Figure 1
where,because of different groove geometry, the dendrite growth
wasmore nearly vertical.
Fig. 2. Sketch Showing Various Kinds of Grain Boundaries in Weld
Metals (Courtesy of Ohio State University [10])
512
-
As the weld metal forms, parallel bundles of dendrites with
nearly identical crystallographic orientation form and grow intothe
melt. The boundaries between these similarly oriented dendrites are
called solidification subgrain boundaries (SSGBs)and tend to have
low angular mismatches, as well as low energy,and are believed to
form paths for PWSCC relatively infrequently.Where different
bundles of dendrites intersect or overlap, largerangular mismatches
often occur between the grains of the bundles. In this case, the
resulting grain boundaries, termed solidificationgrain boundaries
(SGBs), can be high energy and are believed to be more common paths
for PWSCC. However, it has been found that some high-angle grain
boundaries or sections of boundaries have relatively low energy
since they have coincident site lattices(meaning that the crystal
orientations are such that the atomicstructures of the two grains
have a significant level of matching);this causes those particular
high-angle boundaries to be relatively more resistant to PWSCC
propagation [11]. Such low-energy,high-angle grain boundaries may
be present in cases for which thecrack path is kinked, connecting
two planes that are offset by asmuch as 12 mm (even with high-angle
boundaries in closeproximity that would have facilitated a
straighter crack path).
A typical wavy pattern of high energy grain boundaries
isobserved in Alloy 82 and 182 weld metals. The structure of the
grain boundaries is illustrated in Figure 3 [10], which also shows
migrated grain boundaries (MGB) that develop as the result
ofsubsequent weld passes. In multipass welds, the SGBs canmigrate
on cooling after solidification and during re-heating andresult in
a straighter, migrated grain boundary. Visible in this figure are
SSGBs formed between dendrites, SGBs, and MGBs.PWSCC cracks in weld
metals typically follow the higher energySGB and/or MGBs.
The orientation of the crack in the weld, i.e., relative to
thewelds columnar microstructure, has a strong influence on theCGR.
Thus, it is necessary to include the relative crackorientation
(i.e., parallel or perpendicular to the predominantdirection of the
weld dendrites) in the development of a CGRmodel. The convention
used for identifying crack orientation is shown in Figure 4.
Cracks grow fastest along high energy grain boundaries inthe
direction of grain growth (TS and LS orientations), and nextfastest
along high energy grain boundaries perpendicular to thedirection of
grain growth but parallel to the welding direction.Cracks that grow
perpendicular to the high energy grain
boundaries, i.e., perpendicular to the columnar dendrites,
growsignificantly slower.
SSGB
SGB
MGB
Fig. 3. Micrograph Showing Solidification (SGB),
SolidificationSubgrain (SSGB), and Migrated (MGB) Grain Boundaries
in
Weld Metals (Courtesy of Ohio State University [10])
II.B Effects of Chemical Composition on Crack Growth Rate
The nominal chemical compositions of Alloys 82, 182, and132 are
shown in Table I.
The only well explored effect of the compositionaldifferences
among the weld alloys on PWSCC is the influence ofchromium.
Buisine, et al. evaluated the PWSCC resistance ofnickel-based weld
metals with various chromium contents rangingfrom about 15% to 30%
chromium [12]. The results indicatedthat weld metals with 30%
chromium were resistant to cracking,with a threshold for PWSCC
resistance being between 22 and30% chromium.
Crack growth is along (parallel to) the direction of the
dendrites for the TSand LS orientations.
Crack growth is across (perpendicularto) the direction of the
dendrites for the TL, LT, ST, and SL orientations.
Nomenclature for crack orientationThe first letter denotes the
direction normal to the planeof the crack face. The second letter
denotes the direction of crack growth.
Fig. 4. Terminology Used for Orientations of Cracks in Test
Specimen with Respect to Welds
513
-
Table I. Nominal Chemical Composition for Alloy 82, 182, and 132
Weld Metal
wt. % Alloy
Ni Cr Fe Mn Nb Ti
82 71 20 2 3 2.5 0.5
182 67 15 8 7 1.8 0.5
132 70 15 9 1 2.5 ---
Other laboratory investigations related to compositional effects
have been reviewed and the following conclusions can be drawn [9]:
(1) increasing the average chromium concentration of the material
correlates with increasing resistance to PWSCC, (2) the reduction
in local chromium concentration that occurs at grain boundaries as
the result of exposure to sensitizinga heat treatments does not
increase the materials susceptibility to PWSCC, and the reduction
of residual stresses provided by certain heat treatments is
helpful, and (3) there are no known consistent effects of
impurities (Si, P, and S) on PWSCC susceptibility.
II.C Effect of Weld Design and Fabrication on Crack Growth
Rate
Weld design and fabrication can affect CGR in weld metals in
several ways, such as by their effects on residual stresses, local
material composition, strength level of the weld material,
microstructure, and presence of micro-flaws [9]. Residual Stresses.
It is considered that residual stresses
associated with welding can have a strong influence on PWSCC CGR
because residual stresses often are a strong contributor to the
stress intensity factor at crack tips within the welds and
laboratory tests have shown the crack-tip stress intensity factor
to be a key parameter.
Local Material Composition. Depending on welding conditions, the
chemical content of regions of a weld adjacent to a base metal with
lower alloy content may be significantly affected by dilution from
the base metal.
Strength Level. The effect of strength level on CGR of weld
metals is expected to be similar to that of wrought material, and
crack growth rates are expected to increase as the materials yield
and tensile strengths rise.
Microstructure. The major microstructural feature of Alloy
82/182/132 consists of dendritic grains, as discussed above. PWSCC
in those welds involves an intergranular (IG) cracking mechanism,
whereby cracks propagate along high-angle grain boundaries
[13,14,15]. Cracks in welds usually have an undulating or wavy
character that reflects the wavy morphology of the grain
boundaries. In tests at constant load or displacement, unbroken
ligaments often form in the wake of advancing crack fronts because
the most SCC-resistant boundaries tend not to fail. In some
regions, uncracked ligaments can be massive and extend back to the
fatigue precrack commonly produced in the specimen for CGR testing,
thereby resulting in incomplete engagement of the
a Sensitization refers to the precipitation of chromium carbides
leading to low chromium concentration at grain boundaries, making
the material susceptible to rapid corrosion in acid-oxidizing
environments.
stress corrosion crack to the precrack over the full width of
the specimen. Such uneven crack fronts and incomplete engagement of
stress corrosion cracks are sources of potential uncertainty in
making laboratory measurements of CGR for welds.
Weld Defects. Micro-fissures and other weld defects such as
pores and slag inclusions are often present in Alloy 82/182/132.
The PWSCC CGR in the weld metal could plausibly be affected by
latent defects. However, recent investigations [11,13,14] indicate
no discernable effect of hot or ductility-dip cracking on PWSCC. On
the other hand, relatively large and sharp defects, such as some
lack of fusion areas, could potentially promote PWSCC by acting as
stress concentrators and increasing the local stress intensity
factor.
III. Specimen Manufacture and Crack Front Patterns
Special purpose welds are typically fabricated to make compact
tension (CT) specimens for conducting crack growth rate tests for
nickel-based alloy welds used in PWR applications. The majority of
the specimens included in the Alloy 82/182/132 crack growth rate
database were manufactured using a single-sided, V-shaped, butt
weld preparation similar to that commonly used in plants. The
welding parameters corresponded to normal industry practice, with
attention given both to the weld macro- and microstructure (as
described above) and to several additional factors discussed
below.
The crack growth rate data applied directly in the development
of the MRP-115 CGR equation [9] were obtained from CT specimens
having a width of 0.50.6 inches except for the 1-inch CT specimens
used by Studsvik [9]. Practice varied regarding the extent of Alloy
600 base metal present in the CT specimens away from the Alloy 82,
182, or 132 crack plane.
III.A Weld Chemical Composition
Test specimen fabrication utilized typical vendor fabrication
practices and ASME/AWS specified weld metals and are thus
considered representative of weld materials in operating PWRs.
III.B Convention for Identifying Crack Orientation
Because CGR is strongly affected by the direction of crack
orientation relative to the microstructure (as discussed above), it
is important during tests to identify and control the direction of
crack growth relative to the microstructure. The convention used
for identifying crack orientation relative to weld fabrication, and
thus relative to weld microstructure, is shown in Figure 4, which
is an extension of the standard convention for rectangular sections
of wrought material [16].
III.C Restraint
Nickel-based alloys and the other structural materials used with
these welding materials start to liquefy when heated in the range
from approximately 1350 to 1450C. When welds start to cool from
these elevated temperatures, significant levels of thermal
displacements are involved which can lead to considerable stresses
and strains. These stresses and strains depend on the degree of
mechanical constraint in the weld and are
514
-
usually sufficient to plastically deform the underlying weld
metalas well as the base metal.
A J-groove weld will normally be much more constrainedthan a
butt weld. The resulting solidification-strain-inducedstresses are
considered likely to make the material moresusceptible to stress
corrosion cracking, and it is also possible thatthe weld shrinkage
strains increase the strength of the weld metalsomewhat and thereby
also have some effect on CGR. Estimatesof the residual stresses
present in and adjacent to the weld can be determined by various
techniques such as x-ray diffraction orfinite-element analyses, and
estimates of material strengtheningcan be obtained from mechanical
tests, such as hardness andtensile tests.
The stresses and strains induced during the welding process take
the form of macroscopic, local residual stresses and strains aswell
as microscopic strains at the grain boundaries. Typically,
theremoval of the CT specimen from the sample weld is expected to
relieve most of the macroscopic residual stresses in the
sample,although the extent of such relief might be expected to
depend onthe volume of base metal material remaining in the CT
specimen.
In laboratory CGR testing, it is standard practice for the
levelof residual stress remaining in the test specimen after its
removalto be ignored when estimating the stress intensity factor
applied tothe growing crack. However, for the constrained geometry
typicalof plants applications, welding residual stresses often
dominatethe other stresses and cannot be ignored in crack growth
calculations.
III.D In-Process NDE
The ASME Boiler and Pressure Vessel Code setsrequirements for
in-process NDE of PWR pressure boundarywelds. Because of the
favorable laboratory conditions under which weld test samples are
fabricated, it is considered that theirquality may, in general, be
better than that of typical plant welds.However, even under ideal
welding conditions some defects areexpected, and the presence of
defects in some laboratory testwelds has been reported.
III.E Weld Defects
Nickel-based welds made with Alloys 82/182/132 can be affected
by various forms of solidification cracking, liquationcracking, and
ductility-dip cracking during manufacture. Modifications in welding
consumables and procedures have verymuch improved resistance to
solidification and liquation cracking.However, processes such as
ductility-dip cracking can be expected to produce subsurface weld
defects or surface defectssmall enough to be accepted for service
during pre-service NDE. It is seemingly plausible that weld defects
such as hot cracking or ductility-dip cracking could affect the CGR
in those welds.However, recent investigations appear to provide
convincingevidence that such weld defects do not play a significant
role inPWSCC initiation and propagation [11,13,14].
Relatively large and sharp weld defects such as some weld lack
of fusion regions may have the potential to promote PWSCCby
creating a local stress concentrator and a high local
crack-tipstress intensity factor. Lack of fusion areas at the weld
wettedsurface would be expected to be detected during pre-service
NDE.Subsurface defects would necessarily have to become wetted
bythe primary coolant through some cracking process before
theycould grow via PWSCC. Potential types of cracking to cause
a
subsurface lack of fusion region to become wetted include
ductiletearing, environmental or mechanical fatigue, and PWSCC
(crackgrowth in from the wetted surface). Although there is
notuniversal agreement among experts, it is possible that at
leastsome of the cracking observed in BMI nozzles at South Texas
Project Unit 1 in 2003 may have involved the wetting ofsubsurface
weld lack of fusion areas [11,17,18].
III.F Post-Welding Heat Treatment
Post-weld heat treatment (PWHT) of the buttering used in primary
weld preparations is standard practice, and PWHT is in some cases
also performed on the filler metal following finalwelding. In
addition, PWHT of some nickel-based weld metalsoften occurs
indirectly as a consequence of stress relief heat treatments
performed on adjacent low-alloy steel components per ASME Code
requirements. In these latter cases, the stress relieftemperature
is well below that which would be optimum for nickel-based
alloys.
Le Hong et al. [19] report that CGRs are reduced by a factorof
2.0 for stress relieved specimens compared to otherwise
similaras-received specimens.
No stress relief was applied to the test welds for all of the
testspecimen crack growth rate data used directly in the
developmentof the deterministic CGR model for Alloy 82/182/132
welds in thepresent study.
III.G Crack Front Patterns
As mentioned earlier, CGR specimens for Alloy 82/182/132weld
metals often exhibit irregular crack fronts, with regions
ofnon-engagement (no SCC crack initiation) and with large
differences in the extent of SCC crack growth. Photomicrographsof
example fracture surfaces that illustrate some of the variety
ofcrack front types that can occur are shown in Figures 5 and
6.These figures reflect test specimens that are included in the
MRPdatabase. Figure 5 illustrates a case where the crack front
ismoderately well behaved, i.e., it is not very irregular. Figure 6
illustrates a case with a highly irregular crack front.
Theprocedure adopted for evaluating CGRs in the case of uneven
crack fronts is described later in Section IV.C.
Fig. 5. Example of a Fracture Surface of Alloy 182 Weld Metal
with Moderately Uniform Crack Front
515
-
Fig. 6. Example of a Fracture Surface of Alloy 182 Weld Metal
with Highly Irregular Crack Front
IV. Crack Growth Rate Testing Techniques and Consideration of
Incomplete Engagement to SCC
Key testing techniques regarding specimen loading and thetest
environment are discussed below, followed by a discussion of the
effect of incomplete engagement to stress corrosion crackgrowth
across the specimen width.
IV.A Specimen Loading
The welding specimens applied directly to develop the MRP-115
CGR model [9] reflect a number of loading variables,including
stress intensity factor, cyclic loading parameters, andweld
orientation. Test results in the database (after the
screeningprocess) reflect the following: Stress intensity factors
ranging from 19.7 to 60.0 MPam for
Alloy 182/132 specimens and from 28.0 to 56.8 MPam for Alloy 82
specimens.
A combination of purely constant loading and periodic,partial,
cyclic unloading. Cyclic loading parameters includeload ratios, R,
between 0.65 and 0.75 and hold times between3600 and 100,000 s.
Three of the six possible weld orientations (TS, LS, and TL).TS
and LS represent orientations where crack growth isparallel to the
direction of the weld dendrites, and TLrepresents an orientation
where crack growth is across(perpendicular to) the direction of
dendrite growth. During CGR testing, addressing the following
issues can
yield more consistent and meaningful test results: Application
of a stress intensity factor within accepted limits.
All of the laboratories that contributed CGR data to the MRP
database applied linear elastic fracture mechanics (LEFM)validity
criteria (e.g., ASTM E399 [16] and E647 [20]).
Use of side grooving to maintain the crack plane. Limiting large
variations in the stress intensity factor during
constant load testing. Use of periodic, partial cyclic unloading
to maintain a
straight crack front.
Appropriate procedures for precracking the specimen. A
transgranular fatigue precrack is generated first in order to
provide a sharp, linear initiation site for the PWSCC growth
mechanism. The length of the precrack, and loading
detailsassociated with generation of the precrack, must be
considered. In-situ transitioning to an intergranular crackgrowth
path is sometimes used to encourage development of a uniform SCC
crack front.
Use of continuous crack monitoring, which is a valuable toolthat
can: 1) aid in determining when SCC initiated from thefatigue
precrack, and 2) assist in estimating CGRs during different phases
of multi-condition testing. The mostcommonly used technique for
such monitoring is reversedDC potential drop.
Careful control and documentation of machining, surface
condition, and pre-oxidation in high-temperature water (e.g.,to
obtain proper corrosion potentials).
Control of test temperature, with stability ideally
withinr0.5C.
Control and monitoring of water purity and dissolved
gaschemistry.
Measurement of the corrosion potential of the CT specimenitself
and a separate platinum electrode.
Table II. Environments Used for Crack Growth Rate Tests
(MRPDatabase after Screening)
Li BTest Org. Weld ID
Tem
p.(
C)
(ppm)
Diss.H2
(cc/kg)(Note 3)
Ref.
Westinghouse D545/D582,33644,PP751
323to
342
2.0 1200 25 [21]
26B2 343,345
2.0,2.2
1202to
1273
28.8to
30.4
[22]
6892 343 2.2 1212 29.5 [9]
Studsvik
WC05F8 319 2.28 1297 29.6 [9]
Bechtel Bettis A-1,C-1, C-2, C-3, C-4
338 Note 1 50 [13]
LM 182-1 328 Note 2 35 [23]
LM 182-2 338 Note 2 40 [23]
LM 82-1 360 Note 2 40 [23]
LM 82-2 338 Note 2 20, 40 [23]
LockheedMartin KAPL
LM 82-3 316to
360Note 2
30to40
[23]
MHI MG-7,132 Heat
325 3.5 1800 30 [15]
Notes:1) High-temp. hydrogenated water with a room-temp. pH of
10.1 to 10.32) High-temp. hydrogenated water with a high-temp. pH
of 6.63) The unit for dissolved hydrogen of cc(standard temperature
and pressure (STP))/kg H2O is abbreviated as cc/kg
516
-
IV.B Test Environment
The environments for the tests that were directly used to
develop the MRP-115 CGR model are shown in Table II. All of the
environments were hydrogenated high-temperature water, often with
lithium and boric acid additions to simulate the PWR primary
coolant environment. Concentrations of impurities such as chlorides
and sulfates were limited to low levels. The temperature range
covered was 316 to 360C. Because of the sensitivity of PWSCC CGRs
to the hydrogen concentration, special care was taken to screen out
data from tests where careful control of hydrogen was not
demonstrated.
IV.C Derivation of an Appropriate Crack Growth Rate Considering
Uneven Crack Fronts
Alloy 82/182/132 weld specimens tend to exhibit uneven stress
corrosion cracks, which introduce uncertainty into crack growth
rate measurements. For specimens with small to intermediate amounts
of SCC, it is common for intergranular cracking in these materials
to incubate along the transgranular (TG) fatigue precrack front
non-uniformly versus test time, so that only portions of the crack
front may exhibit IG cracking in limited duration tests. Even when
intergranular cracks are almost fully engaged, fingers of SCC often
jut out beyond the overall crack front, thereby demonstrating the
heterogeneous nature of SCC growth in these weld metals. The
incomplete engagement and uneven crack front issues are important
because they can introduce a bias into CGR measurements. Because
these issues are complex, there is no single approach for assuring
that any bias is removed from the CGR database. Specific analysis
methods designed to account for incomplete engagement and uneven
crack fronts were proposed and debated in detail at the EPRI-MRP
CGR Expert Panel meetings. Fundamental differences in expert views
involve the underlying cause of the absence of crack incubation in
some regions of weld metal. At present, it is not known
definitively if the non-incubated regions are associated with
pinned transgranular fatigue precracks, with microstructural
regions of enhanced SCC resistance, or with a combination of both
effects. These differences are potentially important because they
govern how unengaged and shallow crack extension regions are
treated when calculating average CGR. It is noteworthy, however,
that the issue of engagement is quantitatively less important for
average SCC crack lengths of more than about 2 mm based on a
comparison of average and maximum crack extensions for CGR tests
for which both these parameters were available [9]. While no
consensus was reached concerning methods that address the
incomplete engagement and uneven crack front issues for future
testing, a consensus was reached on how to address this issue in
analyzing the existing CGR database. Specifically, there was
agreement that including the zero crack extension values within the
unengaged portion of the precrack produces non-conservative CGR
estimates. To avoid this problem, the EPRI-MRP CGR Expert Panel
agreed that it is appropriate that the average crack extension used
to compute CGR be based solely on the engaged segments of the
stress corrosion crack (i.e., a simple average of all non-zero
crack extension values across the specimen width). In addition,
data points with less than 50 percent engagement and less than 0.5
millimeters of average crack extension were excluded from the
screened database.
V. Development of the Screened MRP Database and Derivation of
CGR Disposition Curves
After a screening process was applied to the set of worldwide
laboratory CGR data for Alloy 82/182/132 collected by the MRP, a
multiple linear regression statistical model was applied in order
to derive recommended deterministic CGR disposition curves for
these weld metal materials.
V.A Screening Criteria
The starting point for screening the available stress corrosion
crack growth database for nickel-based welds Alloy 82/182/132 was
the same as that adopted for the earlier MRP-55 [2,3] study of
Alloy 600. The EPRI expert panel for PWSCC revised those screening
criteria in consideration of the issues that are particularly
relevant to weld metals, and Table III lists the key factors that
were considered during the screening process for the Alloy
82/182/132 weld CGR data. It should be noted that the main reasons
leading to exclusion of Alloy 600 data from further consideration
in the MRP-55 study were: No measurable growth. Less than 50% of
crack front with IGSCC initiated (hereafter
called engagement) or lack of crack front mapping to enable this
feature to be assessed and average growth rates to be calculated in
addition to the maximum rates supplied.
Out of specification PWR primary water chemistry (particularly
hydrogen).
Cyclic or ripple loading with less than 1 hour hold time at
constant load during each cycle.
A similar pattern emerged for the nickel-based weld metal Alloys
82/182/132 with the addition of a few instances of data rejected
because of loading beyond LEFM criteria. However, the second
criterion listed above (relating to crack engagement) assumed much
higher importance for the weld metals. Due to difficulties with
lack of uniform crack initiation from starter fatigue cracks and
the development of irregular crack fronts, an additional
requirement to that of greater than 50% engagement used here was a
minimum crack growth increment averaged across the specimen width
(aave) of at least 0.5 millimeters. A sensitivity study established
that the precise choice of the aavecutoff used as the screening
criterion did not have an arbitrary influence on the acceptable
screened data set and the eventual outcome of the data analysis
[9]. Please see Appendix A to this article for further details
regarding the screening process. A detailed treatment of hydrogen
effects, which are known to be potentially significant for crack
growth in nickel-based weld metals [23], was not possible with the
limited number of results contained in the screened database [9].
It was noted, however, that one set of KAPL data which apparently
illustrates a significant KI dependency actually resulted more from
testing at two distinct H2 levels [23].
517
-
Table III. Key Factors for Consideration in CGR Testing and Data
Reporting
1 Material within specifications including
composition/condition/heat treatment
2 Mechanical strength properties
3 ASTM specimen size criteria and degree of plastic
constraint
4 Pre-cracking technique (including straightness criteria,
plastic zone size, crack morphology)
5 Special requirements for testing welds (e.g. pre-crack
location, residual stresses/strains)
6 Environment (chemistry, temperature, electrochemical potential
(ECP), flow rate at specimen, neutron/gamma flux)
7 Loop configuration (e.g., once-through, refreshed, static
autoclave)
8 Water chemistry confirmation by analysis (e.g., Cl, SO4, O2,
Cr, total organic carbon (TOC), conductivity)
9 Active constant or cyclic loading versus constant displacement
loading (e.g., using wedge)
10 On-line measurement of crack length versus time during test
(including precision)
11 Actual crack length confirmed by destructive examination
(assessment method/mapping)
12 Appropriateness of crack characteristics (fraction SCC along
crack front, uniformity, adequate SCC increment, transgranular
portions within IGSCC fracture surface, etc.)
13 Possible effects of changes in loading or chemistry
conditions during a test (including heat up and cool down)
14 Calculation and reporting of K or K values
15 Reporting of raw a vs. t data and derivation of da/dt
values
16 Reproducibility of data under nominally identical test
conditions
V.B Development of MRP Database for Alloy 82/182/132
After the screening process described above was applied, a
multiple linear regression statistical model was applied to the MRP
database of CGR data for Alloy 82/182/132 to develop theMRP-115
deterministic CGR model. The MRP expert panel concluded that Alloy
182 and Alloy 132 can be regarded assufficiently similar to be
described by one CGR curve.
Figure 7 is a CGR versus stress intensity factor plot showingthe
complete set of available data for which average CGRs werereported,
adjusted to a common reference temperature of 325C assuming a
thermal activation energy of 130 kJ/mole (31.0 kcal/mole). This is
the same activation energy that was applied tothe CGR data for
Alloy 600 in MRP-55 [2,3]. The expert paneljudged that there were
insufficient data to develop reliableactivation energy values for
Alloy 182/132 and for Alloy 82, sothe accepted activation energy
value for Alloy 600, which has a similar composition, was used.
Multiple independent studies ofAlloy 600 have resulted in thermal
activation energy valueswithin about 1015% of the value of 130
kJ/mole (31.0kcal/mole) [2,3].
Figure 8 is the corresponding plot for the available data
forwhich CGRs based on the maximum crack increment across
thespecimen width were reported. Figures 9 and 10 show theaverage
CGR data in the MRP database following the screeningprocess: Figure
9 shows the data for Alloys 182 and 132, and Figure 10 shows the
data for Alloy 82. Note that for referencepurposes, two previously
developed CGR curves are shown in Figures 7 through 10.
V.C Data Reduction
The statistical methodology for developing the deterministicCGR
equation for Alloy 82/182/132 weld metal is describedbelow. The
procedure is similar to that presented in MRP-55[2,3] for Alloy 600
wrought material, but includes a linearizedmultiple regression
model in order to determine a best-fit stressintensity factor
exponent, , while still treating the data on a weld-by-weld
basis:1. Collect data including reported initial or average K,
CGR
based on the crack increment averaged across the entirespecimen
width, average crack increment, test temperature,and percentage
engagement of the crack front to IGSCC(%eng). Average CGR data were
used, rather than themaximum measured CGR across the specimen
width,because it is believed that the average CGR is a
bettermeasure of the fundamental material behavior, whereas
themaximum CGR is more dependent on the spatial variabilityin
resistance to PWSCC. In addition, the maximum CGR appears to be
more dependent on test duration than the average CGR [9]. Finally,
it is standard practice in fatiguetesting of CT specimens to
average the crack extensionacross the specimen width [20].
2. Perform data screening using the key factors listed inTable
III.
3. Modify the reported CGR to account for the effect of
incomplete initiation of PWSCC across the crack front bydividing by
the engagement fraction (effectively excludingzero crack extension
points from the average across thespecimen width):
'%100
CGRCGReng
(1)
518
-
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da
/ dt
(m/s
)
0
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
1mm/yr
All CGRs are reported averageCGRs and are not adjusted toaccount
for percentageengagement across the crack front,alloy type, or
crack orientation
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
Fig. 7. Complete Set of Worldwide Alloy 82/182/132 Average CGR
Data before Screening Process (144 points)
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da
/ dt
(m/s
)
0
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
1mm/yr
All CGRs are reportedmaximum CGRs and arenot adjusted to account
foralloy type or crack growthorientation
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
Fig. 8. Complete Set of Worldwide Alloy 82/182/132 Maximum CGR
Data before Screening Process (158 points)
519
-
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da/ d
t (m
/s)
0
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
1mm/yr
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
All CGRs are adjusted to accountfor percentage engagement
acrossthe crack front but not alloy typeor crack orientation
Fig. 9. Average CGR Data for Alloys 182 and 132 after MRP
Screening (43 Points)
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da/ d
t (m
/s)
0
1mm/yr
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
All CGRs are adjusted to accountfor percentage engagement
acrossthe crack front but not alloy typeor crack orientation
Fig. 10. Average CGR Data for Alloy 82 after MRP Screening (34
Points)
520
-
4. Adjust the data to a common reference temperature of 325C
using an activation energy of 130 kJ/mole (31.0 kcal/mole).
5. Assume no stress intensity factor threshold for PWSCC ofthe
weld metals (i.e., Kth = 0). The EPRI expert panel for PWSCC
concluded that, for the weld metal materials, therewere
insufficient data to justify a stress intensity factorthreshold
other than zero. See Appendix B to this article for discussion of
this assumption.
6. Assume the following form to model the set of screenedaverage
CGR data:
'weld alloy orient
temp
CGR f f f Kf
ED (2)
where = power-law constantftemp = factor adjusting CGR to common
reference
temperature of 325Cfweld = common factor applied to all
specimens
fabricated from the same weld to account forweld wire/stick heat
processing and for weld fabrication (see discussion below)
falloy = factor accounting for effect of compositiondifference
between Alloy 182/132 and Alloy 82(taken as 1.0 for Alloy
182/132)
forient = factor accounting for difference in CGR resulting from
crack growth perpendicular tothe direction of the weld dendrites
versus parallel to the direction of the dendrites (takenas 1.0 for
the parallel case)
K = crack-tip stress intensity factor = power-law exponent
7. Linearize the assumed form of the CGR equation by taking the
natural logarithm of the adjusted CGR.
8. Perform a least-squares multiple linear regression fit
treatingthe weld factor (fweld) as a normally distributed random
variable.
9. Choose the alloy factor (falloy) for Alloy 82 based on
thevalue that makes the log-mean for the set of weld factors forthe
Alloy 182/132 welds equal to the log-mean for the set ofweld
factors for the Alloy 82 welds.
10. Determine the orientation factor (forient) for crack growth
in the direction perpendicular to the weld dendrites (TL, LT,ST,
SL) versus growth parallel to the dendrites (TS, LS) based on the
best-fit from the regression model.
11. Similar to the procedure in MRP-55, base the
deterministicCGR equation on the 75th percentile of the log-normal
distribution for the 19 weld factors.
The fweld factor serves the same function as the heat
factor(fheat) applied in MRP-55 [2,3] to all specimens fabricated
fromthe same heat of Alloy 600 wrought material to account for
theeffect of material processing differences on the CGR. The
weldfactor is necessary in the statistical treatment of the CGR
data to account for the systematic biases associated with
particular testwelds. Because of material and fabrication
differences, differentwelds (of the same alloy type) will display a
range of CGRs evenwhen loading and environmental factors are
identical. In practice,the weld factor for a particular application
is not known, so asdescribed below, the 75th percentile of the
distribution ofcalculated weld factors is adopted for the
recommendeddeterministic CGR disposition equation.
Table IV. Calculated Normalization Factors for Alloy
Type(82/182/132) and Weld Heat/Processing
WeldRank Alloy
AlloyFactor
falloy(Note 1)
WeldFactor
fweld(Note 1) falloyfweld
1 182 1.00 2.17 2.172 182 1.00 2.12 2.123 132 1.00 1.70 1.704
182 1.00 1.25 1.255 182 1.00 1.15 1.156 182 1.00 0.91 0.917 132
1.00 0.89 0.898 82H 0.38 2.04 0.789 82H 0.38 2.03 0.78
10 182 1.00 0.76 0.7611 182 1.00 0.74 0.7412 82H 0.38 1.54
0.5913 82 0.38 1.32 0.5114 82H 0.38 1.32 0.5115 182 1.00 0.51
0.5116 182 1.00 0.38 0.3817 82 0.38 0.61 0.2418 82H 0.38 0.47
0.1819 82 0.38 0.31 0.12
Note:1Assuming form CGR = ftempfalloyfweldforientK1.6
The linearized multiple regression model fit to the set of
77points in the screened MRP database resulted in the following:
the set of 19 weld factors (fweld) tabulated in Table IV and
plotted in Figure 11 an alloy factor (falloy) of 1/2.6 = 0.385
for Alloy 82 an orientation factor (forient) of 0.5 for crack
growth
perpendicular to the direction of the dendrites a stress
intensity factor exponent () of 1.6 a constant factor of
9.82u10-13
Figure 11 shows the log-normal distribution fit to the set of19
weld factors. Because it is fit to the weld factors, this
distribution describes the variability in CGR due to difference
inweld wire/stick material heat processing and weld fabrication.The
75th percentile value of this distribution is a weld factor of1.49.
For the purpose of producing a single deterministic CGR model, the
75th percentile weld factor is absorbed into theconstant factor,
resulting in a value of of 1.510-12.
V.D MRP Disposition Curves
The deterministic CGR curves for Alloy 182/132 and Alloy82 are
shown in Figure 12. The MRP database indicates that theCGR for
Alloy 82 is on average 2.6 times lower than that forAlloy 182/132,
so the MRP-115 curve for Alloy 82 is 2.6 times lower than the curve
for Alloy 182/132. For crack propagationthat is clearly
perpendicular to the dendrite solidificationdirection, a factor of
2.0 lowering the CGR may be applied to the curves for Alloy 182/132
and Alloy 82.
Figures 13 and 14 show the results of the statistical analysisin
comparison with the MRP screened database for Alloys182/132 and 82,
respectively. Note that, unlike for Figures 9 and10, the data in
Figures 13 and 14 have been normalized for theeffect of crack
orientation. The raw CGRs for points for which the crack growth was
perpendicular to the dendrite direction have been increased by a
factor of 2.0.
521
-
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.1 1. 10.
Weld Factor, f weld
Cum
ulat
ive
Dis
trib
utio
nF
9 182 Welds8 82 Welds2 132 WeldsLog-Normal Fit
Weld factors for 19 welds of Alloy 82/182/132material with fit
log-normal distribution(most likely estimator), K th = 0, and best
fit E
25th Percentile
75th Percentile
Median
The Alloy 82 data have been normalized(increased) by applying a
factor of 2.61:1/f alloy = 2.61
Fig. 11. Log-Normal Fit to 19 Weld Factors for Screened MRP
Database of CGR Data for Alloy 82/182/132
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 80Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da/ d
t (m
/s)
The reference temperature for theMRP curves is 325C (617F);
therecommended thermal activationenergy for temperature
adjustmentis 130 kJ/mole (31.0 kcal/mole),the same value
recommended inMRP-55 for base metal.
1 mm/yr
MRP-115 Curve for Alloy 182/132CGR = 1.510-12K 1.6
MRP-115 Curve for Alloy 82CGR = (1.510-12/2.6)K 1.6
For crack propagation that isclearly perpendicular to
thedendrite solidification direction, afactor of 2.0 lowering the
CGRmay be applied to the curves forAlloy 182 (or 132) and Alloy
82.
MRP-55 Curve forAlloy 600 Base Metal
Laboratory testing indicates thatthe CGR for Alloy 82 is on
average2.6 times lower than that for Alloy182/132, so the MRP-115
curvefor Alloy 82 is 2.6 times lowerthan the curve for Alloy
182/132.
Fig. 12. MRP-115 Deterministic Curves for Alloy 182/132 and
Alloy 82 Weld Materials
522
-
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da/ d
t (m
/s)
0
1mm/yr
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182MRP-115 Curve for
Alloy 182/132da/dt = 1.510-12K 1.6
All CGRs are adjusted to accountfor percentage engagement
acrossthe crack front and crackorientation but not alloy type
Fig. 13. Average CGR Data for Alloys 182 and 132 after MRP
Screening (43 Points) Normalized to a Crack Orientation Parallel to
theWeld Dendrites with MRP-115 Curve for Alloy 182/132
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 8Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da
/ dt
(m/s
)
0
1mm/yr
All data adjusted to 325C (617F)using an activation energy of130
kJ/mole (31.0 kcal/mole)
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
MRP-115 Curve forAlloy 82da/dt = (1.510-12/2.6)K 1.6
All CGRs are adjusted to accountfor percentage engagement
acrossthe crack front and crackorientation but not alloy type
Fig. 14. Average CGR Data for Alloy 82 after MRP Screening (34
Points) Normalized to a Crack Orientation Parallel to the
WeldDendrites with MRP-115 Curve for Alloy 82
523
-
The mathematical form of the MRP-115 CGR curve forAlloy 182/132
at 325C (617F) is:
CGR (in m/s) = 1.510-12 K1.6 (3)(for K in MPam)
CGR (in inches/hr) = 2.4710-7 K1.6 (4)(for K in ksiin)
The general form of the MRP-115 equation is as follows:
1 1exp g alloy orientref
Qa
R T Tf f K ED
(5)
where: a = crack growth rate at temperature T in m/s (or
in/h)Qg = thermal activation energy for crack growth
= 130 kJ/mole (31.0 kcal/mole)R = universal gas constant =
8.314u10-3 kJ/mole-K
(1.103u10-3 kcal/mole-R)T = absolute operating temperature at
location of
crack, K (or R)Tref = absolute reference temperature used to
normalize data = 598.15 K (1076.67R)
= power-law constant = 1.510-12 at 325C for a in units of m/s
and K
in units of MPam (2.4710-7 at 617F for ain units of in/h and K
in units of ksiin)
falloy = 1.0 for Alloy 182 or 132 and 1/2.6 = 0.385 forAlloy
82
forient = 1.0 except 0.5 for crack propagation that is clearly
perpendicular to the dendritesolidification direction
K = crack-tip stress intensity factor, MPam (or ksiin)
= exponent = 1.6
The MRP curve may be interpreted as the mean of the upperhalf of
the distribution describing the variability in CGR due tomaterial
heat, orin this caseindividual weld. Therefore, theMRP curve
addresses the concern that welds that are moresusceptible than
average to crack initiation tend to have higherCGRs than average.
Cracking detected in operating plants would tend to be located in
components using such susceptible welds.The use of a conservative
mean CGR is consistent with thegeneral approach of Section XI of
the ASME Boiler & Pressure Vessel Code, in which mean CGRs are
assumed and allowable crack sizes are based on design loads
multiplied by load factors.
It is noted that all of the data points which were derived from1
inch CT specimens fall on or below the MRP-115 line for
Alloy182even after correction for dendrite orientation. This is
incontrast to data points derived from 0.50.6 inch CT
specimens,some of which are above the MRP-115 line. However, it is
judged that insufficiently diverse data are available to
discernwhether specimen size in fact has a significant impact
onmeasured CGRs because only the two sets of Studsvik data reflect1
inch CT specimens (while all others reflect CT specimens of 0.50.6
inches in width).
VI. Comparison of MRP Disposition Curves with Other Data
Several comparisons were made of the MRP-115 deterministic
equation to the data that were not included in thefinal screened
MRP database of laboratory CGR data, as well asto the limited
available field data. These comparisons were madein order to verify
the robustness of the MRP-115 multiple linearregression model,
given the manner in which the data screeningprocess was
implemented. These comparisons were alsoperformed to verify the
absence of any hidden effects in theoverall set of CGR data
collected. The following specificcomparisons and investigations
were performed: comparison with lab data for ex-service weld
material, comparison with available plant data for Alloy
82/182/132
weld metal based on repeat NDE crack sizing, comparison with
available laboratory data investigating the
potential effect of pH, comparison with average CGR data that
were excluded from
use in calculating the MRP-115 equation, comparison with maximum
CGR data including for tests for
which average CGRs were not available, and investigation of
effect of periodic unloading and hold time.
The first two items, and a comparison with previouslypublished
CGR curves for Alloy 182, are presented below.
VI.A Comparison with Laboratory Data Generated for RemovedPlant
Weld Material
As described in more detail in MRP-113 [24], boric acid crystal
deposits led to the discovery of a small hole in the Alloy82/182
butt weld between the low-alloy steel reactor vessel outlet nozzle
and stainless steel primary coolant pipe during the October2000
refueling outage at VC Summer. Destructive examinationsrevealed the
presence of several axial cracks, including a through-wall axial
crack extending essentially the full weld width, as wellas a short,
shallow circumferential crack in the Alloy 182 claddingthat
arrested when it reached the low-alloy steel nozzle.
Samples of both Alloy 182 butter and Alloy 82 filler
materialtaken from this hot leg safe end weld were used in a series
ofcrack growth rate tests completed by Westinghouse [25,26].
Thetest conditions included a test temperature of 325C; a simulated
primary water environment with 3.5 ppm Li, 1800 ppm B, and3035
cc/kg dissolved hydrogen; fatigue pre-cracking in air at astress
intensity factor below 15 MPam; and active loading with anominal
test stress intensity factor of either 20 or 35 MPam.
The specimens were periodically unloaded to a load equal to70%
of the full applied load (R = 0.7) in order to break any oxidesthat
might affect the accuracy of the crack growth measurements.Three
cyclic loading test phases and one constant loading testphase were
conducted on each of the two Alloy 182 and twoAlloy 82 specimens
[9]. Side grooves were included in thespecimens in an attempt to
keep cracking in the intended plane.For the Alloy 82 specimens,
testing was in the TS direction (crackplane parallel to the
dendrites) while for the Alloy 182 specimens,testing was in the TL
direction (crack growth perpendicular to thedendrites).
For each of the four samples, post-test fractography was usedto
determine the overall crack increment, and this increment
wasdivided into four parts on the basis of on-line DC potential
dropmeasurements, thereby facilitating four separate data points
foreach sample corresponding to the four test phases.
524
-
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 80Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da
/ dt
(m/s
)MRP-115 Curve for Alloy 182
MRP-115 Curve for Alloy 82
MRP-55 Curve for Alloy 600
Lab Data for Summer 182-1 and182-2 CT Samples (As Meas.)
Lab Data for Summer 182-1 and182-2 CT Samples (Adj.
forOrient.)Lab Data for Summer 82-1 and 82-2 CT Samples
The Summer data were producedat a test temperature of 325C,which
is the reference temperaturefor the MRP-115 curves for
Alloys182/132 and 82 weld metal
1 mm/yr
MRP-115 Curve for Alloy 182/132CGR = 1.510-12K 1.6
The Summer data for Alloy 182are shown with and without
anadjustment factor of 2.0 to account forcrack orientation
MRP-115 Curve for Alloy 82CGR = (1.510-12/2.6)K 1.6
Fig. 15. Comparison of MRP-115 Curves for Alloys 182/132 and 82
with Westinghouse CGR Data for Weld Material Removed from VC Summer
Reactor Hot Leg Safe End Butt Weld [25,26]
The crack growth rate data produced from these tests areplotted
in Figure 15 along with the MRP-115 curves for Alloy 182and Alloy
82. As noted in the figure, two sets of data points areincluded for
the Alloy 182 specimens: the as-measured CGRs, and CGRs increased
by a factor of 2.0 to account for the crackorientation
(perpendicular to the dendrites). After the crackorientation
correction of the CGRs for the two Alloy 182 weldsamples, these
data are still in reasonable agreement with theMRP-115 lines for
both Alloy 182/132 and Alloy 82.
The VC Summer data were screened from the database usedto
develop the MRP-115 deterministic model because theselaboratory
data were generated during multi-condition tests inwhich the
loading type was changed.
VI.B Comparison with Available Field Data
During the Ringhals Unit 3 refueling outage in 2000, two axially
oriented defects were detected in one of the reactor vesseloutlet
nozzle-to-safe-end Alloy 182 butt welds using a qualifiededdy
current technique [27,28,29].b During the 2000 outage, thedepth of
each defect was measured to be 93 mm and the length1610 mm with
ultrasonic testing (UT). After additionaloperation for
approximately 8000 effective full power hours, thefirst defect
(Crack 1) had grown to a depth of 133 mm while thesecond defect
(Crack 2) measured 163 mm, as shown in Table V.
b Note that the root pass of each of the double-V type welds at
Ringhals is reported to have been produced using Alloy 82 weld
metal. However, the reported cracks did not extend to the root
region. Hence, both cracks were located exclusively in Alloy 182
material.
The left portion of Table V lists the initial and final
crackdepths and corresponding crack extensions associated with
best-estimate, statistical upper- and lower-bound, and worst case
crackgrowth. The best-estimate case assumes that the initial and
finalcrack depths are subject to no error (or, more precisely, that
eachis subject to the same error). For example, for Crack 1, the
best-estimate initial depth, final depth, and extension are 9, 13,
and 4mm, respectively. The worst-case crack growth assumes that
theinitial and final depths are at the extreme values implied by
themeasurement uncertainty (e.g., for Crack 1, the initial
depthwould have been 93 = 6 mm, the final depth 13+3 = 16 mm,
andthe extension 166 = 10 mm). The upper and lower statistical
bounds assume that the initial and final depth measurements are
independent (i.e., that the measurement errors in each case are
notsubject to a common bias). Based on standard
engineeringtolerance stack-up assumptions, this implies that the
uncertainty inthe measurement difference is equal to 32 = 4.24 mm.
If halfof this uncertainty tolerance is assigned to both the
initial andfinal depth measurements, the values in Table V for
Stat. LowerBound and Stat. Upper Bound are obtained (e.g., initial
depthof 92.12 = 6.88 mm, final depth of 13+2.12 = 15.12 mm, and
extension of 15.126.88 = 8.24 mm for the statistical upper boundfor
Crack 1).
The stress intensity factors that apply at the locations of
theRinghals Unit 3 defects were calculated by Efsing and
Lagerstrm[27] based on stresses (including welding residual
stresses)calculated using the finite-element method and fracture
mechanics calculations assuming standard superposition assumptions
and arereported in the rightmost portion of Table V.
525
-
Table V. Data Reported for Ringhals Unit 3 Hot Leg Safe End
Nozzle Weld Cracks
AverageCGR (m/s)
Stress Intensity Factor(MPam)
Crack Statistical Case
InitialDepth
a1(mm)
FinalDepth
a2(mm)
Extensiona
(mm)
Oper.at
319C
Adjustedto
325CInitial
K1Final
K2MeanKave
Stat. Lower Bound 11.12 10.88 No Growth No Growth 32.2 32.2
32.2Best Estimate 9.0 13.0 4.0 1.4E-10 1.8E-10 29.5 33.5 31.5
Stat. Upper Bound 6.88 15.12 8.24 2.9E-10 3.7E-10 24.0 35.5
29.71
Worst Case 6.0 16.0 10.0 3.5E-10 4.5E-10 21.0 36.5 28.8Stat.
Lower Bound 11.12 13.88 2.76 9.6E-11 1.3E-10 32.3 34.6 33.5
Best Estimate 9.0 16.0 7.0 2.4E-10 3.2E-10 29.5 36.5 33.0Stat.
Upper Bound 6.88 18.12 11.24 3.9E-10 5.1E-10 24.0 38.3 31.2
2
Worst Case 6.0 19.0 13.0 4.5E-10 5.9E-10 21.0 39.5 30.3
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 80Stress Intensity Factor, K (MPam)
Cra
ck G
row
thR
ate,
da/ d
t (m
/s)
MRP-115 Curve for Alloy 182/132
MRP-115 Curve for Alloy 82
MRP-55 Curve for Alloy 600
Ringhals 3 / Crack 1 / DepthIncrease from 2000 to 2001
Ringhals 3 / Crack 2 / DepthIncrease from 2000 to 2001
1 mm/yr
MRP-115 Curve for Alloy 182/132CGR = 1.510-12K 1.6
MRP-115 Curve for Alloy 82CGR = (1.510-12/2.6)K 1.6
All curves adjusted to 325Cusing an activation energy of130
kJ/mole (31.0 kcal/mole)
The points for the Ringhals 3 hot leg safe end weld cracks are
based on thedepth measurements made in 2000 and 2001 and the stress
intensity factorscalculated by Ringhals (points shown at average of
initial and final Kcorresponding to best estimate initial and final
depths). The Ringhals datawere adjusted from the operating
temperature of 319C (606F) to thereference temperature of 325C
(617F) using the activation energy of130 kJ/mole (31.0
kcal/mole).
Fig. 16. Field Crack Growth Data for Ringhals Unit 3 Hot Leg
Safe End Alloy 182 Weld
1.E-12
1.E-11
1.E-10
1.E-09
1.E-08
0 10 20 30 40 50 60 70 80Stress Intensity Factor, K (MPam)
Cra
ck G
row
th R
ate,
da/ d
t (m
/s)
MRP-115 Curve for Alloy 182
Ringhals Two-Part Curve(Adjusted to 325C)
EDF Alloy 182 Curve
MRP-21 Curve for Alloy 182
MRP-55 Curve for Alloy 600
All curves adjusted to 325C (617F)using an activation energy
of130 kJ/mole (31.0 kcal/mole)
1 mm/yr
MRP-55 Curvefor Alloy 600
MRP-21 Curvefor Alloy 182
MRP-115 Curve forAlloy 182CGR = 1.510-12K 1.6
Ringhals Two-Part Curve @325C
EDF Alloy 182 Curve
Fig. 17. Comparison of MRP-115 Curve for Alloy 182 WeldMetal
with Other Disposition Curves
Figure 16 shows the Ringhals Unit 3 field data plotted with the
deterministic MRP-115 curve for Alloy 182 weld metal. Thesolid and
open squares in Figure 16 represent the best-estimatecrack growth
rates for the depth increase of the two Ringhals Unit 3 cracks.
These two points have been adjusted to thereference temperature of
325C (617F) using the standardthermal activation energy of 130
kJ/mole (31.0 kcal/mole), and they reflect the stress intensity
factors calculated by Efsing andLagerstrm [27]. (The points are
shown at the average of theinitial and final stress intensity
factors corresponding to the best-estimate initial and final
measured depths.) The tolerance bars on the points illustrate the
uncertainty in the average crack growthbetween the two UT
inspections based on the statistical tolerancefor the crack
extension discussed above (4.2 mm). Note that thestatistical lower
bound crack growth rate for Crack 1 correspondsto no growth because
the 4.2 millimeters tolerance is greaterthan the best-estimate
extension for this crack.
VI.C Comparison with Other Published Curves for Alloy 182
Figure 17 compares the MRP-115 curve for Alloy 182/132with the
following four deterministic curves: The two-part curve developed
by Ringhals [28]. The plateau curve published by EDF based on
laboratory test
data [19,30].
The MRP-21 curve that has previously been applied in theU.S.
based on a limited set of laboratory CGR data [6,7].
The MRP-55 curve for thick-wall Alloy 600 material [2,3].Like
the MRP-115 curve, the MRP-55 equation for Alloy
600 was derived using a multiple regression statistical fit
(basedon a heat-by-heat treatment of the data). Unlike the MRP-115
curve, however, the MRP-55 curve assumes a threshold
stressintensity factor value of 9 MPam and also uses Scotts value
for the exponent (1.16) based on field data for Alloy 600 steam
generator tubes [8] rather than letting the exponent be
determinedby the statistics associated with the regression fit to
the set of available laboratory data.
Examination of the five curves in Figure 17 and theunderlying
data leads to the following observations: The MRP-115 curve is
based on a worldwide database of
CGR measurements for both Alloy 182/132 and Alloy 82from
numerous laboratories. CGRs based on crack extensionaveraged over
the specimen width (excluding any segmentswith zero crack
extension) were used in the statistical modelthat yielded the
MRP-115 curve.
The MRP-115 curve is about 25% lower than the MRP-21 curve for
stress intensity factors greater than about20 MPam. At stress
intensity factors less than 15 MPam,the MRP-115 curve is
higher.
526
-
The MRP-115 curve is nearly parallel to, and about fourtimes
higher than, the MRP-55 curve for stress intensityfactors greater
than 20 MPam.
The MRP-115 curve crosses the Ringhals curve at about22 MPam and
again at 49 MPam. For stress intensityfactors outside this range,
the MRP-115 curve is higher.
Similarly, the MRP-115 curve crosses the EDF curve atabout 9
MPam and again at 27 MPam. For stress intensityfactors outside this
range, the MRP-115 curve is higher.
VII. Example Application
Now that the crack growth model has been developed, it is
helpful to illustrate its application to a typical geometry
whereflaws have been found in the field. Subsequent to this
example, conclusions are presented regarding a series of
examplesinvestigating the effect of the new MRP-115 model with no
stressintensity factor threshold, compared to the previous model
ofMRP-21 [6,7], which had a threshold.
Before proceeding to the example calculations, the
followinggeneral steps constitute a deterministic crack growth
evaluation(additional guidance on the overall approach is provided
inSection XI of the ASME Code [31]): Calculate the stress field in
the region of interest including
the effect of welding residual stresses and normal
operatingstresses. Either a conventional
strength-of-materialsapproach or, alternatively, finite-element
analysis (FEA) can be used to determine the stresses. Use of FEA is
normallyrequired if there are weld repairs to the inside and/or
outsideweld surfaces (see, e.g., MRP-106 [32]).
Determine the stress intensity factor K that corresponds tothe
postulated weld geometry as a function of crack size.References
[33], [34], and [35] provide standard Kexpressions from LEFM that
are often applied to calculatestress intensity factors from the
corresponding stress field.These standard K expressions are based
on LEFMsuperposition assumptions, so they do not take any credit
forrelaxation of the residual stress field as the crack grows.
Choose an initial flaw size based on the size crack that is
detected in the field, the detectability limit for a particulartype
of inspection, or another criterion such as the size crackthat
results in a CGR of engineering significance. Choose afinal crack
size based on criteria such as the size crack thatproduces coolant
leakage, the allowable crack size forcontinued service, or the
critical crack size for pressureboundary rupture. Typically, an
assumption also is made regarding the flaw aspect ratio (length vs.
depth) during the growth process.
Calculate the time for crack growth by integrating a
deterministic CGR equation such as the MRP-115 equation(Equation 5)
for the variable K as a function of crack size. Typically, the
normal operating temperature is assumed, and the number of points
in the numerical integration is selectedto be large enough so that
the result is insensitive to the step size.
VII.A Example Application: PWR Piping Butt Welds
The location chosen for this example is the reactor vesseloutlet
nozzle-to-safe-end weld, corresponding to a plant designedby
Westinghouse, where the nozzle is low-alloy steel and the
piping is stainless steel. The stainless steel safe end is
welded tothe pipe in the field in most applications, and is
connected to thenozzle with an Alloy 182 weld, as shown in Figure
18.
In this example, a flaw is postulated in the Alloy 182
weldmaterial, oriented circumferentially, with a range of aspect
ratios.The calculations discussed here have considered all
theappropriate loadings, including dead weight, thermal
expansion,welding residual stress, and pressure. Since PWSCC is a
long-term phenomenon controlled by steady-state stresses, seismic
loads and thermal transient loads are not included.
Standardresidual stress distributions were assumed from expressions
developed in the 1980s based on residual stress measurements for
test mockups of BWR piping butt welds [36].c Details regardinghow
the strength-of-materials approach may be applied todetermine
stresses for piping butt welds are provided in MRP-109 [38] and
MRP-112 [39].
Results were generated for the example location using the
MRP-115 CGR model (Equation 5) to determine the time for anassumed
initial part-depth, circumferential flaw to grow in thethrough-wall
direction to a depth of 75% of the wall thickness, for a range of
crack aspect ratios (Figure 19). The nominal outsidediameter of the
pipe is 30 inches, and the wall thickness is2.6 inches. The
temperature used for the example was 617F (325C), so no temperature
adjustment was required when applying Equation 5.
Fig. 18. Geometry of Weld Region Used for the Crack
GrowthIllustration (Reactor Vessel Outlet Nozzle-to-Safe-End
Weld)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 60 120 180 240 300
time (months)
a/t (f
law
dep
th/w
all
thic
knes
s)
T=617 F
AR=3AR=6
AR=10
AR=2
MRP-115 CGR Model
AR = Aspect Ratio= flaw length/flaw depth
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 60 120 180 240 300
time (months)
a/t
(fla
w d
epth
/wal
l th
ickn
ess)
T=617 F
AR=3AR=6
AR=10
AR=2
MRP-115 CGR Model
AR = Aspect Ratio= flaw length/flaw depth
Fig. 19. Results of Sample Calculations for a Range of Flaw
Shapes: Time for Through-Wall Growth for a Part-Depth
Circumferential Flaw at a Reactor Vessel Outlet Nozzle Safe
EndWeld (Including Residual Stress)
c Although many BWR piping butt welds are substantially thinner
than PWR piping butt welds of corresponding diameter,
finite-elementcalculations of welding residual stresses in PWR
piping butt welds (in the absence of weld repairs) [37] indicate
that the standard residual stressdistributions are generally
conservatively high.
527
-
Note that the crack growth is most rapid for the largest aspect
ratio, 10, which corresponds to a flaw with length 10 times its
depth. The assumed initial flaw depth was about 2% of the wall
thickness, or 0.05 inches. The assumption of a shallower flaw would
have resulted in a longer growth time, but the goal here was to
illustrate how a realistic flaw would grow. As can be seen in the
figure, the assumed initial flaw size can have a significant effect
on the result.
VII.B Effects of a Stress Intensity Factor Threshold
Assumption
The MRP-115 model contains no stress intensity factor threshold
for crack growth, because no basis could be found for the existence
of one in the MRP database, which includes no data for crack-tip
stress intensity factors less than 19.7 MPam. As such, it differs
from the previous work reported in MRP-21 [6,7], where a threshold
used for Alloy 600 was assumed to apply to the weld metal as well.
Calculations were carried out to examine the effect of the stress
intensity factor threshold assumption at various PWR piping butt
weld locations [9]. The effect of the elimination of the stress
intensity factor threshold on crack growth through the thickness of
the weld was in each case detrimental, in that the time for a flaw
to propagate through the wall was shorter. Once a flaw was through
the wall and leaking, the time required for it to reach a critical
circumferential length was actually calculated to be somewhat
longer with the MRP-115 model, although the difference was
sometimes small. This behavior is the result of the MRP-115 CGR
curve being higher than the MRP-21 curve for low stress intensity
factors but about 25% lower than the MRP-21 curve for stress
intensity factors greater than about 20 MPam (see Figure 17), in
combination with residual stresses that often become compressive
near the center wall region of a thick weldment. These negative
stresses tend to produce low stress intensity factors for
part-depth flaws that are propagating through the wall. On balance,
it may be concluded that the use of a model with no stress
intensity factor threshold value is conservative for the PWR piping
butt weld application. In addition to this advantage, this approach
is more strongly based technically, as discussed in Appendix B to
this article.
VIII. Summary and Conclusions
The following are the key conclusions regarding the MRP study of
stress corrosion CGRs of Alloy 82/182/132 nickel-based weld metals
under PWR primary water conditions: An international expert panel
was formed and collected
detailed laboratory test data for the relevant set of worldwide
laboratory CGR tests using pre-cracked fracture mechanics
specimens.
The expert panel developed screening criteria to qualify data
for use in the development of a deterministic CGR model for Alloy
82, 182, and 132 weld metals. The screening criteria were based
upon the criteria previously applied to Alloy 600 wrought material
in MRP-55 [2,3], but were necessarily extended to cover the special
test considerations associated with the weld metal materials.
Based on a literature review and the laboratory experience of
the expert panel members, a methodology was developed for
considering the potentially non-conservative effect of incomplete
engagement to intergranular SCC across the
specimen width and over test duration. Engagement fractions were
estimated for all the specimens in the screened database, and, in
the case of incomplete engagement, the reported CGRs were adjusted
by dividing by the respective engagement fractions. This approach
is appropriate regardless of whether the incomplete engagement is
caused by isolated islands of more crack-resistant material or is a
testing artifact due to the difficulty of the crack transitioning
from the transgranular fatigue pre-crack to the intergranular
stress corrosion crackor a combination of the two.
The expert panel concluded that there are currently insufficient
data available to include a stress intensity factor threshold in
the deterministic CGR model for the nickel-based weld metals.
Analyses of weld metal cracking that involve the existence of
pre-existing defects (either real or postulated) could be strongly
influenced by assuming an arbitrary stress intensity factor
threshold value.
A linearized, multiple regression statistical model was fitted
to the screened database including an Arrhenius temperature
correction, an alloy factor (Alloy 182/132 or Alloy 82), a crack
orientation factor (parallel or perpendicular to the weld
dendrites), a crack-tip stress intensity factor exponent, and a
weld factor that accounts for the randomness associated with the
heat of weld wire/stick material and welding process. Insufficient
data were available to include dissolved hydrogen concentration (or
electrochemical potential), cold working, post-weld heat treatment
stress relief, or loading type (constant or periodic unloading) in
the model.
For the purpose of producing a single deterministic CGR model,
the 75th percentile weld factor was absorbed into the statistical
model. The MRP recommends that Equation 5 be applied for the
disposition of PWSCC flaws detected in Alloy 182/132 and Alloy 82
in PWR primary circuits (analogous to [4,40]) and used in safety
case calculations that assume hypothetical PWSCC flaws [38,39].
Furthermore, data such as those in Table IV and Figure 11 may be
used to determine statistical CGR distributions for use in
probabilistic fracture mechanics models of the growth of PWSCC
flaws in the weld metal materials.
Detailed comparisons with the available worldwide laboratory CGR
data that were not included in the final screened database used to
produce the MRP-115 deterministic model were performed. These
comparisons verified the robustness of the MRP-115 multiple linear
regression model, given the manner in which the data screening
process was implemented, and verified the absence of any hidden
effects in the overall set of CGR data collected.
Evaluation of the only known set of repeat PWSCC crack sizing
data for nickel-based weld metals in an operating PWR plant (2
cracks in Alloy 182 reactor vessel outlet nozzle-to-safe-end weld
at the Swedish plant Ringhals Unit 3) produced best-estimate CGRs
bounded by the MRP-115 curve for Alloy 182/132, as shown in Figure
16.
In other countries, different approaches have been applied to
develop CGR disposition curves for the nickel-based weld metals,
resulting, as would be expected, in CGR curves somewhat different
than the MRP-115 model (see Figure 17).
The MRP-115 equation (Equation 5) was applied to calculate the
time for flaws in piping butt weldments to grow to larger sizes. As
expected, the assumption of no stress intensity
528
-
factor threshold has a significant effect for relatively small
part-depth flaws.
Acknowledgements
This paper is a summary of work sponsored by EPRI on behalf of
the Alloy 600 Issue Task Group d of the Materials Reliability
Program [9]. Valuable input and review comments were received from
many sources, but the authors wish to express their special
gratitude to P. Andresen, S. Attanasio, W. Bamford, W. Cullen, J.
Daret, P. Efsing, S. Fyfitch, R. Jacko, C. Jansson, A. Jenssen, A.
McIlree, W. Mills, R. Pathania, P. Scott, W. Shack, T. Yonezawa,
and K. Yoon.
Appendix A: Reasons for Data Exclusion
The original set of worldwide laboratory CGR data collected by
the MRP comprised 261 individual data points. The technical issues
that were addressed by the screening process are listed in Table
III. In practice, 184 data points were excluded from the
statistical evaluation for the following objective reasons:
reported CGR based only on the maximum crack increment
along the crack front because the MRP data reduction was based
on the average crack extension (95 points),
no measurable crack growth (24 points), less than 0.5 mm of
crack extension averaged along the crack
front (23 points), hold time less than 1 hour for periodic
unloading tests
(18 points), complex loading changes during the test (16
points), hydrogen concentration outside standard plant range
(12 points with 150 cc/kg), loading exceeding the nominal linear
elastic fracture
mechanics (LEFM) limit (9 points), engagement to intergranular
(i.e., stress corrosion) cracking
along less than 50% of the crack front (4 points), flutter
loading (2 points), and temperature change during the test (1
point). For 20 of the excluded data points, two of the above
reasons applied. Note, however, that the data excluded because only
maximum CGRs were available were not evaluated for compliance with
the screening criteria regarding minimum average crack extension
and minimum engagement to intergranular cracking and might have
been affected by these further considerations.
Appendix B: Assumption of Zero Stress Intensity Factor Threshold
for the Alloy 82/182/132 Weld Metals
For the CGR equation developed for the weld metal materials, it
was concluded that insufficient data exist to justify a stress
intensity factor threshold other than zero. This appendix briefly
describes the factors taken into consideration by the expert panel
in this recommendation. Note that in MRP-55 [2,3] a threshold KISCC
of 9 MPamwas assumed as a curve-fitting parameter for Alloy 600
based on the arrest of cracks in Alloy 600 steam generator tubes
once stress intensity factor values decreased to about 9 MPam [8],
and MRP-55 cautions that use of its equation for Alloy 600 at
low
d Chairman: C. Harrington; EPRI project manager: C. King.
crack-tip stress intensity factors (< approximately 15 MPam
(14 ksiin)) involves assumptions not currently substantiated by
actual CGR data for CRDM nozzle materials. First, it must be
recognized that the threshold stress intensity factor, KISCC, is a
concept that is difficult to implement as a practical engineering
tool for PWSCC of nickel-based alloys. Because SCC is a
time-dependent process, KISCC is not an absolute material property,
but depends on test procedure and duration. KISCC is sometimes
designated at an arbitrary, slow, selected crack growth rate, but
slow crack growth over long periods can, however, be significant.
Furthermore, in nearly all practical cases, SCC initiates in
circumstances where linear elastic fracture mechanics (LEFM) cannot
be applied, and consequently the initiation of short cracks
involves physical processes different from those responsible for
the arrest of longer cracks. No data in the MRP database following
the screening process are available for Alloys 182/132 at stress
intensity factors less than about 20 MPam and for Alloy 82 at
stress intensity factors less than about 27 MPam. In addition, for
the weld metals, no comparable field data to those for Alloy 600
steam generator tubes are available that might allow such a
threshold to be reasonably estimated. Moreover, analyses of weld
metal cracking that involve the existence of pre-existing defects
(either real or postulated) could be strongly influenced by
assuming an arbitrary KISCC. Finally, trial fits of the laboratory
data with and without an imposed threshold did not support the
assumption of a particular KISCC value. Based on these
considerations, the MRP-115 CGR equation for the nickel-based weld
metals does not take credit for a stress intensity factor threshold
greater than zero. This conservative approach has been adopted
until data become available specific to the weld metals that
justify assuming a KISCC threshold.
References
1. W. Bamford and J. Hall, A Review of Alloy 600 Cracking in
Operating Nuclear Plants: Historical Experience and Future Trends,
Proceedings of 11th International Conference on Environmental
Degradation of Materials in Nuclear Power SystemsWater Reactors,
ANS, 2003, pp. 10711081.
2. Materials Reliability Program (MRP) Crack Growth Rates for
Evaluating Primary Water Stress Corrosion Cracking (PWSCC) of
Thick-Wall Alloy 600 Materials (MRP-55NP) Revision 1, EPRI, Palo
Alto, CA: 2002. 1006695-NP. NRCADAMS Accession No. ML023010510.
3. G. A. White, J. Hickling, and L. K. Mathews, Crack Growth
Rates for Evaluating PWSCC of Thick-Wall Alloy 600 Material,
Proceedings of 11th International Conference on Environmental
Degradation of Materials in Nuclear Power SystemsWater Reactors,
ANS, 2003, pp. 166179.
4. Paragraph IWB-3660, Evaluation Procedure and Acceptance
Criteria for PWR Reactor Vessel Upper Head Penetration Nozzles,
ASME Boiler & Pressure Vessel Code, Section XI, ASME
International, July 2004 Edition.
5. R. S. Pathania, A. R. McIlree, and J. Hickling, Overview of
Primary Water Cracking of Alloys 182/82 in PWRs, Fontevraud V
International Symposium on Contribution of Materials Investigation
to the Resolution of Problems Encountered in Pressurized Water
Reactors, SFEN, 2002, pp. 1327.
529
-
6. Crack Growth of Alloy 182 Weld Metal in PWR Environments
(PWRMRP-21), EPRI, Palo Alto, CA: 2000. 1000037.
7. W. H. Bamford, J. Foster, K. R. Hsu, L. Tunon-Sanjur, and A.
McIlree, Alloy 182 Weld Crack Growth, and its Impact on
Service-Induced Cracking in Operating PWR Plant Piping, Proceedings
of 10th International Conference on Environmental Degradation of
Materials in Nuclear Power SystemsWater Reactors, NACE
International, 2002.
8. P. M. Scott, An Analysis of Primary Water Stress Corrosion
Cracking in PWR Steam Generators, Presented at NEA/CSNI Specialist
Meeting on Operating Experience with Steam Generators, Brussels,
Belgium, September 1620, 1991.
9. Materials Reliability Program Crack Growth Rates for
Evaluating Primary Water Stress Corrosion Cracking (PWSCC) of Alloy
82, 182, and 132 Welds (MRP-115),EPRI, Palo Alto, CA: 2004.
1006696. Nonproprietary version (MRP-115NP, 1006696-NP) planned to
be made public through submittal to the US NRC.
10. J. C. Lippold and A. J. Ramirez, Elevated Temperature Grain
Boundary Embrittlement and Ductility-Dip Cracking of Nickel-Base
Weld Metals, USNRC-ANL Conference on Vessel Penetration Inspection,
Cracking, and Repairs,September 29October 2, 2003, Gaithersburg,
Maryland.
11. P. M. Scott, et al., Examination of Stress Corrosion Cracks
in Alloy 182 Weld Metal after Exposure to PWR Primary Water,
Proceedings of 12th International Conference on Environmental
Degradation of Materials in Nuclear Power SystemsWater Reactors,
The Minerals, Metals & Materials Society, 2005.
12. D. Buisine, et al., PWSCC Resistance of Nickel B