NUREG-1061 Vol.1 PDF

"0 NOT^^^weiH
NUREG-1061
Volume 1
Report of the
U.S. Nuclear Regulatory Commission
Piping Review Committee
Ifrvsstigation and Evaluation of Stress Corrosion Cracking
in Piping of Boiling Water Reactor Plants
U.S. Nuclear Regulatory

Commission
Prepared by The Pipe Crack Task Group
o
o
S
_S
*****
DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED
DISCLAIMER
This report was prepared as an account of work sponsored by an
agency of the United States Government. Neither the United States
Government nor any agency Thereof, nor any of their employees,
makes any warranty, express or implied, or assumes any legal
liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its endorsement,
recommendation, or favoring by the United States Government or any
agency thereof. The views and opinions of authors expressed herein
do not necessarily state or reflect those of the United States
Government or any agency thereof.
DISCLAIMER
Portions of this document may be illegible in
electronic image products. Images are produced
from the best available original document.
NOTICE
Availability of Reference Materials Cited in NRC Publications
Most documents cited in NRC publications will be available from one of the follovt/ingsources:
1. The NRC Public Document Room, 1717 H Street, N.W.
Washington, DC 20555
2. The NRC/GPO Sales Program, U.S. Nuclear Regulatory Commission,
3. The National Technical Information Service, Springfield, VA 22161
Although the listing that follows represents the majority of documents cited in NRC publications,
it is not intended to be exhaustive.
Referenced documents available for inspection and copying for a fee from the NRC Public Document Room include NRC correspondence and internal NRC memoranda; NRC Office of Inspection
and Enforcement bulletins, circulars, information notices, inspection and investigation notices;
Licensee Event Reports; vendor reports and correspondence; Commission papers; and applicant and
licensee documents and correspondence.
The following documents in the NUREG series are available for purchase from the NRC/GPO Sales
Program: formal NRC staff and contractor reports, NRC-sponsored conference proceedings, and
NRC booklets and brochures. Also available are Regulatory Guides, NRC regulations in the Code of
Federal Regulations, and Nuclear Regulatory Commission Issuances.
Documents available from the National Technical Information Service include NUREG series
reports and technical reports prepared by other federal agencies and reports prepared by the Atomic
Energy Commission, forerunner agency to the Nuclear Regulatory Commission.
Documents available from public and special technical libraries include all open literature items,
such as books, journal and periodical articles, and transactions. Federal Register notices, federal and
state legislation, and congressional reports can usually be obtained from these libraries.
Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference
proceedings are available for purchase from the organization sponsoring the publication cited.
Single copies of NRC draft reports are available free, to the extent of supply, upon written request
to the Division of Technical Information and Document Control, U.S. Nuclear Regulatory Commission, Washington, DC 20555.
Copies of industry codes and standards used in a substantive manner in the NRC regulatory process
are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, and are available
there for reference use by the public. Codes and standards are usually copyrighted and may be
purchased from the originating organization or, if they are American National Standards, from the
American National Standards Institute, 1430 Broadway, New York, NY 10018.
GPO Printed copy price:
' "^!^'^0"^M
CQVEH
$8.50
13UREG~1061-Vol.l
TI85 900069
Report of the
Piping Review Committee
Investigation and Evaluation of Stress Corrosion Cracking
in Piping of Boiling Water Reactor Plants
Manuscript Completed: August 1984

Date Published: August 1984
Prepared by
The Pipe Crack Task Group

I Washington, D.C. 20555
It has been reprodaeetf from the best

|avatfable cop# to permit the broadest
'possiiiSe availabilitf.
HT IS UHUWITED
DISTRIBUTION OF THIS mm
PIPE CRACK TASK GROUP MEMBERS

Spencer H. Bush, Chairman
Richard A. Becker
Ching-Yao Cheng
William J . Collins
B i l l y R. Crowley
Jacque P. Durr
Warren S. Hazel ton
P h i l l i p R. Matthews
Joseph Muscara
Richard C. Robinson
Jack Strosnider
TABLE OF CONTENTS
Page
FOREWORD
Ix
ACKNOWLEDGMENT
xi
EXECUTIVE SUMMARY
1.0
INTRODUCTION
1-1
1.1
1.2
1.3
1-1
1-3
1.4
1.5
2.0
3.0
4.0
PURPOSE
REVIEW OF PRESSURIZED WATER REACTORS
DOCUMENTS POTENTIALLY AFFECTED BY RECOMMENDATIONS
IN THIS REPORT
THE TASK GROUP ON IGSCC
REFERENCES
1-5
1-7
1-8
CAUSES AND DESCRIPTIONS OF IGSCC PHENOMENA
2-1
2.1
2.2
2.3
2-4
2-7
2-7
MECHANISMS OF IGSCC
CONCLUSIONS
RECOMMENDATIONS
PRESENT STATUS OF PIPE CRACKING IN OPERATING BWRs
3-1
3.1
3.2
3.3
3.4
3-1
3-3
3-5
3-8
BACKGROUND
INSPECTION RESULTS
FOREIGN EXPERIENCE
CONCLUSIONS
NONDESTRUCTIVE EVALUATION OF PIPING WELDS
4-1
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
4-1
4-2
4-7
4-10
4-16
4-20
4-24
4-29
4-32
INTRODUCTION
CURRENT ULTRASONIC EXAMINATION REQUIREMENTS
TECHNICAL PROBLEMS OF ULTRASONIC INSPECTION
FIELD EXPERIENCE
ROUND ROBIN EXPERIENCE
RECENT IMPROVEMENTS IN ULTRASONIC INSPECTION
PERSONNEL AND PROCEDURE QUALIFICATION
AREAS OF RESEARCH
FOREIGN EXPERIENCE
LEAK DETECTION IN LWRs: REVIEW
AND RECOMMENDATIONS
4.11 CONCLUSIONS AND RECOMMENDATIONS
4.12 REFERENCES
iii
4-40
4-51
4-55
TABLE OF CONTENTS (Cont'd)

Page
5.0
6.0
7.0
8.0
9.0
INSPECTION OF PIPING FOR IGSCC
5-1
5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5-1
5-3
5-7
5-11
5-13
5-14
5-15
5-17
INTRODUCTION
INSPECTION RECOMMENDATIONS
INSPECTION PERSONNEL
PLANT INSPECTIONS
FOREIGN EXPERIENCE
CONCLUSIONS
RECOMMENDATIONS
REFERENCES
DECISION AND CRITERIA FOR REPACEMENT, REPAIR, OR

CONTINUED OPERATION WITHOUT REPAIR
6-1
6.1
6.2
6.3
6.4
6.5
6.6
6.7
6-1
6-5
6-36
6-38
6-63
6-72
6-74
INTRODUCTION
FLAW EVALUATION AND LEAK-BEFORE-BREAK
SHORT-TERM SOLUTIONS
LONG-TERM SOLUTIONS
CONCLUSIONS AND RECOMMENDATIONS
SUMMARY
REFERENCES
IGSCC RISK STUDIES
7-1
7.1
7.2
7.3
7.4
7-1
7-2
7-5
7-7
INTRODUCTION
LOSS-OF-COOLANT ACCIDENT (LOCA) RISK
SYSTEMS RISK EVALUATONS
INTERGRANULAR STRESS-CORROSION CRACKING (IGSCC)

VALUE-IMPACT ASSESSMENT
8-1
8.1
8.2
8.3
8-1
8-2
8-7
INTRODUCTION
SUMMARY OF PROBLEM AND PROPOSED SOLUTION
VALUE-IMPACT ASSESSMENT CONCLUSIONS
SUMMARY
APPENDIX
APPENDIX
APPENDIX
APPENDIX
APPENDIX
APPENDIX
A
B
C
D
E
F
9-1
LIST OF TASK GROUP MEMBERS

LIST OF CONSULTANTS AND MAJOR CONTRIBUTORS
10 CFR 50.54 AND 50.59 POSITIONS AND REQUIREMENTS
INTERGRANULAR STRESS-CORROSION CRACKING RISK STUDIES...
FRACTURE MECHANICS EVALUATION OF
BWR RECIRCULATION PIPING
APPENDIX G - LONG-TERM ISSUES
APPENDIX H - REPORT OF INTERNATIONAL REVIEW TEAM
APPENDIX I - CONCLUSIONS AND RECOMMENDATIONS FROM NUREG-0531
iv
A-1
B-1
C-1
D-1
E-1
F-1
G-1
H-1
I-l
LIST OF TABLES
Page
Table 3.1
Table 4.1
Table 5.1
Table 6.1
Table 6.2
Table 6.3
Table 7.1
Table 7.2
Summary of All Inspection Findings on Large

Piping in All Operating BWRs Inspected According
to lEB 82-03 and 83-02
3-4
Capabilities of Leakage Monitoring Methods

(From Reference 4.10)
4-42
Condition and Inspection Required for Weld

Categories
5-8
Results of Fracture Mechanics Analyses for 22and 28-inch-diameter Pipes with Surface Flaws
50% of the Circumference in Length
6-25
Results of Surface Crack Analyses for 12-inch-diameter

Pipe
6-25
Recommended Measurements for Controlling IGSCC in

BWR Piping
6-73
Sensitivity of Core Melt Frequency for BWR PRAs for

a Postulated Increase (lOx) in LOCA Initiator
Frequency
7-3
Risk Worth for Safety Systems* with Respect to Core

Melt Frequency
7-6
Table 8.1
Value-Impact Summary Pipe Replacement
8-3
Table 8.2
Value-Impact Summary IHSI
8-4
Table 8.3
Value-Impact Summary Augmented Inspection and Repair

with HWC
Value-Impact Summary Augmented Inspection and Repair
No HWC
8-6
Sensitivity of Core Melt Frequency for BWR PRAs for

a Postulated Increase (lOx) in LOCA Initiator
Frequency
D-4
Table D.2
RSS Pipe Failure Assessed Values
D-8
Table D.3
Hypothetical Failure Frequency Estimates
D-11
Table 8.4
Table D.l
8-5
LIST OF TABLES (Cont'd)

Pagei
Table D.4
Table D.5
Table D.6
Fatigue and Corrosion Failures as a Function of

Total Reported Failures
D-16
List of Typical Piping Systems Involved in the

Operation and/or Safe Shutdown of BWRs
D-23
Risk Worth for Safety Systems with Respect to Core

Melt Frequency
D-32
Table D.7
ECCS Single-Valve-Failure Analysis
D-34
Table E.l
Average Operating BWR Plant
E-2
Table E.2
UT Inspection Summary Data
E-12
Table E.3
Weld Overlay Summary Data
E-13
Table E.4
Pipe Replacement Summary Data
E-14
Table E.5
IHSI Summary Data
E-15
Table E.6
HWC Summary Data
E-16
Table F.l
Recirculation System Pipe Dimensions
F-16
Table F.2
Recirculation System Typical and Bounding Load

Definition
Detectable Leakage and Makeup Capacity Size
Throughwall Flaws for Typical and Bounding Loading..
Table F.3
Table F.4
F-17
F-18
Tensile Properties for Stainless Steel Weld

Materials, Temperature = 550F
F-19
Jjc Values at 550F for Submerged Arc and Shielded

Manual Metal Arc Weld Metals
F-20
Instability Moment for Bounding Load SLA Leakage

Size (10 gpm) Cracks
F-23
Table F.7
Crack Size for Instability at SLA + ISSE Load
F-24
Table F.8
Estimated Margin Against Unstable Crack Extension

for Part-Through Flaws up to 75% of Wall for Submerged Arc and Tungsten Inert Gas Arc Welds
F-26
Margin Against Instability for Total Load
F-27
Table F.5
Table F.6
Table F.9
vi
LIST OF FIGURES
Page
Figure 4.1
Allowed limits on reactor coolant system leakage

from technical specifications for 74 plants for
a) unidentified leakage and b) for total leakage
4-58
Histograms based on the technical specifications for

74 plants and the number of leakage monitor systems
versus a) system check interval (hours) and b) system
calibration or functional-test Interval
4-59
Figure 6.1
Stress-corrosion crack growth rates
6-10
Figure 6.2
Throughwall distribution of axial residual stress in

large-diameter weldments
6-13
Throughwall distribution of axial residual stress in a

large-diameter weldment
6-13
Assumed throughwall welding residual stress distribution

in small-diameter weldments (<12 in.)
6-14
Comparison of elastic-plastic solutions for throughwall

axial stresses with simple elastic superposition, 5-ksi
applied load
6-16

axial stresses with simple elastic superposition, 18-ksl
applied load
6-16

axial stresses with simple elastic superposition, 26-ksi
applied load
6-16
Collapse curve for a lO-inch pipe with throughwall and

part-through 360 cracks with bounds for crack sizes for
5 gpm. Leak rate under an applied stress of 10-ksi
6-30
Collapse curve for a 24-inch pipe with throughwall and

6-30
Figure 6.10 Collapse curve for a 10-inch pipe with throughwall and
6-31
Figure 6.11 Collapse curve for a 24-inch pipe with throughwall and
6-31
Figure 4.2
Figure 6.3
Figure 6.4
Figure 6.5
Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
vii
LIST OF FIGURES (Cont'd)

Page
Figure 6.12 Conservative estimate of upper-bound stress-corrosion
crack growth rates of Type 304 stainless steel in BWR
environments
Figure D.l
6-32
Leak rate as a function of crack opening area or pipe

diameter (assuming a full circumferential break)
D-24
Figure D.2
Reactor core isolation cooling system
D-26
Figure D.3
Residual heat removal system
D-26
Figure D.4
Residual heat removal system, shutdown cooling function..
D-27
Figure D.5
Residual heat removal system, steam condensing function..
D-27
Figure D.6
Residual heat removal system, pressure coolant injection

function
D-28
Figure D.7
High pressure core spray system
D-28
Figure F.l
Figure F.3
Stability of a part-through circumferential crack under

fully plastic bending or axial loads
Through and part-through geometries used in fracture
mechanics analyses
Illustration of stability assessment diagram
F-32
F-33
Figure F.4
Tungsten inert gas welds, 550F
F-33
Figure F.5
J-R curve for shielded metal arc weld, 550F (Ref. F.23).
F-34
Figure F.6
J-R curve for submerged arc weld, 550F (Ref. F.22)
F-34
Figure F.2
viii
F-32
FOREWORD
The Executive Director for Operations of the U. S. Nuclear Regulatory

Commission (NRG) requested that a comprehensive review be made of NRC
requirements in the area of nuclear power plant piping.
request an NRC Piping Review Committee was formed.
In response to this
The activities of this
Review Committee were divided into four tasks handled by appropriate Task
Groups.
These were:
Pipe Crack Task Group
Seismic Design Task Group
Pipe Break Task Group
Dynamic Load/Load Combination Task Group
Each Task Group will prepare a report appropriate to its scope.
In
addition the Piping Review Committee will prepare an overview document

rationalizing areas of overlap between the Task Groups.
This will be released
as a separate report.
Because of the nature of the current intergranular stress-corrosion

cracking (IGSCC) problems in boiling water reactors (BWRs), the Pipe Crack
Task Group is on an accelerated schedule.
The first draft of this report was
completed March 1984, while the other Task Groups are aiming for August September.
The Review Committee is scheduled to complete its activities prior
to the end of 1984.
ix
The project titles of the five volumes are:
Volume
I - Investigation and Evaluation of Stress-Corrosion Cracking in

Piping of Boiling Water Reactor Plants
Volume
II - Evaluation of Seismic Designs
Volume III - Evaluation of Potential for Pipe Breaks
Volume
Volume
IV - Evaluation of Other Dynamic Loads and Load Combinations
V - Summary - Piping Review Committee Conclusions and Recommendations
ACKNOWLEDGMENT
The Task Group on Pipe Cracking wishes to thank those who assisted in
the preparation of this report.
The Task Group members, consultants, and
major contributors are listed in Appendices A and B.

in preparation of individual sections included:
Others who assisted
from the utility industry,
J. Sinclair (Vermont-Yankee) R. Godby (Hatch), W. Pierce and M. Tucker

(Dresden), and J. McElroy (PECO); from NRC, D. Clark, G. Gear, M. Boyle, and
R. Stark; from ANL, L. Beckford; from PNL, G. Selby; from BNL, C. Auerbach
and C. Czajkowski; also J. Spanner of Spanner Engineering.
Our thanks to Brookhaven National Laboratory for its support in the

use of its facilities and staff to prepare the report, particularly to
Ms. Olga Betancourt for typing coordination and Ms. Mary Rustad for
editorial assistance and report coordination.
The Task Group met with representatives of EPRI, General Electric Co.,
BWR Owners Group, NRC-RES, utilities, ASME code groups, and personnel in
foreign countries.
A special international team of experts reviewed and commented on a draft

version of this report, consisting of the following:
Dr.
Dr.
Dr.
Dr.
Yoshio Ando
Ferenc de Kazinczy
Karl Kussmaul
Brian Tomkins
(Japan)
(Sweden)
(Federal Republic of Germany)
(United Kingdom)
Their comments appear as Appendix H.
xi
EXECUTIVE SUMMARY
The Executive Director for Operations concurred in the recommendations of

the Committee for Review of Generic Requirements (CRGR) that the '^RC should
develop an integrated program plan to deal with the entirety of the stress
corrosion cracking problem in BWR piping in order to avoid a piecemeal
approach to the problem.
Such a program should enlist the aid of recognized
experts in the area and result in a coordinated resolution that takes into
consideration the entire spectrum of BWR plants."
The Task Group on Pipe Cracking has attempted to comply with this
charter. The title of the report mirrors the scope of the study, namely,
"Investigation and Evaluation of Stress-Corrosion Cracking in Piping of
Boiling Water Reactor Plants." A review of cracking incidents in both BWRs and
PWRs since the release of NUREG-0531 (primarily BWRs) in 1979 and NUREG-0691
(PWRs) in 1980 reveals that incidents due to all types of cracking for both
BWRs and PWRs illustrate no unusual trends other than IGSCC in the recirculating system of BWRs (Section 3).
Therefore, as noted in Section 1, the Task
Group on Pipe Cracking made a conscious decision to limit its scope to IGSCC
in BWRs.
In essence the overall scope of this document is quite comparable to that

of NUREG-0531 with one notable exception.
The Task Group has made a conscious
effort to relate recommendations to regulatory documents that would require

changes if such recommendations are implemented.
In fact, a comparison of the
relevant conclusions and recommendations of NUREG-0531, which are repeated in

Appendix I of this report, with those of this report confirms the strong
similarity of the conclusions and recommendations between the two documents.
Stress-corrosion cracking, particularly intergranular stress-corrosion

cracking (IGSCC), is not a new phenomenon in nuclear reactor piping.
For
example, we have experienced IGSCC in a range of piping sizes in BWRs over the
past 25 years.
To our knowledge the most severe leaking crack was at or below
the technical specification limits; therefore high leak rates have not been a
problem and there have been no structural failures induced by IGSCC.
The fact
that the current problem affects the larger recirculation piping has resulted
in increased concern.
As indicated, the Task Group on Pipe Cracking considers that IGSCC is a

serious problem requiring some changes to current regulatory practice.
How-
ever, we do not feel that IGSCC represents such an urgent problem that it
necessitates immediate additional regulatory action.
The ongoing BWR sur-
veillance and monitoring programs are adequate short-term responses to the

problem.
Both value-impact and risk studies indicate that pipe failures, even
assuming the higher rates due to IGSCC utilized in this report, are not a
major contributor to core melt.
However, we believe that the following
recommendations represent good engineering practice that should markedly

reduce levels of inservice inspection and reactor outage time.
As discussed in Section 2, IGSCC occurs because of a synergism between

the three causative factors:
sensitization, an environment conducive to
initiation and propagation of IGSCC, and high tensile stresses (primarily

residual, but other loads also are contributors). The Task Group recommends that at least two of the three causative factors need to be modified
to minimize IGSCC, although best practice calls for modification of all
three.
The preferred action to combat IGSCC is to replace materials such
as Types 304 and 316 with a resistant material such as Type 316NG stainless
steel.
Recommended measures are summarized in the table at the end of this
Executive Summary.
With regard to replacement material, control of carbon to appropriately

low levels, e.g., less than 0.03, appears acceptable and has been used in the
past; the Task Group recognizes that this results in an ASME Code penalty
because of the lower strength of the low carbon stainless steels, which requires thicker pipe for the same circumstances.
A preferable alternative is
an alloy such as Type 316NG where the carbon is low enough that the material
is highly resistant to IGSCC and the strength is retained through controlled
additions of nitrogen.
While Type 316NG is more resistant to IGSCC than Types
304 and 316, greater care must be exercised in using it to limit hot short
cracking by placing controls on composition and fabrication variables related
to hot forming and welding.
The control of water chemistry is discussed in Section 6.
The Task Group
believes that conductivity control to a low level is possible and beneficial,

but that elimination of IGSCC in piping systems containing a sensitized stainless steel will require the use of hydrogen additions to lower the electrochemical potential of the steel to a nonsusceptible level.
recognizes the limited nature of the work in this area.
The Task Group
Even so, it re-
iterates its recommendation cited here and in NUREG-0531 that water chemistry
should be improved.
The Task Group recognizes there are several measures capable of

reducing the tensile residual stress level in the heat-affected zone (HAZ).
With appropriate control of parameters, a ntmber of these processes should be
acceptable; however, the group has a definite preference for induction heating
stress improvement (IHSI), based on a review of the extent and applicability
of the available data.
A major problem cited in NUREG-0531 was the inability of ultrasonics to

detect and size IGSCC in austenitic alloys reliably.
Substantial progress has
been made in the reliability of detection, assuming proper operator training

and procedures; however, more work is required to develop a better understanding of the variability from operator to operator.
The existing tech-
niques are adequate to size the crack length to acceptable limits of accuracy.
With regard to crack depth sizing, the techniques as currently
practiced are clearly inadequate and require more work.
Sections 4 and 5
discuss the problem at length as well as specific measures taken or to be

taken to improve the reliability of crack detection and sizing.
A substantial percentage of welds are limited to single-sided access.

This will remain the case even after replacement of recirculation systems with
a material such as 316NG if the original pumps, valves, vessel nozzles, etc.,
are used and their regions adjacent to the weld are too short for UT.
Attention to stress reduction with IHSI should be considered for these
components.
Every effort should be made to improve accessibility and weld
joint design for ultrasonic testing (UT) when pipes are replaced.
The changes in overall LOCA risk because of IGSCC and a value-impact

analysis of the options available to the utility are discussed.
Both studies
arrived at the same major conclusion, namely, that pipe breaks, even when
higher rates are assumed as a result of IGSCC, represent a minor contribution
to the frequency of core melt.
A deterministic study discussed in Section 6
suggests a similar conclusion; leak-before-break is the dominant mode rather

than catastrophic failure.
The preceding has been a broad overview of the current problem of IGSCC
in BWR recirculating lines and the recommendations of the Task Group on Pipe
Cracking.
Chapter 9 pulls together all the conclusions and recommendations
throughout the document.
The following are conclusions and recommendations
considered to be most significant by the Task Group:
Because mitigating actions addressing only one of the factors may

not be fully effective under all anticipated operating conditions,
mitigating actions should address two and preferably all three of the
causative factors; e.g., material plus some control of water
chemistry, or stress reversal plus controlled water chemistry.
Code minimum UT procedures result in totally inadequate IGSCC

detection.
Easily implementable modifications to these procedures
have resulted in some improvement.

Code Case N-335.
These have been incorporated into
Therefore, it is recommended that Code Case N-335
should immediately be made mandatory for all augmented inspections

until better procedures are developed.
Although IGSCC detection has improved to the point that it is considered acceptable under optimum conditions and procedures, the
detection reliability as impacted by variability in operator procedure
and equipment performance along with field conditions needs further
study and improvement.
While length sizing of cracks is acceptable,
depth sizing is currently inadequate.
It is recommended that advanced
techniques and procedures for crack detection and depth sizing continue to be developed and incorporated into Code requirements to
provide data to reduce the need for extremely conservative fracture
mechanics evaluation.
The current activities in personnel and procedure qualification and

performance demonstration represent steps in the right direction, and
the resultant process that is being implemented is acceptable in the
interim; however, they need further improvement.
Therefore, it is
recommended that ongoing industry and NRC activities to develop

adequate criteria for qualification of the entire inspection process
to achieve more reliable field inspection be completed and implemented
on a high priority basis.
For future plants or for replacement of existing piping systems, the

material, design of pipe joints, and accessibility from both sides of
the weld should be optimized for UT examinations; this requirement
should be mandatory for all components with the exception of existing
items such as pumps, valves and vessels in older plants.
The
uninspectable joints should be subjected to IHSI.
Inspection techniques should be developed for detection and

dimensioning of flaws in pipes repaired by the weld overlay process.
All Type 304 and 316 austenitic piping systems operating over 200 F
(93 C) should receive augmented inservice inspection unless they
have been treated with effective countermeasures.
Flaw evaluation criteria should limit the length of the cracks accepted for continued operation without repair.
The limitation on
acceptable crack length is primarily a result of the lack of confidence in flaw depth sizing capability, and is intended to ensure
leak-before-break conditions.
The maximum allowable throughwall crack
length can be determined based on weld joint specific loads.
The maximum crack length allowable without repair for a specific weld
joint should be the minimum of either 1) the throughwall crack length
demonstrated by elastic-plastic fracture mechanics analyses to be
stable under normal operating plus SSE loading conditions, 2) the
throughwall crack length that would still permit the pipe to withstand
normal operating plus SSE loading conditions as demonstrated by net
section collapse (limit-load) analyses, or 3) the maximum crack length

that would result in a leak rate greater than the plant's normal
makeup capacity.
Shorter cracks can be evaluated using the IWB-3640
criteria as modified by the NRC staff in SECY 83-267C.
Calculations
indicate that in the majority of cases the maximum crack length

acceptable under the above criteria will be approximately 25% to 30%
of the pipe circumference.
On the basis of fracture mechanics evaluation for bounding and typical

stress conditions and weld toughness properties, it is concluded that
IWB-3640 provides an adequate basis for evaluating the majority of the
weld connections in BWR recirculation piping.
This is especially true
because many of the cracks will be in higher toughness zones adjacent

to the lower toughness welds.
Additional fracture mechanics analyses, material properties characterization, and large scale pipe tests be performed to understand
further the implications of stainless steel weld and cast material
fracture toughness properties in flawed pipe evaluations.
Further-
more, in this regard, the Task Group recommends active NRC support of
the ASME Task Group currently evaluating the concerns which have been
raised regarding IWB-3640.
Since operating experience and fracture mechanics evaluations indicate

that leak-before-break is the most likely mode of piping failure, it
is recommended that reasonably achievable leak detection procedures be
in effect in operating plants.
Current sump pump monitoring systems
are sensitive enough to provide additional margin against leak-beforebreak if more stringent requirements on surveillance intervals and
unidentified leakage are imposed (see Section 4 ) .
Therefore, the Task
Group recommends that the limits on unidentified leakage in BWRs be decreased to 3 gpm and that the surveillance interval be decreased to 4
hours or less.
For relatively short axial cracks, analysis can be used to justify

long-term operation with weld overlays, since errors on crack depth
measurement or flaw growth predictions for these cracks will lead at
worst to relatively small leaks, which will be easily detectable long
before the crack can grow long enough to cause failure.
For circum-
ferential cracks weld overlay is considered an acceptable repair

procedure for a maximum of two refueling outages unless reliable
techniques for the sizing of cracks through the overlay or for the
monitoring of crack growth are developed.
Experience with materials to mitigate IGSCC, such as 347NG in Germany

and 304NG and 316NG in Japan has been excellent.
in the U. S. include 304L and 316L.
Other materials used
All of these alloys are more re-
sistant to IGSCC than conventional Types 304 or 316 stainless steel.

Based on U. S. data and prior use. Type 316NG stainless steel offers
an additional margin of resistance to IGSCC and utilities that choose
to replace pipe should be strongly encouraged to use it.
Although low-carbon stainless steels with nitrogen additions have been

successfully fabricated and welded in Japan and Europe, U. S. experience with these materials is limited.
It appears that greater
care must be exercised in the control of composition and fabrication

variables to limit cracking during hot forming or welding.
The use of IHSI on weldments with detectable cracking must be considered on a case-by-case basis.
However, for relatively short cracks
(approximately 20% of the circumference in length), since even large

errors in crack sizing or the prediction of flaw growth will lead only
to small leakage, the decision on whether an additional repair is
required can be determined by analysis.
For longer cracks, repair
will probably be required.
IHSI is considered to be a more effective mitigating action for IGSCC

than HSW and LPHSW in part because more data are available to demonstrate that the process does produce a more favorable residual stress
state.
All the residual stress improvement remedies are considered to
be much more effective when applied to weldments with no reported

cracking.
BWR water chemistry controls should be modified to minimize IGSCC.

These modifications should include both a substantial reduction in the
levels of ionic species entering the primary coolant and a control of
oxygen level.
The current work on reduction of oxygen through
hydrogen additions should be followed closely with the possibility
10
that it may be employed to reduce further the electrochemical potential of the stainless steel to a level at which SCC, either IGSCC or
TGSCC, will not occur.
It appears that hydrogen water chemistry is an
effective IGSCC countermeasure.
However, ongoing work regarding
potential adverse effects on other reactor components should be

closely followed in order to confirm the acceptability of this
countermeasure.
The Task Group recommends that the inspection schedule for welds be
based on the resistance to IGSCC of the materials and the
effectiveness of the mitigating processes applied to the welds.
The
materials which are considered resistant and the categories of

countermeasures processes are:
Resistant Materials
(1)
304L, 316L, 316K, 304NG, 316NG, 347NG, 308L
(2)
low-strength carbon steels
(3)
approved nickel-based materials
(4)
cast low-carbon/high-ferrite austenitic stainless steels
(5)
welds solution heat-treated after fabrication and welding.
(6)
other, as approved by NRC
Countermeasure A Processes - Any combinations of two mitigating processes

which are intended to reduce or minimize any two of the three causes (i.e.,
sensitization, stress and environment) contributing to IGSCC.
summarized below:
11
These are
(1)
IHSI on new pipe or pipe with no reported indications plus hydrogen

water chemistry*;
(2)
HSW on new welds, including new welds used to install short repair
sections ("pup pieces") plus hydrogen water chemistry;
(3)
LPHSW on new pipe welds including new welds used to install short repa
sections plus hydrogen water chemistry.
(4)
Other, as approved by NRC.
Countermeasure B Processes - Any of the following mitigating processes.
(1)
IHSI on new pipe or pipe with no reported indications.
(2)
HSW on new welds including new welds used to install short repair
sections ("pup pieces").
Hydrogen water chemistry is discussed in Section 6.
12
(3)
LPHSW on new pipe welds including new welds used to install short repair
sections.
(4)
Hydrogen water chemistry.
(5)
Other, as approved by NRC.
The recommended inspection requirements are:
Condition and Inspection Required for Weld Categories

Weld
Category
Condition
Inspection Required
Resistant
material or
countermeasure A
applied
25% of the welds of each pipe size in 10 years.

At least one-third of these should be inspected
every three and one-third years or the nearest
refueling outage.
Nonresistant
material with
countermeasure B
applied

refueling outage.
Neither of the
above, and all
welds with detectable cracking
regardless of the
use of mitigating
processes
100% in 6 years. At least one-half of these

should be inspected every three and one-third
years or the nearest refueling outage.
For plants older than 6 years, all uninspected

Category C welds shall be inspected at the next
outage after compliance with IE Bulletin 82-03
or 83-02.*
*Weld inspections not performed in accordance with IE Bulletin 82-03 or 83-02
are considered inadequate; therefore, no credit is given for such inspections.
13
Recommended Measures for Controlling IGSCC in BWR Piping

Near Term
Replace
Long rerm
Ult rasonic Exam.
Residual
Stress Improvement
IHSI, HSW
Hydrogen
Water
chemistry
3 1 6 NG
Piping,
Design Stage
NTOL or Recent Startup
No Cracks Detected
X - --X#
Only Limited Shallow

Cracks
X--
X**
Xthen
BWR Plant Status
Overlay
Weld*
with
Similar
Alloy
New
Base- Accel.
line
UT
Norm.
UT
Enhanced Leak
Detection
(moisture tapes,
AE, etc.)
Operating >5(?) years
Deeper Cracks - Few

or Many
X
maybe
^Limited to two cycles unless convincing evidence is presented

**Also suggested after replacement with 3 1 6 NG
# Prior to mitigation, accelerated UT; thereafter normal UT
X Accelerated UT limited to cracked welds
* With mitigation, accelerated UT limited to cracked welds
X
X(for
316NG)
X (maybe)
X
X Prior to
Replacement
1.0 INTRODUCTION
1.1 PURPOSE
On August 1, 1983, the Executive Director for Operations of the Nuclear

Regulatory Commission (NRC) established a committee to review nuclear piping
in the context of current regulations, regulatory guides, branch technical
positions, and other pertinent documents.
A Task Group of this committee was
assigned the responsibility for reviewing specific incidents of cracking in

both boiling water reactors (BWRs) and pressurized water reactors (PWRs) with
emphasis on intergranular stress-corrosion cracking (IGSCC) reported in
1982-1983 in the BWR recirculating systems.
The charter of the Task Group
included a spectrum of issues such as:
The significance of IGSCC in the context of overall reactor safety;
the reliability of detection and of sizing of IGSCC in BWR piping

using current ultrasonic techniques;
the potential effects of changes in water chemistry on the initiation

and propagation of IGSCC;
the value and/or limitations of repair techniques such as overlay

cladding on piping containing IGSCC;
the long-term value of alternative measures such as induction heating

stress improvement (IHSI) on both uncracked and cracked piping;
1-1
the long-term reliability of replacement piping such as 316NG and the

possible limitations of the material;
a review of information pertinent to pipe cracking occurring in

foreign light water reactors.
Previous Pipe Crack Study Groups have reviewed IGSCC in BWRs [Technical
ReportInvestigation and Evaluation of Cracking in Austenitic Stainless Steel
Piping of Boiling Water Reactor Plants, NUREG-75/067 (1.1) and in LWRs
Investigation and Evaluation of Stress-Corrosion Cracking in Piping of Light
Water Reactor Plants. NUREG-0531 (1.2)].
A Pipe Crack Study Group also
reviewed incidents of cracking in PWRs (1.3).
One Pipe Crack Study Group (1.2) reviewed and commented on the
significance of IGSCC in large-diameter piping typical of BWR recirculating
lines.
Their conclusion is given below:
The cracking experience in a German BWR facility demonstrates that

IGSCC can occur in large-diameter (more than 20 in.) BWR stainless
steel piping and confirms the previous study group's conclusions that,
although cracking in large-diameter pipes has not been observed in the
United States, the conditions for IGSCC do exist in these lines. We
also have reviewed the safety aspects associated with the potential
for IGSCC in large-diameter BWR stainless steel piping.
Based on our
review, we have concluded that, although throughwall and part-through

wall IGSCC can exist in BWR stainless steel piping during plant
operation, it is unlikely that significant cracks would go
1-2
undetected.
In addition, based on our review and analysis, we also
conclude that it is unlikely these cracks will cause unstable crack

growth and excessive loss of coolant.
Further, since the Emergency
Core Cooling System (ECCS) provides protection should a loss-ofcoolant accident occur, we conclude that IGSCC in these lines, while
generally undesirable, will not be a hazard to public health and
safety.
Several of the conclusions and recommendations of reference 1.2 are

particularly pertinent to this study.
Therefore, the relevant conclusions and
recommendations from NUREG-0531, numbered as in the report, are repeated as

Appendix I to this report, to serve as "benchmarks" throughout this document
as to progress or lack of progress since 1979.
1.2 REVIEW OF PRESSURIZED WATER REACTORS
Pressurized water reactor (PWR) piping was reviewed cursorily in 1979

(1.2) and it was concluded that stress corrosion was not a problem for the
primary system.
A more extensive study in 1980 (1.3) examined several failure
mechanisms relevant to PWR piping systems. Again stress corrosion was

considered to be a secondary contributor where attack was limited to lower
pressure, thin-wall secondary or tertiary systems.
1-3
PWR primary piping systems whether ferritic or austenitic have been

essentially immune to any cracking mechanisms.
In fact, probabilistic studies
at Lawrence Livermore National Laboratory (LLNL) yield very small (10~12 to

10~14) direct failure levels and acceptably small (approximately 10"^)
per reactor year failure levels by indirect failures such as cranes falling on
piping during several seismic events.
The LLNL studies and extensive studies
in the Federal Republic of Germany (1.4) both led to the conclusions that
catastrophic failure of PWR primary piping is an incredible event.
A paper by Bush (1.4) at an IAEA Symposium on Reliability of Reactor

Pressure Components reexamined PWR piping failures through 1982.
In addi-
tion, an unpublished report by Bush examined PWR failures through 1982 and
partly into 1983. Both studies confirmed that there have been very few cases
of SCC in any PWR piping systems during the period since reference 1.3.
Since 1980, failures due to vibrational fatigue, thermal fatigue, and

corrosion fatigue have been quite limited.
Corrosion-cavitation has resulted
in severe failures; however, these have not been in safety systems, e.g., the
large steam turbine lines.
The one possible exception has been in the area of
dynamic loads such as water hammer.
There has been a feedwater line failure
at Maine Yankee which is understandable because steam generator J-tubes had

not been installed to prevent such incidents.
On the basis of the preceding comment, the Task Group has concluded that
there are no bases for an extensive review and analysis of PWR pipe cracking.
With regard to inspectability. Section 4 discusses the relative reliability of
NDE of PWR primary materials.
1-4
1.3 DOCUMENTS POTENTIALLY AFFECTED BY RECOMMENDATIONS IN THIS REPORT
The Task Group was directed in its instructions to cite various documents
that might require changes as a result of the recommendations developed in
this report.
The following is a citation of these documents.
Specific
suggestions concerning changes appear in the appropriate sections:
Generic Issues
A-14
Flaw Detection
A-42
Pipe Cracks in BWRs (PGI-8)
No. 34
Reactor Coolant Systems Leakage
Regulations
10 CFR Part 50, Domestic Licensing of Production and Utilization Facilities,

Appendix A
General Design Criteria for Nuclear Power Plants
Criterion 30 - Quality of Reactor Coolant Pressure Boundary

Criterion 31 - Fracture Prevention of Reactor Coolant Pressure Boundary
Criterion 32 - Inspection of Reactor Coolant Pressure Boundary
Criterion 36 - Inspection of Emergency Core Cooling System
1-5
Regulatory Guides
1.31 - Control of Ferrite Content in Stainless Steel Weld Metal

1.44 - Control of the Use of Sensitized Stainless Steel
1.45 - Reactor Coolant Pressure Boundary Leakage Detection Systems
1.46 - Protection Against Pipe Whip Inside Containment
1.56 - Maintenance of Water Purity in Boiling Water Reactors
1.58 - Qualification of Nuclear Power Plant Inspection, Examination, and
Testing Personnel
1.84 - Code Case Acceptability in ASME Section III Design and Fabrication
1.85 - Code Case Acceptability in ASME Section III Materials
1.116- Quality Assurance Requirements for Installation, Inspection, and
Testing of Mechanical Equipment and Systems
1.130- Service Limits and Loading Combinations for Class 1 Plate-and-ShellType Component Supports
1.147- Inservice Inspection Code Case Acceptability ASME Section XI
Division 1
NUREG-0313 - Technical Report on Material Selection and Processing Guidelines

for BWR Coolant Pressure Boundary Piping, Rev. I, 1979, Rev. II, in
draft form
NUREG-0800 - Standard Review Plan for the Review of Safety Analysis Reports
for Nuclear Power Plants
1-6
3.6.1 - Plant Design for Protection Against Postulated Piping Failures in

Fluid Systems Outside Containment
3.6.2 - Determination of Rupture Locations and Dynamic Effects Associated
5.2.1.1 - Compliance with the Codes and Standards Rule 10 CFR 50.55a
5.2.1.2 - Applicable Code Cases
5.2.3
- Reactor Coolant Pressure Boundary Materials
5.2.4
- Reactor Coolant Pressure Boundary Inspection and Testing
5.2.5
- Reactor Coolant Pressure Boundary Leakage Detection
Codes and Standards
ASME XI - UT
ANSI Draft Standard - Leak Detection
1.4 THE TASK GROUP ON IGSCC
The Task Group is composed of ten members from the NRC and one member
from Review and Synthesis Associates, listed in Appendix A.
The ten members
represent the Offices of Nuclear Reactor Regulation (4), Nuclear Regulatory

Research (3), Inspection and Enforcement (1), and one each in Regions 1
and 2. Services of several additional NRC staff members (cited in the
Acknowledgements) were used as required, so that the expertise of the
personnel in the division was factored into the investigation of the Task
Group.
1-7
In addition, the services of outside consultants in the fields of

corrosion chemistry, nondestructive examination, and fracture mechanics
were used.
They are listed in Appendix B.
The Task Group limited its scope to IGSCC of BWR piping on the basis of
its charter and the arguments raised in Section 1.2 regarding PWRs.
The Task Group met with representatives of EPRI, General Electric

Company, BWR Owners Group, NRC-RES, utilities, ASME code groups, and personnel
in foreign countries.
1.5
REFERENCES
1.1
Pipe Crack Study Group.
1975.
Investigation and Evaluation of Cracking
in Austenitic Stainless-Steel Piping of Boiling Water Reactor Plants.

NUREG-75/067, U.S. Nuclear Regulatory Commission, Washington, D.C.
1.2
Pipe Crack Study Group.
1979.
Investigation and Evaluation of Stress-
Corrosion Cracking in Piping of Light Water Reactor Plants.
NUREG-0531,
U.S. Nuclear Regulatory Commission, Washington, D.C.
1.3
PWR Pipe Crack Study Group.
1980.
Investigation and Evaluation of
Cracking Incidents in Piping in Pressurized Water Reactors.

U.S.
1.4
NUREG-0691,
Nuclear Regulatory Commission, Washington, D.C.
S. H. Bush.
1983.
"Pressurized Water Reactors."
In Proceedings of
Reliability of Reactor Pressure Components, pp. 29-56, IAEA-SM-269/73,

International Atomic Energy Agency.
1-8
2.0
CAUSES AND DESCRIPTION OF IGSCC PHENOMENA
As noted in the introduction, intergranular stress-corrosion cracking

(IGSCC) has been occurring in BWR and PWR nuclear power plants for a number of
years.
The recent identification of the problem in a number of operating BWRs
has not indicated the occurrence of any new phenomenon, but rather that the
phenomenon already known and discussed in the previous review group reports is
being found in a wider range of locations, probably because of longer operating times and improved inspection and detection techniques.
Further, as a
result of the considerable research performed in this area since our previous
reviews, our understanding of the causes of IGSCC has improved.
In this
section, therefore, the purpose is to summarize our current understanding of

the problem in order to provide a background for the discussions in later
sections on short- and long-term solutions and remedial actions.
IGSCC in stainless steels arises from the synergistic effects of stress

and the environment on a material that is susceptible to the phenomenon.
Each
of these aspects of the synergism will be examined separately.
A susceptible material:
Types 304 or 316 SS, in a certain
temperature range, will form grain boundary networks of Cr carbides,

accompanied by adjacent areas depleted in Cr.
The production of the Cr
carbides in these areas results in a network of narrow zones adjacent to the

grain boundaries in which the Cr levels in solid solution are sufficiently
lower than those in the bulk of the grains to enable electrochemical cells to
be set up between the lower Cr grain boundary material and the higher Cr bulk
grain material.
Precipitation of the Cr carbides and formation of the Cr-
2-1
depleted zones are a function of the composition of the steel and the time
that the steel is heated in the temperature range at which the kinetics of the
reaction are sufficiently rapid and thermodynamics favor production of the Cr
carbides.
Typically, this occurs between 900 and 1500OF, although some low
temperature sensitization (LTS) can be anticipated at temperatures as low as

the operating temperatures of a BWR. While sensitization can be produced in
stainless steels containing carbon levels as low as 0.02 wt.%, IGSCC has
seldom if ever been seen in BWRs in stainless steels containing significantly
less than 0.04 wt.% carbon; the probability of IGSCC occurring appears to
increase rapidly with carbon concentrations above this level up to 0.055%, and
thereafter to be constant up to the limit for Type 304 or Type 316 SS of 0.08
wt.%.
All stainless steels are specified by the industry and by NRC positions
to be in the solution-annealed state prior to fabrication.
At the solution-
annealing temperatures, i.e., at 2000OF or higher, the Cr carbides are

unstable with respect to solid solutions of Cr and carbon in the stainless
steel, and thus decompose. Rapid cooling following solution-annealing
prevents recombination from occurring. However, the welding of stainless
steel pipe produces a narrow band of material in an area adjacent to the weld
which has been heated into the sensitization range by the heat of welding; the
pipe cracks are occurring primarily in this area.
Although the general phenomenon of IGSCC has been mainly attributed to

the formation and extent of the Cr-depleted zones in the grain boundaries,
other minor constituents of the stainless steel can affect either the kinetics
of formation of the Cr-depleted zones or the initiation and propagation of the
stress-corrosion cracks, as discussed below.
Steels containing alloying
elements that form carbides more stable than Cr carbide (such as niobium
2-2
carbide or Ti carbide) are less prone to sensitization. However, even with

these steels there is a potential area immediately adjacent to the weld fusion
line in which Cr-depleted zones can be produced if the carbon levels are high
(knife-line attack).
The relative freedom of the Swedish and the newer German
BWRs from IGSCC is thought to be related to the use of a 0.05% carbon limit in
Sweden, and the use of a low-carbon, niobium-stabilized stainless steel (often
referred to as Type 347NG) in West Germany.
Stress:
Austenitic stainless steels as a general rule do not deform
elastically to a sharp yield stress and plastically at stresses above it.

Rather, some plastic deformation can occur at stresses considerably below the
nominal "design yield" (0.2% offset) strength.
It is generally considered
that the straining or creep of the stainless steel under tension produces a
rupture in the protective oxide films, followed by a relatively rapid bare
metal corrosion until the film reheals. The lower Cr content in the sensitized grain boundaries results in locally greater metal loss prior to
rehealing.
Repeated rupture of the film, caused either by alternating the
stresses or applying continuous stresses appreciably above the yield point,

produces a crack along the grain boundary, since preferential attack occurs at
the Cr-depleted grain boundaries.
The primary source of tensile stress in the
operating BWR piping is considered to be the residual stresses from welding,

as discussed by the earlier review groups. Added to this, however, are
operational pressure stresses, vibrational stresses, and thermal stresses,
particularly during heatup and cooldown.
The combined effects of all these
tensile stresses on the inner surface of the piping in the sensitized regions
need to be considered in determining the susceptibility of a given welded area
to IGSCC. Generally, sufficient stress to cause repeated rupture of the
2-3
protective oxide film is necessary to initiate cracking.
Although in terms of
conventional engineering design, creep is negligible in austenitic stainless

steels at BWR operating temperatures, the strain rates associated with
stresses at or above "yield" are sufficient to produce rupture of the oxide
film.
This may even be true for stresses below "design yield," particularly
if they are applied intermittently, because of the tendency of the material to

creep.
Once a crack is initiated, a continued tensile stress intensity at the
crack tip is required for crack propagation.
Environment:
In the BWR primary coolant, the presence of residual
oxygen from radiolysis puts the chemical potential of the stainless steel in a
range in which the intergranular stress corrosion cracks can initiate and
propagate in sensitized material.
Oxygen appears to establish a chemical
potential such that the electrochemical crack propagation process can proceed
and is therefore the key environmental cause of IGSCC.
The species
responsible for conducting the electrochemical currents in IGSCC are assumed

to come either from the environment (such as in-leakage of impurities from the
condenser or resin bead breakup and decomposition) or from corrosion of the
metal itself, particularly from bare-metal corrosion following film rupture,
producing salts of sulfur, phosphorus, silicon, or other anions.
These
species can produce a current-carrying capability in the grain boundary

cracks, so that IGSCC can occur in theoretically pure water at sufficiently
high stresses in the presence of an oxidizing potential.
2.1
MECHANISMS OF IGSCC
IGSCC can be described in terms of an initiation phase and a propagation

phase, the mechanisms of which are reviewed briefly below.
2-4
2.1.1
Initiation Phase
The surface of stainless steels is protected from corrosion by a thin,

adherent, passivating protective oxide.
In addition to this passive layer,
the surfaces of the austenitic stainless steel pipes in a BWR have a relatively thick layer of mass-transported oxides deposited on them in the
form of crud, which contains some radioactivity.
These crud deposits can
also be highly specific adsorbents for impurities entering the solution, such
as sulfates or chlorides, and this property may have some influence on the
crack-initiation process.
It is difficult to extrapolate laboratory data to
determine the precise point of crack initiation.
In slow strain rate stress-
corrosion tests, if the protective passivating oxide layers are ruptured

frequently enough, localized corrosion of sensitized grain boundary material
at the root of these ruptures can produce an environment in a relatively short
period of time that is capable of carrying the electrochemical corrosion current, and therefore bringing about IGSCC.
The presence of the crud deposits
and of materials or ions adsorbed in these deposits may also influence the
initiation of IGSCC.
Laboratory tests have shown that very small amounts of
sulfates and chlorides can accelerate the initiation of IGSCC in a constant

load test when the electrochemical potential is held in a range at which IGSCC
is known to occur.
Crack initiation is assumed to occur at the point at which an electrolyte

builds up in the crevice between the ruptured protective oxide and the baremetal surface.
This rather vaguely defined point of transition between
initiation and propagation reactions is difficult to measure experimentally.

In terms of a mechanism for IGSCC, however, it is consistent with the
experimental data.
2-5
2.1.2
Propagation
Once IGSCC has been initiated, the subsequent growth of these cracks
depends upon several factors:
the stress intensity (determined from the
geometry of the crack and the applied tensile stress), the presence in the
crevices of impurities that can stabilize a conducting (usually acidic)
environment in the crevice, and the degree of sensitization of the material
(which affects the potential of the electrochemical cells existing between the
higher Cr and lower Cr areas).
It also depends upon the continued presence in
the bulk solution of an oxidizing species, controlling the overall chemical

potential of the steel surface.
Experimental evidence on the role of impurities, electrochemical potential, sensitization, and stress intensity factors in crack propagation rates
is much more abundant than are the data on crack initiation.
This evidence
has shown that crack propagation can be terminated by a shift in the electrochemical potential (i.e., by removing oxygen from the system), by a change
in the material (i.e., by the crack running into a duplex weld structure containing sufficient delta ferrite to control the chromium-depleted zones), or
by a reduction in the stress intensity factor (i.e., by the crack reaching an
area of residual compressive stresses).
Once a crack has been initiated,
however, the environment within the crack is difficult to change, and will
continue to favor propagation unless the chemical potential is changed by
reducing the oxygen levels in the system.
Thus, given sufficient stress
intensity, a crack may continue to propagate into material normally considered

resistant to IGSCC in a BWR environment.
2-6
2.2
CONCLUSIONS
IGSCC in BWR piping is occurring owing to a combination of material,

environment, and stress factors, each of which can affect both the
initiation of a stress-corrosion crack and the rate of its subsequent
propagation.
In evaluating long-term solutions to the problem, one
needs to consider the effects of each of the proposed remedial actions

in terms of the current understanding of the initiation and
propagation of stress-corrosion cracks.
2.3
RECOMMENDATIONS
Mitigating actions to control IGSCC in BWR piping must be designed to

alleviate one or more of the three synergistic factors:
sensitized
material, the conventional BWR environment, and high tensile

stresses.
Because mitigating actions addressing only one of the factors may

mitigating actions should address two and preferably all three of
the causative factors; e.g., material, plus some control of water
In the event that any of the preceding recommendations of this section

are implemented, there may be a need to modify the following doctiments when
relevant to the specific issue:
NUREG-0313 - Rev. 2.
2-7
3.0
3.1
BACKGROUND
The Nine Mile Point Nuclear Station Unit 1 (a BWR Model 2) was shut down
in March 1982 to replace recirculation pump seals. A normal hydrotest was
performed to test the new seals. During this hydrotest, leaks were noticed at
two of the furnace-sensitized Type 316 SS 28-in.-diameter recirculation loop
safe ends.
Further examination identified that the leaks originated in very
small pinholes or cracks in the heat-affected zones of the safe-end-to-pipe

welds.
Subsequent ultrasonic examinations revealed extensive cracking at many
weld joints in the Type 316 SS recirculation system.
Two boat samples were
removed for metallographic examination, and the cracking was identified to be

IGSCC. The utility decided to replace the recirculation loops with IGSCCresistant Type 316NG material.
This was the first known incident of IGSCC in large-diameter (larger than
10 inches) piping in the U. S., although cracks in large piping had been found
in Japan, and cracks in 24-in. furnace-sensitized safe ends had been found in
Germany.
After the extent of cracking at Nine Mile Point had been determined, IE
Information Notice No. 82-39 was issued on September 21, 1982, to alert all
BWR licensees to the problem. Meetings were held with General Electric, EPRI,
and BWR Owners to discuss the relevance of the Nine Mile Point cracking to
other BWRs.
3-1
On September 27, 1982, the NRC held a meeting with all BWR licensees to
discuss plans for near-term inspections of welds in the large-diameter
recirculation piping.
This was followed on October 28, 1982, by IE Bulletin
No. 82-03, "Stress Corrosion Cracking in Thick Wall, Large Diameter,

Stainless Steel Recirculation System Piping at BWR Plants."
IE Bulletin No. 82-03 required that licensees of eight BWRs with

scheduled outages through January 31, 1983, perform inspections of a
reasonable sample of the welds in the recirculation system during the next
outage.
It further required that effective ultrasonic inspection procedures
be used, and that the effectiveness of the procedures and competence of the UT
examiners be demonstrated on samples of cracked piping from the Nine Mile
Point recirculation system.
After these inspections showed that cracking in large pipes had occurred
in five of the seven plants then inspected, IE Bulletin No. 83-02 was issued
in March 1983, to extend the inspection requirements to all other BWRs.
In
addition, it required that a larger sample of welds be inspected, with

provisions for further increase in the inspection sample if cracks were
found.
These inspections continued to indicate that cracking in most BWRs was

extensive, and many required major repairs by weld overlay reinforcement.
Because the reports indicated that many welds required repair to meet ASME
Code requirements for further operation, the NRC determined that five plants
still to be inspected should be ordered to inspect as soon as practicable, and
should inspect essentially 100% of all welds in the recirculation system and
connecting systems that were considered to be susceptible to IGSCC.
3-2
3.2
INSPECTION RESULTS
The results of all of the BWR inspections conducted under lEB 82-03 and
83-02 are summarized in Table 3.1.
This covers the first round of
inspections, and is complete as of March 1984. The inspection results show

wide variations in extent of cracking reported.
Some plants. Oyster Creek,
Duane Arnold, Fitzpatrick, and Millstone 1, for example, reported very little,
if any, cracking.
Other plants such as Dresden 3, Cooper, Hatch 2, Vermont
Yankee, and Browns Ferry 1 reported very extensive cracking.
At this time no
obvious reasons for this wide variation have been determined.
In addition to the welds listed in Table 3.1, a number of licensees

inspected welds in systems with smaller-diameter piping.
Significant
cracking was also found in these systems (RWCU, Core Spray, etc.), as
expected.
Augumented inspections of these systems were already required by
NUREG-0313, Rev. 1, and are therefore not of primary interest in this report.
As a direct result of this inspection program, utilities are planning to

take positive actions to preclude or minimize IGSCC in the future.
Several
utilities are in the process of replacing susceptible piping with Type 316NG
stainless steel, which is highly resistant to sensitization and IGSCC.
Monticello, Hatch 2, and Pilgrim are currently undergoing such replacement.
Several other utilities are making positive plans, including procuring
replacement material.
The NRC has permitted plants to operate for one fuel cycle with weldoverlay-reinforced cracked welds, or with welds with minor cracking that will
3-3
TABLE 3.1
Plants
Big Rock Point
Browns Ferry 1
Browns Ferry 2
Browns Ferry 3
Brunswick 1
Brunswick 2
Cooper
Dresden 2
Dresden 3
Duane Arnold
FitzPatrick
Hatch 1
Hatch 2
Millstone 1
Monticello
Nine Mile Pt. 1
Oyster Creek
Peach Bottom 2
Peach Bottom 3
Pilgrim 1**
Quad Cities 1
Quad Cities 2
Vermont Yankee
Summary of All Inspection Findings on Large Piping in All

Operating BWRs Inspected According to lEB 82-03 and 83-02
Extent of Inspection
(% of welds inspected)
Recirc.
RHR
20%
98%
27%
98%
25%
100%
100%
47%
100%
42%
47%
47%
94%
11%
100%
82%
39%
100%
91%
(11/59)
(103/105)
(25/91)
(103/105)
(29/115)
(102/102)
(108/108)
(47/101)
(115/115)
(49/117)
(49/106)
(47/100)
(97/103)
(11/100)
(106/106)
(62/76)
(31/80)
(91/91)
(77/85)
8% (9/110)
100% (106/106)
66% (58/88)
Inspection Results
(No. of cracked welds)
Recirc.
RHR
91% (32/35)
92% (35/38)
0
33
2
0
3
15
20
10
53*
0
1
5
36
0
6
53
0
19
10
20% (9/44)
90% (45/50)
7% (2/30)
0
20
33
90%
28%
28%
75%
100%
100%
10%
90%
40%
45%
100%
100%
0%
78%
(36/40)
(9/32)
(9/32)
(3/4)
(5/5)
(7/7)
(4/40)
(45/50)
(2/5)
(5/11)
(11/11)
(11/11)
(0/46)
(18/23)
14
0
0
0
1
0
0
11*
0
0
2
3
0
0
0
0
7
5
0
42
0
0
3
8
13
7
61
0
0
6
27
0
6
0
0
21
15
0
2
1
0
9
22
*It should be noted that 18 welds originally reported to be cracked

were later reevaluated and determined not to be cracked, so are not
included in these totals.
**After inspecting approximately 7 welds, and finding cracks in 4 of
them, the utility decided to replace the piping with Type 316NG, so
has not completed the examination.
3-4
No. of
Welds
Overlay
Repaired
remain well within ASME Code limits for one fuel cycle. Other utilities, with
less extensive cracking, are considering alternative measures.
Some utilities
have performed IHSI on welds with minor cracking to preclude further crack
extension.
Decisions regarding the acceptability of further operation under
the above conditions are needed.
3.3
FOREIGN EXPERIENCE
A special meeting of the Committee on the Safety of Nuclear Installation

(CSNI) was held at the NRC's request in February 1984. The purpose of this
meeting was to review the experience of IGSCC among the various countries with
operating BWRs, and to discuss technical and licensing aspects of the
problems.
3.3.1
A summary of the incidents reported by the representatives follows.
Japan
A total of 43 welds at six plants were found to have IGSCC. Most of the
cracking was discovered before 1978. Thirteen of the 43 cracked welds were
detected by leakage; the others (30) were detected by NDE. Thirty-two cases
of IGSCC were in pipes 4 inches or larger in diameter (up to 22 inches); the
remainder were in the smaller-diameter piping. No cracks have been found
since 1982, probably because of extensive mitigating actions, including
corrosion resistant cladding (CRC), solution heat treatment (SHT), IHSI, or
replacement with 304NG or 316NG.
been replaced.
In all known cases, the cracked pipe has
In plants not yet operating, recirculation piping has been or
is being replaced with material such as 316NG.
In operating plants, replace-
ment with 316NG or similar materials has not been completed but has been done
selectively.
3-5
The water chemistry of Japanese plants, particularly with regard to

residual ions contributing to conductivity, is about one order of magnitude
better than that in U. S. plants, which probably contributes to the lower
incidence of IGSCC in the older sections of Types 304 or 316 SS Deaeration
during startup markedly reduces the level of dissolved oxygen prior to ascent
to power (e.g., from approximately 3.0 ppm to approximately 0.1 ppm).
3.3.2
Spain
During the 1983 outage, a large sample of welds was inspected at the
Santa Maria De Garona Plant.
In the recirculation system, 66 out of a total
of 154 welds were inspected and cracks were reported in 15.
Two additional
cracks were found, one in the LPCI system, and one in the CRD return line.
Actions taken were similar to those being taken in the United States.
Four
welds were repaired by weld overlay, two were repaired by other means, and the
remaining eleven were evaluated as acceptable for further operation.
3.3.3
Germany
In the newer German BWR's, Type 347NG stainless steel is used for piping
less than 12-in.-diameter.
This is an alloy that is stabilized against
sensitization; therefore IGSCC would not be expected.
Larger diameter piping
is carbon or low alloy steel clad with Type 347 weld metal.
The main recir-
culation piping in Gundremmingen Unit A (a dual-cycle BWR) was made of Type

304 stainless steel.
Safe ends were furnace sensitized.
In 1978, cracks were
found in the 24-in.-diameter safe ends and piping at the safe-end-to-pipe

welds.
Germany.
This is the only case of IGSCC in stainless steel reported in

This reactor has now been decommissioned.
3-6
3.3.4
Switzerland
One crack was found and identified as IGSCC in the recirculation bypass
line at the Muehleberg plant.
The bypass line was removed and capped.
Recent
inspections, in 1982 and 1983, disclosed a cracklike indication in a weld to

the main recirculation discharge valve. Although the reported inspections
disclosed no growth during this period, the utility plans to replace this
section of pipe in 1984.
3.3.5
Sweden
Although most of the piping in Swedish BWRs is clad ferritic steel,

several cases of IGSCC have been found in Type 304 SS in Ringhals Unit 1 and
Oskarshamn Unit 1. Most cases were in relatively small-diameter piping, and
were attributed to combinations of fabrication problems.
These were of
concern, because it was believed that the relatively low carbon level (less
than 0.05%) required in Sweden would minimize IGSCC.
In 1982, leaks were noticed in the 4-in.-diameter control rod drive

hydraulic system.
An ultrasonic examination was conducted on all 4-in.-
diameter piping, and 39 indications were found, of which six had to be

repaired.
Because the piping involved had a relatively high carbon content,
(near the 0.05% limit), it was all replaced in 1983 with low-carbon material.
Subsequent examinations showed that most of the indications reported were not
cracks.
3-7
One small leak was found in a bypass line in Barseback 1 during 1983;
again, the problem was attributed to poor fabrication processes.
In summary, there have been only about 20 incidents of IGSCC in Swedish

plants, primarily in 4-in.-diameter piping welds.
The Swedes are actively
pursuing hydrogen water chemistry as a remedial action to prevent future

incidents of cracking.
3.3.6
Italy
The only case of IGSCC reported in Italy occurred at Garigliano in 1978.

A safe-end weld was removed, and destructive examination disclosed IGSCC. No
IGSCC has been found to date in the Caorso plant, but there are plans to
perform an augmented inspection using improved methods during 1984.
3.4
CONCLUSIONS
Although significant IGSCC was found in the United States in the past,
no cracking in large pipes was reported until recently.
As a result
of extensive inspections performed during the past two years, IGSCC

has now been reported in 19 out of 23 plants. All other countries
have experienced IGSCC in piping made of Type 304 stainless steel,
including large-diameter Type 304 stainless steel piping, in those
plants that have it.
3-8
In all other countries the preferred long-term solution is to replace

piping made of Type 304 stainless steel with carbon steel or one of
the low-carbon or stabilized grades of stainless steel. Type 316NG is
considered the preferred choice, except in Germany, where they prefer
their special nuclear grade of Type 347.
Interim or short-term fixes in other countries follow the same approaches as have been used in the United States, and include IHSI,
weld overlay reinforcement, welded "clam shell" reinforcement, lastpass heat sink welding, and interim operation with unrepaired small
cracks.
It should be noted that IHSI on uncracked Type 304 stainless
steel pipe is considered to be a permanent fix in Japan.
Hydrogen water chemistry is being seriously considered by several

countries as a desirable adjunct to other solutions.
3-9
4.0
4.1
INTRODUCTION
Early detection of cracking in primary piping systems is important for

economic as well as safety considerations.
Ultrasonic inspection is one of
the major technologies for early detection of pipe cracks.
Recent field experience and round robin tests have demonstrated a need
for improvement, as well as some possible mechanisms for improving the
reliability of crack detection, interpretation, and characterization.
In
addition, the NRC and industry have supported extensive research efforts aimed
at quantifying the capabilities of nondestructive evaluation techniques.
This chapter will discuss:
Current Ultrasonic Examination Requirements
Technical Problems of UT Inspection
Actual Field Experience before and after the issuance

of IE Bulletins 82-03 and 83-02
UT Round Robin and Laboratory Experience
Recent Improvements in Ultrasonic Inspection
Personnel, Equipment, and Procedure Qualification

4-1
4.2
Foreign Experience
Leak Testing Requirements and Techniques
Conclusions and Recommendations
CURRENT ULTRASONIC EXAMINATION REQUIREMENTS
4.2.1
Applicable Requirements
The preservice and inservice examination requirements for the piping

systems in nuclear power plants are set forth in the Code of Federal
Regulations, Part 50, Section 50.55a, which references the ASME Boiler and
Pressure Vessel Code, Section XI, "Rules for Inservice Inspection of Nuclear
Power Plant Components." Section XI is published every three years, with
semiannual Addenda.
In determining which edition and addenda of Section XI are applicable,

the following rules are used.
(However, utilities may choose to update to
later editions of the Code.)
The edition of Section XI to be used for a plant's preservice

inspection program (PSI) is determined by the date of issuance of the
plant's construction permit.
The edition of Section XI to be used for a plant's inservice

inspection program is determined by the date of issuance of the
plant's operating license.
4-2
During the intervening time
the duration of construction
has typically undergone significant changes.
Section XI
These fundamental differences in
examination criteria undermine the major purpose of the PSI which is to

provide an NDE baseline measurement of the examined components, for comparison
with the findings of subsequent inservice inspections.
4.2.2
Example of Requirement Changes
An example of the far-reaching effects of changing Code editions is

illustrated in the change in ultrasonic calibration reflectors between the
1974 edition of Section XI (Summer 1975 Addenda) and the 1977 edition of
Section XI (Summer 1978 Addenda).
The 1974 edition requires that calibration be performed using sidedrilled holes, located at either 1/2T, or at 1/4, 1/2, and 3/4T (depending
on wall thickness).
The 1977 and later editions require calibration on
end-milled notches that are 10% of the wall thickness in depth.
These
notches are placed on the inner and outer surfaces of the block.
The impact of the change in calibration reflectors between the 1974

edition of Section XI (Summer 1975 Addenda) and the 1977 edition of Section XI
(Summer 1978 Addenda) was investigated by PNL under two NRC programs (4.1,
4.2).
Data from these studies consisted of approximately 540 measurements on
34 piping calibration standards.
The standards were from a boiling water
reactor (BWR) currently under construction and ranged in wall thickness from
0.237 to 2.343 inches.
steel.
Twenty-six were carbon steel and eight were stainless
The samples contained both side-drilled holes and notches in accord-
ance with Code requirements.

4-3
Measurements were performed by establishing a distance amplitude curve

(DAC) using the side-drilled holes and then measuring the response of the
notch relative to this DAC curve.
The notches produced responses that
exceeded the response of the side-drilled holes.
A calibration performed
using notches will, therefore, be less sensitive than a side-drilled hole

calibration.
This reduction in sensitivity depends on wall thickness,
ranging from -6 dB (a factor of 2) at 0.4 in. to -16 dB (a factor of 6.3)

at 2.4-in. wall thickness.
Theoretical calculations were performed to assure that these experimental

results were reasonable.
The theoretical calculations were in close agreement
with the measured results.
Thus, the results of this study show that preservice examination performed to the requirements of the 1974 edition of Section XI may be significantly more sensitive than inservice examinations performed to the 1977
edition of Section XI.
Consequently, preservice and inservice examinations
may have different reliabilities (4.1, 4.2).
4.2.3
Comments on Standard Review Plans and Generic Safety

Issues
Sections 5.2.4 and 6.6 of the U.S. NRC Standard Review Plan (NUREG-0800)
outline the review and compliance criteria for inservice inspection of ASME
Class 1, 2, and 3 components.
The Standard Review Plan criteria referenced in
Sections 5.2.4 and 6.6 are the requirements of Section XI of the ASME Code.
As discussed in the preceding paragraph, changes in Code requirements can
4-4
cause significant variation in inspection reliability.
Therefore, review
criteria in Standard Review Plans may be inadequate because Code requirements

are inadequate.
Generic Safety Issue A-42, "Pipe Cracking in Boiling Water Reactors," was
addressed by NUREG-0313, Revision 1.
The report requires, in part, that all
applicants/licensees of BWR plants provide a program for replacement of

service-sensitive lines.
If design criteria for replacement of lines includes
consideration for ultrasonic inspection, improvement in subsequent ultrasonic

examination reliability would be possible.
Generic Safety Issue A-14, "Flaw Detection," has been dropped.
The
rationale for this is that flaw detection is not by itself a safety issue.
Flaw detection is important on a problem-specific basis, i.e., pressurized
thermal shock.
Research and development work is occurring through the U. S.
NRC, Office of Nuclear Regulatory Research, and the industry to improve flaw
detection reliability.
The Task Group on Pipe Cracking supports these efforts
and supports timely completion of the work for resolution.
4.2.4
Comments on Current ASME Code Requirements and Suggested Improvements
Round robin tests (4.5) and other studies (4.3) have shown inadequacies
in ASME Code requirements.
Research activities in the Office of Nuclear
Regulatory Research and the industry and field experience have identified
methods and means to improve the Code.
The following list details areas where
inspection reliability can be improved and inspection variability can be

reduced.
4-5
Code Case N-335 - This Code case should be made mandatory for all
required inspection.
Calibration Blocks - For austenitic stainless steels, the calibration

blocks should contain welds, and the calibration reflectors should be
located either in the weld or on both sides of the weld.
In addition,
the calibration block and pipe should have the same nominal microstructure.
Calibration Reflectors - When notches are used as calibration

reflectors, compensation should be required for the sensitivity
differences between notches and side-drilled holes.
Search Units - In addition to the currently required 45

examination for welds, an additional 60
be required.
shear wave
shear wave examination should
The potential advantages of the refracted longitudinal
wave technique for dissimilar metal welds, weld metal, and far-side
inspections should be recognized, and this technique should be required as a supplementary examination.
Beam Spread - If beam spread corrections are made as permitted by the

ASME Code, flat calibration reflectors should be used for the corrections to crack and lack-of-fusion-type indications.
Angled Defects - A skewed scan should be required to detect defects

oriented other than parallel or perpendicular to the weld.
4-6
Crack Length Sizing - The 50% DAC method of crack length sizing should
be revised to require that end points of a flaw be determined by loss
of signal amplitude to the background noise level.
4.3
TECHNICAL PROBLEMS OF ULTRASONIC INSPECTION
Ultrasonic inspection of piping is hindered by a number of conditions

that are encountered in the field.
This section considers those conditions
which have the largest adverse impact on the inspection.
4.3.1
Component Access. Geometry, and Environment
The inside surface contour of a pipe weldment affects the ultrasonic

examination.
The presence of a number of conditions such as counterbores,
inner surface cladding, mismatch, weld root drop-through, or suck-back can

produce nonrelevant ultrasonic indications that can be misinterpreted as being
caused by flaws, sometimes resulting in unnecessary repairs.
Conversely,
since at least one of these conditions is usually present, the UT operator

may mistakenly attribute a flaw indication to one of these ID geometric
reflectors.
Usually IGSC cracks occur in the weld heat-affected zone (HAZ) near the
weld root.
Since the root of a stainless steel weld usually returns a signal,
it can be difficult for the UT operator to distinguish between the root indication and an adjacent crack indication. A crack indication may be mistaken for
an irrelevant root indication, or a root indication may be misinterpreted as
being from a crack. An unambiguous discrimination can sometimes be achieved
4-7
by making careful measurements of search unit position and pulse transit time
for plotting purposes, but these measurements increase the time spent in the
radiation area, and are subject to errors caused by beam redirection (bending)
effects.
Ultrasonic inspection is further complicated by inadequate access for

examination.
Usually pipe joint design and installation are performed without
consideration for adequate access for UT weld inspection.
Access is often
available from only one side of the weld, and may be further limited by the
presence of pipe supports or other restraints.
Poor outer surface finish
and/or excessive weld crown contour can also reduce the effectiveness of UT
examination.
Ultrasonic inspection is most effective when applied in a deliberate and

careful manner.
In the environment of a nuclear power plant, however, this is
often not possible.
The inspections may be performed under conditions of
great discomfort due to high temperature and humidity, exacerbated by the

necessity for wearing respiratory equipment and multiple layers of protective
clothing.
Accurate manipulation of the search unit is necessary with
simultaneous visual contact with the UT instrument, but since the pipe runs
are not designed for inspection, the UT operator must often assume very
awkward positions in order to reach the examination areas.
Finally, health
physics considerations frequently require that the examination be performed in

a very short time.
4-8
4.3.2
Effect of Material Properties on Inspection
The acoustic properties of the material affect the propagation of

ultrasound.
Sound beams encountering anisotropic grain structures undergo
reflections and refractions that do not occur in grain structures of isotropic

materials.
The grain size may vary from point to point in the pipe base
material, and will vary between the pipe base metal and the HAZ.
The weld
metal will also have different grain size and structure than either the HAZ or
the pipe base metal.
These variations in grain size and grain structure tend
to increase the attenuation of the beam and also return irrelevant reflections
that may complicate the instrument screen presentation.
These effects are
most severe in cast stainless steel materials and stainless steel welds.
The cumulative effects of refraction at many grain boundaries can cause

the sound beam to travel in a curved path instead of the expected straight
line, resulting in mislocation of indications and sometimes in failure to
insonify* parts of the required inspection volume. This is particularly a
problem in attempting to inspect the far side of a stainless steel weld by
directing the sound beam through the weld metal.
When applying the con-
ventional shear wave examination required by the Code, the sound beam may be
guided by the weld metal grain structure to the weld root, and the far-side
HAZ may not be inspected.
*The definition of insonify is to provide the proper sound field distribution

throughout the volume of material undergoing examination.
4-9
Variable UT attenuation in the base metal causes the sensitivity of the

examination to vary.
Methods of measuring and correcting for these variations
exist, but they significantly increase the inspection time and radiation
exposure.
Many cracked welds have recently received overlays of stainless steel

weld metal on the outer surface, as a means to achieve reinforcement and
stress reversal for a temporary remedy pending pipe replacement.
These weld
overlays present the same inspection problems experienced in cast stainless

steel.
Attenuation in the overlay will be high and may also vary with loca-
tion, the beam angle may be unpredictable, and a high UT noise level will
probably be present.
Detection of new cracking and/or monitoring known cracks
below the overlay may be unreliable when using manual (conventional) UT

equipment.
4.4
4.4.1
FIELD EXPERIENCE
State of UT Practice in ISI Before Mid-1982
Before the issuance of IE Bulletin (lEB) 82-03 in October 1982, UT

procedures reflected the requirements of two documents:
and NUREG-0313, Revision 1.
for examining BWR piping.
ASME Code Section XI,
Section XI defined the basic minimum requirements
NUREG-0313, Revision 1 required augmentation of the
inspection sample and inspection frequency to include a larger sample and more
frequent inspection than specified by Section XI.
This document also required
that UT examinations be performed in accordance with Section XI, but noted

that the Code minimum requirements may be inadequate for IGSCC detection.
4-10
Hence, most utility and ISI vendor UT procedures designed for NUREG-0313,
^Bevision 1 applications are enhanced beyond the Code minimum requirements.
The most common enhancements are added sensitivity, the requirement that
search units be optimized for IGSCC detection, and inclusion of scanning
motions appropriate for detection of IGSCC that is neither parallel nor
perpendicular to the weld.
4.4.2
IE Bulletins 82-03 and 83-02
As discussed in Section 3.1, leaking IGSC cracks were visually detected,

in March 1982, at two recirculation system safe-end welds in Nine Mile Point
Unit 1.
The welds had been ultrasonically inspected nine months earlier,
but no reportable indications were discovered.
After the leakage, axially-
oriented inner surface cracking was confirmed using a modified UT procedure.

Additional examinations of large-diameter recirculation welds disclosed inner
surface cracking adjacent to a large number of the welds.
As a result of the experience at Nine Mile Point, IE Bulletins (lEB)

82-03 and 83-02 were issued in late 1982 and early 1983.
lEB 82-03 was
applicable to nine BWR plants (operated by eight different utilities), and

required that the UT methodology and detection capability for use on
recirculation system welds be demonstrated on service-induced cracked samples.
The demonstration required by lEB 82-03 showed a tendency toward both severe
overcalling, and severe undercalling, since some procedures and equipment were
inadequate for the task.
The major benefit of lEB 82-03 was probably the
demonstration to the NRC, the licensees, and the ISI vendors of the overall
ineffectiveness of the UT/ISI inspections that were conducted prior to late
9^82.
4-11
The issuance of lEB 83-02 had an even greater impact since it applied to
14 additional BWRs, and required that each field UT/ISI team demonstrate its
UT process using a minimum of 80 linear inches of service-induced crack
samples.
An 80% detection rate was required, and a penalty was imposed for
false calls.
The general requirements included a six-hour time limit, and NRC
review of the proposed procedure.
lEB 83-02 also specified an augmented ISI
sampling plan which included (as a minimum) ten welds in recirculation pipes
larger than 20 inch diameter, ten welds in recirculation risers and safe ends,
and two sweepolet-to-manifold welds near end caps.
If cracks were found,
additional sampling per IWB-2450 was required.
The results from the performance demonstration data of IE Bulletin 83-02

were:
More meaningful demonstrations (relative to lEB 82-03) were achieved

owing to the requirements for a time limit, a penalty for false calls,
specified size and makeup of the UT team, additional specimens,
specific pass/fail criteria, and better uniformity between NRC
Regions.
The overall detection capability of the UT process (personnel,

equipment, and procedures) for BWRs was definitely enhanced as a
result of passing lEB 83-02.
The basis for this statement is the
dramatic increase in the number of field indications detected and

classified as IGSCC per inspection.
A fully definitive evaluation of
these results is not yet possible pending completion of the correlation of UT results with further examination of pipe welds removed from
the field.
4-12
since each team's success was the result of their combined efforts,
individual team member performance could not be evaluated.
The limited number of specimens and cracks does not permit a

statistically significant assessment of performance.
A psychological factor is also involved.
Before the Nine Mile Point
experience, an operator claiming to have found a crack in a large

recirculation pipe could expect to meet skeptical resistance, because of the
general perception that these pipes were not susceptible to IGSCC.
Because of field experience, the results from the bulletin performance

demonstrations and the rather discouraging results from round robin studies of
UT crack detection capability, the EPRI/NDE Center initiated development and
implementation of a series of one-week training courses on IGSC crack
detection using test specimens removed from cracked Nine Mile Point piping
(4.3).
The course outline includes UT theory, IGSCC morphology, a generic
procedure, laboratory exercises to improve operator proficiency, and a threepart final examination.
Each student's performance is reported to his
employer for certification use, although the NDE Center does not certify
successful graduates.
As of early January 1984 this course was being offered at least monthly,
with about 15 students per class.
The passing rate for students has been
about 50%, and passage of this course has been considered as one acceptable
4-13
way to meet the requirements of lEB 83-02.
The major benefit of such courses
is that the same information is being distributed to all who need it; hence,
some element of industry-wide standardization is occurring.
However, the
number of specimens being used during the final, practical examination is

insufficient to provide high statistical confidence levels.
On a statistical
basis alone, the performance demonstrations resulting from lEB 83-02 may be
analyzed from two viewpoints:
as a measure of individual inspector
capability, or as a measure of the average capability of the inspector

population.
A probability of detection of at least 42% at a 90% confidence
level is calculated for an individual inspector capability using classical

statistical methods (binary) for a 4 out of 5 test.
Using Bayesian statistics
and assuming a 20% false call probability during the demonstration tests, the
average capability of the inspector population (for those passing the
demonstration test) is calculated at a level of 65% probability of detection.
However, the Nine Mile Point specimens in lEB 83-02 performance demonstrations
may not be typical of IGSCC conditions in other BWR plants and the testing
environment may not be a reasonable approximation to an in-field environment
and therefore complicates the extrapolation of the test results to field
performance.
Even with the foregoing, it is quite evident that IE Bulletins 82-03 and
83-02, and the resulting IGSC crack detection demonstration and training
courses at the EPRI/NDE Center, have produced both positive and tangible
results.
The augmented-type inspections conducted since these have been in
effect have generally resulted in much higher reporting rates for IGSCC in the
4-14
primary piping systems of BWRs, and a corresponding general increase in the

^Hconfidence that can be placed in these examinations.
To maintain this level
of proficiency, there should be a requirement that plant examinations be

conducted using the same equipment and procedures used to pass performance
demonstration tests to eliminate the possibility that a successful inspector
could still be an ineffective operator by using equipment or procedures that
are inherently not capable of detecting IGSCC.
4.4.3
Crack Sizing
In late 1983, the EPRI/NDE Center initiated a "crash" program to develop

and implement an additional one-week training course on UT crack sizing.
This
course has also been structured in the "module" format, and each of several
sizing techniques will involve modules and theory and procedures, plus some
laboratory exercise time. Written and practical examinations will be given to
evaluate student performance.
The techniques to be covered will include:
amplitude drop, sound trapping (full V-path), various crack-tip detection

methods such as shear and longitudinal waves at 1/2 and full V-paths, and
high angle longitudinal waves (also called creeping waves).
Because UT crack sizing often produces questionable results, even under

laboratory conditions, the actual value of this crack sizing course remains to
be seen. Certainly, some direct benefits will accrue since all of the
students will develop a greater appreciation for the current difficulties.
However, it should be recognized that the availability of a course and its
graduates may not necessarily result in a routine and reliable process for
crack sizing.
4-15
4.5
ROUND ROBIN EXPERIENCE
Two UT round robins have been conducted in response to piping problems.

A summary of the results from each follows in the next two subsections.
4.5.1
PNL Round Robin
A round robin test of pipe inspection reliability was initiated in 1981

at the Pacific Northwest Laboratory (PNL) under NRC sponsorship.
The
objective was to measure the crack detection and sizing reliability of

inspection teams from commercial ISI vendors (4.4).
Three-man inspection teams from six commercial ISI vendors participated

in the Pipe Inspection Round Robin.
Each team consisted of Level I, II, and
III examiners, selected to represent neither the best nor the worst levels of
ISI experience available.
Each team spent three weeks at PNL, working eleven hours per day six
days per week to complete a carefully designed test matrix of 253 separate
inspections per team.
A variety of inspection conditions were simulated, but
in almost all cases the specimens were masked to permit inspection access from
only one side of the weld.
The results and conclusions drawn from this pipe inspection round robin
test were as follows:
4-16
UT detection of cracks in clad ferritic main coolant pipe can be 100%

effective, if adequate sensitivity is used per ASME Code Case N-335.
Section XI minimum sensitivity (50% of the notch amplitude) is not
adequate.
Detection of cracks in clad ferritic pipe is almost equally effective

with and without weld metal in the sound path.
UT detection of cracks in centrifugally cast stainless steel (for

PWRs) is ineffective using the conventional manual techniques currently applied in the field.
The false call rate was almost identical
to the probability of detection and correct interpretation (PODI)

rate.
Section XI minimum requirements do not provide effective inspection of

wrought stainless steel pipe welds.
Increased sensitivity and selec-
tion of optimized search units improves detection reliability.
For
both IGSC and thermal fatigue cracks in stainless steel, the six teams
achieved an average PODI of 50-60% when using their own procedures
for cracks 15% throughwall or greater.
When the sound beam must pass through the butt weld in wrought stainless pipe, UT inspection using current field techniques is ineffective.
4-17
Variability in crack detection reliability is significant from

operator to operator, even when identical equipment and procedures,
are used.
Crack detection reliability in UT inspection of stainless steel pipe

welds should be qualified by test.
Crack length measurements were in general quite good.
There was a
trend to oversize very small cracks and to undersize very long cracks
by small amounts.
A conservative approach would be to record length
based on signal reduction to the background noise level.
Crack depth measurements by all the teams using the Code-advocated

method of amplitdue drop was totally ineffective.
4.5.2
EPRI Sizing Study
EPRI conducted a round robin test of IGSC crack depth measurement

capability in 1983 (4.3).
The objective of the exercise was to generate
information necessary to assess the current state of practice of ultrasonic

depth measurements of IGSCC found in the piping of some BWRs.
Seventeen teams
from ISI vendors, utilities, research organizations, NRC, and one foreign
utility participated.
Thirteen IGSC cracks and three EDM notches were
included in the test.
Two of the cracks were removed from the Nine Mile Point
BWR; the rest were laboratory-cracked pipe from PNL and IHI in Japan.
All of
the deep reflectors were either notches or wide open laboratory-induced

IGSCC.
After the UT measurements were finished, nine of the specimens were
destructively evaluated to determine the true crack depths.
4-18
The results showed a general tendency to oversize the cracks that were
less than 20% throughwall and to undersize those deeper than 20%.
some teams definitely performed better than others (4.3).
techniques were also included in the round robin.
However
Two advanced UT
Their effectiveness was
similar to that of the better manual teams.
The EPRI report contained the following conclusions (direct quote):
"The results show the following:
The number of teams doing an adequate job is much lower than

anticipated.
The range of performance from best to worst is very large.
The advanced techniques provide verification that the crack-tip

diffraction sizing approach is the most viable method.
The influence of the ultrasonic measurement uncertainty to the flaw

evaluation results must be assessed on case-by-case basis.
Corrective actions are needed."
Analysis of the EPRI sizing study results in the following additional

observations:
4-19
The most important conclusion of the EPRI sizing study was that none
of the teams in the study could size accurately.
Even though some of
the teams showed a significant relation between measured and true

crack size, the scatter of the measured data is so large that crack
depths could not be determined accurately; at best l/3t.
A relatively small number of nonrepresentative reflectors were used

during the EPRI sizing study.
This reduces the accuracy with which
the team's performance can be determined, particularly for deep

cracks.
Furthermore, the thick-walled specimens contained only
shallow cracks, so the regression results were largely determined from

thin-walled pipe specimens which contained a broader range of crack
sizes.
The use of EDM notches and wide cracks could tend to aid crack sizing
performance, particularly for techniques relying on crack-tip signals.
Since most field cracks have been found by destructive analysis to be
tight, basing sizing performance on EDM notches and wide cracks is
nonconservative.
4.6
RECENT IMPROVEMENTS IN ULTRASONIC INSPECTION
4.'6.1
Qualification of Personnel and Procedures
The major impact of IE Bulletins 82-03 and 83-02 has been in demonstrating inspection capability.
While the concept of demonstrating inspection
capability is not new, the IE Bulletins marked the first effort to demonstrate
inspection capability on a national scale.
4-20
The results of the demonstrations
required by lEB 83-02 showed a dramatic increase in crack reporting during inservice inspection.
cost.
This increase in crack reporting has not been without
The results of inservice examinations at Dresden Unit 3 and other
plants has confirmed several cases of false calling.
4.6.2
Automated Ultrasonic Systems
A variety of automated UT systems have been manufactured.

systems are commercially available.
Few of these
Data appropriate for comparing automated
system performance to manual UT performance are rare.
The inspection functions which have been automated vary from system to
system.
For purposes of comparing one automated system to another, the
inspection functions may be broken out into data collection, data recordings,
and data interpretation.
Automated data collection by mechanical scanners improves the accuracy

and repeatability of indication location and facilitates comparison of the
data from successive ISIs.
The use of mechanical scanners reduces the need
for highly skilled UT operators to remain in radiation zones for general data
collection and thereby, expanding the capability for taking whatever limited
special manual data that an operator may choose.
In some instances, mechanical scanners may be ineffective or even

unusable.
Current design limitations generally restrict scanners from use on
complex weldment geometries such as those found in branch connections,

sweepolets, etc.
Some automated scanners are not capable of skewing the
4-21
search unit from side to side, which is necessary for detecting IGSC cracks
oriented at an angle to the weld, and for optimizing the UT response from
other obliquely oriented planar reflectors.
Physical access for mounting some
scanner designs may be precluded by the presence of pipe supports, restraints,

branch connections, etc.
Automatic techniques can record information on signal properties and

physical coordinates many orders of magnitude more quickly than manual
techniques.
More complete information can be recorded while still reducing
the UT operators' radiation exposure times.
Automatically recorded data are
more accurate and less error-prone than manually recorded data.
Automated
techniques tend to record raw, unreduced data, thus allowing the possibility
of reprocessing archived raw data using new interpretation schemes as they are
invented.
While data interpretation can be broken out logically as an inspection

function, it should be kept in mind that it requires compatible data records
and in some cases data collection using scanners with exacting specifications.
Thus, the systems that automatically interpret data also tend to collect and
record data automatically.
Automated interpretation eliminates the occurrence of random errors and

nonuniform interpretations caused by humans.
Depending on the interpretation
algorithm used and its implementation, a speed advantage over human interpretation may result; on the other hand, some current algorithm implementations
are not compatible with real-time operations.
4-22
Automated data interpretation involves manipulation of the data, either

^ B by directly extracting waveform features, or by first using the waveforms in
an infaging scheme and then extracting image features.
In either case, these
features are correlated with features previously associated with particular

types, sizes, shapes, and orientations of defects.
Southwest Research Institute developed a computer program called FLAWSORT

that interprets waveforms to be either from cracks or from geometry on the
basis of a preset algorithm.
Adaptronics sells an Adaptive Learning Network
(ALN) that performs similarly, on the basis of an algorithm that it generates

while "learning" the difference between data it is told represent cracks and
data it is told represent geometry.
Several systems perform imaging:
the Synthetic Aperture Focusing
Technique (SAFT) developed under NRC sponsorship at the University of

Michigan, Southwest Research Institute and currently at Pacific Northwest
Laboratory; the Advanced Ultrasonic Testing System (AUTS) at EG&G Idaho; the
Projection Image Scanning technique (P-Scan) developed by the Danish Welding
Institute; the Ultrasonic Data Recording and Processing System (UDRPS) at
Dynacon Systems in California; the Ultra-Image III developed by General
Dynamics; the ALN 4060 developed by Adaptronics; and acoustic holography as
practiced at several organizations.
Most of these imaging systems have not
yet implemented fully automated image interpretation.
A research program sponsored by the NRC is currently evaluating the

merits of several of the above advanced UT systems.
is scheduled for June 1984 publication.
A report from the program
Preliminary data are encouraging
4-23
particularly for some of the imaging systems.
Difficulties with the mechani-
cal scanners are a major limitation for application of the systems in the
field.
4.7
PERSONNEL AND PROCEDURE QUALIFICATION
Although a general upgrading of qualification requirements does not

automatically assure more effective field inspections, it does increase the
intrinsic capability for a reliable inspection and intensifies the awareness
of the field-inspection personnel to the significance of their work.
4.7.1
Qualification Document
Improvements in the detectability of intergranular stress-corrosion

cracking (IGSCC) on reactor piping systems by the ultrasonic inservice
inspection teams following implementation of IE Bulletins 82-03 and 83-02
showed that upgraded qualification requirements can provide definite
benefits.
However, further improvements in overall effectiveness of the
ultrasonic inservice inspection (UT/ISI) processes are considered to be both

necessary and achievable.
Toward this end, the NRC Office of Nuclear
Regulatory Research is developing a document that specifies more rigorous

qualification requirements than those now in effect.
The scope of the document will include qualification requirements and

criteria for the critical elements of a UT/ISI system; namely, personnel,
equipment, and procedures.
The document will specify both general and
specific requirements and criteria for the qualification and requalification

of UT personnel, equipment, and procedures.
4-24
The most significant features of the qualification approach specified in

^^^this document are:
1.
A statistically significant performance qualification demonstration

is required.
2.
The UT personnel, equipment, and procedures are all considered to be

critical inspection elements, and all three are involved in the
qualification process.
3.
Classroom training is required prior to initial qualification, and

annually thereafter.
4.
Guidelines are provided so that users of the document can generate

application supplements.
The performance qualification demonstrations are intended to assure

minimum levels of inspection capability.
Limited qualification will be
possible; for example, a UT operator may be qualified to interpret indications, but not to characterize those indications which are interpreted to be
flaws.
This document will specify requirements and processes for conducting

"blind test" performance demonstrations to quantify the flaw detection
probability and characterization accuracy for candidate UT/ISI systems.
The
document will also specify the flaw detection probability and characterization
accuracy that must be demonstrated as a qualification for performing UT/ISI.
4-25
Although the implementation of such a program is not a trivial consideration, little emphasis has been placed on this aspect to date.
Rather,
the initial approach has been to focus most of the effort on developing
and refining the technical aspects of the program.
Once the program has been
defined and refined, the emphasis will be shifted towards addressing the
implementation aspects.
4.7.2
Ad Hoc Utility Committee Document (NUC-MR-IA)
In 1982, the Ad Hoc Committee for Development of Qualification

Requirements for Nuclear Utility Examination Personnel was established for the
purpose of developing minimum requirements for qualifying NDE personnel, and
recommendations for implementing these minimum requirements.
This group
submitted a proposed minimum requirements document (NUR-MR-IA) to the NRC in

September 1983.
The Ad Hoc Committee document (NUC-MR-lA) represents both an
expansion of the "recommended practice" document (SNT-TC-IA) that is published

by the American Society for Nondestructive Testing, and an effort to establish
a set of '\ninimum requirments" in lieu of the general "guidelines" that are
contained in SNT-TC-IA.
The result, however, is still a document that will require each user
(utility or ISI contractor) to develop an individualized "Written Practice"
to specify how that employer will comply with the '\ninimum requirements"
contained in NUC-MR-IA.
This approach could be viewed as a basic weakness
of the NUC-MR-IA document since users may be more inclined to tailor their
written practices to their current programs, rather than to the spirit and
intent of NUC-MR-lA.
Although the Ad Hoc utility committee document did
4-26
increase some of the basic guidelines from SNT-TC-IA with respect to ex^^Krience, nimber of exam questions, etc. , many of the additional requirements in NUC-MR-IA can be regarded as rather superficial.
The document,
NUC-MR-IA, does contain an added section (beyond SNT-TC-IA) entitled

"Qualification for Special Applications," wherein the need for special
qualifications are recognized.
However, this section (consisting of three
brief paragraphs) simply states that 'Vhen circumstances require the

demonstration of additional examination capabilities, the Level III examiner
shall determine the need for qualification of personnel for such special
applications." This section goes on to say that special techniques or
equipment or problems may require special procedures, examinations, and
training.
Furthermore, such special qualifications must be documented.
If, and when, this document is implemented it is expected that the status
quo will not change because most employer's written practices will require
only minor changes and will therefore, be inadequate.
4.7.3
Proposed Appendix to ASME Section XI
The Section XI Working Group on Nondestructive Examination (WGNDE) is

developing a procedure and personnel qualification requirements Code Case for
Section XI.
The proposed approach is patterned after qualification processes
performed under lEB 83-02.
4-27
The document defines the requirements for qualification of ultrasonic

examination procedures and personnel for detection and sizing of IGSCC.
The qualification requirements apply to personnel who record, interpret,
characterize, and size IGSCC flaws.
The document requires that both procedures and personnel be qualified on

five IGSC cracks and provide requirements for the other aspects of the
qualification test matrix.
The following requalification requirements are
being proposed:
When an examiner has not found confirmed IGSCC flaws or participated

in a blind test on a single approved test specimen for a period of six
months.
When there is specific reason to question his ability.
The document provides the following list of essential variables.
Instrument or system - make and model.
Search unit - size, type, angle, frequency, manufacturer's model.
Methods and criteria for establishing examination sensitivity levels

including instrument controls to be used.
Scanning surface limitations, if applicable.
The volumes to be
examined, and the types and orientation of flaws to be detected.
4-28
Detection and sizing techniques including angles and modes of wave

propagation in the material, directions, maximum speed, minimum
overlap, and extent of scanning, method of calibration, and method of
establishing scanning sensitivity levels.
Data to be recorded and method of recording.
Method and criteria for the discrimination of indications (e.g.,

geometric versus flaw indication).
The document requires requalification if any essential variable is

changed.
If the examination procedure allows a range of any essential
variables, the document requires qualification over said range.
The Code Case being developed by Section XI ensures qualification of

procedures and personnel on realistic flaws. However, the number of flaws
required for qualification by the document is inadequate and will not provide
a statistically significant performance demonstration. Adopting the proposed
appendix will only incorporate current personnel qualification requirements
(i.e., lEB 83-02) into the Code.
4.8
AREAS OF RESEARCH
There are active research programs currently being pursued by the

industry and the NRC.
The intent here is not to describe these programs but
to focus on the issues that need to be emphasized or added to the existing

programs to address issues identified in this section and to provide timely
resolutions.
4-29
4.8.1
Applied Research
Recent laboratory research and field experience indicates that the need
for improvement is still critical for both detection and characterization
(dimensioning) of cracks.
Specifically, the following inspection issues need
to be addressed immediately:
Develop and validate effective and reliable manual UT inspection

methods to correctly interpret UT indications.
Develop ultrasonic techniques for dimensioning flaws in the throughwall plane.
Develop inspection techniques for detection and dimensioning of flaws

in pipe repaired by the weld overlay process.
The following issues need resolution on a less urgent basis:
Develop reliable automated UT equipment to detect, interpret, and

characterize cracks.
Determine the effect on detection and characterization of cracks in

pipe that has had the induction heat stress improvement (IHSI), or
last pass heat sink (LPHS) treatments.
4-30
Develop a practical means for implementing a "transfer method" to

compensate for response differences between the calibration block and
the piping material.
This concept was recently eliminated from the
Code requirements since no practical means existed for its

implementation.
However, the concept has merit, and a serious
development effort is warranted.
Develop inspection techniques for examination of austenitic butt weldi

through the weld metal (i.e., far-side access which is typical for a
pipe-to-component weld).
Develop reliable inspection techniques for piping with inner surface

cladding, for welds in CCSS piping (for PWR) and for dissimilar metal
welds.
4.8.2
Establish the reliability of advanced techniques.
Basic Research
Additional research is needed on the interaction of ultrasound with

materials and cracks. Excellent modeling work has been completed for
predicting the results obtained from the Program for Inspection of Steel
Components (PISC) experiments. However, these modeling efforts need to be
expanded.
Specifically, finite element analysis modeling should be done for
cracks in austenitic (wrought and cast) material.

predict results and complement work in:
4-31
This work should be used to
Developing the inspection techniques listed above.
Predicting conservative crack characteristics that could be used in

qualification test samples.
4.8.3
Reliability of advanced techniques.
Human Factors Research
Research is needed to identify, isolate, and analyze the human factors

aspects of inservice inspection, and their influence on the overall
effectiveness of NDT/ISI. After these human factors aspects have been
identified and quantified, they should be prioritized with respect to relative
influence and relative correctability.
Finally, recommendations should be
developed toward mitigating the consequences of the major negative human

factors aspects.
4.9
4.9.1
FOREIGN EXPERIENCE
Experience in Sweden
Because of leaks found in the 4-in.-diameter control rod drive

hydraulic system of Ringhals Unit 1, UT was performed for 60 welds of
4-in.-diameter piping in April 1983 (4.5, plus private communication with
4-32
Dr. Ferenc de Kazincy of SA). Twenty-four welds were reported to contain 39

^^r crack indications. On the basis of this finding and higher-than-normal carbon
content in two heats from which the piping was made, the utility decided to
replace piping containing 247 welds with 316NG material.
On the cut-out
pipes, only 6 of the 39 indications could be metallographically verified as

being due to IGSCC.
Two more cracks were discovered that were not detected by
UT (one of which had initiated in the weld root).
Crack lengths ranged
between 10 and 70 mm. Most of the other UT indications were attributed to

geometrical reflectors such as weld roots and poor penetration.
the weld ID also produced UT indications.
Grinding on
Because of the high rate of false
calls in UT, it was decided that from 1984 on operators must be trained before
performing inservice UT on cut-out pipe welds which contain IGSCC.
An additional 90 welds from the cut-out pipes which had not been
ultrasonically tested previously were tested by liquid penetrant and 19 crack
indications were found, 15 of which were confirmed by radiography to be
cracks.
Further examinations are under way.
A comparison of UT response amplitude versus actual crack depth showed no

systematic correlation.
Length measurements, however, turned out to be rather
accurate using the 50% DAC technique.
The following tentative classification
of the indications is being used.
1.
Indications with high amplitude (300-1200% DAC) and a length of

60 mm (2.36 in.).
These are judged to have a relatively high
probability for causing unacceptable leakage rates.
4-33
2.
Indications with a high amplitude (300-1200% DAC) and a length of

40 mm (1.6 in.). These cracks are judged to have potential for
small, but acceptable leakage rates.
3.
Indications with a low amplitude (200% DAC) and a length of

40 mm (1.6 in.). These cracks are judged not to be harmful during
the next operating period (1000 hours).
All Type 1 UT indications are repaired, and Type 2 UT indications are

repaired, depending upon the possible consequences of a leak.
The Swedish Nuclear Power Inspectorate (SKI) has initiated development

projects to provide better discrimination (detection and interpretation)
procedures, and to develop more reliable sizing techniques.
A "consultative"
group has been formed from representatives of the utilities, SKI, the Swedish
Plant Inspectorate (SA), and a testing company.
This group is coordinating a
research program to improve UT techniques which will include:
Evaluation of all IGSCC examinations to date.
Qualification programs for UT operators.
Procedures for better discrimination of cracks and geometrical

indications.
Optimization of equipment (search units with different frequency,

damping, and refraction angles will be studied).
4-34
In l a t e 1983, SKI i n i t i a t e d a project for investigation of the ' c r a c k - t i p ^^Bef lection-technique. '
4.9.2
Experience in Japan
In a recent Japanese paper (4.6), it was reported that five techniques

were investigated during a study to measure crack depth by UT.
specimens were 5.3-cm-thick weld joints with natural defects.
The test
The UT
techniques investigated were tandem, tip echo, mode-converted surface waves,

tomography, and the ALOK method (amplitude and time-of-flight locus).
The
results showed no significant differences between the relative capabilities of

these five techniques for measuring crack depth.
This study will continue
until at least 1985.
The Nuclear Power Engineering Test Center (NUPEC) is conducting a program

to evaluate the reliability of ISI methods (UT and ET), and to modify the
inspection procedures to reduce the radiation exposure of inspectors (4.6).
This major program (8 years and 12 million dollars) was initiated in 1978, and
includes various full-scale mockups of portions of a variety of nuclear power
plant components.
The artificial defects to be introduced include EDM
notches, drilled holes, fatigue cracks, weld defects, and stress-corrosion

cracks.
For piping, the depth of the flaws will be from 5 to 15% of the
nominal wall thickness, and the flaw lengths will be 2 to 6 times the depth.
Most of the flaws will be located in the weld heat-affected zone.
Evaluations
of flaw detectability, overall technique performance, and inspection time will

be conducted for both manual and automatic UT techniques.
4-35
A comparison of the UT response from side-drilled holes and notches was

conducted (4.6).
These results showed that for 45
provided the greatest response; whereas, for 60
shear waves, the notches
and 70
shear waves, the
response was about the same for thicknesses up to about 20 mm, and the sidedrilled holes provided slightly greater response when the section thickness
was 30 mm.
The Japanese concluded that when the UT calibration sensitivity is
established using side-drilled holes, the flaw detection sensitivity will be

higher for 45
shear waves and lower for 60
and 70
shear waves, compared to
the sensitivity established using notches. Additional work was conducted to

correlate UT response amplitudes with the area (depth x length) of defects.
Although the data exhibited considerable scatter, usable correlations were
obtained and it was also found that the response amplitudes were influenced
more by depth than by length (with a constant area defect).
The influence of
applied stress on the UT response from fatigue cracks was also investigated.
The echo amplitudes varied approximately 25 dB owing to tensile and compressive loading for non-oxidized crack surfaces, but this difference was only
about 5 dB when the crack surfaces were oxidized (the situation they expect in
an actual pipe containing high temperature coolant water).
The Tokyo Electric Power Company (TEPCO) and the Toshiba Corporation have
reported on their experience at the Fukushima Unit No. 3 (4.7).
Investiga-
tions were conducted to evaluate six pipe branches removed from the PLR system
during the fifth outage in 1982, and the cracking was determined to be caused
by intergranular stress corrosion.
The response from UT indications (100% DAC
or greater) were compared before and after IHSI treatments were performed (in
April and May of 1981), and again in May 1982 after 7000 hours of plant operation.
The UT measured length of these indications appeared to decrease
4-36
during the period between the IHSI treatment and one year later.
The UT re-
^^Fsponse amplitudes also tended to decrease over this same time period.
A com-
parison of UT and liquid penetrant (PT) results indicated that UT detected

about 74.1% of the indications detected by PT.
For large UT indications (100%
DAC or more) the UT measured length was almost the same as the length measured
by PT.
The Japanese concluded from these results that the applied UT tech-
nique was "exact enough for indication evaluation." They also concluded that
the IHSI performed in 1981 had been an effective countermeasure for IGSCC
since no crack propagation or initiation was observed after an additional 7000
hours of plant operation.
Their comparison of the UT and PT results showed
that the UT techniques had been effective.
In a more recent informal report (4.9), the Japanese concluded that UT

examination of the welds in plants completed prior to 1980 is difficult
because:
1) the weld crown inhibits the UT scanning process, 2) weld suckback
and ID counterbore cause UT indications which inhibit evaluation of valid UT

signals, 3) the thickness and weld cross section are not accurately shown on
the drawings.
In view of these factors, very small search units (10 x 10 mm)
are used for longitudinal wave, 45 , and 60
angle beam scanning, both
parallel and perpendicular to the weld axis, and indications greater than 20%
DAC are recorded.
Indications that exhibit the following characteristics
receive special attention:
1) indications oriented parallel to the weld axis,
2) indications in the weld heat-affected zone (HAZ), 3) high amplitude

indications with lengths shorter than 100 mm, and 4) indications whose
amplitude is much higher than during the previous inspection.
4-37
A frequency analysis technique is used to distinguish IGSCC from

metallurgical indications at the weld fusion line, and irrelevant indications
caused by the dendritic grain structure.
This Hitachi report also discussed UT sizing techniques for IGSCC (4.8).
Crack length is measured using a focused, 45 , shear wave search unit, and the
20 dB amplitude drop technique.
The techniques being used to measure crack
depth include: a) the longitudinal wave scattering technique, b) the peak echo
technique (basically, crack-tip diffraction), and c) the surface-wave, modeconversion technique.
Because of overall unreliability, more than one of
these two techniques are usually applied.
The peak echo technique was
considered effective for fatigue cracks, but was less effective for IGSCC
because of the branched nature of this type of cracking.
Another Japanese report described a study to compare the ultrasonic

response from artificial notches and IGSCC in pipes subjected to IHSI (4.9).
The UT measurements included the conventional ASME Code techniques for
measuring depth and length (i.e., recording peak amplitude response).
depth measurements were supplemented by the flaw-tip echo technique.
The
Very
little, if any, difference in UT response amplitudes was observed before and

after IHSI for the EDM notches.
Although similar results were obtained for
the IGSC cracks, the data point spread was much greater (1.2 dB for EDM
notches versus 2.7 dB for IGSCC).
Similarly, few if any changes were observed
for the estimated flaw length and flaw depth for either EDM notches or IGSCC,
before and after IHSI treatment.
than for EDM notches.
Again, the data spread was greater for IGSCC
These authors concluded that no significant changes in
4-38
_echo height or defect dimensioning had occurred as a result of the IHSI

^^rreatment based on the ultrasonic response data.
Furthermore, they concluded
that the flaw-tip echo technique provided an effective means for measuring
flaw depth based on a correlation of UT predicted depths with destructive
analysis.
4.9.3
Experience in Switzerland
The KKM plant in Muhleberg, Switzerland (a 334-MWe BWR) has been in operation since October 1972.
A plant visit by an NRC staff member and an NRC
consultant provided the following information.
Preservice examination of the piping system involved only dye penetrant

and radiography - a UT baseline was not conducted.
The initial construction
process did not include grinding of the weld crowns; however, it is planned to
manually grind all crowns for future UT exams.
Their experience to date has
shown that the shop welds contain very few UT indications, but the field welds
contain many UT indications.
Throughwall leaks were visually detected in the 4-in. bypass lines during
routine post-outage pressure tests (to operating pressure) before startup.
The throughwall cracks in the 4-in. bypass lines were found by destructive
evaluation to be 2 to 3 in. long.
In 1980, all 4-in.
bypass lines were
removed and 304L stainless steel end caps were installed.
4-39
Several UT indications (interpreted to be cracks) have been found

adjacent to welds on the pump discharge line. These cracks appear to extend
360
around the pipe, and vary from 2 to 6 mm in depth. After 3000 hours of
plant operation, another UT inspection was performed and no detectable crack

growth was found.
They plan to operate until August 1984, at which time the
pipe will be replaced with 316NG stainless steel pipe.
Currently, all UT inspections in Switzerland are performed manually.

Sensitivity calibrations are based on a 2-mm side-drilled hole. Depth sizing
is performed using various techniques, and the information from each technique
is compared before a final sizing call is made. In addition, they do not feel
that effective examinations are being performed from the component side of
static cast components such as elbows and valves.
4.10
LEAK DETECTION IN LWRs:
4.10.1
REVIEW AND RECOMMENDATIONS
Introduction
Early detection of leaks in nuclear reactors is necessary in order to

detect deteriorating or failed components and minimize the release of radioactive materials.
Consequently, the NRC requires that operational leak-
detection systems of various kinds be installed.
Since equipment cannot be
perfectly leak-tight, allowance is made for identified leakage from valve

packing, shaft seals, and other equipment.
Thus, even during normal oper-
ation, there may be some accumulation of water in the sumps with an increase
in the level of radioactivity.
For most BWRs the plant technical specifica-
tions require that plant shutdown be initiated for leak rates of 5 gpm
(unidentified) or an increase of 2 gpm in 24 h (unidentified).
4-40
4.10.2.
Standard Practice
The American National Standards Institute (ANSI) has prepared a draft

standard (4.10) which reviews several leak-detection methods and indicates
their capabilities for detection, location, and measurement.
This information
is summarized in Table 4.1.
As Table 4.1 indicates, no single currently used leak-detection method

combines optimal leakage detection sensitivity, leak-locating ability, and
leakage measurement accuracy.
Consequently, Regulatory Guide 1.45 suggests
that at least three different detection methods be employed in the reactor.

Sump flow and airborne particulate radioactivity monitoring are mandatory.
third method can involve either monitoring of condensate flow rate from air
coolers or monitoring of airborne gaseous activity.
Although the current
methods used for leak detection reflect the state of the art, other techniques
may be developed and used.
Regulatory Guide 1.45 also suggests that flow
rates from identified and unidentified sources should be monitored separately

to an accuracy of 1 gpm, and indicators and alarms for leak detection should
be provided in the main control room.
Because the recoimnendations of Regulatory Guide 1.45 are not mandatory,

the technical specifications for 74 operating plants (including PWRs) have
been reviewed to determine the type of leak detection methods employed, the
range of limiting condition for operation, and the surveillance requirements
for the leak detection systems.
4-41
Table 4.1
Capabilities of Leakage Monitoring Methods (from Reference 4.10)
Method
Leakage
Detection
Sensitivity
Leakage
Measurement
Accuracy
Leak
Location
Sump Monitoring
G^
P"
Condensate Flow Monitors
F^
Radiogas Activity Monitor
Radioparticulate Activity
Monitor
Primary Coolant Inventory
Humidity Dew Point
Tape Moisture Sensors
Temperature
Pressure
Liquid Radiation Monitor
Visual
G (Good) - can generally be applied to meet intent of this standard if

properly designed and utilized.
F (Fair) - may be acceptable, marginal, or unable to meet intent of this
standard depending upon application conditions and the number
of measurement points or locations.
"P (Poor) - not normally recommended but might be used to monitor specific
confined locations.
For PWR during steady-state conditions.
'For detection of intersystem leakage; may also be used for location function
in sump or drain monitoring.
Provided that the leakage area is visible.
4-42
All plants use at least one of the two systems recommended by Regulatory
Guide 1.45.
All but eight specify sump monitoring as one of the leakage de-
tection systems, and all but three use particulate radioactivity monitoring.
Monitoring drywell air cooler condensate flow rate and atmospheric gaseous
radioactivity are also frequently used.
also been tabulated.
Leakage limits for most plants have
The allowed limits on reactor unidentified coolant
leakage are shown in Figure 4.1a.

limit for most BWRs is 5 gpm.
The limit for all PWRs is 1 gpm and the
The limits for total leakage (Figure 4.1b) are
generally 10 gpm for PWRs and 25 gpm for BWRs.
(Regulatory Guide 1.45 does
not specify leakage limits, but does suggest that the leakage detection system
should be able to detect a 1-gpm leak in 1 hour.) In some cases limits for
rates of increase in leakage are stated in the plant technical specifications.
On an hourly basis they are either 0.1 gpm/h (2 BWRs) or 0.5 gpm/h (4 BWRs).
Additional limits for rates of increase in leakage (2 gpm/24h) were temporarily imposed on five BWRs as part of the five orders (IGSCC inspection
orders confirming shutdown) of August 26, 1983.
Surveillance periods are indicated in Figure 4.2a.
Leakage in most PWRs
is checked every 12 hours, and in most BWRs every 4 or 24 hours.
One BWR
specifies that a continous monitor with control room alarm shall be operational.
The intervals between system calibration and checks are indicated
in Figure 4.2b.
For BWRs, calibration is generally performed at 18-month
intervals and functional tests every month.
4-43
In general, sump pump monitoring is used to establish the presence of

leaks.
Other methods appear to be less reliable or less convenient.
In most
reactors the suirveillance periods are too long to permit detection of a 1-gpm
leak in one hour as suggested by Regulatory Guide 1.45, but it appears that
this sensitivity could be achieved if monitoring procedures were modified.
None of the systems provide any information on leak location, and leaks must
be located by visual examination after shutdown.
Since cracks may close when
the reactor is shut down, reducing flow rates considerably, it would be

desirable to be able to locate cracks during plant operation.
The estimated sensitivity of leakage monitoring systems is occasionally

addressed in the technical specifications.
For example, one specification
indicates that air particulate monitoring can in principle detect a 0.013-gpm

leak in 20 min., that the sensitivity of gas radioactivity is 2-10 gpm, and
that of condensate flow monitoring is 0.5-10 gpm.
Sump pump monitoring
appears capable of detecting a 1-gpm leak in 10-60 minutes (with continous

monitoring).
The impact of Reactor Coolant Pressure Boundary (RCPB) leakage detection

systems on safety was evaluated for 8 reactors as part of the Integrated Plant
Safety Assessment-Systematic Evaluation Program (NUREG 0820-0827).
In 4 of
the 8 reactors a 1-gpm leak would not be detected in 1 hour nor did they have
three leakage monitoring systems, as suggested by Regulatory Guide 1.45. The
fracture mechanics and leak rate calculations in the SEPs are consistent with
the studies reported in Section 6, which indicate that current leak detection
systems and leakage limits will detect and require plant action for throughwall cracks of 4 to 10 in. in length in 12 to 28-in.-diameter piping in one
day.
Since these cracks are much smaller than those required to produce
4-44
failure in tough reactor piping, improved leak detection systems may offer
little safety benefit for this particular class of flaws when crack growth
occurs by a relatively slow mechanism.
However, the SEPs note that local leak
detection systems may be desirable for some postulated break locations where
separation and/or restraint is not practical to remove the effects of a high
energy pipe break.
Although current leak detection systems are adequate to ensure leakbefore-break in a great majority of cases, the possibility of large cracks
resulting in small leaks must be considered.
This could arise because of the
relatively uniform growth of a long crack before penetration or the corrosion

plugging or fouling of relatively slowly growing cracks.
In such cases the
time from a small leak to a significant leak or rupture could be short depending on crack geometry, pipe loading, and transient loading (sesmic or
water hammer event).
The shortcomings in existing leak detection systems are not simply a

matter of conjecture.
The Duane Arnold safe end cracking incidents indicate
that the sensitivity and reliability of current leak detection systems are
clearly inadequate in some cases.
The plant was shut down on the judgment of
the operator when a leak rate of 3-gpm was detected; however, this rate is
below the required shutdown limit for almost all BWRs.
Examination of the
leaking safe end showed that cracking had occurred essentially completely
around the circumference.
The crack was throughwall about 20% of the circum-
ference and 50-75% throughwall in the nonleaking areas.
The other seven riser
safe ends were also severely cracked, but since the cracks were not throughwall no leakage resulted.
4-45
simply tightening the current leakage limits may not be an adequate

solution to these shortcomings, since it is possible that this may produce
an unacceptably high number of spurious shutdowns due to the inability of
current leak detection systems to identify leak sources.
One other safety-related aspect of improved leak detection systems is in

the area of radiation exposure to plant personnel.
Improved systems with leak
location capability could reduce the exposure of personnel inside containment

and present an attractive alternative to augmented ISI. Some welds are inaccessible either for inspection or replacement, and improved leak detection
would provide additional margin against possible failure in these locations.
Improved leak detection is consistent with the defense-in-depth philosopy of
the NRC and would lead to earlier detection of system degradation.
4.10.3
Improved Leak Detection Systems
In order to improve detection of leaks through IGSCC, some utilities have

installed either acoustic emission monitors or moisture-sensitive tape (MST)
for better localized leak detection at specific welds. Acoustic monitors have
been installed at Dresden (Commonwealth Edison) and Hatch (Georgia Power
Co.).
MST has been installed at Browns Ferry-2 (TVA), Peach Bottom 2 and 3
(Philadelphia Electric), and Vermont Yankee (Vermont Yankee Nuclear Power).
4-46
In general, MST or acoustic emission monitors (AEM) have been installed

^^lar welds that have crack indications. At Peach Bottom, a MST sensor in the
vicinity of a leaking valve triggered an alarm. No other incidents of leak
detection by MST or AE in-reactor have been noted at the time of this report.
MST systems are sensitive if a path exists for leakage to reach the detector;
however, the incident at Peach Bottom points out the need for quantitative
information regarding leak characterization, location, and flow rate to avoid
false alarms which lead to unscheduled shutdowns and unnecessary exposure to
plant personnel during subsequent inspection inside containment.
No leak has been indicated by the AE system installed at Hatch and no

leaks have been found during shutdown periods.
Tests on a virtually identical
system at Argonne National Laboratory indicate that because the acoustic background level is very low in this particular case, leaks as small as 0.002 gpm
could be detected.
However, the system has a limited dynamic range, satu-
rating at approximately 0.006 gpm. Thus it could also be prone to alarms

generated by relatively large leaks from sources other than cracks far from
the crack location being monitored (an instance might be valve leakage).
High temperature piezoelectric accelerometers have been used at the

Dresden 2 reactor on safety relief valves and on a main recirculation line
(28 in.) elbow.
The system detects signals from leaks in the 20 to 50-kHz
range and employs a spectrum analyzer to verify if a leak is present.

in a specific frequency window suggest the presence of a leak.
this report no alarm signals have been detected.
4-47
Signals
At the time of
This system is considered
experimental and the main leak detection system is still the sump monitor.
High temperature accelerometers have also been employed by Philadelphia
Electric (PECO) since 1974 to monitor valves for leakage.
of plant shutdown has been packing gland valve leaks.
The primary cause
Leaks as low as 0.5 gpm
can be detected by the PECO system.
Several plants in Sweden and Finland have installed television monitors

in containment to aid in identifying and locating leaks.
Although lacking in
sensitivity and completeness of coverage, such systems appear to be useful,

inexpensive supplements to current leak detection systems.
Although current installations probably do not fully utilize the best

available technology, the use of acoustic monitoring or MST to improve leak
detection capability at specified sites is technically feasible and has been
studied by a number of investigators (4.11-4.17).
However, current acoustic
monitoring techniques provide no source discrimination (i.e., pipe crack vs.

valve leak) and no flow rate information (i.e., the system may saturate on a
small leak).
MST does not provide quantitative leak rate information and
gives no specific location information other than the location of the tape;
moreover, its usefulness with the new porous "soft fiberglass" insulation
which permits escape of steam through the insulation, needs to be
demonstrated.
Leak detection techniques need further improvement in the following

areas:
1) improving reliability and eliminating false calls through location
information and leak characterization to identify the source; 2) quantifying

and monitoring leak rates; and 3) minimizing the number of installed
transducers required for a "complete" system (see also reference 4.17).
4-48
A number of studies have been carried out to evaluate the feasibility of

using acoustic leak detection to achieve these goals.
have studied acoustic leak detection in PWRs.
studies were conducted.
Smith et al. (4.11)
Both field tests and laboratory
The results suggest that leaks as small as 0.02 gpm
could be detected (near the source) in a PWR by acoustic emission techniques,

and that acoustic emission monitoring of valve and pipe leakage would be
useful.
They suggest that valve leakage could be located by measuring the
amplitude decay of the resultant acoustic signal.
Researchers at Battelle-
Columbus (4.12) have studied the acoustic signals generated by the flow of
initially saturated or subcooled liquid through narrow slots and laboratorygrown IGSCC.
Acoustic data have been acquired from several transducers for
-4
flow rates of approximately 10
to 1 gpm.
Studies at KWU, Erlangen, West
Germany (4.14, 4.15), have also tried to establish the sensitivity of acoustic
leak detection in the presence of acoustic background noise.
These studies
indicate that it is possible to detect leaks in primary loops of a PWR with a

leak rate of approximately 0.4 gpm by acoustic means, leak location on a
specific component is possible, and relatively few transducers are needed to
monitor a loop for leaks.
loop.
A prototypical system will be tested on a primary
The system is being considered as a redundant alternative to ISI.
Acoustic leak detection has also been studied at the Japanese Central Research
Institute of the Electric Power Industry Energy and Environment Laboratory
(CRIEPI (4.15, 4.16).
They have concluded that acoustic leak detection is
useful for detection of leakage from boiler tubes and that the techniques are
applicable to piping in nuclear power plants.
4-49
The ongoing U.S. NRC program on leak detection at Argonne National

Laboratory (ANL) is directed at the development of an acoustic leak detection
(ALD) system for complete system monitoring.
The program will determine
whether meaningful quantitative data on leak rates and locations can be

obtained from acoustic signatures of leaks from IGSCCs and fatigue cracks in
low and high pressure lines, and whether these can be distinguished from other
types of leaks.
The program will also establish calibration procedures for
acquiring acoustic data and will determine whether advanced signal processing
can be employed to enhance the adequacy of ALD schemes.
Based on the
laboratory measurements at the Hatch and Watts Bar reactors, estimates of the
sensitivity of acoustic leak detection systems can be made (4.18).
Under
subcontract to ANL, GARD Inc. has established a system configuration and

developed a breadboard system, which will provide a basis for a prototype leak
detection location, and quantification system.
The breadboard system will be
evaluated in FY 1984.
Complete monitoring systems are desirable to provide additional margin

for undetected or unsuspected cracking while avoiding spurious shutdowns.
The
installation of television monitoring may be a relatively simple, worthwhile

addition to current leak detection systems.
not yet available.
developed.
Complete monitoring systems are
However, prototype ALD systems and MST systems are being
Testing of these prototype systems in-reactor to determine the
reliability of hardware and accuracy of these techniques for leak detection

should be encouraged.
4-50
4.11
Previous subsections in this section have described significant problem

areas relative to the uncertainty of UT inspection during both preservice and
inservice examination.
The following items in this subsection represent a
synthesis of important conclusions and recommendations.

detection.
Easily implementable modifications to these procedures
have resulted in some improvement.

Code Case N-335.
These have been incorporated into
Therefore, it is recommended that Code Case N-335
should immediately be made mandatory for all augmented inspections

until better procedures are developed.
Although IGSCC detection has improved to the point that it is considered acceptable under optimum conditions and procedures, the detection reliability as impacted by variability in operator procedure
4-51

performance demonstration represent steps in the right direction, and
Therefore, it is
recommended that ongoing industry NRC activities to develop adequate

criteria for qualification of the entire inspection process to achieve
more reliable field inspection be completed and implemented on a high
priority basis.

items such as pumps, valves and vessels in older plants. The

Improved leak detection systems would permit more stringent requirements on unidentified leakage without increasing the occurrence of
spurious shutdowns due to relatively benign leakage, and their
development should be pursued.
Acoustic and moisture-sensitive tape leak detection systems for local

leak detection are available, and their use is recommended where
inspection is difficult or impossible or operation for more than one
fuel cycle is considered for long cracks with weld overlay repairs.
4-52
An effort is also needed to be made to accumulate information on

current acoustic leak detection (ALD) and moisture sensitive tape
(MST) field installations.
Detailed reports on the operability,
maintenance, and reliability of these systems should be acquired,

assembled, and distributed for review by utility, government, and
research personnel interested in improving reactor leak detection
technology.
Sump pump monitoring systems have sufficient sensitivity to detect

leak rates as small as 1 gpm in one hour.
Improved leak detection
systems would provide additional margin against leak-before-break

without increasing the number of false alarms by providing information
about leak location and leak source discrimination.
It is, therefore,
recommended that improved leak detection systems under development

be completed and field tested.
Since the major purpose of the preservice inspection (PSI) is to

provide a reliable NDE baseline for comparison with subsequent ISI, it
is recommended that the latest edition of Section XI acceptable under
CFR 50.55(a) be used for the PSI.
4-53
Expand modeling work to predict UT crack response for guiding

development of UT techniques and guiding qualification test sample
selection.
Since the UT examiner is one of the more erratic inspection variables,

it is recommended that human factors research be performed to lead to
a reduction in the human variability.

are implemented there may be a need to modify the following documents when
NUREG-0800, Sections 5.2.4 and 6.6; Draft NUREG-0313, Revision 2; Generic

Safety Issue 14, A-42 and A-14.*
ASME Section XI; Regulatory Guide 1.45; and
the development of new regulatory guides on qualification of UT personnel,

equipment and procedures, and on pipe inspection.
*This has been dropped as a Generic Safety Issue because it is not a safety
issue per se, but it impacts specific problems which the Office of Nuclear
Regulatory Research is addressing.
4-54
12
REFERENCES
S. K. Edler, ed. November 1980.
"Integration of Nondestructive
Examination Reliability and Fracture Mechanics." Reactor Safety Research

Programs. Quarterly Report.
April-June 1980. NUREG/CR-1454, Vol. 2,
Pacific Northwest Laboratory, Richland, WA.
T. T. Taylor and G. P. Selby.
April 1981.
Evaluation of ASME Section
XI Reference Level Sensitivity for Initiation of Ultrasonic Inspection

Examination.
NUREG/CR-1957, Pacific Northwest Laboratory, Richland, WA.
Electric Power Research Institute.
August 1983.
Ultrasonic Sizing
Capability of IGSCC and Its Relation to Flaw Evaluation Procedures.

EPRI/NDEC 82-10, Electric Power Research Institute, Palo Alto, CA.
S. R. Doctor, G. P. Selby, P. G. Heasler, and F. L. Becker.
1983.
"Effectiveness and Reliability of Inservice Inspection, a Round Robin

Test." Proceedings of the Fifth International Conference on Quantitative
NDE in the Nuclear Industry. San Diego. CA. May 10-12. 1982.
Am. Soc.
Metals, Metals Park, OH.
L. Kiessling, F. Hilsson, M. Trolle, and L. Skanberg. 1983.
Licensing
and Research Aspects on Stress-Corrosion Cracking in Swedish Boiling

Water Reactors.
Swedish Nuclear Power Inspectorate, Stockholm, Sweden.
Y. Ando "Reliability Assessment Activities for Non-destructive

Examination and Structural Integrity for Reactor Components in Japan."
Paper obtained from S. H. Bush, undated.
4-55
4.7
Tokyo Electric Power Company, Inc. , Toshiba Corporation, and Ishikawajima

Harima Heavy Industries.
January 1983.
Fukushima Unit No.3 Work Report
for Pipe Branches Replacement.
4.8
Hitachi Works of Hitachi Ltd.
July 1983.
Ultrasonic Examination for
Detection and Dimensioning of IGSCC in BWR Austenitic Piping.
4.9
S. Shibata, H. Yoneyama, S. Tanaka, and M. Kishigami.
October 1982.
"Effects Brought by Induction Heating Stress Improvement (IHSI) Measures

on Ultrasonic Signals Reflected from Intergranular Stress Corrosion
Cracking (IGSCC)." IHI Eng. Rev. 15. No. 4.
4.10
The American National Standards Institute.
1978.
Standard for Light
Water Power Reactor Coolant Pressure Boundary Leak Detection.
ISA
Standard S67.03, ANS Standard N41.21.
4.11
J. R. Smith, G. V. Rao, and R. Gopal.
1979.
"Acoustic Monitoring for
Leak Detection in Pressurized Water Reactors."

Monitoring of Pressurized Systems.
In Acoustic Emission
W. F. Hartman and J. W. McElroy,
eds., ASTM STP 697 177-204.
4.12
R. Collier, Battelle-Columbus Laboratories.

D. Kupperman, Argonne National Laboratory.
4-56
Personal coimnunication to
^^^.13
K. Fischer et al. "Leak Detection and Location by Means of Acoustic

Methods." In European Nuclear Conference. May 6-11, 1979.
Hamburg,
Federal Republic of Germany, KWU Kraftwereke KWU 0017 TEXTSEIT E 50.000

4.79 1788.
4.14
K. Fischer and G.Stipsitz.
1981.
"Leak Monitoring of Pressure
Boundaries in LWR's Using Acoustic Emission." Yearly Symposium on Nuclear

Reactors, Dusseldorf.
4.15
A. Kitajima. June 1980. Acoustic Leak Detection in Piping Systems

(Part 3). CRIEPI Research Report 279050, Japan.
4.16
A. Kitajima and N. Naohara. Acoustic Leak Detection in Piping Systems.

CRIEPI Research Report preprint.
4.17
D. 0. Harris, R. G. Brown, D. Dechhia, and D. E. Leaver.
December 1980.
Acoustic Emission Leak Detection and Location Systems Technology

Review.
4.18
Electric Power Research Institute Report NP-80-7-LD.
D. Kupperman and T. N. Claytor.
1984. Acoustic Leak Detection and
Ultrasonic Detection of Cracks.
Proceedings of the Eleventh Water
Reactor Safety Research Information Meeting.
4-57
Vol. 4, NUREG/CP-0048.
.50
V)
\-
0BWR
40
_J
CL
30
<
li_
O
tr
20
CD
10
LLI
2
3
PWR
I
5
UNIDENTIFIED LEAKAGE (gal/min)

(a)
if)
50
\<
40
_J
CL
U.
0BWR
30
PWR
or 20
UJ
CD
2
"D
2
10
i_jdbL
10
20
30
TOTAL LEAKAGE (gal/min)

(b)
Figure 4.1
Allovjed Limits on Reactor Coolant System Leakage

from Technical Specifications for 74 Plants for
a) Unidentified Leakage and b) for Total Leakage
4-58
VO
r
Ut
JS
(D
3 en
g
a- W
3
r t E: s
g
to j :
3
HD- en
rt
'-t o
3-05
rH to t i
n
3(D
n
?r
< t<i
to
(-
M
3
rt
(D
i-t
rt
m
m
ro
*~
I-l
ft)
CO
3
tr*
t i ro
en to
iti
3
to
3
n
rt
H-
-^J
rt
fD
3
rt
CD
'J
M
to
c ên *~
i-ri Vi
o
^-^ on
M
to
en en
n
Hi-ti
Ho
< ; to
(li r t
t ( Hen o
C
3
rt
n>
3
en
he Technical Spe
kage Monitor Sys
) and b) System
(-3
to
hl-icr
ti
to
rt
H-
rt
ased o
er of
al (ho
CM
4J
en
tA
33
"D
3J
CD
o
c
3)
<
>
o
m
o
T:
m
CALIBRATION
i^^^^^SS^l FUNCTIONAL TEST
RADIOACTIVITY CHANNEL
CHECK
f\5
CD
OL
Oi
o
c
U)
ro
NUMBER OF LEAKAGE
MONITORING SYSTEMS
ro
O
O
ro^wwwwwww^
ro
Ci
cn
^\v\\^\\^
UI
ro Ol
UI O
33
Ol
u>
NUMBER OF LEAKAGE
MONITORING SYSTEMS
33
03
5.0
5.1
INTRODUCTION
The extent and frequency of inspection of susceptible piping welds

recommended in this section are based primarily on the reinspection program
specified in SECY-83-267c (5.1) for those plants inspected under IE Bulletins
82-03 and 83-02 (5.2, 5.3), and the inspection programs recommended in the
draft NUREG-0313, Rev. 2 (5.4).
The reinspection program specified in
SECY-83-267C was reviewed and approved by this task group and is included in
Sub-section 5.4.2 for completeness because these plants will be reinspected
within the next year or so. This task group has also reviewed the overall
inspection program presented in the draft NUREG-0313, Rev. 2, and has
developed the recommended inspection program in this section.
One positive result of the extensive inspections performed on largerdiameter ( 2 4 in.) BWR piping is that no other significant mode of degradation has been noted.
This means that inspections can focus on those
approaches best suited for detecting and evaluating IGSCC. A less favorable
finding is that special methods and specific operator training are required to
detect IGSCC reliably.
Further, differentiating UT signals arising from geo-
metric configurations from signals caused by cracks is perhaps even more difficult than detection of indications.
5-1
It is not the intent of this section to provide specific guidance to

operators regarding details of equipment and procedures.
This function is
best handled by Code activities in which industry and regulatory participants

reach a consensus.
It is not a simple problem; finding and recognizing IGSCC
by UT is still as much an art as it is a science. UT examiners must also

characterize quantitatively the cracks they find.
IWB 3600 rules for flaw
evaluation (ASME Code, Section XI) are based on the assumption that the depth
and length of flaws are known reliably. Although the length (circumferential
extent) of cracks is believed to have been determined with sufficient accuracy
and reliability by inspectors capable of detecting cracks, accurate measurement of the depth (throughwall dimension) cannot be relied upon. While some
UT examiners may do an adequate job, most will require special training to
ensure sufficient accuracy for use as the basis for Code flaw evaluations.
Nevertheless, the experience gained during the inspections performed

under IE Bulletins 82-03 and 83-02, UT sizing round robin tests, and
particularly the insight into the problems gained through the performance
capability demonstrations specified by the bulletins, can be used to formulate
requirements that will significantly upgrade the reliability of future
inspections.
The intent of the recommendations covered by this section is to ensure

that (1) an adequate sample of susceptible pipe welds will be inspected on a
sufficiently frequent basis to detect IGSCC before it can jeopardize the
integrity of the piping system and (2) the UT operators inspecting BWR piping
for IGSCC can reliably detect and recognize IGSCC in the welds they inspect.
I
5-2
5.2
INSPECTION RECOMMENDATIONS
All BWR Types 304 and 316 piping systems operating at temperatures over
200 F (93 C) have shown susceptibility to IGSCC; therefore, all systems are
considered to be equally prone to cracking, and the recommendations for
augmented inspections beyond those required by ASME Code, Section XI, will
apply to all Type 304 and 316 austenitic piping operating over 200 F (93 C).
Differences in the extent and frequency of examinations will depend on the
resistance to IGSCC of the materials and the effectiveness of any processing
used to prevent cracking.
5.2.1
Categorization of Welds
All welds in BWR systems will be categorized according to how likely they
will be to crack.
Three categories will be used:
Category A - Welds very unlikely to have IGSCC, because the piping is made
of resistant materials, or welds made with, or subjected to, a^
countermeasure A mitigating process.
Category B - Welds with some degree of improved resistance to IGSCC, because

the piping is not made of resistant material, and welds are made
with or subjected to a countermeasure B mitigating process.
Category C - Welds likely to be subject to IGSCC because they are neither made
of resistant material nor subjected to a mitigating process.
5-3
5.2.2
Bases for Weld Categories
The bases for categorizing welds are directly related to the materials
and the welding or mitigating processes used.
These are summarized below:
Resistant Materials
(1) 304L, 316L, 316K, 304NG, 316NG, 347NG, 308L

(2) low-strength carbon steels
(3) approved nickel-based materials
(4) cast low-carbon/high-ferrite austenitic stainless steels
(5) welds solution heat-treated after fabrication and welding
(6) other, as approved by NRC
Countermeasure A Processes - Any combinations of two mitigating processes

which are intended to reduce or minimize any two of the three causes (i.e.,
sensitization, stress and environment) contributing to IGSCC.
These are
summarized below:
(1) IHSI on new pipe or pipe with no reported indications plus hydrogen
water chemistry*;
(2) HSW on new welds, including new welds used to install short repair
sections ("pup pieces") plus hydrogen water chemistry;
*Hydrogen water chemistry is discussed in Section 6.
5-4
(3) LPHSW on new pipe welds including new welds used to install short repair
sections plus hydrogen water chemistry.
(4) Other, as approved by NRC.
Countermeasure B Processes - Any of the following mitigating processes.
(1) IHSI on new pipe or pipe with no reported indications.
(2) HSW on new welds including new welds used to install short repair
sections ("pup pieces").
(3) LPHSW on new pipe welds including new welds used to install short repair
sections.
(4) Hydrogen water chemistry.
(5) Other, as approved by NRC.
5.2.3
Sample Selection
Results of inspections conducted to date under lEB 82-03 and 83-02

indicate that all Types 304 and 316 stainless steel piping welds in systems
operating over 200 F (93 C) are susceptible to IGSCC,
In addition, field data
show that the cracking experience does not correlate well with the Stress Rule
5-5
Index (SRI) or the carbon content.
Therefore, the primary basis for sample
selection should be field experience; where other factors such as weld prep,
excessive grinding, extensive repairs, or high-stress locations are known to
exist, they should also be considered in the sample selection.
The results of inspections conducted under lEB 83-02 show that the
specified sampling scheme (i.e., an initial sample size of about 20% in each
pipe size and the logic for expanding the inspection sample size when cracks
are found) has been found to be reasonably effective.
For example, because of
expansion of sample size under the bulletin, the 83-02 plants averaged an
inspection of 62% of welds.
By specifying a limited initial sample, the inspection resources in terms

of occupational exposures can be used more effectively.
As more experience is
gained in detecting IGSCC, the staff anticipates that this sampling scheme
would be even more effective than in lEB 83-02.
Allocation of examiner
resources must always be made in determining the initial sample size so that
the inspection resources are available, not only for the detection and the
necessary sample expansion but also for the depth sizing and post-replacement
baseline inspections.
5.2.4
Inaccessible Welds
For those categories B and C welds that are not inspectable because of
physical access or poor inspectibility, an appropriate local monitoring system
such as a leak detection system, should be employed to ensure the continuous
integrity of the weld.
These local monitoring systems should be approved by |
NRC before field applications.

5-6
5.2.5
Summary of Inspection Recommendations
Inspection requirements are dependent on the material and processes used

for each weld.
The piping categories and inspection required thereon are
presented in Table 5.1.
5.3
INSPECTION PERSONNEL
The recent problems regarding the capability to detect and size IGSCC in
BWR piping, in accordance with Section XI procedures and requirements, can be
primarily attributed to either one or both factors of UT technology; namely,
the ineffectiveness of UT procedures and of examination personnel.
This was clearly demonstrated by the inspection results of recirculation

system piping at Nine Mile Point Unit 1, and more recently by the results of
IE Bulletins 82-03 and 83-02 performance capability demonstrations, PNL round
robin tests, and EPRI sizing studies. Both UT procedures and examiners are
responsible for the recent poor showing of UT reliability.
However, the fact
that there was a large disparity in the demonstrated performance capability

among inspection teams or examiners highlights the need to address the issue
of the qualification and certification of NDE personnel in general, and UT
examiners in particular.
5.3.1
Personnel Qualifications
Many very experienced and competent UT examiners have had little or no

experience with interpreting IGSCC signals.
trained in this very specialized area.
5-7
They may not be adequately
Still, they are "qualified" under
Table 5.1
Weld
Category
Condition and Inspection Required for Weld Categories
Condition
Inspection Required
Resistant
material or
countermeasure A
applied

refueling outage.
Nonresistant
material with
countermeasure B
applied

refueling outage.
Neither of the
above and all
welds with detectable cracking
regardless of the
use of mitigating
processes
100% in 6 years. At least one-half of these

should be inspected every three and one-third
years or the nearest refueling outage.
For plants older than 6 years, all uninspected
Category C welds shall be inspected at the next
outage after compliance with IE Bulletin 82-03
or 83-02.*
*Weld inspections not performed in accordance with IE Bulletin 82-03 or 83-02

are considered inadequate; therefore, no credit is given for such inspections.
governing Code requirements to perform examinations of BWR piping.
They may
be very reluctant to take tests to prove their competence (which they feel
should be taken for granted), and may also find it somewhat demeaning to be
required to take special training. Nevertheless, no UT examiner should be
permitted to perform inspections of BWR piping without proving his competence,
even if it requires him to take special training to gain the specific skills
and knowledge required to perform these inspections.
The performance
demonstrations specified in the bulletins were major steps in the direction

required.
They set a precedent, but more importantly, they made clearly
evident the need for such actions.
The utilities quickly understood the
problem and were cooperative and supportive.
They did not want to pay for
inspections that were ineffective and certainly would not want to shut down in
midcycle to repair a leak in a weld recently pronounced to be sound.
5-8
^ ^
Inspections will continue to emphasize crack detection and discrimination, in addition to the anticipated improvement in the sizing capability.
All level 1, 2, or 3 UT examiners performing operations (general scanning
observations and discrete signal interpretation and sizing) should demonstrate their field performance capability in a manner acceptable to the NRC.
In addition, all examiners performing evaluations must be able to view on CRT
display for the entire time that the transducer is in contact with the pipe
for scanning, either in real time, remotely, or on tapes.
5.3.2
Performance Demonstration Tests
It is clear that the performance capability demonstration programs

similar to those conducted under IE Bulletins 82-03 and 83-02 must continue.
It is also clear that the NRC now can strengthen the program in several ways.
Experience has also shown that the NRC must upgrade the program if results of
inspections are to be relied upon.
In addition, a similar program must be
developed for sizing cracks once they are detected.
Details of an upgraded program should be worked out with industry, as the

original program was with representatives from the Office of Nuclear Reactor
Regulation (NRR), the Office of Inspection and Enforcement (IE), the Office of
Nuclear Reactor Research (RES), and all NRC Regions involved.
The IE and
Region-based inspectors who must ensure that inspections are properly

performed must be directly involved in working out requirements and monitoring
actual performance demonstrations.
^^^
Although all aspects of an upgraded program cannot be covered in this

section, there are several important aspects to be highlighted.
5-9
(1)
Additional suitable samples of service-induced cracked and uncracked pipe

should be obtained and properly characterized as to crack depth and
length.
(2)
The grading of performances should be tightened up, particularly in the

allowance for detection and discrimination (false calls) and sizing
uncertainties (undersizes and oversizes).
indication a crack if it is not.
A false call is calling an
A miscall is falsely identifying a real
crack as a geometric discontinuity.
Performance should be judged on the basis of the importance of the call.

For example, no crack should be missed (not reported) if it is large
enough to require evaluation or repair in accordance with the requirements given later on in this section.
Further, severe penalties should
be assessed if a nonexistent crack is reported as large enough to require

evaluation or repair.
Flaws with the throughwall thickness less than
that given in Table IWB-3514-2 of Section XI Code are considered to be

acceptable without evaluation or repair.
5.3.3
Availability of Inspection Personnel
Since the issuance of IE Bulletin 82-03, the EPRI NDE Center has been conducting an IGSCC detection and discrimination training course using cracked
samples.
Consequently, the number of qualified IGSCC examiners, i.e., those
people who have successfully completed the performance demonstration test required by IE Bulletin 83-02, has been steadily increasing.
5-10
However, there is
also a sign that the NDE industry is losing a substantial number of qualified
UT examiners because of a labor dispute.
In addition, after a few trial
courses on sizing, the EPRI NDE Center is planning to initiate a formal

training and qualifying course in the near future for crack sizing.
Because of the continued efforts of the EPRI NDE Center, it is expected

that sufficient qualified inspection personnel will be available for the
reinspection program required by SECY-83-267C.
However, in the long term, the
allocation of qualified UT examiners for detection, depth sizing, and postreplacement baseline inspections should be considered in all future inspection
programs.
5.4
PLANT INSPECTIONS
All BWR pipe weld examinations should be performed in accordance with the
latest version of Code Case N-335 and with the specific equipment and procedures used, and personnel passed in the performance demonstration tests.
BWR pipe weld examinations should be monitored by Regional Inspectors,

who must be satisfied (1) that the inspection is performed satisfactorily and
(2) that equipment and procedures used and personnel have passed the performance demonstration tests.
In the event that a Regional Inspector is not satisfied with the

adequacy of the inspection, the inspector may require that an additional
independent qualified inspection agency perform a total or partial
reinspection.
This may include check inspections by NRC personnel.
5-11
5.4.1
All Plants
The extent and frequency of examinations for all operating BWR plants
containing Types 304 and 316 austenitic piping operating over 200 F (93 C)
should follow the inspection recommendations specified in Section 5.2 of this
report.
The future inspection program after reinspection for those plants
that are required to be reinspected in accordance with Section 5.4.2 of this

report should also follow the inspection recommendations specified in Section
5.2.
5.4.2
Plants Inspected Under IEB-82-03 and 83-02
The reinspection program discussed in this subsection is taken from

SECY-83-267c and is included here for completeness because the affected plants
will be reinspected within the next year or so.
The scope for the inspections which follow the Bulletins 82-03 and 83-02
inspections should include the following stainless steel welds, susceptible to
IGSCC, in piping equal to or greater than 4 inches in diameter in systems
operating over 200 F (93 C) that are part of or connected to the reactor
coolant pressure boundary out to the second isolation valve:
(1) Inspection of 20% (but no fewer than 4 welds) of each pipe size of IGSCC
sensitive welds not inspected previously and reinspection of 20% (but no
fewer than 2 welds) of each pipe size inspected previously and found not
to be cracked.
This sample should be selected primarily from weld
locations shown by experience to have the highest propensity for

cracking.
5-12
(2) All unrepaired cracked welds.
(3) Inspection of all weld overlays on welds where circumferential cracks

longer than 10% of circumference were measured.
Disposition of any
findings will be reviewed on a case-by-case basis.
Criteria for
operation beyond one cycle with overlaid joints are under development.
(4) Inspection of any IHSI-treated weld which has not received post-treatment
UT acceptance testing.
(5) In the event new cracks or significant growth of old cracks* are found,
the inspection scope should be expanded in accordance with lEB 83-02.
The inspection requirements for these plants after the reinspection

specified above should follow the generic inspection recoimuendations discussed
in Section 5.2.
5.5
FOREIGN EXPERIENCE
The inspection program for austenitic piping for IGSCC in foreign

countries is generally similar to that of ASME Code Section XI.
The only
major difference, at least in Japan, is that once IGSCC is detected in a given

size piping system, all welds in that piping system and those welds in other
size piping systems under the same environment are also inspected.
*Significant growth of the old crack is defined as growth to a new crack size
that cannot be accepted without repair for the remaining period of the current
or a new cycle of operation, in accordance with the criteria in SECY-83-267C.
5-13
This extensive inspection can be accomplished without encountering much|

difficulty in the availability of inspection personnel because there is no
widespread IGSCC problem in most of the foreign countries.
In isolated cases
such as in Japan, the utility was encouraged to replace the affected piping
either iirauediately or after one cycle of operation with more corrosion
resistant piping material in order to avoid unnecessary inspections.
5.6
CONCLUSIONS
Augmented inspections beyond those required by ASME Code Section XI

should apply to all Types 304 and 316 austenitic piping systems operating
over 200 F (93 C) unless they have been treated with effective
countermeasures.
The degree of augmented inspection is dependent on the material and

processes used for each weld.
The most frequent inspections are required
for welds fabricated from nonresistant material.
Field data show that the cracking experience does not correlate well with
the stress rule index and the carbon content.
Therefore, the primary
bases for sample selection should be field experience coupled with other
factors such as weld prep, excessive grinding, extensive repairs, or
high-stress locations.
5-14
RECOMMENDATIONS
All Type 304 and 316 austenitic piping systems operating over 200 F
(93 C) should receive augmented inservice inspection unless they have
been treated with effective countermeasures.
The extent and frequency of examinations should depend on the resistance

of materials to IGSCC and the effectiveness of any processes used to
prevent cracking.
The primary basis for sample welds selected for examination should be
field experience, not the stress rule index and carbon content. Other
high-stress locations should also be considered in the sample selection.
The performance capability demonstration program used to demonstrate the

effectiveness of UT equipment, procedure, and examiner combination on the
service-induced cracked samples should be continued and strengthened.
The effectiveness and reliability of detecting IGSCC with all future UT
procedure/examiner combinations should be demonstrated on the cracked
samples before field application.
5-15
All UT examiners should attend the industry-sponsored UT detection and

sizing training courses and should continue to be evaluated on the
cracked samples under the witness of a third party personnel before
participating in the field inspection.
All BWR pipe weld examinations should be performed in accordance with the
latest version of Code Case N-335 and with the specific equipment and
procedures used, and personnel passed in the performance demonstration
tests.
An appropriate local monitoring system such as local leak detection

system should be used to monitor the continuous integrity of the
uninspectable welds in Categories B and C.
These local monitoring
systems should be approved by NRC before field application.

are implemented there may be a need to modify the following documents when
A-42; NUREG-0313, Rev. 1; Generic Issue No. 34; 10CFR50, App. A 30 and
32; Regulatory Guides 1.58 and 1.147; SRP 5.2.4 and 5.2.5; 10CFR50.55a(g).
5-16
5.8
1.
REFERENCES
NRC.
November 7, 1983.
Staff Requirements for Reinspection of
BWR Piping and Repair of Cracked Piping.
2.
SECY-83-267C.
Office of Inspection and Enforcement (IE), NRC.
October 1982.
Stress
Corrosion Cracking in Large-Diameter Stainless Steel Recirculation

System Piping at BWR Plants.
3.
IE Bulletin No. 82-03, Rev. 1.
Office of Inspection and Enforcement (IE), NRC.
March 4, 1983 Stress
Corrosion Cracking in Large-Diameter Stainless Steel Recirculation

System Piping at BWR Plants.
4.
W. S. Hazelton et al. 1984.
IE Bulletin No. 83-02
"Technical Report on Material Selection and
Processing Guidelines for BWR Coolant Pressure Boundary Piping." Draft

Report, Rev. 2, NUREG-0313.
Available from the Materials Engineering
Branch, Division of Engineering, Office of Nuclear Reactor Regulation,

U.S. Nuclear Regulatory Commission, Washington, DC
5-17
20555.
6.0
6.1
DECISIONS AND CRITERIA FOR REPLACEMENT. REPAIR

OR CONTINUED OPERATION WITHOUT REPAIR
INTRODUCTION
This section is a relatively complex interaction of the several factors

pertaining to the acceptance or rejection of cracked weldments, short-term
mitigative measures, and longer-term permanent fixes.
The opening section
covers fracture mechanics as it relates to the rates of growth of IGSCC, a

discussion of leak-before-break, and an evaluation of the various fracture
mechanics models that have been applied to cracked austenitic piping.
The second major subsection covers short-term measures, such as weld

overlay, not considered to be a permanent fix.
The final subsection represents permanent fixes or longer-term fixes.
mitigative operation such as induction heating stress improvement falls into

both short-term and long-term fixes.
Controlled water chemistry is considered
a long-term fix complementing mitigation; e.g., IHSI replacement of piping

with a material such as 316NG is considered a long-term fix.
Even so, the
Task Group on Pipe Cracking considers that hydrogen water chemistry, reduction
in residual stresses, and replacement with a material such as 316NG represent
a desirable "belt-and-suspenders" approach.
An event tree, shown below provides an overview of the options available

in the event that IGSCC is detected in the piping of a BWR.
As noted in the
heading, only short-term solutions are addressed in this event tree.
6-1
SHORT-TERM SOLUTIONS FOR REPLACEMENT. REPAIR OR CONTINUED OPERATION WITHOUT REPAIR

EVENT TREE*
NDE |UT)
CRACK
EXCEEDS STANDARDS. ASME XI.
APPENDIX D AND/OR NRR
CRITERIA USED
MEETS ASME XI
FLAW STANDARDS
NO CRACK
MAY BE UNDER ACCELERATED UT
PER SECTION 5
^^
CONTINUE ISI ON
ACCELERATED SCHEDULE
CAN REPLACE WITH < -

316 NG. ETC.
(LONG TERM FIX)
ACCEPTABLE
MARGINAL
UNACCEPTABLE
CAN REPLACE
WELD OVERLAY OR OTHER APPROACH

(LIMITED TO TWO FUEL CYCLES)
IHSI
{CATEGORY B
PER SECTION 51
NO IHSI
(CATEGORY C
PER SECTION 5|
> REPLACE CRACKED SECTION WITH 304. 316. ETC.
I
N3
LATEST DESIGN CRITERIA
WELD DESIGN OPTIMIZED FOR UT
ORIGINAL WELD DESIGN
WELD DESIGN OPTIMIZED FOR UT

NO IHSI
PRESERVICE UT
PRESERVICE UT
ISI TO NRC CRITERIA
ISI TO NRC CRITERIA
*The following assumptions apply to the event tree:

Repairs and replacements per original version of ASME Construction Code or. alternatively, updated to latest edition.
Section 5 refers to this report.
The intent is to examine the options available and suggest criteria for
the acceptance, conditional acceptance, or rejection of the options based on a
safety assessment of each option.
The following assessments in essence consider each portion of the event

tree and the regulatory significance of the various decisions.
Detection of very small cracks at the threshold of sizing reliability

would require no repair; however, the presence of these small cracks
below ASME XI standard sizes, would require UT at shorter intervals
until decisions could be made as to additional action.
If no change
in size is observed over two or three inspection intervals, the option

exists to revert to the standard ISI schedule.
If predicted crack sizes at the end of the next inspection interval

are marginally acceptable, the probable action would be reassessment
of the margins with weld overlay, and the application of a weld
overlay.
There are various potential problems with weld overlay that
are addressed more fully later in this section.
If predicted crack sizes at the end of the next inspection interval

are unacceptable, various options exist, such as weld overlay,
mechanical clamps, and partial replacement of the piping.
6-3
Two options exist with regard to criteria applied on replacement,

per the ASME XI Code and previous NRC policy:
1) Application of the
construction version of ASME III or other Code to the replacement.

This presumes a similar design assumption on loads, layout, etc.
2)
Use of the latest version of ASME XI/III to update design and
fabrication requirements.
Reduction of residual stress from use of IHSI, HSW, or LPHSW should

be promoted because such measures should substantially extend the
time to initiation of IGSCC.
Possibly, it would not initiate for the
remainder of plant life.
Material selection, if truly short term, is a second-order effect;

however, available data indicate the lower levels of carbon in the
L grades of 304 and 316 are highly resistant to IGSCC.
A penalty
exists in terms of lower stress allowables so the designer needs to

consider the advantages and disadvantages of replacing with L grades
and thicker piping.
As noted in the event tree, replaced weldments require a new baseline

UT examination.
While design updating to the latest versions of the Code should not be
a requirement, every effort should be made to promote its use.
This
should include relaxation of damping criteria on a case-by-case basis

to permit a more flexible and more inspectable system.
I
6-4
All steps related to the weld prep, welding, and weld finishing, and
HSW or IHSI should be validated through certification and through use
of an appropriate Q.A. manual. As noted previously, the weld should
be optimized for future ISI.
Future preservice and inservice examinations should utilize the

advanced techniques developed under ASME XI or NRC, including
certification of the NDE examiner.
If the original stainless steel considered to be susceptible to IGSCC

has not cracked but most systems have several years of operation, the
leak detection systems should be reviewed with regard to both sensitivity and reliability on the assumption that delayed IGSCC may
occur.
If the leak detection systems are deemed inadequate, they
should be improved.
6.2
FLAW EVALUATION AND LEAK-BEFORE-BREAK
The occurrence of intergranular stress-corrosion cracking (IGSCC) in

operating BWRs raises two important issues related to fracture mechanics
analysis.
These are flaw evaluation and leak-before-break.
The first of
these concerns the disposition of pipes containing cracks detected during

service. When a crack is detected in service an evaluation must be performed
to determine if repair or replacement of the pipe is necessary or if the crack
is sufficiently benign to return the cracked pipe to operation for some specified operating time. Repair of cracked pipes, generally by weld overlay, is
S-5
costly in terms of both personnel exposure and extended plant outage times.
Also, the weld overlay repair may induce undesirable stresses at other locations in the piping system.
Thus prudent decisions must be made regarding
the necessity for repairs.
The leak-before-break failure mode in piping refers to the concept that a

crack propagating through the wall of a pipe, by mechanisms such as SCC or by
ductile crack extension due to applied loads, will result in a stable
throughwall crack that can be reliably detected by leakage. Acknowledging
that some cracking which occurs may not be detected, it is desirable to
demonstrate that, unchecked, IGSCC will progress in a fashion that leads to a
leak-before-break mode of failure.
This section presents a critical review of
the currently used methods of flaw evaluation and the application of the leakbefore-break concept.
6.2.1
Flaw Evaluation Criteria
The criteria currently used by the Nuclear Regulatory Commission's Office

of Nuclear Reactor Regulation (NRR) for performing crack evaluations is summarized in SECY-83-267C (6.1) and presented in greater detail in draft NUREG
Report 0313, "Technical Report on Material Selection and Processing Guidelines
for BWR Coolant Pressure Boundary Piping-Revision 2" (6.2).
The evaluation of
cracks found in service requires a reasonably accurate knowledge of:
1) the
ability of ultrasonic testing (UT) to size the length and depth of stresscorrosion cracks accurately, 2) the applied and residual stress states, 3) the
relation of stress-corrosion crack growth rate to stress and environment, and
i
6-6
4) the ultimate load-carrying capacity of the cracked pipe.
Section 4 of this
report is devoted to the discussion of UT techniques, including crack sizing

capabilities.
The following two sections discuss the methods for predicting
crack growth in piping weldments and determining the ultimate load carrying
capacity of flawed pipes and the associated acceptable crack dimensions.
6.2.1.1
Prediction of Corrosion Crack Growth in Piping Weldments
Analyses of stress-corrosion crack growth in weldments are usually based

on linear elastic fracture mechanics (LEFM) and assume that the crack growth
is controlled by the stress intensity factor, K.
Stress-corrosion crack
growth rates have been measured in laboratory as a function of stress intensity.
The stress intensity factors associated with cracks in piping can
be calculated, if the crack size and shape and the stress configuration
including residual and applied stresses are known.
These calculated stress
intensities are then combined with the measured crack growth rates to predict
the growth of cracks in reactor piping systems. Direct quantitative confirmation of this approach is not available.
It has been argued that it is
inapplicable in the plastically deformed region near a weld (6.3).
However,
considering the variability in crack growth rates for different heats of

material and enviromental conditions and the variability in residual stresses,
the predictions of the model seem reasonably consistent with field experience
and laboratory measurements on large-diameter pipes (6.9).
6-7
since for a given weldment containing a crack the applied stresses are ^M
presumed known from the stress report, the degree of conservatism in the
predicted crack growth depends on the choice of crack growth rate law, on the
choice of throughwall residual stress distribution, and to a lesser degree on
the assumptions used in computing the stress intensity.
In the approach used
currently by NRR, conservative assumptions are made for both the crack growth
rate law and the throughwall residual stress distributions, and the stress
intensity factor is computed assuming a complete 360o circumferential crack.
Although it is conceivable that a particular weldment could have both a
material with an unusually high crack growth rate and an unfavorable residual
stress state, the probability of this is very low, and the overall crack
growth in weldments predicted using the NRR assumptions appears to bound that
observed in-reactor.
Stress-Corrosion Crack Growth Rates
In large-diameter piping the fatigue crack growth associated with design

loading histories is very small, and in most cases crack growth will be due
primarily to IGSCC.
Cracks in smaller-diameter pipes where transients con-
tribute more significantly to crack growth are generally repaired.
The
available data suggest that the contribution of the conventional design

operating transients to crack growth is negligibly small (because they
comprise such a relatively small fraction of the life) and that most of the
crack growth occurs under the nominal steady-state operating conditions.
However, it should be noted that even under "steady" load, the stresses in the
piping may not be truly constant because of pressure and temperature fluctuations and mechanical vibrations.
Although in most cases the associated
6-8
stresses are very small and are negligible from a fatigue standpoint, small
variations in stress can have significant effects on stress-corrosion crack
growth (6.5).
The available data for constant and "nearly constant" loads are
summarized in Figure 6.1 along with the correlation currently used by NRR in
the analysis of flaw growth in reactor piping.
Much of these data has been obtained on rather heavily sensitized

materials in environments with 8 ppm dissolved oxygen.
The sensitization
level is higher than would be considered typical in as-welded material, and

the test environments contain higher dissolved oxygen levels than the 0.2 ppm
typical of operating BWRs.
However, most of the tests have also been carried
out in high purity water with impurity anion levels (sulfate, chloride,
carbonate, etc.) substantially below those possible in BWRs operating under
Regulatory Guide 1.56 limits on BWR water chemistry.
Recent data from slow
strain rate tests (6.6, 6.7) suggest that although there are significant
differences in susceptibility to IGSCC in high purity environments with 0.2
ppm and 8 ppm dissolved oxygen, additions of very low levels of impurities
substantially reduce these differences.
Similarly the addition of impurities
seems to diminish the differences in susceptibility associated with different

degrees of sensitization (6.6).
Even the sensitization levels may not be too
conservative, since for weldments with relatively low levels of sensitization

after welding, low temperature sensitization may occur during service. The
uncertainties associated with degree of sensitization and water chemistries
appear to be no larger than the heat-to-heat variability (e.g. , the two heats
of material represented by the solid triangles in Figure 6.1, which have
identical heat treatments, measured sensitization values typical of weldments.
6-9
10
,-4
10
^ 10"^
ppm Oj; sensitized at 1160*F/2 h
(EPR 15 C/cm^) - GE
2 ppm 0,; sensitized at 1150*F/2 h
(EPR 10 C/cm^) - GE
2 ppm 0-; sensitized at n50'F/24 h
cr
; severely sensitized
sensitized at n50*F/24
HITACHI
ANL
o
cr 10"
o
<I
l - ^ H S ppm 0,; sensitized by welding;

LTS at 932*F/24 h (SRI,
.^
O 8 ppm 0^; sensitized at (1292'F/10 m i n P
(932"F/2 h) (EPR 4 C/cm^)
A
8 ppm 0,; sensitized at (1292*F/10 min)
2'
(932*F/24 h) (EPR 4 C/c')
f 0.1 Hz. R - 0.94
W
8 ppm O^; sensitized at (1292*F/10min)
(842*F/257 h) (EPR - 15 C/cm^)
O 8 ppm 0^; sensitized at (1292*F/10niin) )RECENT
+ (842'F/257 h) (EPR 15 C/cm^)
ANL
f - 0.1 Hz, R 0.94
DATA
#
8 ppm 0^: sensitized at 1292*F/14 h
- 0 04 in /yr
Q:
I
10,-7
(EPR - 20 C/cn^) f - 0.008 Hz,

R 0.95
8 ppm 0^; sensitized at 1292*F/14 h
(EPR 20 C/cm^) f 0.08 Hz,
R 0.95
10
20
30
40
50
60
70
STRESS INTENSITY, K(ksi>/in.)
Figure 6.1
Stress-corrosion crack growth rates
6-10
and were tested in the same laboratory, give crack growth rates very close to
the upper and lower bounds on all the data).
Thus on the basis of the
available information, the data presented in Figure 6.1 are not considered to
be an unduly conservative representation of the crack growth rates expected in
a reactor environment.
Current evaluations by NRR utilize the power law curve identified in

Figure 6.1 which appears to be conservative for most heats of material under
typical reactor operating conditions.
The roll-off in crack growth rate with
increasing stress intensity predicted by the power law representation does not
appear consistent with the data in all cases, but this is of little consequence in the analysis of throughwall crack growth, since the stress intensity
factors in this case are not large enough to require extrapolation beyond the
available data.
SCC growth rates in many material/environment systems show a plateau or

limiting growth rate.
In mechanistic terms (6.3) a plateau could arise
because there is a limiting rate for environmental attack at the crack tip.
In phenomenological terms crack branching may occur which limits the stress
intensity at the crack tip.
In the stainless steel/BWR coolant system, the
mechanistic estimates of plateau rates are very high.
Crack branching is
observed in both the field and the laboratory tests; however, the conditions
that produce branching are not well understood.
Since the mechanical loading
in constant extension rate tests (CERT) is much more severe than could be
encountered in any actual piping system, use of the crack growth rates
observed in these tests should provide a very conservative estimate of the
highest crack growth rates that could occur due to SCC in piping systems.
6-11
To
compute the growth of throughwall cracks for an analysis of leak-before-break

the very conservative representation of the crack growth rate dependence on
stress intensity by a straight line in the semi-log plot of Figure 6.1, which
uses CERT data to estimate a plateau rate, is used.
Residual Stresses in BWR Pipe Weldments
Residual stresses in the weldment arise because of the thermal and

mechanical loading produced during the welding process.
The thermal expansion
and contraction associated with the temporal and spatial variations in

temperature give rise to plastic strains, and these incompatible strains
produce residual stresses upon final cooling.
A substantial effort has been
made to experimentally measure the residual stresses in BWR pipe weldments.

However, there is a great deal of variability in residual stress distributions
from weldment to weldment even for large-diameter pipes.
The available
throughwall data for axial residual stresses in large-diameter pipes are

smmnarized in Figures 6.2 and 6.3 (6.10).
As can be seen, there is a wide
range of data, presumably reflecting the variability in the welding techniques

used to prepare these test weldments and the variability to be expected in the
field.
The distribution of residual stress used by NRR for evaluation of flaw
growth is also shown in these figures.
It is seen to be a conservative repre-
sentation of the available data for large-diameter weldments.
Much of the information available is based on finite element calculations, since experimental measurements of residual stresses in every pipe size
and weld condition of interest are not available.
The experimental data serve
as a "benchmark" for the finite element calculations, and the validity of the
6-12
INSIDE WALL
OUTSIDE WALL
n
O
D
A
0
\\\mrq
6E26in NP944-I
6E26 in IHSI ref pipe (4 ozimuths)
ANL 26 in ND 9 4 4 - 2 (2 azimuths)
ANL 26 in KRB
^
\
ANL 20 in T-II4
/ ^
FINITE ELEMENT
DO
D
NRR
EVALUATION
CURVE
01
0 2 0 3 0 4 0 5 0 6 07 0 8 0 9
o/t
10
Figure 6.2 Throughwall distribution of

axial residual stress in large-diameter
weldments.
04
06
DEPTH/THICKNESS
Figure 6.3 Throughwall distribution of

axial residual stress in a large-diameter
weldment.
6-13
calculations at least for semiquantitative prediction has been established.

Although a number of finite element models for the welding process have been
developed, most of the modeling work directly related to multipass welds on
heavy-walled piping has been based on a finite element code developed at
Battelle Columbus Laboratories (6.8) under NRC sponsorship and subsequently
refined and modified under EPRI sponsorship (6.9).
The finite element calculations, which are supported by the limited

experimental measurements available (6.11, 6.12), suggest that there are
significant differences in the throughwall distributions of residual stress in
small-diameter and large-diameter pipe weldments.
Stress distributions in
10- to 12-in. weldments appear to be quite dependent on the weld heat input.
The assumed stress distribution used for evaluation of flaws in small-diameter
weldments (less than 12 inches) is shown in Figure 6.4.
It is a conservative
representation of the available data and computational results.
INSIDE WALL
50
OUTSIDE WALL
w
LLI
I-
Figure 6.4 Assumed throughwall

welding residual stress distribution in small-diameter weldments (<12 in.).
6-14
All of the available experimental measurements have been performed on

pipe-to-pipe welds, which represent only a portion of the welds in a reactor
piping system. However, similar results can be expected for pipe-to-elbow
welds, since the local geometry near the weld is similar.
Finite element
calculations for a variety of weld geometries have been carried out (6.9).
The results of these calculations indicate that the axial stress distributions
used for evaluation appear to be very conservative for pipe to component
welds.
Elastic Superposition and the Calculation of Stress Intensity Factors
The use of linear elastic fracture mechanics assumes that the residual
stresses due to welding (or IHSI or a weld overlay) can be linearly superposed
on the operating stresses to determine the total stress acting on the weldment.
The assumption of linear superposition is compared with an elastic-
plastic finite element analysis in Figures 6.5 to 6.7.
For stress levels
typically encountered in BWR piping systems the assumption gives satisfactory

results, although it should be used with some caution with high applied
stresses.
Stress intensity factors for circumferential cracks in piping can be

calculated by a variety of methods. Probably the most convenient is the
influence function technique (6.13, 6.14).
For finite aspect ratio cracks,
growth in both the throughwall and circumferential directions must be considered.
However, the assumption of a complete circumferential crack, which
is currently the basis for most assessments by NRR, give a conservative

estimate of the stress intensity for throughwall crack growth (6.13).
6-15
AXIAL STRESS (MPo)
AXIAL STRESS (MPo)

-300-200HOO 0
-300-200H00
l(X) 2(X) 300 4(X)
100 200 300 400
ELASTIC
PLASTIC FEA
ELASTIC
SUPERPOSITION
3t/4
3t/4
t/2
t/4
ID
-40
-40
-20
0
20
AXIAL STRESS Iksi)
-20
0
20
AXIAL STRESS (ksi)
Figure 6.6 Comparison of e l a s t i c p l a s t i c solutions for throughwall

axial s t r e s s e s with simple e l a s t i c
superposition, 18-ksi applied load,
Figure 6.5 Comparison of e l a s t i c p l a s t i c solutions for throughwall

axial s t r e s s e s with simple e l a s t i c
superposition, 5-ksi applied load.
AXIAL STRESS (MPa)

-300-200H00
100 200 300 400
ELASTIC
PLASTIC FEA
ELASTIC
SUPERPOSITION
3t/4
t/2
('
)
/
t/4
Figure 6.7 Comparison of elasticplastic solutions for throughwall

axial stresses with simple elastic
superposition, 26-ksi applied load.
ID
-40
-20
0
20
AXIAL STRESS (ksi)
6-16
In addition to the assumption that the residual and operating stresses

can be linearly superposed, it is also assumed that the stress intensity due
to the residual stress can be calculated based on the residual stresses acting
on the crack plane in an uncracked pipe. Clearly the residual stresses will
redistribute and relax as the crack grows.
The influence function solutions
do account for this relaxation at least approximately.
However, the validity
of this approach in the presence of steep axial gradients of residual stress

and in a region which has undergone extensive plastic deformation is still a
somewhat open question for deep cracks (6.13) (i.e., over halfway through the
wall).
6.2.1.2 Determination of Allowable End-of-Operation Crack

Dimensions: Current Approach
To establish the acceptable crack sizes at the end of a specified

operating interval, NRR allows only two-thirds of the crack depths provided in
Paragraph IWB-3640 of Section XI of the ASME Code.
The two-thirds allowable
is intended to account for uncertainties in crack sizing and roughly translates to a factor of 2 on crack size.
(if the original crack depth is
actually twice the value reported from UT data, the crack will still remain
within the IWB-3640 acceptable values at the end of 18 months of operation.)
The description of the flaw acceptance criteria in SECY-83-267c further states
that the staff criteria would likely require repair for cracks greater than
30% of the circumference in length and cracks with a reported depth 25% or
greater of the wall thickness.
6-17
The acceptable flaw sizes presented in IWB-3640 are intended to provide ^

factor of safety against pipe failure of at least 2.77 for nominal operating
conditions and at least 1.38 for emergency and faulted conditions.
The
critical crack dimensions in IWB-3640 were determined using net section

collapse, also referred to as limit load or plastic collapse analysis. The
applicability of net section collapse analysis to the evaluation of flaws in
the wrought base metal of stainless steel piping was demonstrated in a series
of pipe fracture experiments conducted at Battelle Columbus Laboratories (BCL)
for the Electric Power Research Insitute (EPRI) (6.15) and the analytical
studies reported in EPRI NP 2661.
The tests performed by BCL included 18 quasistatic experiments performed

on Type 304 stainless steel at ambient temperatures.
Eight of the experiments
involved throughwall circumferential cracks in 2-, 4-, and 16-in.-diameter

pipes.
A total of ten circumferential surface crack experiments were per-
formed in 4- and 16-in.-diameter pipe. All of the experiments involved

bending loads.
In an earlier program at BCL (6.16), two 4-in.-diameter
throughwall crack tests were performed on stainless steel pipes under combined
pressure and bending loads. The largest difference between the maximum load
predicted by net section collapse analysis and the maximum load in the experiments was 14%, and the differences in the remaining tests were all within a
few percent.
The one exception was the test of a 16-in.-diameter surface
flawed pipe.
This experiment involved a surface crack approximately 66%
throughwall and 170o around the pipe circumference.
The maximum load
observed in the experiment was approximately 14% below the failure load
predicted by net section collapse. Overall, the tests conducted by BCL
convincingly demonstrated the validity of net section collapse analyses under .
the range of crack dimensions, materials, and load combinations considered.
6-18
^^
Recently, several concerns have been raised, from various sources,
^^regarding IWB-3640 and the use of net section collapse analysis. These
concerns can be suimnarized briefly as follows.
Large uncertainties in the ability of ultrasonic testing techniques to

size the depth of IGSC cracks accurately raise serious concerns about
the ability to perform a reliable flaw evaluation.
IWB 3640 is based on primary loads only and did not include secondary
loads (e.g., thermal, support failures) in determining allowable flaw
sizes.
Flaw evaluation criteria should include consideration of "preservation

of structural ductility" in piping systems with cracks present.
Ductile fracture toughness data show much lower resistance, to crack

extension in stainless steel welds than in wrought stainless steel
base metal.
This lower toughness could invalidate the use of net
section collapse analysis.
The net section collapse formulation for circumferential cracks used

in IWB-3640 can predict only gross failure of the cracked pipe cross
section.
There is a possibility that a surface crack could break
through the remaining wall thickness before reaching net section

collapse conditions and become a throughwall crack.
This condition is
referred to as a ligament instability.
These concerns are addressed in the following discussions.

6-19
Flaw Sizing Considerations
Section 4 of this report discusses the difficulties associated with the

sizing of IGSC cracks.
Results of tests conducted by Pacific Northwest
Laboratory and EPRI suggest large deficiencies in the ability to measure the
depth of IGSCC accurately.
Furthermore, the trend in these round robins was
to undersize deep cracks (greater than 20% throughwall).
The current NRC flaw
evaluation procedures address this issue by applying a factor of two-thirds to

the flaw depths allowable by IWB-3640.
In marginal cases, independent and
additional crack depth measurements are required.
Another proposal which has
been made in an attempt to circumvent the problem of crack sizing is to

establish an acceptance criterion based on crack length.
The idea is that
flaw length measurements of IGSCC are more reliable than depth measurements
and a maximum acceptable crack length can be established such that even if the
crack broke through the wall of the pipe it would be small enough in length
that gross failure of the pipe would not occur and the resulting leak rate
would be acceptably small.
Assumed Loads and Structural Ductility
The Task Group has the following opinions regarding the second and third
issues described previously.
6-20
In general, secondary stresses should be included in the flaw evaluation

if, as described in the next paragraph, the material toughness is low enough
to preclude reaching limit load prior to crack instability.
The Task Group's
fracture mechanics analyses described in subsequent sections of this report

include thermal expansion stresses, which are conservatively included as
primary stresses.
The Task Group further believes that other secondary
stresses (e.g., through-the-thickness stresses) do not contribute significantly to crack driving potential and, in view of the conservative treatment
of thermal expansion stresses, can be neglected in the flaw evaluation.
The secondary stresses resulting from support failures and the ductility
issue apparently result from a concern that displacements during seismic
events may be so large that loads on the supports and at locations in the
piping system may exceed the ultimate load.
It has been suggested that piping
systems be shown to be able to tolerate added loads due to support failure and
demonstrate that cracked sections have sufficient ductility (have net section
plasticity) to absorb the energy associated with the postulated extreme displacements.
It is the Task Group's opinion that the conditions associated with the
suggested extreme displacements do not represent a credible event.
Further-
more, the suggested displacement and loading conditions are inconsistent with
the ASME Code philosophy, which has been deemed acceptable for the design and
operation of reactor components.
Therefore, the Task Group has concluded that
the loading conditions defined in the ASME Code are acceptable for performing
evaluations of cracked piping.
The suggested approach of postulating large
displacements is acceptable but is considered overly conservative and should

not be a general requirement.
6-21
Low Toughness and Ligament Instability Considerations
Regarding the issue of lower resistance to crack extension in weld

deposits relative to wrought materials and the potential for ligament instability, the concern is that the assumptions associated with net section collapse may be invalid.
Specifically, the assumption that no crack extension
will occur prior to reaching maximum load may be incorrect.
Destructive ex-
amination of a limited number of cracked weldments removed from service has

revealed that cracks which initiate in the HAZ by SCC may grow into the weld
metal.
This type of cracking behavior also has been observed in laboratory
tests.
Although crack growth into welds is rare and, in service, has been
associated with low-ferrite-content repair welds, the possibility that IGSC

cracks may grow into the weld material, regardless of the ferrite level, cannot be ignored.
In relation to IWB-3640, this issue must be addressed for
both weld metal and cast stainless steel materials which also have reduced
fracture toughness.
To address the issues of low fracture toughness and the potential for
ligament instability, a series of fracture mechanics calculations using the
J-integral tearing instability methodology was performed.
analyses are presented in Appendix F.
Details of the
The assumptions made in the analyses
and the results are summarized here.
The first set of calculations was performed for postulated elliptically

shaped, inside-surface cracks half the pipe circumference in length and having
assumed maximum crack depths of 25%, 50%, and 75% of the wall thickness.
6-22
^ ^ These flaw geometries were evaluated for 12-, 22-, and 28-in.-diameter pipes
with wall thicknesses typical of BWR recirculation system piping.
The flaw
geometry was selected because it is believed to be representative of actual

stress-corrosion crack geometries and because this flaw geometry with the
assumed 50% crack depth is believed to bound the most severe end of operating
cycle crack geometry predicted for those cracked pipes which have gone back
into operation with no repair.
Loading conditions evaluated include service level A (pressure, thermal,

and deadweight loads) and service level A plus additional bending moments up
to twice SSE load.
These loading conditions correspond to stress ratios of
0.7 under normal operating conditions in 28- and 22-in.-diameter pipes and
0.9 in 12-in.-diameter pipes.
The stress ratios for the service level A plus
one SSE loads are 1.3 in 28- and 22-in.-diameter pipes and 1.5 in the 12-in.diameter pipe.
These loads represent reasonable upper bounds on actual plant
loading conditions.
It is expected that the majority of the stress-corrosion cracks that

occur in service will not lie solely in the weld metal but will be in zones of
material having substantially higher resistance to crack extension (aside from
IGSCC) than the weld metal.
However, the potential for ductile extension of
the surface cracks was evaluated for the above loading conditions using the
lower bound of the ductile fracture toughness properties generated for the NRC
for stainless steel welds.
These data were generated from welds made using a
hot wire tungsten inert gas welding process and are considered a lower bound
of the fracture toughness properties for the majority of the circumferential
^ ^ w e l d s in BWR piping.
However, the Task Group has recently learned that a
6-23
significant percentage (approximately 30%) of the circumferential welds in BW|

piping are shop fabricated submerged arc welds.
It has also been learned that
some fraction of these welds, the exact number of which has not been tabulated, were not solution annealed.
These findings are important because
limited data show that ductile fracture toughness properties for submerged arc
welds are significantly lower than those used in this evaluation.
The Task
Group is therefore initiating a high priority effort to evaluate the impact of

the submerged arc weld fracture toughness data on the calculations presented
here.
The Task Group also recommends that NRC and industry initiate high
priority efforts to generate more fracture toughness data for the range of
weld types existing in BWR plants.
The results of the surface crack analyses are presented in Tables 6.1 and
6.2.
Because of limitations in the available calculational techniques,
analyses were not performed for cracks 75% of the wall thickness in depth at
loads greater than service level A.
These limitations in the analysis
techniques did not allow a determination of the exact margins against fracture
to be made. However, for the service level A normal operating condition
loads, no ductile crack extension was predicted for crack depths up to 75%
throughwall. At normal operating loads for the 50% deep crack (which is
deeper than any known crack in service), the factors of safety (based on load)
against ductile crack extension are at least 2.7 for the 22- and 28-in.diameter pipes and 2.3 for the 12-in.-diameter pipe.
The factors of safety
against actual pipe failure will be larger than these factors of safety
because there is additional load-carrying capacity beyond the initiation of
crack extension to crack instability.
A factor of safety against failure of
6-24
TABLE 6.1
Results of Fracture Mechanics Analyses for 22- and 28-Inch-Diameter
Pipes With Surface Flaws 50% of the Circumference in Length
Loading
Condition
SLA
SLA + 1 SSE
SLA + 2 SSE
Stress
Ratio
PM + fR.
SM
0.7
1.3
1.9
Maximum Crack Depth in % of Wall

25%
50%
no initiation
no initiation
no initiation
no initiation
no initiation
no initiation
75%
no initiation
*
*
TABLE 6.2
RESULTS OF SURFACE CRACK ANALYSES FOR 12-INCH-DIAMETER PIPE
Stress
Ratio
Loading
Condition
SLA
SLA + 1 SSE
SLA + 2 SSE
NOTES:
SM
0.9
1.5
2.1
Maximum Crack Depth in % of Wall

25%
no initiation
no initiation
no initiation
50%
no initiation
no initiation
no initiation
75%
no initiation
*
*
SLA = Service Level A; SSE = Safe Shutdown Earthquake.

* = This case was beyond the limits of the analysis technique.
6-25
approximately 2.7 or greater would be predicted by IWB-3640, which is based
net section collapse analysis for this crack geometry and normal operating
condition loads.
For the 50% deep crack under service level A plus 1 SSE
loading the factors of safety are 1.5 for the 22-in.-diameter pipes and 1.4
for the 12-in.-diameter pipe.
These factors of safety are approaching those
intended in IWB-3640.
Tne results described in the previous paragraph were obtained for service
level A and seismic stress conditions that represent the largest stress
reported for the population of welds in 12-, 22-, and 28-in. BWR piping.
These loads are associated with a relatively small percentage of the weld
connections in the recirculation system.
To illustrate the increased margins
that exist for the majority of weld connections in the BWR recirculation
system, surface flaw analyses were performed for a weld connection in a
28-inch pipe having a typical rather than bounding stress condition. The
results from this analysis indicate that margins of approximately 3.3 for
service level A conditions are obtained for most weld joints compared to
values of 2.7 for the bounding stress location.
Because of the limitations in the surface crack analysis techniques, the

factors of safety against ductile crack extension under emergency and faulted
conditions cannot be determined for the 75% and deeper crack depths. Tearing
instability calculations for throughwall cracks, presented in the next section
on leak-before-break, give some insight into margins for these crack sizes and
loading conditions.
6-26
6.2.2
Leak-Before-Break-Evaluations
The leak-before-break failure mode in piping refers to the concept that a

crack propagating through the wall of a pipe, by mechanisms such as SCC or by
ductile crack extension due to applied loads, will result in a stable throughwall crack that can be reliably detected by leakage.
To evaluate these
margins for leak-before-break, IGSCC analyses were performed assuming stresscorrosion cracks exist in the wrought stainless steel base metal.
In addi-
tion, ductile fracture mechanics analyses were performed for cracks assumed to
exist in the lower toughness stainless steel weld metal.
These calculations
are summarized below, and in Appendix F.
6.2.2.1 Margin for IGSCC Leak-Before-Break in Wrought Stainless Steel
The inherent toughness of Type 304 stainless steel is expected to lead to

leak-before-break behavior, and to the knowledge of the Task Group, no sudden
catastrophic failures of this material have occurred in the nuclear industry
or in a wide variety of other industries. However, IGSC cracks in largediameter pipe weldments are generally very long in relation to their depth;
therefore, the possibility that cracking will occur 360o around the circumference before a throughwall crack develops must be considered.
Unless IGSCC growth is arrested by a favorable residual stress distribution, failure of the piping system could occur by continued growth of the
crack by stress corrosion without the necessity for postulating loads on the
piping beyond the normal operating and upset loads.
6-27
The determination of the margin for leak-before-break requires an
analysis of the failure behavior of the pipe for a particular crack geometry,
calculation of the crack opening area, and a calculation of the flow through
the crack opening.
In the results presented here it is assumed that the crack
is not in the weld metal but lies in a zone of relatively high toughness
material and the failure of the cracked pipe is assumed to be adequately
described by a net section stress approach similar to that used in IWB-3640.
This assumption is expected to be valid for the majority of the cracks that
occur in operating BWRs.
The calculation of the leak rate through the crack
is subject to considerable uncertainty.
The crack opening area has been
estimated on the basis of elastic solutions for a throughwall crack in an

infinite plate with an approximate correction for plasticity effects based on
the Dugdale model (6.17).
More complete elastic-plastic solutions for the
crack opening area suggest that this solution is conservative (6.18).
The flow of steam-water mixtures through tight cracks is a complex

function of crack geometry, crack surface roughness, temperature, and pressure. Limited experimental measurements of the fluid flux through IGSCC and
simulated cracks have been carried out (6.19, 6.20).
Comparisons of these
data with the flow predicted by the homogeneous critical flow model developed
by Henry (6.21) indicate that the actual flux is from 1-1/10 of the corresponding frictionless flux predicted by the Henry model. Although models which
attempt to account for the frictional losses have been developed (6.6), the
limited data base and the uncertainty about the internal crack geometry, corrosion product deposition, and roughness make it difficult to make accurate
predictions of the flow through an IGSCC.
I
6-28
This uncertainty has surprisingly little effect on the assessment of the

leak-before-break margin in piping.
10-in. pipe.
Figure 6.8 shows a typical case for a
It is assumed that the crack can grow completely around the
pipe circumference before a throughwall crack occurs. The collapse curve in

Figure 6.8 assumes that the net section stress at failure equals 3 Sjj,
The
observed variations in flux are used to estimate upper and lower bounds on the
crack sizes necessary to obtain a 5-gpm leak rate. Based on crack size there
is a significant leak-before-break margin unless the circumferential crack is
quite deep.
Because of the steep nature of the collapse curve for deep cir-
cumferential cracks, the size of the crack that can be postulated before
violation of leak-before-break is not very dependent on the value chosen for
the fluid flux (6.22).
The corresponding case for a 24-in. pipe is shown in Figure 6.9.
The
size of the crack needed to obtain detectable leakage is only very weakly
dependent on pipe diameter, while the size of the crack needed to produce
collapse is roughly proportional to pipe diameter. Hence the margin for leakbefore-break generally increases with pipe size. As Figures 6.10 and 6.11
show, however, the leak-before-break margins are smaller at highly stressed
joints.
Estimates of the time from the onset of unallowable leakage to structural

failure depend strongly on the mechanisms of crack growth and the corresponding crack growth rate assumptions.
In most cases most of the remaining
crack growth will occur by stress corrosion.
Figure 6.12 summarizes the
available data on IGSCC growth rates in BWR type environments.
6-29
Both data from
16
14
^v
10- in. PIPE A

u - 10 ksi
15
13-^^
30
\X
y
/
1
24-in. PIPE
0- = 10 ksi
24 1
.
/COLLAPSE
^ CURVE
22
20 r
18L
r
16
7/
^
i_ UPPER BOUND
6
^
5 qol/min
5
3-
26 L
CURVE
'
28 1
^'COLLAPSE
12
II _>I0DAYS \ \
FROM ONSET \\
10- OF LEAKAGE^ \ \
TO FAILURE ^
9
<r
cc
o
>7o\ \
-^-^DAYS^
LOWER BOUND
^
^
5 gol/min
>I5 DAYS FROM

\
ONSET OF LEAKAGE \
TO FAILURE
\
14
1 UPPER BOUND
\
12 5 qoi/min
\
\
10
8 h LOWER BOUND
6 \ 5 qol/min
4 |
20
DAYS 1
\^
"
1
0.2
1
0.4
1
0.6
0.8
1.0
0.2
0.4
0.6
' 1
0.8
1.0
FRACTIONAL DEPTH OF 360" CRACK
Figure 6.8 Collapse curve for a

10-inch pipe with throughwall and
part-through 360 cracks with bounds
for crack sizes for 5 gpm. Leak rate
under an applied stress of 10-ksi.
6-30

for crack sizes for 5 gpm. Leak rat
16
10-in. PIPE
a --17 ksi
15
14
28
13
26
12
24
II
22
.COLLAPSE
CURVE
10
20 :
cc
o
o
t
cc
o
UPPER
5 qol/min
'COLLAPSE
CURVE
14
" UPPER BOUND

12 5 gol/min
10
> 1J UAib
6 _ LOWER BOUND^
5 gol/min
LOWER BOUND
5 qol/min
0.2
1
24-in. PIPE
0- = 17 ksi
16
6
5
18
> 10 DAYS FROM

ONSET OF LEAKAGE
TO FAILURE
8
<t
30
0.4
2
0.6
0.8
1.0
I ^^^-^
1
0.2
FROM ONSET
OF LEAKAGE
"
, TO FAILURE
_
0.4
0.6
0.8
1.0

6-31

fracture mechanics tests under controlled load and slow strain rate tests in
which the load is varied to maintain the nominal strain rate constant are
included.
It is unlikely that the strain rates commonly used in slow strain
tests can be maintained under reactor loading conditions, and therefore an

upper bound for in-reactor loading based on growth rates in slow strain tests
should be very conservative.
The crack growth curve selected in Figure 6.12
bounds all available fracture mechanics data (the plateau rate is ten times
the highest reported rate in a constant load fracture mechanics test), and the
plateau rate is typical of slow strain rate tests.
10-2>CERT
CGR
10,-3
10"
10-='-
n-6
10',-7
_L
0
20
40
_L
60
_L
Figure 6.12 Conservative estimate of

upper-bound stress-corrosion crack
growth rates of Type 304 stainless
steel in BWR environments.
80
STRESS INTENSITY, K (ksiyjn)
6-32
Using this crack growth rate curve and using the usual linear elastic
fracture mechanics approach, the time from the onset of leakage to failure can
be estimated for different crack geometries.
For simple throughwall flaws
conservative estimates of the times range from two months for the 10-inch pipe
to three months for the 24-in. pipe.
Even for the case of the compound crack
with a complete circumferential part-through crack in addition to the throughwall crack, the time is relatively long for most crack geometries.
Figure 6.8
shows the bounding envelope of crack geometries for which there are at least
ten days before the onset of unallowable leakage and failure.
Leak-before-
break in this sense is violated only in the case of very deep, very long partthrough cracks.
Leakage from cracks that are not large enough to cause collapse can be
very large.
For a 24-in. pipe, the leak rate through a 22-in. crack (ap-
proximately 30% of the circumference) under typical applied stresses is

approximately 100 gpm, which is close to the makeup capacity of most BWRs.
Hence, while leak-before-break is still valid, in many cases with cracks
greater than 30% of circumference, the resulting leakage could require use
of the ECCS.
6.2.2.2
Margins for Leak-Before-Break in Stainless Steel Weld Metal
For cracks located in the stainless steel weld metal, the lower
toughness relative to the wrought material may invalidate the use of limit
load analysis.
For this reason a series of tearing instability calculations
6-33
were performed for simple throughwall crack geometries.

these calculations were twofold.
The objective of
First, to determine if throughwall cracks of
a length reliably detectable by leakage have adequate margins against fracture

and, second, to determine the maximum length of crack that could become
throughwall and remain stable.
For these analyses the same loading conditions and material properties
used for the tearing instability analyses in the section on flaw evaluation
procedures were used.
crack.
The crack geometry considered is a simple throughwall
Details of these calculations are presented in Appendix F.
The loads
for crack instability were calculated for throughwall crack lengths corresponding to a 10-gpm leak rate under normal operating conditions (service
level A) in 12-, 22-, and 28-in.-diameter pipes.
7.2 in., 9.6 in., and 10 in., respectively.
These crack lengths were
The 10-gpm leak rate represents a
safety factor of 2 relative to the 5-gpm limit for unidentified leakage

currently imposed on most operating BWRs.
The results of the calculations showed that under normal operating

conditions (service level A) the factor of safety on stress against unstable
fracture of the throughwall cracks ranges from a minimum of 2.5 for the
12-in.-diameter pipe to approximately 3.5 for the 22- and 28-in.-diameter
pipes.
For faulted conditions (service level A plus one SSE) the factors
of safety are approximately 1.5 for the 12-in.-diameter pipe and 1.9 for
the 22-and 28-in.-diameter pipes.
Thus, for 12- to 28-in.-diameter pipes
significant margins against unstable fracture exist for cracks which should
be reliably detected by leakage.
6-34
^^
In addition to the above, calculations were performed to determine ap-
^^proximately what length of throughwall crack would become unstable under the
assumed bounding loading conditions (service level A plus 1 SSE). The
calculations are important for two reasons.
First, they show the large
difference between the throughwall crack length that should be reliably

detected by leakage and the throughwall crack size that could result in
unstable fracture of the pipe.
This difference in crack length indicates that
there is a large amount of time in which to detect leakage from a throughwall

crack before it reaches a critical crack length, thus providing additional
confidence in the applicability of leak-before-break.
The calculations are also important because of the large uncertainties

associated with the ability to size the depth of IGSC cracks (see Section 4).
These uncertainties suggest that it would be desirable as a flaw evaluation
procedure to limit the length of acceptable IGSC cracks such that if they
became throughwall cracks under normal or postulated accident conditions they
would not result in unstable pipe fracture or unacceptably large leak rates.
The limit on leak rates stems from the desire not to challenge emergency core
cooling systems.
Tearing instability calculations were performed for throughwall cracks

extending around 30% and 26% of the circumference for 22- and 28-in.-diameter
pipes, respectively.
The crack length resulted in factors of safety on stress
of approximately 1.2 for the two pipe diameters under faulted loading conditions.
These calculations indicate that even in the event that a surface
crack, of the type evaluated in the section on flaw evaluation, suddenly

^ ^ became a throughwall crack under accident loading conditions, it would remain
6-35
stable if the crack were less than approximately 30% of the circumference in
length.
On a weld joint specific basis the length of the critical throughwall
crack based on fracture mechanics would be larger than 30%, due to lower
applied stresses than assumed in the analysis.
6.3
SHORT-TERM SOLUTIONS
This subsection is concerned with measures considered acceptable by the

NRC for relatively limited periods; e.g., 1 to 2 fuel cycles. Inmost instances weld overlays or similar approaches are used to "buy time" where the
assumption is that many of the piping systems containing several cracks may
ultimately have permanent corrective measures. Measures such as IHSI
described in subsection 6.4.3 may represent a short-to-long-term mitigative
measure.
The event tree approach in Section 6.1 attempts to display the
interaction of the various factors impacting on a decision to continue to

operate or to make limited or longer-term repairs.
A variety of techniques have been used or proposed for short-term repair

of flawed piping including replacement of a short portion of the piping, a
welded "clamshell," mechanical clamping devices, and reinforcement by deposit
of weld metal on the outer surface of the weldment (weld overlay).
In this
country the vast majority of repairs have been done by weld overlay.
The weld
overlay obviously provides structural reinforcement; in addition the radial

shrinkage induced by the overlay produces strongly compressive residual
stresses on the inner surface of the weldment, and the low-carbon, highferrite level. Type 308L weld metal used for the overlays is more resistant to
IGSCC than sensitized Type 304 stainless steel.
6-36
The compressive stresses induced by the overlay should be effective in

preventing the initiation of new intergranular stress-corrosion cracks and
inhibiting the growth of existing flaws.
Calculations suggest that for crack
depths up to 60% of the wall thickness the stresses ahead of the crack will be
compressive for applied stresses up to 9 ksi, which is typical of the stress
levels in large-diameter piping.
However, because of uncertainties in crack
depth measurement, it is difficult to assure that any particular crack is

under a favorable stress condition that would inhibit further IGSCC growth.
At present it is also not possible to rely on the weld overlay itself to
arrest IGSC crack growth, since although it has been clearly demonstrated that
this type of material is very resistant to crack initiation, its ability to
arrest crack growth in the presence of high stress intensity factors has not
been convincingly demonstrated.
Since it is also very difficult to carry out
reliable inservice inspections through the overlay, the Task Group concurs
with the current NRR position that the weld overlay be considered as a shortterm remedy used primarily to allow time to prepare for the replacement of
severely cracked pipe.
In cases in which the weld overlay is applied to
relatively short deep cracks to prevent leakage rather than to prevent

structural failure, longer-term operation with weld overlays can be
considered.
Because of the resultant difficulties in inspection, routine
application of weld overlays to induce favorable residual stresses in

weldments with no detectable cracking as an alternative to IHSI is not
recommended.
6-37
6.4
6.4.1
LONG-TERM SOLUTIONS AND SAFETY ASSESSMENT
Introduction
The proposed long-term remedy procedures for pipe cracking in BWRs can be
classified into three groups:
alternative materials, residual stress improve-
ment remedies, and alternative water chemistries.
The alternative materials
that have been considered include both new piping materials such as nuclear
grade Types 304, 316 and 347, Types 304L and 316L, Types 347, CF3, and Types
304LN and 316LN stainless steels and remedies such as corrosion-resistant
cladding which introduce only a localized region of resistant material in the
vicinity of the heat-affected zone.
Residual stress improvement remedies such
as Induction Heating Stress Improvement (IHSI), Last-Pass Heat Sink Welding

(LPHSW), and Heat Sink Welding (HSW) are intended to produce compressive
residual stresses (or at least lower tensile stresses) on the inner surface of
a weldment.
The most promising alternative water chemistry utilizes hydrogen
additions to feedwater to reduce the dissolved oxygen produced by radiolysis

in the core.
Reactor experiments in both the U.S. and Sweden have indicated
that lower oxygen levels are achieved, and laboratory studies indicate that
the lower oxygen levels can reduce or even eliminate susceptibility to IGSCC.
6-38
^ ^
6.4.2
Alternative Materials
The alternative materials which have been considered for BWR piping
systems include low-carbon stainless steels, stabilized stainless steels,
higher chromium steels such as the Nitronic alloys, and cast stainless steels
with duplex austenitic-ferritic structures.
For a variety of reasons [ease of
fabrication, familiarity, ASME Code acceptability, ease of inspection, etc.

(6.23)] most attention has focused on low-carbon stainless steels.
The regulatory position on the actual procedures for the replacement of

piping is outlined in Appendix C-2.
Low-carbon stainless steels such as Types 304L and 316L have long been
widely used in a variety of industrial applications to avoid sensitization and
the associated cracking problems.
However, because of their relatively low
carbon levels (less than 0.03%), they have lower design allowable stresses
than conventional Types 304 and 316 stainless steels and hence cannot in
general be used as a one-to-one replacement for the higher carbon stainless
steels.
The nuclear grade stainless steels have even lower maximum allowable
carbon levels (less than 0.02%), but nitrogen levels are also controlled
within specified levels (0.06% to 0.10% N) to ensure that materials have
strength levels that meet the ASME Code requirements for conventional
stainless steels, and thus they can be used as a one-to-one replacement for
conventional stainless steels without reanalysis or redesign of the piping
^ ^
systems (at least for piping systems which were designed to meet ASME Code
6-39
requirements).
It should be noted that the nitrogen levels for the nuclear
grade steels are within the Code specifications for Types 304 and 316
stainless steels (N less than 0.1%).
'
The LN-grade steels are nitrogen-
strengthened steels somewhat similar to the nuclear grade steels, although

the allowable carbon and nitrogen limits are slightly different (less than
0.035% C and 0.06% to 0.16% N).
They have recently been recognized by the
ASME Code as acceptable materials for primary pressure boundaries.
Both Types 304 and 316 nuclear grade materials have performed well in a
wide range of laboratory tests including full-scale pipe tests, slow strain
rate tests, simulated crevice corrosion tests, and cyclic crack growth tests
carried out in the U.S. and Japan (6.24, 6.25).
Some of the data suggest that
low levels of molybdenum play an important role in the stress-corrosion

cracking resistance of these materials (6.27).
The higher nickel and the
molybdenum additions in Type 316 also make it much less susceptible to

martensite formation during cold work than Type 304 (6.28).
Thus, use of Type
316 nuclear grade material seems to provide additional protection against

IGSCC produced in sensitized regions and cracking associated with cold work,
and it appears to be the material which will be used by most utilities in
replacement programs and for new piping.
The major laboratory testing program to qualify the nuclear grade stainless steels has been the pipe testing program carried out by General Electric
under EPRI and BWR Owners Group (BWROG) sponsorship.
The pipe test specimens
were fabricated by butt welding eleven 100-mm-long segments of 4-in. Schedule

80 pipe together to form a 1.1-m-long test section. Transition pieces and end
caps were welded to the test section.
During the test the specimens were
cyclically loaded while high purity water at 550oF (288oc) containing

6-40
approximately 6 ppm dissolved oxygen was circulated through the test section.
^^kThe conductivity of the water was maintained below 1 micro Siemens per centimeter, but detailed records of the conductivity levels during the tests and
the nature of the impurities are not available.
The load in these tests was increased linearly over a 5-min period until
the stress reached a value of approximately 34 ksi, held at that level for
75 min, reduced linearly to a value of approximately 4 ksi over a period of
0.5 min, and held there for 5 min; the cycle was then repeated.
For conven-
tional stainless steels with 0.06-0.07 C contents, these tests produced

throughwall intergranular cracks in approximately 70 cycles (6.29).
No inter-
granular failures were observed in corresponding tests on 9 heats of Type

316NG and 8 heats of Type 304NG stainless steels in tests ranging from 20006000 cycles (6.30).
One Type 316NG specimen did fail after 2885 cycles. This
crack was reported to be outside the weld heat-affected zone near the weld
counterbore and primarily transgranular in nature (6.30).
A similar failure
was observed in a Type 316L (0.024 C) specimen after 1256 cycles. Because of
the character of the failures, these specimens were considered test anomalies
by General Electric.
Even including these particular cases, the materials
show much superior performance in the pipe tests.
Extensive constant load, U-bend, and slow strain rate tests on five heats
of Type 316NG in high purity, high temperature, oxygenated water have also
been performed by the Japanese (6.31).
The specimens in these tests were
given a wide variety of heat treatments and surface treatments, including

grinding.
No stress-corrosion cracking was observed in these tests. Creviced
bent beam tests (with graphite wool in the crevice) and double-U-bend tests
^^k
(with a Teflon film between the two U-bends to form a tight crevice) were
6-41
carried out to evaluate the susceptibility of these materials to crevice

corrosion.
Some intergranular cracking was observed in specimens subjected to
furnace heat treatments [1100-1400F (600-750C)/30 h ] , but none was observed

in welded specimens, even with an additional low temperature heat treatment
[932F (500C)/24 h ] .
Hot tear tests, hot ductility tests, and varestraint tests (i.e., bend
tests during welding) were also carried out to evaluate the weldability of
Type 316NG stainless steel (6.31).
These tests indicated that the weldability
of the nuclear grade material was equivalent or in some cases superior to that
of conventional Type 316 stainless steels.
The lack of any adverse effect of
nitrogen additions on weldability has been noted by other investigators

(6.32).
Recently, shallow (10-20 mil), intergranular cracking has been
observed near seam welds in rolled and welded Type 316NG piping.
appears to be confined to one fabricator.
The problem
The current hypothesis proposed by
EPRI (6.33) is that the problem is due to copper contamination produced by the
use of air-arc cutting with copper-coated carbon electrodes during fabrication
of the pipe.
During the seam welding process the copper oxide left on the
surface during the cutting is reduced and the molten copper metal produces
some grain boundary cracking.
The explanation seems plausible and consistent
with the absence of the problem at another fabricator using mechanical

cutting, but additional study is needed to resolve the problem.
Cracking of Type 316NG during hot forming has been observed in some
cases.
This has been attributed to nonuniform heating and improper
temperatures during the forming operation.
Nitrogen-strengthened stainless
steels have been widely used in Europe for many years and seem to have good
6-42
fabrication and welding properties. Japanese tests and experience are

^ônsistent with this. However, care should be taken to ensure that the
materials (in terms of impurity content etc.) and fabrication procedures used
in this country are comparable to those used in Europe and Japan so that this
experience is truly relevant.
The qualification tests performed by General Electric also included

corrosion fatigue tests. Cantilever beam fatigue initiation tests were performed on welded specimens in high purity water with approximately 6 ppm
dissolved oxygen at 550 F (288 C) using fully reversed loading at
approximately 4 cph. One heat of Type 316NG (7 specimens), one heat of
conventional Type 316 (7 specimens), and two heats of conventional Type 304
(6 specimens) stainless steels were tested.
failed intergranularly:
transgranularly.
The Type 304 stainless steels
The Type 316NG and Type 316 materials failed
At 1.3% strain the life of the nuclear grade material was
approximately four times that of the conventional materials, and at 0.75%

cyclic strain it was approximately twice the life of the conventional Type 316
material and approximately four times the life of conventional Type 304
material.
Cyclic crack growth rate tests were carried out on one specimen
each of Types 316NG, 347, and 304 stainless steels using a trapezoidal
waveshape at a cyclic frequency of 4.4 cph and a load ratio R = 0.
morphologies observed were transgranular in all cases.
The crack
The crack growth rates
for the Types 316NG and 347 materials were approximately 40% less than for the
Type 304 stainless steel.
6-43
Intergranular crack growth has been observed in Type 316NG stainless

steel in laboratory tests on precracked specimens (6.34).
In tests on bolt-
loaded fracture mechanics-type specimens with fatigue precracks, intergranular

crack propagation was observed in a variety of alloys including Types 347,
316L, 316NG, 304L, XM-19 (Nitronic 50), and 304 stainless steels. The tests
were carried out in high purity water (i.e, conductivity less than 1.5 micro
Siemens per centimeter, but no deliberate additions of impurities and no
detailed history of the actual water chemistry) with approximately 6 ppm
dissolved oxygen at 288 C.
The nominal initial stress intensity factors (K)
were quite high in these tests, ranging from 40-70 ksi-in.1/2.
The final
measured stress intensities were lower (20-55 ksi-in.1/2) presumably because

of the load relaxation produced by crack growth and time-dependent plasticity
at the crack tips of the specimens.
Microstructural, sensitization (EPR), and
slow strain rate tests indicated no sensitization (carbide precipitation) of

the nuclear grade steels. However, fully intergranular cracking was observed
on the specimen fracture surfaces. General Electric has postulated that the
presence of the intergranular cracking is attributable to the crevice cracktip chemistry as well as to the extremely slow strain rate in the crack
propagation process.
The higher nitrogen levels in both the nuclear grade and the LN materials
raise the possibility that these materials may be more susceptible to transgranular cracking, since deleterious effects of nitrogen on the transgranular
chloride cracking of austenitic stainless steels (primarily in MgCl2 environments) were observed previously (6.37), although for materials close to
the composition of the nuclear grade stainless steels the available results
are somewhat ambiguous and little work is available in BWR-type environments.
6-44
The behavior of Type 316NG material in environments containing chlorides

^^ând sulfates has been considered in pipe tests by General Electric and in slow
strain rate tests at Argonne National Laboratory (6.35).
The water chemis-
tries in the pipe tests generally simulated high impurity levels that are well
beyond the normal range of BWR water chemistries and are characteristic of
events like the Millstone chloride intrusion or the 1974 resin intrusion at
Duane Arnold.
The limited number and duration of the pipe tests make it
difficult to draw conclusions from this work.
In all cases the nuclear grade
materials performed as well as the reference Type 304 materials, but in most
cases the tests were ended without failures in either the nuclear grade or the
conventional materials so that it is difficult to assess the relative performance of the two types of materials.
The tests at Argonne have focused on water chemistries with much lower
levels of impurities that would be more typical of usual reactor operation.
In no case has intergranular cracking been observed in nuclear grade materials
even after fairly severe furnace heat treatments [122 F (650 C)/24 h) or
in weldments given an additional low temperature heat treatment [932 F
(500 C)/24 h]. However, transgranular stress-corrosion cracking has been
observed in slow strain rate tests (6.36) in water chemistries with 0.2 ppm
dissolved oxygen and with sulfate and chloride impurity levels consistent with
the BWR water chemistry limits outlined in Reg. Guide 1.56 (approximately
0.1 ppm sulfate or chloride added as acid).
For the same weight concentra-
tions sulfate additions appear to be somewhat more detrimental than chloride

additions.
6-45
In environments in which the nuclear grade materials exhibit transgranular cracking, sensitized Types 304 and 316 stainless steels exhibit intergranu
lar cracking.
Depending on the environmental conditions the transgranular
crack growth rates in the nuclear grade steels are roughly factors of 3-10
lower than the intergranular crack growth rates in the conventional steels.
The transgranular cracking appears to be associated with the impurity
additions and is not observed in tests in high purity water. However, the
minimum impurity levels required to produce cracking are not known at this
time.
The relative susceptibility of unsensitized conventional austenitic
steels (which seem to have exhibited satisfactory resistance to transgranular

cracking in-reactor) compared to the nuclear grade steels is also unknown at
this time. It is also possible that the transgranular cracking observed in
the slow strain rate tests requires unrealistically large plastic strains for
initiation.
This too is an area which requires additional testing.
It is well known (6.38, 6.39) that long-term exposure of Type 316

stainless steel at elevated temperatures can cause decomposition of the
austenitic matrix into various intermetallic phases (e.g. , sigma, laves, chi)
that can have a very detrimental effect on the mechanical properties. The
available data (6.39) on conventional Type 316 stainless steel suggest that
this is unlikely to occur at BWR operating temperatures. Lower carbon levels
seem to promote the formation of the intermetallic phases (6.38).
If, as
seems likely, this is due to the role of carbon as an austenite stabilizer,

then the higher nitrogen levels of the nuclear grade may be beneficial, since
nitrogen is also a strong austenite stabilizer [e.g., nitrogen additions
6-46
appear to suppress the formation of martensite during cold working of 18/8

^^tainless steels (6.32)].
Thus, it seems unlikely that long-term service at
BWR operating temperatures will cause significant degradation of the

mechanical properties of the Type 316NG material.
Overall the nuclear grade materials appear to be a much better choice for
BWR piping systems than the conventional austenitic stainless steels, but
careful control and monitoring of water chemistry may be necessary to ensure
total protection from both intergranular and transgranular stress-corrosion
cracking.
The low-carbon austenitic stainless steels have performed very
satisfactorily in the EPRI/GE pipe tests and have performed well in-reactor in
some repair situations.
Similar performance is expected from the LN grades,
since nitrogen additions do not appear to affect susceptibility to sensitization adversely (6.31, 6.32).
However, as noted for the nuclear grade mate-
rials, nitrogen additions have been observed to increase susceptibility to

transgranular cracking in some systems, and recent laboratory studies have
shown that small levels of impurities greatly increase the relative susceptibility of very lightly sensitized materials (6.35).
This suggests that the
performance of the L grade materials (with C greater than 0.02%) may not be
satisfactory in an actual reactor environment as the laboratory tests in high
purity environments would indicate, and the lower carbon nuclear grade materials offer an additional margin against cracking that may be significant in
actual reactor environments.
Similarly, it seems prudent to restrict nitrogen
additions to levels as low as possible while providing adequate mechanical

properties.
6-47
Corrosion-resistant cladding in which the region in the vicinity of the

weld is clad with duplex Type 308L weld metal with ferrite levels greater than
or equal to 8% also appears to be a reasonably satisfactory remedy (6.40).
In
many cases it is not possible to solution-heat-treat piping after the cladding

is applied so that a sensitized region remains on the inner surface of the
weldment at the end of clad region. However, this sensitized region is
limited in depth into the pipe wall. Hence, although the procedure is
unlikely to produce complete immunity to cracking, it does significantly
reduce the amount of sensitized material exposed to the environment.
It also
complicates the ultrasonic inspection of the weldment, and good baseline data
are essential for satisfactory resolution of the ultrasonic reflections
associated with the weld clad layer from possible cracking indications.
Because of the inspection difficulties and the possible existence of sensitized regions on the inner surface, use of an alternative piping material is
preferable.
6.4.3
Residual Stress Improvement Remedies
A number of residual stress improvement remedies have been proposed

including Induction Heating Stress Improvement (IHSI), Last Pass Heat Sink
Welding (LPHSW), and Heat Sink Welding (HSW).
These procedures are intended
to produce compressive residual stresses (or at least lower tensile stresses)

in the inner surface of a weldment.
(HSW may also provide an additional
benefit by reducing the degree of sensitization near the weldment, although

since the weldment is "seeded" with carbides, low temperature sensitization
can significantly reduce the benefit associated with the initially lower
6-48
degree of sensitization.)
For operating plants, HSW is typically useful only
^ ^ P i n repair situations to reduce the susceptibility to cracking in the welds

joining the new replacement piping to the original susceptible material.
and LPHSW can be applied to weldments in situ.
IHSI
IHSI has been widely used on
existing plants in Japan.
It has been demonstrated that both IHSI and LPHSW produce highly compressive residual stresses on the inner surface of butt piping weldments (6.41 6.43).
Analytical studies indicate, however, that under applied loads of the
magnitude expected in-reactor the total stresses on the inner surface may be
tensile (6.44) but that in most cases the total stresses will be much less
tensile than in a conventional weldment.
Because the stresses may not be
compressive, IHSI may not completely eliminate the possibility of IGSCC, but
the lower stresses should mitigate the problem.
Experimental verification that the compressive residual stresses produced

by IHSI are actually effective in inhibiting stress-corrosion cracking has
been provided by pipe tests on 4-in. welded pipes carried out by General
Electric under EPRI/BWROG sponsorship.
The pipe specimens, test environment,
and loading histories were similar to those described in Section 6.4.2 for the
alternative materials. To prevent relaxation of the residual stresses induced
by the IHSI, somewhat lower applied stresses (approximately 30 ksi) were used
in these tests.
These stresses are still substantially higher than would be
found in most reactor operating situations, and would be expected to reduce

significantly the compressive residual stresses induced by the IHSI process,
and hence the expected benefit.
However, pipes without preexisting flaws that
were treated by IHSI still showed significantly longer lives than comparison
6-49
pipes without IHSI (6.41).
At applied stresses of 28.1 and 30.6 ksi, IHSI-
treated pipes failed at lives ranging from 573 to 991 cycles. At higher
stresses IHSI-treated pipes showed no improvement, which is consistent with
the idea that the improvement is due to the favorable residual stresses.
Tests on pipes with initial defects showed no significant benefit of the

IHSI process even for relatively shallow (less than 10% throughwall) precracks (6.44).
General Electric attributes this to the relatively high loads
in the pipe tests, and they argue that with actual plant loads a significant
benefit is likely for the shallow cracks that might escape detection.
Their
conclusions are consistent with available finite element results. Detailed

analytical calculations (6.45, 6.52) suggest that at least for shallow cracks
(less than 10% throughwall) the stresses ahead of the crack tips are compressive under most applied loads. Even for 40% throughwall cracks, the
stresses ahead of the crack are compressive under typical applied loads,
although they become tensile for applied stresses approximately greater than
10 ksi.
The Japanese have treated precracked weldments with IHSI and then
inmiersed them in boiling MgCl2 solution (6.42).
No additional crack
extension was observed, which suggests that the stresses ahead of the crack
tip were compressive or at least very low. However, this test does not
address the possibility of the redistribution of stresses ahead of the crack
under additional applied loads.
The Japanese have applied IHSI to piping with UT indications (6.53) in an

operating reactor. After a year's operation, there was no apparent crack
growth. However, the indicated flaws were very small (less than an inch in
length).
6-50
Although the finite element results and the limited Japanese experience
^^Pre encouraging, uncertainities in UT sizing and the stress levels at particular joints require that the effectiveness of IHSI on flawed joints be
evaluated conservatively.
However, the evidence indicates that IHSI should be
effective for small flaws that have a relatively low probability of detection
and, in fact, it offers some improvement even for somewhat larger flaws.
For precracked pipes there is also the possibility that cracks may be
extended mechanically during the IHSI process (although the inner surface of
the weldment is in compression after completion of the process, it undergoes a
tensile loading during the process).
Metallographic studies on precracked
weldments subjected to IHSI by the Japanese (6.42) and General Electric

(6.41) suggest this does not occur at least for relatively shallow cracks
(less than 25% throughwall).
Finite element studies (6.52) indicate that
deeper cracks will not propagate mechanically during IHSI (the calculations
were done for 40% throughwall cracks).
Both analytical results (6.44) and the pipe test results (6.41) suggest
that under the primary loads expected in piping systems the stresses on the
inner surface of weldments treated by IHSI will be lower than for nontreated
welds, which should have a mitigating effect on IGSCC. However, since only
limited amounts of plastic strain are necessary to relieve the favorable
residual stresses introduced by IHSI, it is probably also prudent to consider
the possibility that secondary and peak stresses which produce small localized
plastic strains may also relax the stresses induced by IHSI.
A comparison of
plastic strain levels suggests that local plastic strains of the order of
6-51
those produced globally by primary stresses in the finite-element analyses can

be expected in regions with high secondary and peak stresses.
Thus it is
possible that the IHSI-induced stresses will be relaxed in these regions.

However, in such cases the high tensile residual stresses produced by welding
will also be relieved, and the situation at least from the standpoint of the
stresses at the weld, will be no worse after IHSI.
In addition to relaxation under applied loads, it is also possible that

the stresses induced by IHSI will be relaxed by creep deformation.
However,
Japanese studies (6.43) indicate that very little relaxation will occur at
the operating temperature of a BWR during the 40-year design life.
The
available limited measurements on residual stresses in weldments from operating reactors (6.46) also suggest that little relaxation occurs at these
temperatures.
The effect of the critical process parameters (coil length, heating rate,
peak temperature of the outer surface, etc.) on the residual stresses produced
by IHSI has been extensively studied both in the U.S. and Japan (6.41-6.43,
6.47).
The effectiveness of the process parameters currently recommended by
the process vendors in producing high compressive residual stresses on the

inner surface of piping has been convincingly demonstrated for pipe-to-pipe
butt weldments (6.41, 6.42).
ponent weldments.
Much less data are available for pipe-to-com-
Japanese data suggest that the process is less effective
in inducing compressive axial residual stresses in these cases (6.42).
How-
ever, these data were obtained using an IHSI procedure which differs in some
respects from the current procedures recommended by the IHSI vendors, and
6-52
recent experimental and analytical results suggests that significant

^^ômpressive stresses can be obtained, at least for some pipe-to-component
configurations (6.41, 6.47).
It should also be noted that the data do
indicate a substantial reduction in the circumferential stresses, which lead

to axial cracks, for the component configurations, and in no case do the data
indicate that the IHSI process produces a less favorable residual stress state
than the as-welded condition.
During the IHSI process, the treated weldment is subjected to a significant plastic strain.
Since plastic strain is known (6.48) to accelerate low
temperature sensitization, the possibility exists that IHSI may have a somewhat detrimental effect on the susceptibility of treated welds over a long
period of time from a metallurgical viewpoint. However, the chief concerns in
assessing the effectiveness of IHSI are the possibility that in some locations
and under some upset conditions the favorable residual stress state produced
by IHSI may shake down and the possibility that preexisting cracks may be too
deep for IHSI to produce significant benefits.
Despite these possible draw-
backs, it appears from the available analyses and experimental data that for
most piping locations IHSI can significantly reduce susceptibility to IGSCC.
The Task Group also recommends the use of IHSI on weldments joining a
resistant piping material such as Type 316NG with pumps, valves, vessel
nozzels, etc. of a nonresistant material.
6-53
Although both IHSI and LPHSW share the same fundamental basis as remedy
procedures and experimental measurements (6.46, 6.49, 6.50) and analytical
results (6.51) indicate that LPHSW and HSW are effective in producing compressive residual stresses on the inner surface of piping butt weldments,
the existing data base of experimental results and finite-element analyses is
much greater for IHSI than for LPHSW or HSW and more work is needed to demonstrate the applicability of LPHSW and HSW to other weldment geometries.
In
addition, both LPHSW and HSW appear to be more strongly dependent on weldment
geometry and weld process parameters than IHSI.
consider LPHSW and HSW in a generic fashion.
Thus it is difficult to
The use of HSW in making repairs
and in fabrication of piping assemblies appears to be good engineering practice, and it has been widely used in Japan especially for partial pipe replacements (6.53), but because of the sensitivity to actual practice both
LPHSW and HSW can be considered only partially effective remedies for IGSCC in
conventional Type 304 stainless steel piping systems.
6.4.4
Evaluation of Water Chemistry Improvement
Since environment is one of the three synergistic contributors to IGSCC,

the possibility of modifying the BWR water chemistry to prevent or reduce the
chances of future occurrences of IGSCC has received considerable attention
over the last several years (6.54).
Modifications to the environment that
have been proposed fall into two general classes, which should be considered
both independently and in conjunction with one another: first, controls over
the ingress of ionic species that can accelerate both initiation and propagation of cracks; and second, control of the electrochemical potential of
the stainless steel by reducing the concentration of oxygen in the coolant
through injection of hydrogen.
6-54
6.4.4.1 Current Regulatory Positions
In 1973 and 1974, the staff issued R.G. 1.56, "Maintenance of Water
Purity in Boiling Water Reactors," and in 1977, issued Revision 1 to this
Guide for comment.
In each of these versions of the Guide, it was accepted
that the oxygen concentration in BWRs was fixed by the radiolysis of water in
the reactor core and the boiling process.
Controls were placed on the ingress
of impurities in the water to reduce the occurrence of SCC of sensitized

stainless steel. The Guide specifies maximum conductivities of the coolant
for long-term operation and gives absolute limits for short-term operation; if
the latter are exceeded, the reactor must be shut down.
These positions are
generally consistent with the Technical Specification requirements for most

operating BWRs.
No mention is made in either version of the Guide on
techniques for controlling oxygen.
6.4.4.2 Role of Impurities on Initiation and Propagation of IGSCC
Recent research at General Electric (6.55) and at Argonne National

Laboratory (6.56) has shown that, within the water purity limitations spelled
out in Regulatory Guide 1.56, there are significant effects of dissolved
anionic impurities on both the initiation and the propagation of IGSCC.
The three most common anions that can concentrate in films and produce
electrolytes in the crack are chloride, which frequently enters a BWR coolant
through condenser leakage, sulfate or sulfite, which can enter the coolant
from condenser leakage or resin breakup, and carbonate, which can enter the
coolant from air or water leakage into the condenser.
The material itself
may also provide a sufficient source of impurities, as noted in Section 2.
6-55
Further, crack-tip chemistry may differ significantly from bulk chemistry,

through corrosion processes. Breakup of demineralizer resins can cause injection of the sulfonate cation resin beads, which decompose in the heat and
radiation in the reactor to sulfites or sulfates, depending upon the oxidizing
potential.
If resin beds have been used for an extended period of time,
particularly on a coastal plant, their breakup may inadvertently introduce

chlorides that had been adsorbed in the resin beds, along with the sulfur
species, to the coolant.
All three of these species have been shown to have significant accelerating effects on both initiation and propagation of IGSCC (6.54-6.56).
Although tightening the allowable limits on these species (or on conductivity)
in the BWR coolants could have a beneficial effect on the occurrence of IGSCC,
it may not always be easy for a utility to operate with significantly tighter
controls over a long period of time. EPRI has initiated a three-year program
(6.57) which together with NRC-sponsored research (6.56) may provide a fundamental basis upon which improved water chemistry specifications can be
established.
It is known from laboratory tests that chloride ions, for
example, once introduced into the coolant, will adsorb in the films near the
growing cracks and are not easily removed following elimination of the
chloride in-leakage into coolant. How much chloride, sulfate, and even
carbonate are codeposited or adsorbed in the crud deposits is not known at
the present time, nor is their role on subsequent cracking well understood.
Savannah River (6.58), in a slightly different reactor some years ago,
identified high concentrations of chloride in the crud deposits (which in
those reactors were primarily hydrated aluminum oxides), despite maintenance
6-56
of less than 0.2 ppm CI" in the reactor coolant.
They did not establish,
^^mowever, that Cl~ was a major contributor to the IGSCC that occurred in the
Savannah River reactors. Consequently, it may not be possible to reduce
significantly the occurrences of IGSCC solely by reducing the steady-state
levels of chloride, sulfate, and carbonate in solution.
Further, because the
mechanisms of chemisorption and reactivation of the adsorbed species are not

well understood, it is not possible to establish at this time whether a short
period of high impurity levels in an otherwise impurity-free environment is
more or less detrimental to a system than a relatively low but continuous
source of impurities over a long period of time.
It is difficult to demonstrate a clear benefit from operation with low

impurity levels from current plant operating data.
Some plants using
Powdex demineralizers which are known to operate with their water chemistry
close to the allowable chloride or conductivity limits have fewer incidents of
IGSCC than other units with "better water chemistry" (i.e., lower average
conductivity levels) achieved using deep bed demineralizers (6.57, 6.59).
Differences in materials and stress patterns may override differences in water
chemistry between operating units. However, the laboratory data clearly
indicate that for a given material and loading conditions, additions of
sulfates, chlorides, or carbonates have a significant detrimental effect.
6-57
6.4.4.3
Role of Oxygen Control
Since IGSCC of sensitized stainless steel can occur in high purity water
containing oxygen, chemical controls for prevention of IGSCC in a BWR require
a reduction in the levels of both the ionic species and the oxygen in the
primary coolant.
If sufficiently low levels of oxygen are achieved, the
electrochemical potential of the stainless steel will be shifted to a region

at which IGSCC will neither initiate nor propagate.
The electrochemical
potentials required to prevent IGSCC depend strongly on the level of ionic

impurities present (6.56).
Current data suggest maintaining the electro-
chemical potential to less than -350 mV on the standard hydrogen scale

and the conductivity of the coolant to less than 0.2yS/cm (6.56).
In
Dresden-2, reducing the dissolved oxygen level to approximately 20 ppb

brings the potential into this range.
Large shifts in potential have been
achieved by the injection of hydrogen into the BWR coolant (6.54, 6.59).
Demonstrations of this technique have been made in Sweden at Oskarshamn and
Ringhals and in the U.S. at Dresden-2 (6.54, 6.59-6.61) in which a corrosion
test loop is deliberately tapped into one of the recirculation lines to
monitor the corrosion potential of sensitized stainless steel and the behavior
of stress-corrosion cracks in this environment.
In the hydrogen water chemistry approach, hydrogen is injected into the

feedwater lines to reduce the residual oxygen level to approximately 20 ppb.
When the coolant is pumped into the core of the reactor, this hydrogen
suppresses the radiolytic production of oxygen in much the same way that
hydrogen overpressure in a PWR prevents the accumulation of residual oxygen in
6-58
that coolant. We have many plant-operating years of experience with a PWR

^^Fprimary coolant that demonstrate conclusively that reducing the oxygen and
therefore the electrochemical potential of the stainless steel by this
technique (together with strict controls on halide ions) eliminates the
occurrence of IGSCC. However, the oxygen levels in PWR coolant systems are
significantly lower in most cases (typically less than 5 ppb), than are
practically achievable in U. S. BWRs (typically less than 20 ppb).
Both in
Sweden and at Dresden-2, short-term tests have shown that the SCC can be
stopped: after seven months' operation at Dresden-2 with known UT indications
in the piping, subsequent inspections showed no change in the ultrasonic
signals from this IGSCC, but whether this can be attributed to the hydrogen
water chemistry is not certain. Also, in the corrosion test loops, cracks
already initiated stopped propagating when the hydrogen was injected (6.59).
Long-term demonstration tests are planned under EPRI sponsorship at Dresden
(and possibly at another BWR yet to be selected) to determine the overall
effects on the plant system of hydrogen water chemistry (6.57).
Already it is
known that the same levels of hydrogen concentration to the feedwater have a
different effect on the corrosion potential at Dresden than they do at
Ringhals (6.54, 6.59).
An argument has been made that these differences may
be traced back to the fact that the Swedish BWRs do not have jet pumps
(6.59).
There are also possible differences due to other chemical impurities
in the water, since a number of ionic species in the coolant are known to
affect the ability of hydrogen at a given concentration to suppress production
of radiolytic oxygen (6.62).
Plant-specific water chemistry should therefore
be strongly considered in the selection of a second plant for a demonstration

test, and for selection of the hydrogen levels needed for mitigating IGSCC.
6-59
The principal known side effect of the hydrogen chemistry is that it

produces increased levels of hydrogen and nitrogen-16 in the steam. While
nitrogen-16 is present in all reactor primary coolants, from n,p reactions on
the oxygen-16 in the water, the normal excess of oxygen in the coolant reacts
with it to form an oxide of nitrogen or nitric acid, which is soluble and
relatively nonvolatile. With the hydrogen chemistry, however, a reducing
potential exists, which causes a portion of this nitrogen to be reduced to
volatile ammonia gas, which enters the steam. Radiation levels in the turbine
area at Dresden-2 have increased routinely by a factor of 5 when the hydrogen
chemistry is operating.
In fact, GE suggests that monitoring the radiation
levels in the turbine provides a good indication of the effectiveness of the

hydrogen treatment (6.59).
For those plants with currently unshielded
turbines, the introduction of hydrogen water chemistry may require turbine

shielding. Access to the turbine (or unshielded areas) may require that the
hydrogen be turned off. The stress-corrosion tests in the Dresden loop have
shown that hydrogen interruptions lasting up to six hours do not appear to
reinitiate IGSCC. However, during hydrogen interruptions of 18 hours or more,
some IGSCC propagation can occur, but at a rate no greater than would occur in
the absence of the hydrogen (6.59).
Therefore, even if the hydrogen is
operable only 90% of the time, the extent of crack propagation should be
reduced by at least a factor of 10.
Other side effects of the hydrogen chemistry that need consideration are
effects on the fuel cladding, reactor internals, and other plant components.
Some years ago, PWRs switched from zircaloy-2- to zircaloy-4-clad fuel to
minimize the hydrogen uptake in the fuel. Whether or not the hydrogen water
6-60
chemistry approach would require a similar change with the zircaloy fuel
cladding alloys in BWRs remains to be demonstrated.
It is planned that fuel
from Dresden-2, which will have been exposed to three full operating cycles
under the hydrogen water chemistry, will be examined, under EPRI sponsorship
in a cooperative program between EPRI and the utility, to determine the
effects of hydrogen chemistry on the fuel cladding material (6.57).
These
effects are not anticipated to be large.
Reduction of the oxygen levels in the feedwater lines could lead to

increased corrosion of the carbon steel components, which can accelerate the
crud transport problem in BWRs.
With the generally leaky condensers that
exist in this country, there should probably be sufficient oxygen (20 ppb) in
the feedwater to continue the passivation of the carbon steel.
In Japan,
where tight condensers appear to be the rule, they have had to inject oxygen
into the demineralizer outlet to maintain sufficient levels for passivation of
carbon steel even in the absence of hydrogen water chemistry.
Increased
amounts of hydrogen in the steam have also caused a potential for hydrogen
fires, and oxygen injection into the steam lines to react with the hydrogen in
the recombiners has been proposed.
6.4.4.4
Startup Deaeration
The probability of IGSCC may be greatest during the startup of a BWR when
thermal stresses can cause creep of the pipes, relatively high oxygen levels
are normally present, and the temperature is in the range of maximum susceptibility to IGSCC (150 - 250OC) (6.63, 6.64).
6-61
Deaeration of the coolant
during startup can be achieved by pumping on the hot well. While this alone
cannot prevent IGSCC, startup deaeration in combination with the hydrogen
chemistry should go a long way to minimize the occurrence of this problem in
BWRs.
6.4.4.5 Foreign Experience
(6.65)
The activities in Sweden towards development and testing of improved

water chemistries have been discussed above. While most regulatory agencies
set limits on impurity levels, similar to those in Regulatory Guide 1.56,
actual operating levels of ionic species are generally lower in other
countries than in the U. S. because of greater attention to condenser design
and tighter controls on condenser leakage.
The Japanese believe their controls on water purity, combined with their
emphasis on stress-improvement and use of alternative materials, are adequate,
and they do not at present propose to use hydrogen injection to control oxygen
in BWRs.
6.4.4.6 Role of Other Impurity Intrusions; Decontamination
From the above discussions as well as from those made earlier in Section
2, it is apparent that any action which disturbs chemically the protective
layers of oxides on the surface of the stainless steel piping in the vicinity
of sensitized stainless steels can have an effect on IGSCC. Although research
has shown that relatively mild decontamination solutions, such as those used
6-62
in the CANDECON process, do not increase the propagation rates of IGSCC in

^^Fsensitized stainless steel in the laboratory (6.66), the effects of decontamination (removing the crud deposits and activating any ionic species that
might be adsorbed in them) on initiation of SCC following return to service of
the cleaned unit need to be much more thoroughly investigated.
Many utilities
faced with high radiation fields and frequent in-service inspections have felt
it prudent to reduce personnel exposures by decontaminating the piping before
the inspections. The long-term effects of this practice need to be investigated.
6.5
The conclusions and recommendations from this chapter are grouped below
under the major subheadings of the chapter.
6.5.1
Flaw Evaluation and Leak-Before-Break
Conclusions
The evaluation of flaws found in service and the validation of leakbefore-break concept requires reasonably accurate knowledge of the ability to
size the length and depth of SCC accurately, the applied and residual states
of stress, the relation of crack growth rate to stress and environment, and
the ultimate load-carrying capacity of the cracked pipe. A realistic evaluation of flaw evaluation procedures and the concept of leak-before-break
requires an integrated approach considering all the above factors.
6-63
The analysis of stress-corrosion crack growth in weldments using the

techniques of linear elastic mechanics with "typical" values for the
distribution of throughwall residual stresses and stress-corrosion
crack growth rates yields results which are consistent with field and
laboratory experience with welded pipes.
Thus it is a reasonable
approach to use for the analysis of the remaining life of flawed

weldments in-reactor and for detailed probabilistic models such as
the work by Lawrence Livermore National Laboratory.
The combination of residual stress distributions and crack growth

rates currently being used by NRR should yield predictions of flaw
growth in weldments that are conservative in the great majority of
cases.
However, because of the wide variability in residual stress
distributions near welds and IGSC crack growth rates, the current
approach used by NRR is not considered unduly conservative for current
BWR piping materials and environments.
The Task Group endorses the
guidelines for the evaluation of flaw growth in weldments given in

draft NUREG-0313, Rev. 2.
The Task Group also wishes to point out, however, that although conservative estimates of the throughwall growth of stress-corrosion
cracks can be made with a high degree of confidence given an initial
crack size, there is a high degree of uncertainty in the measurement
of the depth of existing cracks, as discussed in Section 4, and this
uncertainty must be adequately addressed in the development of acceptance criteria for flawed piping.
6-64
The use of loading conditions defined by the ASME Code in performing

flaw evaluations is acceptable, with the exception that SLA thermal
expansion stresses should be included in the flaw evaluation when
limit load is not achieved prior to significant crack extension.
Use
of these loading conditions is consistent with the approach which has

been accepted by the NRC for the design and operation of other reactor
components and systems.
Analyses for large surface cracks, conducted using the relatively low
ductile fracture resistance properties of stainless steel weld metal,
show that the margins against fracture may not be as great as those
believed to be associated with IWB-3640.
Nonetheless, a substantial
margin against fracture can be demonstrated for these flaws under

noirmal operating conditions.
Evaluation of throughwall cracks, using weld metal properties, shows

that a surface crack breaking through the pipe wall under bounding
accident loading conditions would have to be approximately 30% of the
circumference in length, for large-diameter pipes, to result in unstable fracture of the pipe.
However, leaks from cracks not large
enough to cause gross failure of the pipe may be undesirably large.


6-65
Specific assessments of piping integrity will have to recognize weld

I
conditions having lower toughness than the 304 stainless steel welds
used in the analyses described in this report.
Operating experience suggests that leak-before-break is the most

likely mode of failure for the vast majority of cracks occurring in
service.
This is a result of the asymmetry of the weld residual
stresses and applied loads and the variability in material properties.
Evaluations using conservative crack growth rate predictions
and net section collapse analyses, applicable to cracks in very high

toughness material, indicate that for the vast majority of possible
crack geometries there exists significant time to allow for detection
of leakage and implementation of corrective actions. Evaluation using
the fracture resistance properties of the weld material show substantial margins against failure, under normal and accident loading conditions for throughwall cracks which should be reliably detected by
leakage.
Recommendat ions
Based on the above discussions and conclusions, the Task Group has
developed the following recommendations.
6-66
The limitation on
makeup capacity.
Calculations

The recommendation for a limit on acceptable crack length is primarily

a result of the lack of confidence in ultrasonic depth sizing capabilities.
In this respect, the fracture mechanics calculations
presented in this report demonstrate that there are acceptable margins

against fracture for relatively long, deep surface cracks.
Demonstra-
tion of reliable sizing of the part-through flaw depth would most

likely allow rfelaxation of the limits on crack length.
6-67
Flaw evaluation criteria should limit the length of the cracks

accepted for continued operation without repair.
Additional fracture mechanics analyses, material properties characterization, and large-scale pipe tests should be performed to
further our understanding of the implications of stainless steel weld
and cast material fracture toughness properties in flawed pipe
evaluations.
Furthermore, the Task Group recommends active NRC
support of the ASME task group currently evaluating the concerns which
have been raised regarding IWB-3640.

are sensitive enough to provide additional margin against leak-beforebreak if more stringent requirements on surveillance intervals and
unidentified leakage are imposed (see Section 4 ) .
Therefore, the Task
Group recommends that the limits on unidentified leakage in BWRs be

decreased to 3 gpm and that the surveillance interval be decreased to
4 hours or less.
Short-Term Solutions

long-term operation with weld overlays, since errors in crack depth
6-68

For circum-
ferential cracks, weld overlay is considered an acceptable repair

Long-Term Solutions and Safety Assessment

and 304NG and 316NG in Japan, has been excellent.
used in the U. S. include 304L and 316L.
Other materials
All of these alloys are more
resistant to IGSCC than conventional Types 304 or 316 stainless

steel.
Based on U. S. data and prior use. Type 316NG stainless steel
offers an additional margin of resistance to IGSCC and utilities that

choose to replace pipe should be strongly encouraged to use it.
The use of additional mitigating procedures such as IHSI or hydrogen

water chemistry to provide additional margin against environmentally
assisted cracking is strongly recommended even after replacement with
Type 316NG stainless steel.
Utilities should be encouraged to adopt
an ALARA approach to coolant impurity levels.
6-69

It appears that greater care
must be exercised in the control of composition and fabrication

V
state.
be more effective when applied to weldments with no reported cracking.
IHSI is an acceptable mitigating procedure for weldments with no

reported cracking even in older plants. However, to be fully
effective a suitable hydrogen water chemistry also should be
implemented.
The use of IHSI on weldments with detectable cracking must be

considered on a case-by-case basis. However, for relatively short
cracks (approximately 20% of the circumference in length), since even
large errors in crack sizing or the prediction of flaw growth will
lead only to small leakage, the decision on whether an additional
repair is required can be determined by analysis. For longer cracks,
repair will probably be required.
6-70
All steps related to the weld prep, welding, and weld finishing and
of an appropriate QA manual. As noted previously, the weld should be
optimized for future ISI.

oxygen level. The current work on reduction of oxygen through
effective IGSCC countermeasure. However, ongoing work regarding

countermeasure.
Either reduction of ionic species or hydrogen water chemistry, used

alone, will have some beneficial effects. Used in combination, they
show great promise for mitigating IGSCC.
The effects of hydrogen water chemistry on the balance of the system

and on overall long-term plant operation need to be explored in
greater depth before they can be recommended without reservation for
adoption on a wide scale.
6-71
Regulatory Guide 1.56 needs to be modified to reflect these water

chemistry recommendations to provide tighter controls on water
conductivity and provisions for oxygen control as an acceptable
option.
The long-term effects of decontamination procedures on IGSCC need to

be investigated.

are implemented, there may be a need to modify the following documents when
NUREG-0313 Revision 2; Regulatory Guides 1.45
and 1.56; Standard Review Plans 5.23 and 5.4.8, Plant Technical Specifications
Governing Unidentified Leakage Rate Limits and Leakage Surveillance Requirements; 10 CFR Appendix A, Nos. 31, 32; Regulatory Guides 1.31, 1.44, 1.116;
SRP 3.6.1, 3.6.6, 5.2.3; ASME XI.
6.6
SUMMARY
Section 6 represents a complex interaction of cracking analyses,

decisions on short-term remedies and longer-term permanent remedies.
Table 6.3 is an attempt to collect the available options into one table,
covering the various conditions of plants not yet built, plants ready for
licensing and plants operating for several years, typical of these now
experiencing IGSCC where the options include no evidence of cracking cracks
believed to be shallow in the sense of ASME XI and deep cracks.
The purpose
of Table 6.3 is to permit the various options to be examined concurrently.
6-72
TABLE 6.3
R e c o m m e n d e d Measures for Controlling IGSCC in BWR Piping

Near Term
Replace
Residual
with
Similar
Alloy
Stress Improvement
IHSI, H S W
Hydrogen
Water
chemistry
3 1 6 NG
Piping,
Design Stage
NTOL or Recent Startup
No Cracks Detected
Only Limited Shallow

Cracks
X*
X**
Xthen
X Ifor
316NG)
BWR Plant Status
Long Term
Overlay
Weld*
Ult rasonic Exam.

New
Base- Accel.
line
UT
Norm.
UT
Enhanced Leak
Detection
(moisture tapes,
AE, etc.)
X
X
Operating >5(?) years
Deeper Cracks -- Few

or Many
X
maybe
'Limited to t w o cycles unless convincing evidence is presented

" A l s o suggested after replacement with 3 1 6 NG
# Prior to mitigation, accelerated UT; thereafter normal UT
X Accelerated UT limited to cracked welds
* With mitigation, accelerated UT limited to cracked welds
X (maybe)
X
X Prior to
Replacement
REFERENCES
NRC. November 7, 1983. Staff Requirements for Reinspection of BWR

Piping and Repair of Cracked Piping. SECY 83-267c.
W. Hazelton et al. 1984. "Technical Report on Material Selection and

Processing Guidelines for BWR Coolant Pressure Boundary Piping."
Draft Report, Rev. 2, NUREG-0313.
M. F. Kanninen, F. W. Brust, J. Ahmad, and I. S. Abou-Sayed. 1982.

"The Numerical Simulation of Crack Growth in Weld-Induced Residual
Stress Fields." In Residual Stress and Stress Relaxation.
E. Kula
and V. Weiss, eds. Plenum, New York.
J. D. Gilman and N. J. Olson. "Full Scale Testing of a Residual Stress

Modification to Control BWR Pipe Cracking." International Symposium on
Environmental Degradation of Materials in Nuclear Power SystemsWater
Reactors. Myrtle Beach, SC, August 22-25, 1983.
F. P. Ford. 1982. Mechanisms of Environmental Cracking in Systems

Peculiar to the Power Generation Industry.
EPRI-NP-2589, Electric
Power Research Institute, Palo Alto, CA.
W. J. Shack et al. 1983. Environmentally Assisted Cracking in Light

Water Reactors: Annual Report October 1981-September 1982. NUREG/CR3292, Argonne National Laboratory.
M. Indig and G. M. Gordon. "The Role of Water Purity on Stress

Corrosion Cracking." International Symposium on Environmental
Degradation of Materials in Nuclear Power SystemsWater Reactors.
Myrtle Beach, SC, August 22-25, 1983.
E. F. Rybicki et al. 1977. Residual Stresses at Girth-Butt Welds

in Pipes and Pressure Vessels.
NUREG-0376, Battelle Columbus
Laboratories.
E. F. Rybicki et al. 1982. Computational Residual Stress Analysis for

Induction Heating of Welded BWR Pipes.
EPRI NP-2662-LD, Electric Power
Research Institute, Palo Alto, CA.
10 R. Horn et al. 1982. The Growth and Stability of Stress Corrosion

Cracks in Large Diameter BWR Piping.
EPRI-NP-2472, Vol. 2, Electric
11 W. J. Shack, W. A. Ellingson, and L. Pahis. 1980. Measurement of
Residual Stresses in Type 304 Stainless Steel Piping Butt Weldments.
EPRI NP-1413, Electric Power Research Institute, Palo Alto, CA.
12 W. J. Shack. 1982. Measurement of Throughwall Residual Stresses in
Large-Diameter Type 304 Stainless Steel Piping Butt Weldments.
ANL-82-15, Argonne National Laboratory.
6-74
6.13
(
D. 0. Harris, E. Y. Lim, and D. D. Dedhia. 1981. "Probability of Pipe

Fracture in the Primary Coolant Loop of a PWR Plant," Vol. 5:
Probabilistic Fracture Mechanics Analysis.
UCID-18967,NUREG/CR-2189,
Lawrence Livermore Laboratory.
6.14
D. D. Dedhia and D. 0. Harris. 1982. Stress Intensity Factors for

Surface Cracks in Pipes: A Computer Code for Evaluation by Use of
Influence Functions.
EPRI NP-2425, Electric Power Research Institute,
Palo Alto, CA.
6.15
M. F. Kanninen et al. 1982. Instability Predictions for

Circumferential Cracked Type 304 Stainless Steel Pipes Under Dynamics
Loading. Volumes 1 and 2.
EPRI NP-2347, Electric Power Research
Institute, Palo Alto, CA.
6.16
M. F. Kanninen et al. 1976. Mechanical Fracture Predictions for

Sensitized Stainless Steel Piping with Circumferential Cracks.
6.17
S. Ranganath. 1982. "Prediction of Leak Rates." In The Growth and

Stability of Stress Corrosion Cracks in Large Diameter BWR Piping.
Vol. 2, EPRI NP-2472, Electric Power Research Institute, Palo Alto, CA.
6.18
C. F. Shih et al. 1981. Methodology for Plastic Fracture.

6.19
D. Kupperman, W. J. Shack, and T. Claytor. "Leak Rate Measurements and

Detection Systems." CSNI Leak-Before-Break Conference, Monterey, CA,
September 1-2, 1983.
6.20
R. P. Collier et al. 1982. Study of Critical Two-Phase Flow Through

Intergranular Stress Corrosion Cracks. Draft Report to the Electric
Power Research Institute, Battelle Columbus Laboratories.
6.21
R. E. Henry. 1970. "The Two-Phase Critical Discharge of Initially

Saturated or Sub-cooled Liquid." Nucl. Sci. Eng.
41. 336-342.
6.22
D. Abdollahian and B. Chexal. 1983. Calculation of Leak Rates Through

Cracks in Pipes and Tubes.
EPRI NP-3395, Electric Power Research
6.23
J. E. Alexander. 1980. "Alternate Alloy Qualification." In

Proceedings:
Seminar on Countermeasures for Pipe Cracking in
BWRs.
Vol. 3, EPRI WS-79-174, Electric Power Research Institute,
Palo Alto, CA.
6.24
Electric Power Research Institute. 1980. In Proceedings:

Seminar on
Countermeasures for Pipe Cracking in BWRs.
Vol. 3, EPRI WS-79-174,
Electric Power Research Institute, Palo Alto, CA.
6.25
J. E. Alexander et al. 1978-1980. Alternate Alloy for BWR Pipe

Applications.
NEDC-23750-1 to 8, General Electric Co.
6-75
6.26
J. E. Alexander et al. 1979. Alternate Alloy for BWR Pipe

Applications Fourth Semiannual Progress Report April-September 1979.
NEDC-23750-6, General Electric Co.
6.27
M. Kowaka et al. 1980. "Development of Low Carbon 316 Stainless

steel." In Proceedings:
Seminar on Countermeasures for Pipe Cracking
in BWRs.
Palo Alto, CA.
6.28
Tryggve Angel. 1954. "Formation of Martensite in Austenitic Stainless

Steels." J. Iron Steel Inst. 165-172.
6.29

Applications Third Annual Progress Report October 1978-March 1979.
6.30

Applications Second Annual Progress Report April 1978-October 1978.
6.31
T. Kobayashi et al. 1980. "Stress Corrosion Cracking Resistance

Mechanical Properties and Weldability of Nuclear Grade 316 Stainless
Steel." In Proceedings;
Seminar on Countermeasures for Pipe Cracking
in BWRs.
Palo Alto, CA.
6.32
J. J. Eckenrod and C. W. Kovach. 1979. "Effect of Nitrogen on the

Sensitization, Corrosion, and Mechanical Properties of 18 Cr - 18 Ni
Stainless Steels." In Properties of Austenitic Stainless Steels and
Their Weld Metals.
C. R. Brinkman and H. W. Garvin, eds. ASTM Special
Technical Publication 679.
6.33
J. Danko (EPRI) to W. J. Shack, (ANL).

Coimnunication.
6.34

Applications Sixth Semiannual Progress Report April-September 1980.
NEDC-23750-8.
6.35

Water Reactors; Annual Report October 1981-September 1982.
NUREG/CR3292 or ANL-83-27, Argonne National Laboratory.
6.36
W. J. Shack et al. Environmentally Assisted Cracking in Light Water

Reactors; Annual Report October 1982-September 1983.
NUREG/CR-3806,
in preparation.
R. M. Latanision and R. W. Staehle. 1969. "Stress Corrosion Cracking
of Iron-Nickel-Chromium Alloys." In Fundamental Aspects of Stress
Corrosion Cracking.
R. W. Staehle, A. J. Forty, and D. van Rooyen,
eds. NACE.
6.37
6-76
February 1984.
Personal
6.38
B. Weiss and R. Stickler. 1972. "Phase Instabilities During High

Temperature Exposure of 316 Austenitic Stainless Steel." Metall.
Trans. 3., 851-865.
6.39
J. K. Lai. 1983. "A Study of Precipitation in AISI Type 316 Stainless

Steel." Mat. Sci. Eng. 58. 195-209.
6.40
N. R. Hughes and A. J. Gianuzzi. 1979. Evaluation of Near-Term BWR

Piping Remedies.
Vol. 2, EPRI NP-1222, Electric Power Research
6.41
N. R. Hughes et al. 1981. Induction Heating Stress Improvement for

Stainless Steel Piping.
NEDE-25394, General Electric Co.
6.42
N. Ohki et al. 1980. "Actual Proof Test of Induction Heating Stress

Improvement for Large Diameter Pipe." In Proceedings;
Seminar on
Countermeasures for Pipe Cracking in BWRs.
Vol. 1, EPRI WS-79-174,
Electric Power Research Institute, Palo Alto, CA.
6.43
A. J. Gianuzzi Ed. 1981. Residual Stress Improvement by Means of

Induction Heating.
EPRI NP-81-4-LD, prepared by Ishikawajima-Harima
Heavy Industries Co.
6.44

Water Reactors; Annual Report October 1981-September 1982.
NUREG/CR3292 or ANL-83-27, Argonne National Laboratory.
6.45
E. F. Rybicki.
Laboratory.
6.46
W. J. Shack. 1982. Measurement of Throughwall Residual Stresses in

Large-Diameter Type 304 Stainless Steel Butt Weldments.
ANL-82-15,
Argonne National Laboratory.
6.47
E. F. Rybicki et al. 1982. Computational Residual Stresses in LargeDiameter Type 304 Stainless Steel Butt Weldments.
ANL-82-15, Argonne
National Laboratory.
6.48
M. J. Fox, ed. International Workshop on Low Temperature

Sensitization. January 21-22. 1982.
6.49
M. J. Fox. 1980. "An Overview of Intergranular Stress Corrosion

Cracking in BWRs." In Proceedings;
Seminar on Countermeasures for Pipe
Cracking in BWRs.
Vol. 1, EPRI WS-79-174, Electric Power Research
6.50
R. Sasaki et al. 1980. Mitigation of Inside Surface Residual Stress of

Type 304 Stainless Steel Pipe Welds by Inside Water Cooling Method."
Seminar on Countermeasures for Pipe Cracking in BWRs.
Vol. 1, EPRI WS79-174, Electric Power Research Institute, Palo Alto, CA.
May 1983.
Interim Report to Argonne National
6-77
6.51
E. F. Rybicki, P. M. McGuire, and R. B. Stonesifer. 1971. "The

Improvement of Residual Stresses in Girth-Butt Welded Pipes Through an
Internal Heat Sink." In Transactions Fifth International Conference on
Structural Mechanics in Reactor Technology Berlin 1979. Vol. F, NorthHolland, Amsterdam.
6.52
E. F. Rybicki. March 1984. Interim Report to Argonne National

Laboratory,
6.53
K. Suzuki. 1983. "TEPCO Replacement Experience." Presented at the

Second Seminar on Countermeasures for Pipe Cracking in BWRs. Electric
6.54
EPRI Seminar on Remedial Actions for IGSCC in BWRs, 1983.
6.55
P. L. Andresen and F. P. Ford. 1983. Paper presented at EPRI Seminar

on Remedial Actions for IGSCC in BWRs. Also, P. Andresen, private
discussions, September 1983.
6.56
W. Shack et al. 1984. "Evaluation of Stainless Steel Pipe Cracking;

Causes and Fixes," Proceedings of the Eleventh Water Reactor Safety
Research Information Meeting. NUREG/CP-0048, Vol. 4.
6.57
R. Jones and D. Cubicciotti, private discussions, December 1983.
6.58
R. Ondrejcin. 1969. A Mechanism for SCC of Stainless Steels in

Reactor Systems. DP-1089.
6.59
R. Cowan and M. Indig, private discussions, December 1983.
6.60
J. M. Aimer, E. Rowley, J. Schroger, and J. Thuot.

43, 323
6.61
J. Magdalinski and R. Ivars. 1982. Trans ANS 43. 323.
6.62
A. 0. Allen, private communication, 1979.
6.63
P. L. Andresen and F. P. Ford. 1984. Countermeasures for Pipe

Cracking in BWRs.
Vol. 1, paper 7. EPRI-WS-79-174, Electric Power
6.64
B. Vyas and H. S. Isaacs (BNL) 1979-80. Unpublished work.
6.65
Information presented at CSNI Meeting on the Regulatory Basis for

Actions on BWR Piping, Paris, February 7-9, 1984.
6.66
H. Takaku, H. Kusanagi, H. Kirano, T. Tomizawa, and K. Miyamaru. 1982.

Decontamination of Nuclear Facilities. Vol. 1, pp. 1-65 to 1-78,
American Nuclear Society.
6-78
1982. Trans. ANS
7.0
7.1
IGSCC RISK STUDIES
INTRODUCTION
Earlier studies on IGSCC in BWR piping systems approached the impacts

primarily on a deterministic basis which defines a spectrvim of design basis
accidents (DBA) considered credible.
The nuclear power plant must be capable
of accommodating these accidents within the envelope of acceptable consequences to the public.
With sufficiently reliable engineered safety features,
the risk of an unmitigated accident will be an acceptably small contribution

to the residual risk, which lies outside the envelope defined by the design
basis.
Probabilistic Risk Assessments (PRA) and risk-related methodologies
are being increasingly utilized to quantify the residual risk to the public.
PRA provides a systematic method to examine how this residual risk is
structured and for identifying the important contributors.
Several plant PRAs have been completed since the pioneer Reactor Safety
Study (RSS) in late 1975.
Risk methodologies have been utilized in a number
of useful ways in the regulatory process.
These risk studies are intended to
evaluate the LOCA risk impact of IGSCC as a disciplined approach to relate

field and inservice inspection experience to risk impact, and as one input to
the value-impact studies of Section 8.
7-1
A rigorous understanding of the impacts of IGSCC on LOCA frequency is not

known.
IGSCC is only one cause of pipe failure and the proportionality
between these causes can only be inferred from field experience at this time.
Lawrence Livermore National Laboratory (LLNL) is currently conducting a
study for the NRC which should improve our understanding of how IGSCC affects
pipe failure frequency, including its relationship to other causes. In the
absence of such an understanding, it was decided to initiate these studies by
testing the sensitivity of representative plant PRAs to the LOCA contribution.
This led to a confirmation of similar studies performed by the Electric Power
Research Institute (EPRI) as well as NRC in-house investigations. With the
understanding of the sensitivity of risk to LOCA contribution, an attempt was
made to assess whether the evidence - analytical and/or statistical (field
data) - implies an IGSCC contribution to risk that is not accounted for in
current analyses.
7.2
LOSS-OF-COOLANT-ACCIDENT (LOCA) RISK
The risk methodology for LOCA calculations was defined in the RSS and has
remained relatively unchanged in subsequent PRAs.
Risk is defined as the
product of the probability of an event and the consequences of the event. It

has become common practice in sensitivity studies to assume that risk is
proportional to the core damage frequency, rather than to trace the radioactivity release pathways for each differential change to the core damage
frequency.
This approach has been taken in this study.
In Appendix D.2, a simple sensitivity model based on the RSS methods is

developed.
Review of pertinent BWR PRAs, as shown in Table 7.1, demonstrates
that LOCA contributions to the core damage frequency are not considered
7-2
Table 7.1
Sensitivity of Core Melt Frequency for BWR PRAs

for a Postulated Increase (lOx) in LOCA Initiator Frequency
Plant A*
Sponsor
Total
Core Melt
Frequency**
(per Rx yr)
Frequency of Core
Melt Due to LOCA
Initiated Sequence
(per Rx yr)
Increase in the Frequency

of Core Melt Due to
Postulated Increase (lOx)
in LOCA Initiator Frequency
Limerick (Rev 5)
PE Co.
1.5 E-5
1.3 E-7
1.2 E-6
Shoreham
LILCO
5.5 E-5***
1.9 E-6
1.7 E-5
Millstone 1
IREP
3.0 E-4
3.0 E-6
2.7 E-5
Browns Ferry
IREP
2.0 E-4
1.2 E-6
1.2 E-5
2.9 E-5
1.2 E-6
1.1 E-5
Peach Bottom (RSS) NRC
*Big Rock Point was considered but deemed atypical, since it is a BWR-1 and has a less
comprehensive ECCS. GESSAR (Grand Gulf) was also considered, but was judged
inappropriate for this list because it is a BWR-6, among other reasons.
**Total core melt frequencies reported in the probabilistic analyses used for this study
resulted from internal initiators only. Analyses including external initiators would
yield higher core melt frequencies and, thus, a lower percentage contribution from
LOCAs.
***The Shoreham value for "core vulnerable frequency" is believed to be representative of
core melt frequency and has been used here.
dominant contributors.
The five plants exhibit an average of 2% LOCA
contribution to the total core melt frequency. A sensitivity analysis shown

in Table 7.1, using the sensitivity model developed in Appendix D, indicates
that even an order of magnitude underestimation in the LOCA frequency would
not cause the LOCA risk to be a dominant contributor to core damage
frequency.
For example, for Millstone 1, the LOCA contribution would change
from a 1% core melt frequency contribution to a 10% contribution.
The field data impact on the frequencies of the three traditional BWR
LOCA sizes was examined to see if the precursor IGSCC has resulted in a statistically significant error in the LOCA frequencies used in PRAs and thereby
cause underpredicted risk contributions.
To date there have been no large
primary recirculation pipe failures in the U.S. commercial nuclear power

industry.
The implication of a significant underprediction of the large LOCA
frequencies is examined in Appendix D.3.
It is found that a significant
(factor of 10) increase in the large LOCA frequency, say, due to IGSCC or
other causes, would have been evidenced by several failures for the 273
reactor years of U. S. BWR operation.
Currently, 700+ reactor years of opera-
tion in the U. S. commercial nuclear power program have resulted in no large

pipe failures in the primary coolant system.
The likelihood of an unobserved
underprediction of large LOCA frequencies is even further reduced if the

foreign BWR experience is utilized.
The aggressive inservice inspection and
repair/replacement program obviously modifies the empirical failure frequencies to even lower values, but the extent of this effect is not clear.
7-4
In the case of intermediate and small LOCAs, the data base is examined by
a review of both current studies and the LER system to determine if statistically significant time trends would indicate an unanticipated IGSCC contribution to these frequencies. Although the quality of the data base as given
by the LER system is considerably below experimental standards, no significant
time trend increase could be attributed to IGSCC.
From these studies it is
concluded that it is highly unlikely that the appraisal of the public risk is
significantly (an order of magnitude) in error.
7.3
SYSTEMS RISK EVALUATIONS
The risk implications on a systems basis is considered in Appendix D.4.

Systems contribute in two ways to PRA considerations.
Those portions of
systems which are nonisolable from the primary coolant system comprise part of
the LOCA sensitive piping, and those systems which mitigate accidents
(engineered safety features) directly contribute to the reduction of risk.
Risk methods can now rank the importance of systems according to their contribution to reducing risk, such as shown in Table 7.2. Risk achievement ratio
is defined as the factor by which core melt frequency would increase if the
system were never operable.
This table indicates the importance of the Resid-
ual Heat Removal System (and thus the integrity of its piping) in risk reduction.
Long-term heat removal is required in all sequences.
7-5
Table 7.2
Risk Worth for Safety Systems* With Respect to Core Melt Frequency
System
Risk Achievement R^tio
Reactor Protection System
2.0 x 10
Residual Heat Removal System (RHR)
1500
Standby Service Water System
1500
Low Pressure Coolant Injection (LPCI) System
23
Automatic Depressurization System (ADS)
14
Low Pressure Core Spray (LPCS) System
7.6
High Pressure Core Spray (HPCS) System
4.4
Reactor Core Isolation Cooling (RCIC) System
2.2
*Based on Grand Gulf, a BWR-6, which has HPCS rather than HPCI as in
BWR-4 as well as some minor differences, but safety systems/functions are similar. Some systems not directly pertinent to LOCA
concerns have been omitted from this list.
7-6
1. Conclusions
Intergranular stress-corrosion cracking (IGSCC) has caused no significant loss-of-coolant events (LOCE) or loss-of-coolant accidents (LOCA)
in BWRs even though its presence in Type 304 stainless steel coolant
piping has been found in many operating plants. This may be due to an
aggressive inservice inspection and repair program and/or a tendency
to leak-before-break (with effective leak detection).
The LOCA con-
tribution to the core damage frequency as calculated by probabilistic

risk assessment (PRA) studies for BWRs is a minor contribution (generally less than 10%) for all LOCA sizes. Because the LOCA contribution to the core damage frequency is small, it would require a
considerable error, at least an order of magnitude, from IGSCC before
LOCAs would be a dominant contributor to the core damage frequency.
IGSCC is but one component of the LOCA occurrence rate. Design and
construction errors, maintenance errors, cyclic and thermal fatigue,
etc. all contribute to the LOCA probability.
It is estimated that
IGSCC may be 30% or more of the LOCA frequency in BWRs as deduced from
experience, the total industrial data base, and engineering judgment.
Therefore, an error in the IGSCC LOCA contribution of 20-30 would be
required for an order of magnitude error in the LOCA probability.
7-7
The pipe failure data base and available studies were reviewed for
evidence of IGSCC impact.
To a considerable extent, uncertainties
have always existed in determining piping failure rates. Largely,

this uncertainty is intrinsic in attempts to measure low-frequency
events from experience, and it makes the effect of trends difficult
to analyze.
It was statistically reasoned that an error of an order
of magnitude in the large pipe failure frequencies (presumably from

IGSCC) would have produced, with one chance in four, at least one
failure during this interval.
Therefore, it was concluded that the
frequency for large pipe failure adequately accounts for the

contribution due to IGSCC.
For intermediate and small pipe failures, the incidence of reported

and identified corrosion events in the data base, although increasing
somewhat over time (owing to improved reporting and other causes), was
judged to be not statistically indicative of an error in the failure
frequencies.
The systems aspects of IGSCC risk were revised using Hatch-2 as a

model.
It was concluded that the plants should be required to be
designed for the most severe large LOCA without significant core
damage.
The ECCS system must be able to survive the most severe
single failure in the system and still perform its intended function(s).
Pipe failure is no more severe than other failures in
the system (valves, pumps, etc.) and is probably less severe.

Coupling this with the ECCS redundancy and conservative margin in
7-8
both design and calculational procedures, it is concluded that IGSCC

would cause little increase in risk from the standpoint of systems
important to safety.
Multiple failures were briefly considered.
IGSCC may be a common mode
failure in the recirculation piping and other systems.
It is con-
cluded that a large LOCA and ECCS system failure both caused by IGSCC
is adequately protected by the single-failure criteria and systems
redundancy.
Multiple failures causing a LOCA larger than a DBA are
likely to add little to the risk because it appears that multiple

failures are several orders of magnitude less in probability than a
large single failure.
Overall it is concluded that the presence of IGSCC in BWRs may reduce

safety margins believed to exist, but there is no apparent evidence to
conclude that IGSCC has increased public risk significantly.
7.4.2
Recommendations
Piping failure rates are difficult to determine because failures of high

quality piping occur infrequently and require a data base acquired over a
large number of reactor years, since normal experimental procedures are
impractical.
This difficulty is unlikely to be overcome now or in the fore-
seeable future. What results is an estitoated frequency with large uncertainties from aggregated events acquired from poor-quality documentation.
To
partition this frequency according to contributing phenomena in a meaningful

ay is very difficult.
#
7-9
It is recommended that a formal study and updating of the pipe failure

data base be undertaken.
Some statistics may be only meaningfully
derived or assessed analytically.

code PBIAISE ( P iping
R eliability
The NRC has developed a computer

A nalysis
1 ncluding
^ eismic
E vents) which is currently being modified to include a model for

IGSCC based on current understanding.
It is recommended that PEIAISE be used to a) investigate the impact of

IGSCC in primary piping according to pipe size, b) study the failure
frequencies of components due to IGSCC, fatigue, etc., and c)
calculate the conditional probability of multiple failures associated
with IGSCC to check the point value and to test the assumption of a
lognormal distribution.
7-10
^^8.0
8.1
INTERGRANULAR STRESS CORROSION CRACKING (IGSCC) VALUE-IMPACT ASSESSMENT
INTRODUCTION
This report presents a value-impact assessment of alternative plant

modifications and operating procedures to prevent and mitigate the effects of
IGSCC in operating BWRs.
The analysis was performed by Battelle Memorial
Institute-Pacific Northwest Laboratories (PNL) and used methods developed in

A Handbook for Value-Impact Assessment (8.1).
The format and additional
details on the development of the methods are discussed in the reference and
not repeated here.
This assessment performs analysis on postulated generic scenarios for

preventing or mitigating IGSCC in a typical BWR plant based on licensee actions and plans to date.
Basically, three scenarios each with options involv-
ing use of Hydrogen Water Chemistry (HWC) have been selected for analysis.
Two of the scenariospipe replacement and IHSI of uncracked pipeare considered as potential long-term "fixes."
The third scenarioaugmented UT
inspection and weld overlay repairis considered only a short- or intermediate-term control.
In general, the mitigation scenarios are graduated to
the extent that the greater the assurance the scheme will prevent future
IGSCC, the less stringent are the assumed post-mitigation inspection scope and
frequency requirements.
It is recognized that many plants may, in fact, adopt
a long term mitigation plan which employs a combination of IHSI of uncracked

pipe and replacing cracked pipe.
It is anticipated that the actual scenarios
will fall within the range of those considered here or some combination of
^^|hem.
8-1
The basis of the impact assessment is utility, EPRI and vendor estimates
of incurred costs. Occupational Radiation Exposure (ORE), and incremental
plant downtime for implementation of inspection and repair activities performed to comply with IE Bulletins 82-03 and 83-02 plus similar estimates of
utilities planning to replace IGSCC susceptible pipe.
The summary results and the conclusions from the value impact analysis
are presented below. Appendix E presents a description of the scenarios and
associated assumptions and discusses the analysis of each factor analyzed.
8.2
SUMMARY OF PROBLEM AND PROPOSED SOLUTION:
The results of the value-impact assessment of the postulated scenarios

are in Tables 8.1 to 8.4 below.
The attributes considered in the analysis are
shown in the following chart:
AFFECTED ATTRIBUTES
Decision Factors
Causes
Quantified
Change
Public Health
Occupational Exposure (Accidental)
Occupational Exposure (Routine)
Public Property
Regulatory Efficiency
Improvements in Knowledge
Industry Implementation Cost
Industry Operation Cost
NRC Development Cost
NRC Implementation Cost
NRC Operation Cost
Causes
Unquantified*
Change
X
X
X
X
X
X
X
X
X
X
*In this context, "unquantified" means not readily estimated in dollars.
8-2
No Changi
X
X
Table 8.1
Value-Impact Summary
Pipe Replacement
Summary of Problem and Proposed Solution;

Pipe replacement is proposed to eliminate IGSCC in a BWR. Estimates are
listed without hydrogen water chemistry. The HWC option has minimal impact on
this scenario and these estimates are tabulated in the conclusions. A discount rate of 5% was used.
Dose Reduction (person-rem)
Attribute
Best
Estimate
High
Estimate
Low
Estimate
Evaluation (S) x 10
Best
Estimate
High
Estimate
6(a)
Low
Estimate
Estimate
Public Health
151
605
Occupational Exp.
Accidental
NQ
NQ
Routine
-1810
2225
Offsite Property
Onsite Property
Reg. Efficiency
Imp. in Knowledge
Industry Cost
Implementation
Power Replacment (]mplementation)
Operation
NRC
Development
Implementation
Operation
TOTAL
RATIO:
-1660
76
NQ
-1520
-1620
-1445
NQ
NQ
NA
NA
NQ
NQ
NA
NA
NQ
NQ
NA
NA
-44
-41
-57
-68
-35
-27
-0.12
-0.23
-0.088
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
-85
-125
-62
Public dose Reduction/Sum of All NRC and Industry Costs

(person-rem/SlO^)^^^ =1.8
NA = Not affected.
NQ = Not quantified.
(a) Note: Favorable or beneficial consequences of a proposed action have a
postive sign. Unfavorable or adverse consequences have a negative sign.
For instance, an increase in industry or NRC operating costs would be
considered an unfavorable consequence and should be entered in the table
with a negative sign.
Strictly speaking, because the ratio would be expressed as a positive
number, the analyst should use the absolute value of the sum of all costs
(industry, NRC, and other) in the denominator.
8-3
Table 8.2
IHSI
Suimnary of Problem and Proposed Solution:
IHSI and hydrogen water chemistry are proposed to eliminate IGSCC in a BWR.
The estimates listed are with water chemistry using a 5% discount rate. The
estimates for IHSI without HWC are not significantly different and are
tabulated in the conclusions.
6(a)
Evaluation (S) x 10

Best
Estimate
Attribute
High
Estimate
Low
Estimate
Best
Estimate
High
Estimate
Low
Estima
Estimate
76
Public Health
151
605
Occupational Exp.
Accidental
NQ
NQ
Routine
-690
-1060
Offsite Property
Onsite Property
Reg. Efficiency
Imp. in Knowledge
Industry Cost
Implementation
Power Replacment (Implementation)
Operation
NQ
-500
NQ
NQ
NA
NA
NQ
NQ
NA
NA
NQ
NQ
NA
NA
-4.0
-9.0
-13.9
-23
-32
-4.4
-1.7
-7.3
NQ
NQ
NQ
NQ
NQ
NQ
-36.4
-47.6
NRC
Development
Implementa tion
Operation
TOTAL
RATIO:
-539
-424
-445

NA = Not affected.
(b) Strictly speaking, because the ratio would be expressed as a positive
8-4
NQ
NQ
NQ
-11.3
Table 8.3
Augmented Inspection and Repair with HWC
Summary of Problem and Proposed Solution:
Augmented inspection, weld repair, and the use of HWC are considered a partial
intermediate solution of IGSCC in a BWR. The ranges listed are for 20%-40%
inspection each outage. Money was discounted at 5%.
Best
Estimate
Attribute
High
Estimate
Low
Estimate
6(a)
Evaluation (S) x 10
Best
Estimate
High
Estimate
Low
Estimate
Estimate
76
Public Health
151
605
Occupational Exp.
Accidental
NQ
NQ
Routine
-1094
-4504
Offsite Property
Onsite Property
Reg. Efficiency
Imp. in Knowledge
Industry Cost
Implementation
Operation
NQ
-249
NQ
NQ
NA
NA
NQ
NQ
NA
NA
NQ
NQ
NA
NA
-2
0
-60
-3
0
-87
-1
0
-36
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
-62
-90
-37
NRC
Development
Implementation
Operation
TOTAL
RATIO:
-1059
194
-4575

(person-rem/SlO^)^^^ = 2.4
NA = Not affected.
Strictly speaking, because the ratio would be expressed as a positive
'
8-5
Table 8.4
Augmented Inspection and Repair - No HWC
Summary of Problem and Proposed Solution:

Augmented inspection, weld repair without HWC is considered only short-term
control of IGSCC in a BWR. The ranges listed are for 40%-60% inspection each
outage. Money was discounted at 5%.
Attribute
Best
Estimate
High
Estimate
Low
Estimate
Evaluation (S) x 10
Best
Estimate
High
Estimate
Low
Estimate
Estimate
Public Health
151
605
Occupational Exp.
Accidental
NQ
NQ
Routine
-2255
-7747
Offsite Property
Onsite Property
Reg. Efficiency
Imp. in Knowledge
Industry Cost
Implementation
Operation
NRC
Development
Implementation
Operation
TOTAL
RATIO:
-2250
76
NQ
-611
-8095
-535
NQ
NQ
NA
NA
NQ
NQ
NA
NA
NQ
NQ
NA
NA
0
0
-125
0
0
-167
0
0
-85
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
NQ
-125
-167
-85

NA = Not affected.
^
(b) Strictly speaking, because the ratio would be expressed as a positive
8-6
VALUE-IMPACT ASSESSMENT CONCLUSIONS

#
Cost and ORE results for the three scenarios are summarized below and in
the summary sheets. Public safety impacts are the same for all cases and not
repeated here.
ORE (man-rem)
Present Value Cost (SM)

5%
10%
H
L
B
H
L
PIPE REPLC.
W HWC
1830
2245
1540
91
130
70
90
129
68
W/0 HWC
1810
2225
1520
85
125
62
85
125
62
IHSI
W HWC
W/0 HWC
690
738
1060
1198
500
498
36
37
48
56
11
3.3
35
35
47
52
9
3.2
249
611
62
125
90
167
37
85
42
82
61
109
25
56
AUGMENTED
INSPECTION/REPAIR
W HWC
1094 4504
2255 7747
W/0 HWC
8-7
The results support the following conclusions:
The analysis indicates that this issue has low public risk impact, based
on risk sensitivity studies and a review of pipe failure data including
statistical estimates of IGSCC contribution to failure.
Each of the scenarios has low value-impact ratios. However, when considering 30-35 affected BWR plants, the total potential public risk reduction
indicates a low to at most medium safety priority ranking using the
guidance in NUREG-0933, "A Prioritization of Generic Safety Issues" (8.2).
However, economics dictate taking some action to reduce IGSCC impacts.

IHSI with or without hydrogen water chemistry (HWC), HWC alone, or pipe
replacement are the order of cost preference. All options will reduce
cost below the inspect and repair strategy alone.
The results are not
sensitive to discount rates except in the case of inspect and repair

costs.
HWC is not cost effective with pipe replacement and is marginal with IHSI.
HWC in plants with existing cracking problems that prevent the use of IHSI
is advantageous.
Occupational Radiation Exposure (ORE) commitments for all strategies are

significant.
IHSI will minimize the impact.
Inspection and repair, with-
out HWC, will exceed pipe replacement in man-rem.
8-8
|8.4 REFERENCES
8.1
S. W. Heaberlin et al. 1983. A Handbook for Value-Impact

Assessments.
8.2
NUREG/CR-3568.
R. Emrit et al. 1983. A Prioritization of Generic Safety Issues.

NUREG-0933.
8-9
9.0
SUMMARY
The current PCTG has reconsidered the subject of pipe cracking in nuclear
power facilities and has reached conclusions and recommendations consistent
with those reported in NUREG-0531 and NUREG-0313, Revs. 1 and 2.
The conclu-
sions and recommendations that follow present new ideas or amplify areas
addressed in the above-listed documents.
The PCTG believes that the following are the most important results of
this study;
2.0
CAUSES AND DESCRIPTION OF IGSCC PHENOMENA
Conclusions
IGSCC in BWR piping is occurring owing to a combination of material,

environment, and stress factors, each of which can affect both the
initiation of a stress-corrosion crack and the rate of its subsequent
propagation.
In evaluating long-term solutions to the problem, one
needs to consider the effects of each of the proposed remedial actions

in terms of the current understanding of the initiation and
propagation of stress-corrosion cracks.
9-1
Recommendations
Mitigating actions to control IGSCC in BWR piping must be designed to

alleviate one or more of the three synergistic factors;
sensitized
material, the convention BWR environment, and high tensile stresses.
Because mitigating actions addressing each one of these factors may

mitigating actions should address two and preferably all three of
the causative factors; e.g., material plus some control of water
3.0
Conclusions
Although significant IGSCC was found in the United States in the past,
no cracking was reported in large pipes until recently.
As a result
of extensive inspections performed during the past two years, IGSCC

has now been reported in 19 out of 23 plants.
All other countries
have experienced IGSCC in piping made of Type 304 stainless steel,

including large-diameter Type 304 stainless steel piping, in those
plants that have it.
9-2
In all countries, the preferred long-term solution is to replace

piping made of Type 304 stainless steel. Type 316NG is considered the
preferred choice, except in Germany, where they prefer their special
nuclear grade of Type 347.
Interim or short-term fixes in other countries follow the same approaches as have been used in the United States, and include IHSI,
weld overlay reinforcement, welded "clam shell" reinforcement, last
pass heat sink welding, and interim operation with unrepaired small
cracks.
It should be noted that IHSI on uncracked Type 304 stainless
steel pipe is considered to be a permanent fix in Japan.
Hydrogen water chemistry is being seriously considered by several

countries as a desirable adjunct to other solutions.
4.0
Conclusions and Recoimnendations
(There are) significant problem areas relative to the uncertainty of UT

inspection during both preservice and in-service examination.
The following
items represent a synthesis of important conclusions and recommendations;
9-3

detection.
Easily implementable modifications to these procedures can
result in some improvement.

Case N-335.
These have been incorporated into Code
Therefore, it is recommended that Code Case N-335 should
immediately be made mandatory for all augmented inspections until

better procedures are developed.
Although IGSCC detection has improved to the point that it is

considered acceptable under optimum conditions and procedures, the
detection reliability as impacted by variability in operator procedure

perfonuance demonstration represent steps in the right direction, and
Therefore, it is
recommended that ongoing industry and NRC activities to develop

adequate criteria for qualification of the entire inspection process
to achieve more reliable field inspection be completed and implemented
on a high priority basis.
9-4

items such as pumps, valves and vessels in older plants.
The

Improved leak detection systems would permit more stringent requirements on unidentified leakage without increasing the occurrence of
spurious shutdowns due to relatively benign leakage, and their
development should be pursued.
Acoustic and moisture-sensitive tape leak detection systems for local

leak detection are available, and their use is recommended where
inspection is difficult or impossible or operation for more than one
fuel cycle is considered for long cracks with weld overlay repairs.
An effort is also needed to be made to accumulate information on

current acoustic leak detection (ALD) and moisture sensitive tape
(MST) field installations.
Detailed reports on the operability,
maintenance, and reliability of these systems should be acquired.
9-5
assembled, and distributed for review by utility, government, and

research personnel interested in improving reactor leak detection
'
technology.
Sump pump monitoring systems have sufficient sensitivity to detect

leak rates as small as 1 gpm in one hour.
Improved leak detection
systems would provide additional margin against leak-before-break

without increasing the number of false alarms by providing information
about leak location and leak source discrimination.
It is, therefore,
recommended that improved leak detection systems under development be

completed and field tested.
Since the major purpose of the preservice investigation (PSI) is to

provide reliable NDE baseline for comparison with subsequent ISI, it
is recommended that the latest edition of Section XI acceptable under
10 CFR 50.55(a) be used for the PSI.
Expand modeling work to predict UT crack response for guiding development of UT techniques and guiding qualification test sample selection.
Since the UT examiner is one of the more erratic inspection variables,

it is recommended that human factors research be performed to lead to
a reduction in the human variability.
9-6
5.0
Conclusions
Augmented inspections beyond those required by ASME Code Section XI

should apply to all Types 304 and 316 austenitic piping systems
operating over 200''F (93*'C), unless they have been treated with
effecting countermeasures.
The degree of augmented inspection is dependent on the material and

processes used for each weld.
The most frequent inspections are
required for welds fabricated from nonresistant material.
Field data show that the cracking experience does not correlate well
with the stress rule index and the carbon content.
Therefore, the
primary bases for sample selection should be field experience coupled

with other factors such as weld prep, excessive grinding, extensive
repairs, or high-stress locations.
Recommendat ions
All Types 304 and 316 austenitic piping systems operating over 200''F
(93*0) should receive augmented inservice inspection, unless they have
been treated with effective countermeasures.
9-7
The extent and frequency of examinations should depend on the resistance of materials to IGSCC and the effectiveness of any processes
used to prevent cracking.
The primary basis for sample welds selected for examination should be
field experience not the stress rule index and carbon content.
Other

high-stress locations should also be considered in the sample
selection.
The performance capability demonstration program used to demonstrate

the effectiveness of UT equipment, procedure, and examiner combination on the service-induced cracked samples should be continued and
strengthened.
The effectiveness and reliability of detecting IGSCC
with all future UT procedure/examiner combinations should be demonstrated on the cracked samplea before field application.
All UT examiners should attend industry-sponsored UT detection and

sizing training courses and should continue to be evaluated on the
cracked samples under the witness of a third party personnel before
participating in the field inspection.
All BWR weld examinations should be performed in accordance with the

latest version of Code Case N-335 and with the specific equipment and
procedures used and personnel passed in the performance demonstration
tests.
9-8
An appropriate local monitoring system should be used to monitor the

continuous integrity of the uninspectable welds in categories B and
C.
These local monitoring systems should be approved by NRC before
field applications.
6.0
DECISION AND CRITERIA FOR REPLACEMENT. REPAIR. OR CONTINUED

OPERATION WITHOUT REPAIR
6.1
Flaw Evaluation and Leak-Before-Break
Conclusions
The evaluation of flaws found in service and the validation of leakbefore-break concept requires reasonably accurate knowledge of the ability to
size the length and depth of SCC accurately, the applied and residual states
of stress, the relation of crack growth rate to stress and environment, and
the ultimate load-carrying capacity of the cracked pipe.
A realistic evalu-
ation of flaw evaluation procedures and the concept of leak-before-break

requires an integrated approach considering all the above factors.
The analysis of stress-corrosion crack growth in weldments using the

techniques of linear elastic mechanics with "typical" values for the
distribution of throughwall residual stresses and stress-corrosion
9-9
crack growth rates yields results which are consistent with field and
laboratory experience with welded pipes.
Thus it is a reasonable
approach to use for the analysis of the remaining life of flawed

weldments in-reactor and for detailed probabilistic models such as
the work by Lawrence Livermore National Laboratory.
The combination of residual stress distributions and crack growth

rates currently being used by NRR should yield predictions of flaw
growth in weldments that are conservative in the great majority of
cases.
However, because of the wide variability in residual stress
distributions near welds and IGSC crack growth rates, the current
approach used by NRR is not considered unduly conservative for current
BWR piping materials and environments.
The Task Group endorses the
guidelines for the evaluation of flaw growth in weldments given in

draft NUREG-0313, Rev. 2.
The Task Group also wishes to point out, however, that although conservative estimates of the throughwall growth of stress-corrosion
cracks can be made with a high degree of confidence given an initial
crack size, there is a high degree of uncertainty in the measurement
of the depth of existing cracks, as discussed in Section 4, and this
uncertainty must be adequately addressed in the development of acceptance criteria for flawed piping.
9-10
The use of loading conditions defined by the ASME Code in performing

flaw evaluations is acceptable, with the exception that SLA thermal
expansion stresses should be included in the flaw evaluation when
limit load is not achieved prior to significant crack extension.
Use
of these loading conditions is consistent with the approach which has

been accepted by the NRC for the design and operation of other reactor
components and systems.
Analyses for large surface cracks, conducted using the relatively low
ductile fracture resistance properties of stainless steel weld metal,
show that the margins against fracture may not be as great as those
believed to be associated with IWB-3640.
Nonetheless, a substantial
margin against fracture can be demonstrated for these flaws under

normal operating conditions.
Evaluation of throughwall cracks, using weld metal properties, shows

that a surface crack breaking through the pipe wall under bounding
accident loading conditions would have to be approximately 30% of the
circumference in length, for large-diameter pipes, to result in unstable fracture of the pipe.
However, leaks from cracks not large
enough to cause gross failure of the pipe may be undesirably large.
9-11


Specific assessments of piping integrity will have to recognize weld

conditions having lower toughness than the 304 stainless steel welds
used in the analyses described in this report.
Operating experience suggests that leak-before-break is the most

likely mode of failure for the vast majority of cracks occurring in
service.
This is a result of the asymmetry of the weld residual
stresses and applied loads and the variability in material properties.
Evaluations using conservative crack growth rate predictions
and net section collapse analyses, applicable to cracks in very high

toughness material, indicate that for the vast majority of possible
crack geometries there exists significant time to allow for detection
of leakage and implementation of corrective actions.
Evaluation using
the fracture resistance properties of the weld material show substantial margins against failure, under normal and accident loading conditions for throughwall cracks which should be reliably detected by
leakage.
9-12
Recommendat ions
Based on the above discussions and conclusions, the Task Group has
developed the following recommendations.
The limitation on
makeup capacity.
Calculations

9-13
The recommendation for a limit on acceptable crack length is primarily

a result of the lack of confidence in ultrasonic depth sizing capabilities.
In this respect, the fracture mechanics calculations
presented in this report demonstrate that there are acceptable margins

against fracture for relatively long, deep surface cracks.
Demonstra-
tion of reliable sizing of the part-through flaw depth would most

likely allow relaxation of the limits on crack length.
Flaw evaluation criteria should limit the length of the cracks

accepted for continued operation without repair.
Additional fracture mechanics analyses, material properties characterization, and large-scale pipe tests should be performed to
further our understanding of the implications of stainless steel weld
and cast material fracture toughness properties in flawed pipe
evaluations.
Furthermore, the Task Group recommends active NRC
support of the ASME task group currently evaluating the concerns which
have been raised regarding IWB-3640.

are sensitive enough to provide additional margin against leak-before-
9-14
break if more stringent requirements on surveillance intervals and

unidentified leakage are imposed (see Section 4).
Therefore, the Task
Group recommends that the limits on unidentified leakage in BWRs be

decreased to 3 gpm and that the surveillance interval be decreased to
4 hours or less.
Short-Term Solutions

long-term operation with weld overlays, since errors in crack depth
For circum-
ferential cracks, weld overlay is considered an acceptable repair

Long-Term Solutions and Safety Assessment

and 304NG and 316NG in Japan, has been excellent. Other materials
used in the U. S. include 304L and 316L. All of these alloys are more
resistant to IGSCC than conventional Types 304 or 316 stainless
9-15
steel.
Based on U. S. data and prior use. Type 316NG stainless steel
offers an additional margin of resistance to IGSCC and utilities that

choose to replace pipe should be strongly encouraged to use it.
The use of additional mitigating procedures such as IHSI or hydrogen

water chemistry to provide additional margin against environmentally
assisted cracking is strongly recommended even after replacement with
Type 316NG stainless steel.
Utilities should be encouraged to adopt
an ALARA approach to coolant impurity levels.

It appears that greater care
must be exercised in the control of composition and fabrication


state.
be more effective when applied to weldments with no reported cracking.
IHSI is an acceptable mitigating procedure for weldments with no

reported cracking even in older plants.
However, to be fully
effective a suitable hydrogen water chemistry also should be

implemented.
9-16
The use of IHSI on weldments with detectable cracking must be

considered on a case-by-case basis.
However, for relatively short
cracks (approximately 20% of the circumference in length), since even

large errors in crack sizing or the prediction of flaw growth will
lead only to small leakage, the decision on whether an additional
repair is required can be determined by analysis.
For longer cracks,
repair will probably be required.
All steps related to the weld prep, welding, and weld finishing and
of an appropriate QA manual.
As noted previously, the weld should be
optimized for future ISI.

oxygen level.
The current work on reduction of oxygen through

effective IGSCC countermeasure.
However, ongoing work regarding

countermeasure.
9-17
Either reduction of ionic species or hydrogen water chemistry, used

alone, will have some beneficial effects.
Used in combination, they
show great promise for mitigating IGSCC.
The effects of hydrogen water chemistry on the balance of the system

and on overall long-term plant operation need to be explored in
greater depth before they can be recoimnended without reservation for
adoption on a wide scale.
Regulatory Guide 1.56 needs to be modified to reflect these water

chemistry recommendations to provide tighter controls on water
conductivity and provisions for oxygen control as an acceptable
option.
The long-term effects of decontamination procedures on IGSCC need to

be investigated.
9-18
IGSCC RISK STUDIES
Conclusions
Intergranular stress-corrosion cracking (IGSCC) has caused no significant loss-of-coolant events or loss-of-coolant accidents (LOCA) in
BWRs even though its presence in 304 stainless steel coolant piping
has been found in many operating plants.
This may be due to an
aggressive in-service inspection and repair program and/or a tendency

to leak-before-break (with effective leak detection).
The LOCA con-
tribution to the core damage frequency as calculated by probabilistic

risk assessment (PRA) studies for BWRs is a minor contribution (generally less than 10%) for all LOCA sizes.
Because the LOCA contri-
bution to the core damage frequency is small, it would require a

considerable error, at least an order of magnitude, from IGSCC before
LOCAs would be a dominant contributor to the core damage frequency.
IGSCC is but one component of the LOCA occurrence rate.
Design and
construction errors, maintenance errors, cyclic and thermal fatigue,

etc. all contribute to the LOCA probability.
It is estimated that
IGSCC may be 30% or more of the LOCA frequency in BWRs as deduced from
experience, the total industrial data base, and engineering judgment.
Therefore, an error in the IGSCC LOCA contribution of 20-30 would be
required for an order of magnitude error in the LOCA probability.
9-19
The pipe failure data base and available studies were reviewed for evidence of IGSCC impact.
To a considerable extent, uncertainties have
always existed in determining piping failure rates.
Largely, this
uncertainty is intrinsic in attempts to measure low-frequency events

from experience, and it makes the effect of trends difficult to
analyze.
It was statistically reasoned that an error of an order of
magnitude in the large pipe failure frequencies (presumably from

IGSCC) would have produced, with one chance in four, at least one
failure during this interval.
Therefore, it was concluded that the
frequency for large pipe failure adequately accounts for the

contribution due to IGSCC.
For intermediate and small pipe failures, the incidence of reported

and identified corrosion events in the data base, although increasing
somewhat over time (owing to improved reporting and other causes), was
judged to be not statistically indicative of an error in the failure
frequencies.
The systems aspects of IGSCC risk were revised using Hatch-2 as a

model.
It was concluded that the plants should be required to be
designed for the most severe large LOCA without significant core
damage.
The ECCS must be able to survive the most severe single
failure in the system and still perform its intended function(s).

Pipe failure is no more severe than other failures in the system
9-20
(valves, pumps, etc.) and is probably less severe.
Coupling this
with the ECCS redundancy and conservative margin in each design and
calculational procedures, it is concluded that IGSCC would cause
little increase in risk from the standpoint of systems important to
safety.
Multiple failures were briefly considered.
IGSCC may be a common mode
failure in the recirculation piping and other systems.
It is con-
cluded that a large LOCA and ECC system failure both caused by IGSCC
is adequately protected by the single-failure criteria and systems
redundancy.
Multiple failures causing a LOCA larger than a DBA are
likely to add little to the risk because it appears that the probability of multiple failures is several orders of magnitude less than
that of a large single failure.
Overall it is concluded that the presence of IGSCC in BWRs may reduce

safety margins believed to exist, but there is no apparent evidence to
conclude that IGSCC has increased public risk significantly.
Recommendations
Piping failure rates are difficult to determine because failures of high

quality piping occur infrequently and require a data base acquired over a
large number of reactor years, since normal experimental procedures are
impractical.
This difficulty is unlikely to be overcome now or in the fore-
9-21
seeable future.
What results is an estimated frequency with large uncertain-
ties from aggregated events acquired from poor-quality documentation.
To
partition this frequency according to contributing phenomena in a meaningful

way is very difficult.
It is recommended that a formal study and updating of the pipe failure

data base be undertaken.
Some statistics may be only meaningfully
derived or assessed analytically.
The NRC has developed a computer
code PRAISE (_Piping Reliability Analysis including Sêismic

JIvents) which is currently being modified to include a model for
IGSCC based on current understanding.
It is recoimnended that PRAISE be used to a) investigate the impact of

IGSCC in primary piping according to pipe size, b) study the failure
frequencies of components due to IGSCC, fatigue, etc. , and c)
calculate the conditional probability of multiple failures associated
with IGSCC to check the point value and to test the assumption of a
lognormal distribution.
8.0
INTERGRANULAR STRESS-CORROSION CRACKING (IGSCC) VALUE-IMPACT ASSESSMENT
Conclusions
The analysis indicates that this issue has low public risk impact,
based on risk sensitivity studies and a review of pipe failure data
including statistical estimates of IGSCC contribution to failure.
9-22
Each of the scenarios has low value-impact ratios.
However, when
considering 30-35 affected BWR plants, the total potential public risk
reduction indicates a low to at most medium safety priority ranking.
However, economics dictate taking some action to reduce IGSCC impacts.

IHSI with or without HWC, HWC alone, or pipe replacement are the order
of cost preference.
All options will reduce cost below the inspect
and repair strategy alone.
The results are not sensitive to discount
rates except in the case of inspect and repair costs.
HWC is not cost effective with pipe replacement and is marginal with
IHSI.
HWC in plants with existing cracking problems that prevent the
use of IHSI is advantageous.
ORE commitments for all strategies are large.

impact.
IHSI will minimize the
Inspection and repair, even with HWC, will eventually exceed
pipe replacement in man-rem.
9-23
APPENDIX A
LIST OF TASK GROUP MEMBERS
Spencer H. Bush
(chairman)
RSA
President
Review and Synthesis Assoc.
Richland, WA
Richard A. Becker
NRC
Office of Nuclear Reactor Regulation

Clinch River Breeder Reactor Program Office
Technical Review Section
Ching-Yao Cheng
NRC

Division of Engineering
Materials Engineering Branch
William J. Collins
NRC
Engineering & Generic Communications Branch

Office of Inspection & Enforcement
Billy R. Crowley
NRC
Engineering and Operations Program Division

Regional Administrators Office, Region II
Jacque P. Durr
NRC
Regional Administrators Office, Region I
Warren S. Hazelton
NRC

Division of Engineering
Philip R. Matthews
NRC

Division of Safety Technology
Safety Program Evaluation Branch
Joseph Muscara
NRC
Office of Nuclear Regulatory Research

Division of Engineering Technology
Richard C. Robinson
NRC

Division of Risk Analysis
Reactor Risk Branch
Jack Strosnider
NRC

Division of Engineering Technology
A-1
APPENDIX B
LIST OF CONSULTANTS AND MAJOR CONTRIBUTORS
William Andrews
Battelle Memorial Institute

Pacific Northwest Laboratories
Value-Impact
Steven R. Doctor

Nondestructive Examination
Ronald M. Gamble
IMPELL Corporation
Fracture Mechanics
David Kupperman
Argonne National Laboratory
William J. Shack
Argonne National Laboratory
Materials Engineering
Tom T. Taylor

John R. Weeks
Brookhaven National Laboratory
Corrosion Mechanisms
and Water Chemistry
Gery Wilkowski
Battelle Columbus Laboratories
Fracture Mechanics
B-1
APPENDIX C
This appendix consists of two parts.
The first, C-1 is a copy of the
10 CFR 50.54 position pertaining to inspection measurements.
Basically, the
same information was contained in SECY-83-267c.
The second, C-2, is a copy of the 10 CFR 50.59 requirements pertaining t

the replacement of piping in BWRs.
C-1
UNITED STATES
NUCLEAR REGULATORY COMMISSION

WASHINGTON, D. C. 20555
April 1^, 1984

TO ALL LICENSEES OF OPERATING REACTORS, APPLICANTS FOR OPERATING
LICENSE, AND HOLDERS OF CONSTRUCTION PERMITS FOR BOILING WATER REACTORS
Gentlemen:
SUBJECT:
INSPECTIONS OF BWR STAINLESS STEEL PIPING (Generic Letter 84-//)
Inspections conducted at several boiling water reactors (BWRs) revealed

intergranular stress corrosion cracking (IGSCC) in large-diameter recirculation
and residual heat removal piping. These inspections were conducted pursuant
to IE Bulletins 82-03, Revision 1 and 83-02, and the NRC August 26, 1983 Orders.
The Conuiission believes that the results of these inspections mandate an
ongoing program for similar reinspections at all operating BWRs. Where IGSCC
is discovered, repairs, analysis and additional surveillance may also be
required to ensure the continued integrity of affected pipes.
Staff efforts to date on this issue include review of the Electric Power
Research Institute (EPRI) report dated August 4, 1983, establishment of a pipe
crack study group within the staff, evaluation of the results of IGSCC
inspections already conducted, and discussions with licensees and industry
groups. As a result of these considerations, the staff has concluded that
the following actions would be considered an acceptable response to the
current IGSCC concerns:
1.
A reinspection program of piping susceptible to IGSCC should be

undertaken. The reinspection should commence within about two
calendar years, adjusted to coincide with the next scheduled
outage, from the previous inspection performed under IE Bulletins
82-03, 83-02, or our August 26, 1983 Order.
2.
These reinspections should include the following stainless steel welds,

susceptible to IGSCC, in piping equal to or greater than 4" in
diameter, in systems operating over 200F, that are part of or connected
to the reactor coolant pressure boundary, out to the second isolation
valve as follows.
(a)
Inspection of 20? of the welds in each pipe size of IGSCC sensitive

welds not inspected previously (but no less than 4 welds) and
reinspection of 20% of the welds in each pipe size inspected
previously (but not less than 2 welds) and found not to be cracked.
This sample should be selected primarily from weld locations shown
by experience to have the highest propensity for cracking.
(b) All unrepaired cracked welds.

(c)
Inspection of all weld overlays on welds where circumferential

cracks longer than 10% of circumference were measured. Disposition
of any findings will be reviewed on a case-by-case basis. Criteria
for operation beyond one cycle with overlaid joints are under
development.
C-3
(d)
Inspection of any weld treated by induction heating stress

improvement which has not been post treatment UT acceptance
tested.
(e)
In the event new cracks or significant growth of old cracks* are

found, the inspection scope should be expanded in accordance with
lEB 83-02.
NOTE:
[Results of inspections conducted to date under lEB 82-03 and 83-02

indicate that all stainless steel piping welds in systems operating
over 200F are susceptible to IGSCC. In addition, "field data shows
that the cracking experience does not correlate well with the Stress
Rule Index (SRI) and the carbon content. Therefore, the primary
index for sample selection should be field exoerience, where other
factors such as weld preparation, excessive grinding, extensive
repairs, or high stress locations are known to exist, they should
also be considered in the sample selection.]
3.
All level 2 and level 3 UT examiners should demonstrate competence in

accordance with lEB 83-02 and level 1 examiners should demonstrate field
performance capability.
4.
Leak detection and leakage limits should be sufficiently restrictive to

ensure timely investigation of unidentified leakage. See Attachment 1.
5.
For crack evaluation and repair criteria see Attachment 2.
Accordingly, pursuant to 10 CFR 50.54(f), BWR operating reactor licensees and

applicants for an operating license (this letter is for information only for
those utilities that have not applie?~for an operating license) are requested,
in order to determine wnether your license should be modified or suspended,
to furnish, under oath or affirmation, no later than 45 days from the date of
this letter, your current plans relative to inspections for IGSCC and interim
leakage detection. Your response should indicate whether you intend to follow
the above staff recommended actions or to propose an alternative approach to
resolving IGSCC concerns. In either case, your response should address:
(a)
Scope and schedule of planned inspections
(b) Availability and qualification of examiners

(c)
Description of any special surveillance measures, in effect or

proposed, for primary system leak detection, beyond those measures
already required by your Technical Specifications
(d)
Results of the Bulletin inspections not previously submitted to NRC
(e)
Remedial measures, if any, to be taken when cracks are discovered
Significant growth of the old crack is defined as growth to a new crack

size that cannot be accepted without repair for the remaining period of
the current or a new cycle of operation, in accordance with Attachment 2.
C-4
- 3 -
The staff considers the IGSCC problem to be generic for all BWRs. Therefore,
your response may incorporate by reference materials furnished by an Owners'
Group. To the extent practicable. Owners' Group and EPRI participation in
the IGSCC effort is encouraged.
Licensees and applicants may request an extension of time for submittals of
the required information. Such a request must set forth a proposed schedule
and justification for the delay. Such a request shall be directed to the
Director, Division of Licensing, NRR. Any such request must be submitted
no later than 15 days from the date of this letter.
This request for information was approved by the Office of Management and
Budget under clearance number 3150-0011 which expires April 30, 1985.
Coninents on burden and duplication may be directed to the Office of
Management and Budget, Reports Management Room 3208, New Executive Office
Building, Washington, D. C. 20503.
Licensing
C-5
ATTACHMENT 1
LEAK DETECTION AND LEAKAGE LIMITS
The reactor coolant leakage detection systems are operated in accordance
with the Technical Specification requirements to assure the discovery of
unidentified leakage that may be caused by throughwall cracks developed in
austenitic stainless steel piping.
A.
The leakage detection system shall be sufficiently sensitive to

detect and measure small leaks in a timely manner and to identify
the leakage sources within practical limits. Particular attention
should be given to upgrading and calibrating those leak detection
systems that will provide prompt indication of an increase in
leakage rates
Other equivalent and/or local leakage detection and collection

systems will be reviewed on a case-by-case basis.
B.
Plant shutdown shall be initiated for inspection and corrective

action when any leakage detection system indicates, within any period
of 24 hours, an increase in rate of unidentified leakage in excess
of 2 gpm or its equivalent, whichever occurs first.
For sump level
monitoring systems with a fixed-measurement interval method, the

level shall be monitored at 4-hour intervals or less.
C.
At least one of the leakage measurement instruments associated with

each sump shall be operable, and the outage time for inoperable
instruments shall be limited to 24 hours or immediately initiate
an orderly shutdown.
c-6
- 2 -
Unidentified leakage should include all leakage other than
(1) leakage into closed systems, such as pump seal or valve

packing leaks that are captured, flow metered, and
conducted to a sump or collecting tank, or
(2) leakage Into the containment atmosphere from sources

that are both specifically located and known either
not to Interfere with the operations of unidentified
leakage monitoring systems, or not to be from a
through-wall crack in the piping within the reactor
coolant pressure boundary.
A visual examination for leakage of the reactor coolant piping shall

be performed during each plant outage In which the containment is
deinerted.
The examination will be performed consistent with the
requirements of IWA-5241 and IWA-5242 of the 1980 Edition of

Section XI of the ASME Boiler and Vessel Code. The system boundary
subject to this examination shall be in accordance with IWA-5221.
C-7
ATTACHMENT 2
CRACK EVALUATION AND REPAIR CRITERIA
Background
(a) Code Requirements
The ASME Boiler and Pressure Vessel Code Section XI has rules for
evaluating the acceptability of flaws for further operation.
Table IWB 3514-3 provides rules for acceptability of flaws
without further evaluation; although the specific dimensions of
such acceptable flaws depends on both the length and depth of
the flaws, the practical effect is that flaws less than about
10% of the wall thickness are acceptable for further operation
without analysis or repair.
A new section has recently been added to the Code, IWB 3600.
This extends the Code flaw evaluation rules for piping to
Include specific rules whereby flaws deeper than those allowed
by IWB 3514-3 can be accepted for further operation without
repair.
Section IWB 3600 also requires that these acceptable flaw sizes
include considerations of crack growth by stress corrosion and
fatigue.
In other words, if a crack is to be considered
acceptable for further operation without repair, it must be

shown that it will not grow to be larger than the IWB 3640
limits during the time period for which the evaluation is
performed.
C-8
- 2 -
Crack Growth Assessment

IGSCC at welds in BWRs is primarily initiated by very high
tensile residual welding stresses on the inside surface at
the heat affected (sensitized) zones of the base metal
very near the welds. This tensile residual stress changes
to a compressive stress toward the middle of the pipe wall;
this reduction in stress reduces the crack growth rate
through the center portion of its pipe wall. As the crack
progresses further through the wall, the relative effect
of the pressure and bending stresses increases, and the
crack growth rate will increase.
The residual stress patterns and calculational methods for

crack growth rates are fairly well established by considerable
research and correlations with service experience. The staff
has selected parameters that should lead to overprediction
of growth. This is intended to compensate for uncertainties
discussed in more detail below.
Staff Treatment of Uncertainties
One of the main uncertainties associated with the evaluation of
pipe cracks is the uncertainty of crack sizing, both depth and
length of IGSCC cracks. Although this technology is being
improved, the uncertaintiy in crack sizing will likely remain.
c-9
- 3 -
The staff has used a relatively simple approach to cover sizing

uncertainties.
In practice, the staff approach permits operati
of unrepaired cracks but only if calculations show that they

would not exceed Code limits even if the crack at the start of
operation were actually twice as large as reported.
2.
Staff Acceptance Criteria

(a) Criterion for Operation without Repair
Plant operation is permitted with cracked welds only for the
time period that the cracks are evaluated to not exceed 2/3*
of the limits for depth and length provided in ASME Code
Section XI, Paragraph IWB-3640. Crack growth analyses must
include any additional stress Imposed on the weld by other
weld repair operations, and each analysis must be approved
by the NRC.
(b) Criteria for Cracked Repairs
(1)
If cracked welds are repaired by weld overlay, the

thickness of the overlay must be sufficient to
provide full IWB-3640 margin during the proposed
operating period, assuming that the cracks are or
will grow completely through the original pipe wall
and the first overlay layer to the low carbon and
low ferrite portion of the overlay, unless it Is
demonstrated that the crack(s) are shallow enough
to be arrested by the weld overlay.
This criterion allows for an uncertainity of up to 100% In crack depth

for reported cracks up to 25% of wall thickness.
c-io
- 4 -
Effective overlay thickness is defined as the

thickness of overlay deposited after the first
weld layer that clears dye-penetrant testing (PT)
inspection.
(11)
The minimum effective overlay thickness permitted,

even for very short cracks In either longitudinal
or circumferential direction. Is two weld layers
after the first layer to clear PT Inspection,
(ill) Full Structural strength weld overlays must be

provided for long cracks with total circumferential
extent approaching the length that would cause
limit load failure If they were actually throughwall,
(iv)
Multiple short circumferential cracks are to be

treated as one crack with a length equal to the
sum of the circumferential lengths.
Discussion of Staff Acceptance Criteria

Since the period of operation between inspections could vary from plant
to plant and the applied stress level varies from location to location,
use of a fixed simplified repair criterion established on the bases of
crack size prior to the period of operation would be difficult.
In
any case, however, flaws less than about 10% of the wall thickness are
acceptable for further operation without repairs. For a typical 18
month operating cycle, the staff criteria would generally require that
c-u
- 5 -
cracks greater than 30% of the circumference and cracks with reported
depth of 25% or greater of the thickness will likely need some form
of repair. For the same 18 month cycle, cracks of smaller size down
to 10% of wall thickness may be acceptable without repair but would
require evaluation in accordance with the staff criteria.
C-12
UNITED STATES
NUCLEAR REGULATORY COMMISSION

WASHINGTON, D. C. 20555
**-
March 14, 1984
TO ALL LICENSEES OF BOILING WATER REACTORS (BWRS)

Gentlemen:
SUBJECT:
PROCEDURAL GUIDANCE FOR PIPE REPLACEMENT AT BWRS

(Generic Letter 84-07)
This letter provides guidance to licensees planning to replace recirculation

system piping (or other reactor coolant system pressure boundary piping) with
material that is less susceptible to intergranular stress corrosion cracking.
In particular, guidance is provided regarding NRC reviews and approvaTs that
may be necessary.
10 CFR 50.59 specifies the conditions that would require prior NRC approval
of changes in the facility. In your compliance with 10 CFR 50.59, we recognize
that the decision on whether your planned replacement program involves an
unreviewed safety questions can be difficult, and that an understanding of
the NRC position on this issue would be helpful in planning your program.
The purpose of this letter is to provide as clear a statement as possible of
our views on this issue.
We encourage programs to replace piping so as to minimize the potential for
cracking and we will expeditiously review any submittals provided to us so
as to not delay this important improvement program. We encourage early
submittal of appropriate requests for review for those situations that
require prior approval. Prior NRC approval is not necessary unless the
proposed change to the facility involves an unreviewed safety question or
a change in Technical Specifications.
In all cases, licensees must perform and document appropriate reviews and
analyses in accordance with 10 CFR 50.59 and the facility Technical
Specifications. These analyses should be maintained by the licensee, in
accordance with Coimiission regulations and the applicable license, to permit
the staff to audit such evaluations, as necessary. In those cases where
licensees determine that their program for pipe replacement does not involve
an unreviewed safety question, there remains the concern that the NRC may,
at a later date, disagree with that determination, thereby potentially
delaying the program. To minimize that possibility, we have developed a
position regarding the major considerations in a pipe replacement program
which licensees can use in determining the necessity or desirability of
seeking prior NRC approval. That position is contained in the enclosure
to this letter.
Replacement of recirculating system piping may involve individual and collective
radiation exposure to plant workers beyond that in other routine maintenance
work. 10 CFR Part 20 requires that licensees "make eêry reasonable effort to
maintain radiation exposures, and releases of radioactive materials in effluents
to unrestricted areas, as low as is reasonably achievable." We request that a
c-13
8403150275
- 2 -
description of your radiation protection program for the pipe replacement

effort be furnished to us as early as possible before initiating the
replacement program. Your submittal should include a description of
appropriate pre-planning procedures, shielding, equipment, personnel training,
estimated total cumulative dose, and other measures to be initiated that will
keep exposures as low as reasonably achievable. We anticipate that most pipe
replacment programs can be accomplished through suitable controls so as to
limit cumulative exposures to less than about 2000 person-rem. We will plan
to meet with licensees whose programs involve greater dose estimates.
This request has been approved by 0MB Clearance Number 3150-0011, which
expires April 30, 1985.
^JuV. f iênh'ui
Division ofViicensing
Enclosure:
Procedural Guidance
c-14
ENCLOSURE
PROCEDURAL GUIDANCE FOR PIPE REPLACEMENT AT BWR'S
As a result of the occurrence of intergranular stress corrosion cracking

(IGSCC) in the recirculation system piping at a number of Boiling Water
Reactors (BWRs), several licensees have indicated their intention to replace
the recirculation piping with material that is less susceptible to IGSCC.
The purpose o-f this document is to present NRC staff guidance regarding
procedural requirements applicable to the replacement of recirculation
system piping. Depending on the specific changes that may be involved,
such changes may be authorized in various ways under Commission regulations.
Replacement of a reactor coolant system component by another component of
the same kind (not changing any feature described in the FSAR) would not
be prohibited by the Commission regulations.
Replacement which involves changes, but which does not involve an amendment
to the license or Technical Specifications, may be carried out without prior
Commission approval, if the changes involve no unreviewed safety questions.
The definition of an unreviewed safety question is set forth in 10 CFR 50.59.
Most Technical Specifications contain provisions governing the way a
licensee is to assess changes to determine whether they involve unreviewed
safety questions. Not only should each aspect of change be carefully considered
in making a determination under 10 CFR 50.59, but the overall cumulative effect
of the various changes, considered as a whole, should also be assessed.
There may be cases involving difficult questions concerning whether some
aspect may be an unreviewed safety question. Such cases may be processed
as a license amendment authorizing the replacement program*. Changes which
entail changes in Technical Specifications may be authorized only by license
amendment**. License amendments involving no significant hazards considerations
will be processed expeditiously - after 30 days notice in the Federal Register
and sufficient time for staff review.
*The case of Nine Mile Point, the first major pipe replacement case, was
handled by the amendment process.
**If the changes in Technical Specifications are limited in number and can
be evaluated as a matter distinct from the replacement program, such
changes may be processed separately from the replacement program.
c-15
- 2 -
If there is a request for a hearing concerning an amendment for which there is ^

a final determination that the amendment involves no significant hazards, the
amendment may be granted notwithstanding such request. The hearing, if any,
would be held after the amendment has been issued. If the changes are such as
to involve a significant hazards consideration, and a hearing is requested, the
license amendment would await the outcome of the proceeding.
We have reviewed information concerning the programs under consideration by
some licensees, and conclude that certain proposed features of such programs
having the characterisitics discussed below, subject to the results of the
plant specific evaluations required by 10 CFR 50.59, do not involve unreviewed
sa-fety questions.
Design Specifications
Design specifications are prepared covering all aspects of construction of the
replaced piping system, including the Codes designated for use. Construction
as used above is an all-inclusive term comprising materials, design, fabrication,
installation, examination, testing and inspection.
Codes
Either:
(1) The codes and standards described in the FSAR, consistent with
current regulations, as the original licensing basis for issuance of the
facility operating license are utilitized;
Or, and preferably:
(2) The Code edition used for the construction of the piping system is, to
the extent practical, the latest Code edition and addenda. If the later Codes
are used, practical adjustments are made in using upgraded Code provisions to
accommodate the limitations of design, geometry and items of the original piping
systems and its supports which are not replaced. The quality and overall margins
required in the original design are not to be impaired in determining the extent
to which adjustments are made in using upgraded Code provisions.
Materials
The materials used are those that have received staff reveiw and approval as
documented in NUREG-0313, Rev. 1.
Design Changes
Replacement components and their resultinq effect upon system performance provide
an equivalent overall safety margin as provided in the originally licensed
design. Within this constraint, there may be design changes that clearly enhance
the safety function of the system under replacement, such as, changes to reduce
the number of welds or changes to facilitate inspection. Piping, pipe supports
and any remaining original components are covered in the specifications of l o a ^ *
combinations and jippropriate allowable limits provided for the replaced s y s t e m M
c-16
- 3 -
^Ânalj^sj^
A stress analysis for the recirculation and other replaced piping systems is
to be performed, that demonstrate that allowable limits have not been exceeded.
The loads and loading combinations used in the FSAR or as described in SRP
Section 3.9.3 should be used in the stress analysis.
Whip restraints and jet impingement design must follow FSAR criteria or that
of Standard Review Plan Section 3.6.2.
Fabrication, Installation, Examination and Testing
Fabrication, installation, examination and testing is to be performed in
accordance with the applicable Code designated under Codes.
Inspection/Quality Assurance
Programmatic quality assurance and independent (third party) inspection
requirements appropriate to the replaced piping are to be followed.
System Characteristics
Flow rates, temperatures, and pressures must not be significantly different
than those evaluated in the FSAR. The performance of Engineered Safety
Features must not be degraded by the replaced piping systems.
Discussion of 10 CFR 50.59
The NRC staff has developed this guidance without incorporating the results
of any detailed review of a specific plant or specific plant Technical
Specifications. A detailed case-specific review could lead to a positive
unreviewed safety question conclusion even though the general guidance
provided above would suggest a negative conclusion. A positive conclusion
v/ould be reached if, for example, the combination of FSAR codes and updated
codes leads to a reduced safety margin for se plant structure, system or
component, and this reduced margin either increases the possibilities or
consequences of an accident or malfunction of equipment important to safety
which was evaluated in the FSAR, or created the possibility of some new accident
or malfunction, or pertains to a margin of safety in the basis for any
Technical Specification. The margin of safety defined by a Technical Specification would comprise several safety margins within a given system. In such
a circumstance the margins of safety for Individual components within the
system could well be adjusted without any effect on the margin of safety for
the system as defined in the Technical Specifications. Thus each licensee
must per-form the specific, detailed review required by 10 CFR 50.59(a)(2) and
in all cases the results of the detailed, case-specific review are controlling
over the general guidance provided in this enclosure. Not all repairs involve
changes and if they do not, no plant-specific review pursuant to 50.59 is
required.
C-17
APPENDIX D
INTERGRANULAR STRESS-CORROSION CRACKING RISK STUDIES
D.l
INTRODUCTION
Early efforts concerning the impact of IGSCC on nuclear power plant

safety relied on the application of a deterministic approach, the established
licensing methodology.
In general, this procedure utilizes the established
engineering sciences to form engineering judgments regarding the impact of a

phenomenon on safety margins.
Some degree of quantification of unlikely events through probabilistic

assessments did not come fully into the regulatory process until the Reactor
Safety Study (D.l) was published in late 1975.
Several controversial years
followed before probabilistic risk assessment (PRA) techniques were accepted

and applied in a fairly routine manner.
Acceptance does not imply total
reliance or that the uncertainties in the methodology are ignored, but PRA is
generally accepted as a useful tool.
Currently, a number of PRAs have been
performed on operating plants either as full-scope PRAs or through the Interim

Reliability Evaluation Program (IREP).
The mounting concern, as more IGSCC is detected by BWR inservice inspections, is discussed elsewhere in this report.
More frequently, inservice
inspections have discovered IGSCC in the large-diameter recirculation piping

both in domestic and foreign BWRs.
As more piping cracks attributable to
D-1
IGSCC are discovered, the question of IGSCC implications to public risk

logically arises.
One facet of the answer to the implications of IGSCC to
"
public risk may be addressed by examining the implications inferred from

existing risk assessments.
The remainder of this Appendix examines these risk
implications as they relate to BWRs only.
D.1.1
Loss-of-Coolant Accident (LOCA)
All current BWRs operate on a direct cycle:
i.e., water removes the heat
from the nuclear fuel by boiling in the core, the steam and water are
separated in the vessel, the steam is taken directly from the vessel to drive
the turbine-generator, and the remaining water is recirculated through the
core.
If the primary pressure boundary fails, a LOCA results.
of a pipe, weld, or boundary component results in a LOCA.
The failure
It has been
common practice (D.l) to define LOCA accidents in ranges of rupture sizes.

Nominally, the rupture sizes correspond to area of the opening, but have been
given equivalent piping diameters.
This categorization for BWR liquid piping*
is 8-1/2 in. and greater, between 8-1/2 and 2-1/2 in., and between 2-1/2 and
1/2 in.
This structuring is made loosely on system considerations which will
be discussed more fully in a later section.
In LWRs this structuring defines
large LOCA b(A), small LOCA b(Si), and small-small LOCA b(S2) sequences.
*WASH-1400 defines different ranges for steam piping, but these variations are
unimportant for this discussion. In reading WASH-1400, confusion between the
two sets of limits can result since the suction side of the RCS is liquid and
the discharge side is steam.
i
D-2
D.1.2
Risk Assessment Methodology
PRA methodology attempts to examine all the accident sequences which

could potentially endanger the general public.
By quantifying the likelihood
of these sequences and the potential consequences should they occur, the
overall risk to public safety and health may be estimated.
Although there are
other sequences, the dominant contributors to public risk are those accident
sequences that lead to reactor core damage and, consequently, release of
radioactivity to the surroundings.
Since the release of radioactivity after
core damage is common to most sequences, the relative contribution to public

risk is assumed in this study to be proportional to the contribution an
accident sequence makes to the core damage frequency.
The first question to be addressed is:

sequences to the core damage frequency?
What is the contribution of LOCA
Table D.l presents the results taken
from a number of BWR PRAs, including WASH-1400.
It can be seen that the LOCA
contribution in all cases is a small fraction of the core damage frequency.

With the detection of more IGSCC in BWR piping, the question of impact on risk
focuses on the impact IGSCC has on failure rate.
Before proceeding further, a
sensitivity model to assess the impact of the failure rate on core damage
frequency proportionate to risk is necessary.
be developed from the risk methodology.
D-3
A simple sensitivity model can
Table D.l
Sensitivity of Core Melt Frequency for BWR PRAs

for a Postulated Increase (lOx) in LOCA Initiator Frequency
Plant A*
Sponsor
Total
Core Melt
Frequency**
(per Rx yr)
Frequency of Core
Melt Due to LOCA
Initiated Sequence
(per Rx yr)
Increase in the Frequency

of Core Melt Due to
Postulated Increase (lOx)
in LOCA Initiator Frequency
(per Rx yr)
Limerick (Rev 5)
PECO.
1.5 E-5
1.3 E-7
1.2 E-6
Shoreham
LILCO
5.5 E-5***
1.9 E-6
1.7 E-5
Millstone 1
IREP
3.0 E-4
3.0 E-6
2.7 E-5
Browns Ferry
IREP
2.0 E-4
1.2 E-6
1.2 E-5
Peach Bottom (RSS)
NRC
2.9 E-5
1.2 E-6
1.1 E-5
u
I
*Big Rock Point was considered but deemed atypical, since it is a BWR-1 and has a less comprehensive ECCS. GESSAR
(Grand Gulf) was also considered, but was judged inappropriate for this list because it is a BWR-6, among other
reasons.
**Total core melt frequencies reported in the probabilistic analyses used for this study resulted
from internal initiators only. Analyses including external initiators would yield higher core
melt frequencies and, thus, a lower percentage contribution from LOCAs.
***The Shoreham value for "core vulnerable frequency" is believed to be representative of core melt
frequency and has been used here.
D.2
SENSITIVITY MODEL
As described in WASH-1400 (p. 166, Main Report), the LOCA accident

sequence probability is given by
^lE ^ ^SF ^ ^CFM "^ ^WC ^ ^PD
'
^^^
where IE is the initiating event; SF is system failure; CFM is containment

failure model; WC is weather conditions; and PD is population density.
Because (1) is a simple linear product, the only term which is germane is
P I E , which in the case under consideration is the frequency of pipe failure
initiating the LOCA.
P J E may be expressed as
PjE = 1 - Rp
where Rp is the reliability of the plant piping.
(2)
If the plant piping is
considered simply as characterized by a failure rate A, (2) can be written
PjE =1 - e ~^^
^Xt
(3)
assuming Xt is small.
To a first approximation, the LOCA probability is directly proportional

to the failure rate of the reactor coolant system (RCS) boundary.
Assuming
that all other factors are constant, the risk is then linearly proportional to
the RCS failure probability.
As noted earlier, it is sufficient to consider
contributions up through the core damage frequency without losing generality.

The A of (3) does not represent the precise failure rate of the RCS piping for
D-5
several reasons. First, the lack of rigor inherent in representing the entim
reactor complex as a single simple component introduces approximations.
^^
Second, the failure rate is a function of failure definition, which is another

way of describing partitioning by break size as discussed earlier. Neither of
these concerns would negate the use of Eq. (3) as a sensitivity model or a
parameter to examine the data base from which it was derived. Utilizing
linear proportionality, it can be noted in the last column of Table D.l that
increasing the LOCA frequency by a factor of 10 is required in most cases,
before the LOCA contribution is a significant fraction of the total core melt
frequency.
An NRC internal study (D.2) of a similar sensitivity nature, which also

considered transient induced LOCA, multiple pipe breaks, and seismically
initiated LOCA, concluded that significantly large increases in frequency of
pipe failure would be required before LOCAs would be dominant contributors to
risk.
It can be concluded that:
(a) the LOCA contribution in BWRs has not been
found to be a dominant contributor to public risk in PRAs performed to date,

and (b) that a substantial error in the frequency of pipe failures would be
necessary before LOCAs could be considered a substantial contributor to public
risk.
These conclusions may vary on a plant-specific basis; however, current
PElAs (like Grand Gulf) indicate that there is considerable conservatism in the
LOCA contribution to risk in BWRs.
The next step is to evaluate these failure
rates and the data base(s) from which they are derived.
D-6
D.3
LOCA FREQUENCIES
The failure rates in current use for most studies appear to stem from the
data base treatments in the RSS, WASH-1400, where the LOCA failure rate was
partitioned into three contributions, apparently dictated by the mitigating
capabilities of the Engineered Safety Features (ESF).
This partitioning of
the failure rate may be somewhat artificial with respect to the data base, but
similar partitioning has been followed in most of the studies to date.
WASH-1400 partitioning is shown in Table D.2.
The
The range can be viewed as
associated with the uncertainty in the parameter.
The uncertainty is large at
the 90% confidence level, at least two orders of magnitude for each LOCA
class.
D.3.1
Large LOCA
The failure frequency for large pipes resulting in a large LOCA is

perhaps the most understandable of the LOCA failure frequencies because it
represents large openings in the primary pressure boundary up to and including
double-ended guillotine rupture of pipes of 8-1/2 in. equivalent diameter or
larger.
The data bases of references D.l and D.3, in addition to the License
Event Reporting (LER) system regarding piping breaks, were reviewed.
No
ruptures of this magnitude have occurred in the primary system in the approximately 700 LWR years amassed through November 1983 for the U. S. commercial
nuclear power industry.
A careful use of definitions is necessary here.
First, there have been large pipe failures in the secondary system of PWRs as
noted in reference D.3.
However, the effective opening of the failure must
correspond to an equivalent 8-1/2 in. pipe size to be consistent with the

^^reatment in reference D.l, not simply any failure in a pipe of that size.
D-7
Table D.2**
RSS Pipe F a i l u r e Assessed Values
LOCA I n i t i a t i n g
Rupture Rates
for P l a n t Per Year
WASH-1900
Pipe Rupture Size
(inches)*
90% Range
Median
1/2 - 2 1/2
Ixio"^ -- IxlO"^
IxlO"^
3xl0"^ -- 3xl0~^
3x10
1/2-6
Ixio"^
>6
-4
Ixio"^
IxlO"^
*Rupture size area equivalent to pipe area of this diameter.

**Reference D.l.
When the RSS was performed, 17 reactors were selected for the collection of failure data during one year of operation (1972).
Since 17 reactor-
years of operational experience do not provide an adequate data base, a pooled

data approach was adopted which utilized data bases from other nuclear
operations, foreign nuclear operations, and diverse industrial experience.
[See references D.l and D.3 for comprehensive discussions of this procedure
and the resulting failure rate statistics.]
There was no attempt, and
probably no need, to partition the relative contribution by cause or differentiate by explicit reactor type within the water-cooled reactor category.
For the demands of the RSS, the pioneering nature of its objectives and
its relation to the existing experience base, these procedures were
appropriate.
However, treatment for LOCAs has been used with essentially no
modification in the BWR PBlAs reviewed.
This is true for LWR PRAs in general
D-8
(see reference D.4).
Pooled data have the advantage of providing an estimated
^^âilure frequency with less statistical uncertainty.
Pooled data have the
disadvantages of contaminating the data base by mixing samples from disparate

populations and the inability to partition the resulting failure rates by
reactor type and cause.
The cross-contaminants appear to be of roughly two
broad and overlapping categories:
(1) Phenomenologically induced failures
predominantly associated with distinct populations, e.g., IGSCC in BWRs and

other corrosive processes in the chemical industry; and (2) differing quality
control and assurance standards.
Both the advantage and disadvantages are
significant problems in their own right and are not directly related to the
objectives of this study.
Discussions of data base concerns are in many of
the cited references, but references D.l, D.3, D.5, and D.6 present more
direct discussions of these problems.
The PCSG was confronted with the problem
of the existing PRA methodology
(pooled data) and its relevance to the population of completed PRAs opposed to
IGSCC and its occurrence in BWRs.
The following argument relating to a
postulated error in the large pipe failure rate is predicated on 273 reactor
years of U. S. commercial BWR experience through November of 1983. This
approach has been adopted to avoid any basically irrelevant arguments about
the data base quality.
An estimate of failure rate can be calculated by

''
A =
-~
(4)
where A is the estimate of the failure rate or point value, r is the number of
failures, and T is the sum of the operating times accumulated by all units.
^J^here there are no failures, a realistic point estimate of the frequency may
^ ^ e obtained from
D-9
^
2
A = X 0.50i 2r + 1
(^\
2f
It can be shown that the unknown failure rate lies in the interval
X^ a/2; 2r
2f~
< X <
which has a confidence level of 1-a.
x^ l-a/2; 2r + 2
2T~
(6)
It must be noted that Eq. (6) does not
establish what the true failure rate is, only that it lies in the defined
interval with probability 1-a.
The best estimate of the failure frequency is
the point value calculated by Eq. (4). It may also be shown that the median
value of the chi-squared interval may be calculated by Eq. (5) regardless
of the number of failures. Both the point and median value results have
been included in Table D.3. Both methods produce comparable estimates.
Equation (6) is calculated in Table D.3 for 7 hypothetical number of
failures.
The apparent increase in IGSCC in BWR piping raises the question:
Is it
likely that the failure frequency might be higher by a statistically substantial amount than the value currently used for large LOCA failure rates?
From
Table D.3, the implication can be drawn that if the failure frequency were in
error, by an order of magnitude, several failures would be expected to have
occurred.
This might lead one to conclude, therefore, that there is little
significant evidence that an error of that magnitude exists in the failure

frequency regardless of the ostensible cause. However, there exists the (statistical) possibility that the failure rate could be in error by an order of
D-IO
Table D.3
Hypothetical Failure Frequency Estimates*
(Total U. S. BWR Operating Time (T) = 273 Reactor-Years)
95% Confidence Limits
pothetical* Chi-Squared
bserved"
Table Value
ilures (r) L
U
Failure Frequency
Point Value
(IQ-^)
Chi-Squared
Confidence Limits
7.38
Chi-Squared
Median
(10-3)
-2
1.35x10
0.84
0.05
11.1
3.7
9.15x10"^ -
1.4x10"^
4.4
0.48
14.4
7.4
8.8x10"^ -
2.6x10"^
8.0
1.24
17.5
11.1
2.2x10 ^ - 3.2x10"^
11.6
2.18
20.5
14.8
4.0x10"^ - 3.8x10"^
15.2
6.91
31.5
29.6
1.3x10"^ -
5.3x10"^
29.8
10
9.59
36.8
37.0
1.8x10"^ -
6.7xl0~^
37.2
;r0.025; 2r
X
= x' 0.975; 2r + 2
icall, no large-pipe LOCA have been reported for primary systems.
D-U
magnitude and yet no failures have been observed in the interval T.

be seen by the use of standard reliability concepts.
This can
The reliability of a con^^B
ponent represents the probability that there will be no failure during the
interval T.
If the reliability of a component population n is given by

R(n) = e n^ .
(7)
The ratio of two component populations is then given by
If A
= 10 A
R(l)
_ (^2"^^^
K(2)
~ ^
(8)
Eq. (8) reduced to

R(l)
R(2y -
5AjT
(9)
^
_3
A^ = 10 ; then
Assume that T = 273 reactor years, and

R(l) ~
R(2T "
2.46
^
~
-
,
12
However, if the total number of reactor years in the U. S. comercial

program are used (representing pooled U. S. data), it is 500 times less likely
to produce 700 failure-free years of operation for an order of magnitude underprediction error in the failure rate.
Further, if worldwide reactor data are
pooled (representing approximately 1000 failure-free operational years), it

becomes even more unlikely.
It can be noted that this argument is sensitive
to the absolute value of the failure rate.
However, the thrust of the
argument is that there is little statistical support for the hypothesis that
there is a significant error in the failure rate even though the presence of
IGSCC may point to a greater failure precursor population.
Statistically, it
has already been pointed out that the true unknown failure rate lies in a wide
uncertainty band, in this case at least two orders of magnitude in length.
D42
what is really being asked is whether there is any evidence that there is a
^^k
significant difference in the medians of the two populations.
Since there
have been no failures of large pipes, it can be statistically concluded that

there appears to be no evidence of any difference, at least in a bounding
sense.
It might be tempting to conclude that, since the frequency of large LOCA

failures does not appear to be increasing, IGSCC has not been an over-looked
factor in large LOCA failure rates.
However, there are no data to support
this conclusion directly since in-service inspection and repair/replacement

may actually be responsible for avoiding large LOCA failures and reducing the
failure rate (reference D.7, pp. 199-200).
Although more reactor years of operation continue to be amassed, it would

be improper to conclude much beyond that the failure rate is in about the correct range consistent with the data.
The failure rate could be much lower,
for example, but the physical situation would appear to impose bounds at about
1x10"^ failures per year as a minimum failure rate that could be statistically justified on the basis of the following bounding argument.
Assume that
in a few years there will be about 100 commercial nuclear power plants in
operation and they all contribute to the data base for their full 40-year
design life.
test time.
This would produce only 4000 reactor years of total amassed unit
Perhaps that could be doubled by the world experience, which may
or may not be appropriately comparable data.

comparison is only A = l/8000"'lxl0-4.
The point estimate for this
The implication, of course, is that
low failure rates are difficult to measure without an extremely large number
of units on test to accelerate the required large test time.
^ ^ that one can be achieve^d is a (l-oi) upper bound.
D-13
The best result
D.3.2
Intermediate and Small LOCA Failure Frequencies
(
Theoretically, it should be possible to construct a plot or table of
failure frequency against break size which will range from guillotine large
pipe rupture to hairline crack.
However, the discussion on large LOCA failure
frequency has demonstrated at least one of the practical limitations encountered when attempting to attain these parameters.
It may be compounding the
uncertainty on failure frequency by trying to examine a partitioned (large,

intermediate, and small LOCA) failure frequency when attempting to determine
if there is an unaccounted for effect of IGSCC in the failure rate.
Because the absolute value is not of interest for the failure frequency
for any failure size, but only the relative comparison of the causes contributing to the failure frequency, it is more important to examine the data base
rather than the partitioned failure frequency.
The examination of the parti-
tioned failure frequency is done thoroughly in a recent Canadian Study (D.3)

which is discussed in the next section.
D.3.3
Pipe Failure Data Base
Since publication of the safety study WASH-1400 in 1975, there have been
few systematic reviews or updating of the pipe failure rate data base.
The
timing and resource limitations of this study did not permit a full and
detailed updating of the data base, which was inconsistent with the committee
objectives.
However, it was possible to review available studies and the
U. S. nuclear data base as represented in the LERs available in computerized

data banks of the current Sequence Coding and Search (SCSS) data base at ORNL.
i
D-14
The most comprehensive study of the U. S. nuclear pipe failures, as

represented by LERs, was done by the Canadians and reported in reference D.3.
This represented the experience base in commercial U. S. nuclear power plants
through December 31, 1978, for a total of 409 reactor years of operation.
Although it was not possible, to extend the in-depth analysis performed by the
Canadians, it was possible to review the reported pipe events and their causes
between 1978 and November 1983 to see if any trends in causes could be
obtained.
Reference D.3 noted the difficulties with analyzing the data base, as did
S. Bush in a forthcoming publication (D.5).
It must be realized that the data
base results from conditions far from controlled experimental practice.
The
first problem, common to both the event recorders and event analysts alike, is
common definitions for "failure," "rupture," and "severance." There are other
conditions discussed later which may well make the determination of true
failure rate highly uncertain.
However, these data (presented in Table D.4)
do represent the best available information and, with the limitation clearly
in mind, should provide insight into the ratio of causes.
Only two major contributors to failure have been separated from the
total:
vibration/fatigue and erosion/corrosion.
Although there does not
appear to be a progressive increase in the category designated erosion/corrosion (an abbreviated category as analyzed in reference D.3) which includes
IGSCC, there probably is little statistical significance for the following
reasons:
a) improved reporting requirements may well have some impact on the
results; b) there are a large number of events with causes unknown; and c) in
D-15
Table D.4
Fatigue and Corrosion Failures

As a Function of Total Reported Failures
(U. S. Commercial Nuclear Power to Nov. 1983)
To 12/78
Total Events
12/78 to 10/81
804
10/81 to 11/83
138
98
Vibration/Fatigue
10%
25%
24%
Erosion/Corrosion
13%
15%
22%
Operating History (Rx-yr) 409
540
700
some events the causes are speculative because reporting deadlines do not
allow sufficient time for analysis to specify the cause accurately.
Failure
analysis which lags the LER submission is rarely updated and correlated with
revised LER.
In addition, a sizable number of IGSCC events are uniquely
associated with the PWR boron injection system.
Therefore, there is large
uncertainty associated with the pipe break causes. Further, the uncertainty
in the failure frequency as inferred from the chi-squared test is very large
by comparison with any trend in the erosion/corrosion contribution. To
attribute statistical significance to a 25% change in a partitioned component
of a failure rate which has two orders of magnitude uncertainty in its value
cannot be statistically supported.
It should be noted that the removal
of incipient failures by UT or other in-service inspection and subsequent

repair is not included in these data.
D-16
^^D.3.4
Leak-Before-Break
The small LOCA failure rate represents a difficult problem.
In general,
defects including IGSCC have a greater likelihood of forming a small opening

(crack or hole) and leaking well in advance of a major failure (break).
If
leak detection systems function properly, this leakage is detected and corrected well in advance of becoming even a minor problem.
This is generally
referred to as the "leak-before-break" concept and appears to be borne out

both in theoretical calculations (reference D.7, p. 196) and in the field data
(reference D.3, p. 68). However, the methodology defined by reference D.l,
and apparently followed in most other PRAs, defines the BWR small-break LOCA
as less than 2-1/2- and greater than 1/2-in. equivalent pipe diameter.
Most
failures recorded are very difficult to categorize, at least with any

precision as to size, because of the reporting system.
The leak-before-break may theoretically precede any size break and the
leak itself could be large enough to qualify for a small or intermediate size
break (although, this is unlikely assuming the leak detection system is functioning properly), but the quality of the data base is inadequate for one to
make such distinctions.
Because of the data base quality and the artificial
division of break sizes, no conclusions about absolute failure rates will be

attempted.
However, inferences about relative causes for failures can be
estimated even though the data base may be uncertain for absolute failure
rates.
D-17
D.4
SYSTEMS RISK ASSESSMENT
D.4.1
Plant Model Selection
There are approximately 40+ BWR Plants licensed to operate or in the

licensing process.
As the BWR population has grown, the reactor manufacturer
has introduced improved models in his product line so that there are currently
6 classes of BWRs in the 40+ plant population.
In order to draw generic
conclusions from a study of this sort within the resource and time limits
available, it is necessary to select a characteristic plant for study.
Gen-
erically, this is not too difficult because the type and number of systems
have not changed markedly for the last three models (BWR 4-6) which comprise
about three fourths of the population.
BWR-48.
Roughly half this population are
Therefore, a BWR-4 was selected as a model.
This is also consistent
because the RSS selected a BWR-4, Peach Bottom-2, for the BWR risk assessment.
The methodology for PRA developed in the RSS, at least in the LOCA sequences,
has been used in all other plant-specific BWR PRAs reviewed, which include
Peach Bottom-2 (RSS), Millstone 1, Browns Ferry, and GESSAR (BWR-6).
The
plant selected as a model was the Georgia Power Company's Edwin I. Hatch
Nuclear Power Plant Unit 2, which went into commercial operation in 1979.
The Hatch-2 plant is also significant because Georgia Power Co. has elected to
replace their recirculation system with a new system, using nuclear grade 316
stainless steel, which is designed to reduce the number of welds in the
system.
D-18
From a systems standpoint, the changes which have occurred during the
development of the BWR-3 through -6 have caused only nominal impact on the
number and type of piping systems significant to safety.
The BWR-3 was the
first large direct cycle system utilizing jet pumps internal to the reactor
vessel and improved ECCS, including core spray and flood capability.
BWR-4 was much like the BWR-3 with increased power density.
The
The BWR-5 and -6
product lines incorporated flow control, changes in fuel design, and changes
in pressure suppression/containment design.
The BWR-4 also incorporates the
major primary engineered safety and auxiliary systems followed in later designs.
This, in conjunction with the fact that more BWR-4s are operating
than other models, makes the selection of a BWR-4 the logical choice.
D.4.2
Previous Pipe Crack Study Groups
The efforts of previous pipe crack study groups (PCSGs) were reviewed.
Each PCSG emphasized certain areas according to its charter, the field
experience to that point, and the resources marshalled for the study scope.
In some cases, pipe cracking in both PWR and BWR reactor types (D.8) was
considered and in others only one reactor type (D.9 and D.10).
The most
recent PCSG focused on the cracking implications in PWRs (D.10).
In its study
(D.10), the PCSG placed major emphasis on systems safety implications of

cracking, essentially auditing the deterministic licensing methods in use at
that time.
The systems approach for the current PCSG centers on the IGSCC
implications in BWRs and its risk implications.
Therefore, the systems
assessments are made in the framework of the RSS risk methodology.
D-19
This
reflects the trend of assimilating risk methods into the licensing process.
The RSS methodology, in this context, has some powerful attributes and it also
has some deficiencies.
Examples of both strength and weakness will illustrate
this point.
Risk methods allow an evaluation beyond the deterministic concepts of

single failure and redundancy.
The risk methodology allows not only a deter-
mination of significance to safety, but also quantifies what is meant by the

statement.
On the other hand, risk methodology is statistical in nature
and is applied in the aggregate to a plant.
This becomes very difficult when
trying to partition the risk to the subsystem level such as pipe failure
frequencies.
In its global concept, it also yields such things as what
action(s) are necessary to retain that importance to safety throughout the

life of the plant with respect to test, maintenance, and operator actions in
emergencies.
When the RSS was published, the peer review produced much comment about
the intrinsic large uncertainties associated with the methodology and the
unreliability of the absolute values.
These are two well-publicized diffi-
culties with risk assessment techniques.
In applying these techniques to the
cracking of pipes, it was difficult to partition the risks to both phenomenon

and subsystem.
For example, the failure frequencies for pipes normally in-
clude all causes:
fatigue, corrosion, erosion, design defect, etc.
Determi-
nistically, existing procedures and approaches are intended to reduce the

causes of failure to an acceptably low level.
Statistically, however, it is
necessary to quantify these very infrequent events.
D-20
In order to gather a data
base on these failure frequencies with the lowest uncertainty possible, the
events are aggregated with no attempt to develop failure frequencies for each
cause.
Finally, the LOCA-sensitive piping is lumped together in only two categories,

large and small, and not by subsystem.
This is primarily for convenience; but
to make relative comparisons, the studies made under existing methodology must
be used.
As PRA becomes a more acceptable tool, a consistent LOCA initiation
probability by subsystem piping within the RCS may be made.
The following discussions of the systems implications of risk assessment

will rest heavily on the Hatch-2 plant.
It should be kept in mind that these
discussions cannot be taken in isolation from other sections of this report

and that certain significant plant specific deviations from these discussions
may exist.
D.4.3
Systems Evaluations
Up to this point, attention has been directed primarily to the initiating

event in the reactor accident sequence.
In the cases of IGSCC (or other
contributors), the pipe failure or rupture is considered to be the initiating

event without regard to precursor mechanism.
This has been common PRA
methodology from the RSS through current risk assessments reviewed.
It is not
within the scope of this study to argue either for or against the validity of
this initial condition, but for most conditions pipe failure is initiated by
the imposition of some pressure, temperature, or mechanical load which, in
turn, will cause the defective pipe to fail.
This approach implies that all
initiating causes are included when the consequences of pipe failure are
D-21
calculated using the piping failure frequency.

loss of coolant.
The pipe failure leads to a
If the failure of the pipe has certain characteristics,* a
loss-of-coolant accident (LOCA) occurs.
The systems of concern for a BWR are listed in Table D. 5, which is

extracted from NUREG-0531 (D.9) and pertains to a large BWR-4 as noted
earlier.
The most severe LOCA, in terms of peak clad temperature (when the ECSS
works) is the DBA.
The BWR DBA is a double-ended guillotine (DEG) severance
of the suction line of B main recirculation pump (D.l).
The BWR is designed
to be able to sustain the DBA without unacceptable damage to the reactor fuel
and bring the plant to a cold shutdown by use of the ECCS and long-term heat
removal capabilities.
The ECCS is an integrated group of systems described in
the FSAR of all plants and is designed to provide coolant makeup for a variety
of LOCA conditions up to and including the DBA.
the ECCS include the following:
The systems which comprise
High Pressure Coolant Injection (HPCI),
Automatic Depressurization System (ADS), Core Spray (CS), and Low Pressure
Coolant Injection (LPCI).
Briefly, each system will be discussed to illus-
trate how these systems provide adequate coolant makeup for any LOCA
including the DBA.
In order to focus the reader's attention. Figure D.l is a
composite plot which shows the LOCA size, flows, and leak detection limits as
taken from the appropriate documentation (D.ll).
The following LOCA
descriptions pertain to Figure D.l.

*For generic discussion, it is not necessary to stipulate what all these
characteristics are; for example, that if the leak is large enough, it may go
undetected, etc.
D-22
Table D.5
List of Typical Piping Systems Involved in the Operation and/or

Safe Shutdown of BWRs*
Number
Function
Reactor Recirculation
Main Steam
Feedwater
Reactor Water Cleanup
Reactor Core Isolation Cooling
Core Spray
Residual Heat Removal
Containment Spray
Reactor Head Spray
10
Standby Liquid Control
11
High Pressure Coolant Injection
12
Low Pressure Coolant Injection
*This list is taken from the FSAR of Peach Bottom Units 2 and 3.
D-23
LARGE
LOCA ( A)
SMALL
LOCA (S|)
..<^
..<<
.<^
^^y.^<^y
SMALL
LOCA (Sg)
^^
<^-
UNIDENTIFIED LEAK
DETECTION LIMIT
0.01
Figure D.l
0.1
I
10
CRACK OPENING AREA, A(in.2)

1
I
I
L_
0.5
I
2
4
PIPE DIAMETER (in.)
100
10
20
Leak rate as a function of crack opening area or pipe

diameter (assuming a full circumferential break).
D-24
For small-small LOCA (S2), the coolant loss can be made up by non-ECC
systems of feedwater and/or control rod drive cooling water until the plant
can be shut down and the leak located and repaired.
For small LOCA (S^), the HPCI system will be activated to supply
coolant until the plant can be secured and the leak found and repaired.
A large LOCA (A) is of sufficient magnitude that adequate coolant level

cannot be maintained (automatically sensed and initiated by low reactor vessel
water level or high drywell pressure), and the ADS reduces the reactor vessel
pressure through selected safety relief valves so that coolant can be supplied
by both the LPCI and/or core spray.
The ECCS is designed to cool the reactor core for LOCAs which range over
the entire spectrum up to and including the DBA.
After the initial phases of
the LOCA (as well as many other events), the long-term cooling is accomplished
by the Residual Heat Removal (RHR) system.
The RHR system is a multi-
functional system which includes the LPCI mode of operation.

other RHR functional modes are:
Briefly, the
1) steam condensing and 2) suppression pool
cooling, in addition to the shutdown cooling and LPCI functions already

noted.
The RHR systems functions are shown in Figures D.2 through D.7 for a
BWR-6, Mark III containment.
For the purpose of this discussion, there is no
functional difference compared with a BWR-4, Mark II containment.
The steam
condensing function is unimportant to this discussion because the possible

IGSCC-LOCA concern deals with systems using 304 stainless steel.
piping in all plants reviewed is low-carbon steel.
D-25
The steam
It is of interest to note.
1 CONTAINMENT
7 SUt-PfiESiiON ÔOL
T CONTAINMENT
2 ORVWELL
3 MAIN STEAM
9 l U H B i N E D R I V E N M A K E U P PUMP
FEEDWATER
SAfETV/BELttr
TURBINE
DRVWEUL
6 SUPPRESSION K X ) L
1 HEAT EXCHANGERS
I
10 TURBINE E X H A U S T
4 SYSTEM PUMPS
11 CONDENSATE EHOM C O N D E N S A T E STORAGE

TANK OH RHR HEAT E X C H A N G E R
5 FULL FLOwr e v P A S S
Figure D.2. Reactor core isolation

cooling system.
Figure D.3.
D-26
C I C PUMP S U C T I O N
9 SERVICE W A T E H
Residual heat removal

system.
"l-ir-^
1 CONTAINMENT
5 S Y S T E M PUMP
2 DRYWELL
HEAT E X C H A N G E R S
3 HPV
7 SERVICE W A T E R
1 CONTAINMENT
2. D R Y W E L L
7 PRESSURE R E D U C I N G SYSTEM
3 M A I N S T E A M L I N E TO T u H B I N E
SUPPRESSION X i L
4 SAFE T Y / R E L l i F
TO RCIC PUMP S U C T I O N
& M A N U A L REMOTE C O N T R O L
VALVE
4 SUPI R E S S I U N POOL
F U L L FLOW BYPASS
10 SERVICE WATER
Figure D.5. Residual heat removal system,

steam condensing function.
Figure D.4. Residual heat removal system,

shutdown cooling function.
D-27
1 CONTAINMENT
4 SYS I EM PUMP
5 S u P P H E S i i O N POOL
1 CONTAINMENT
6 SUPPHESSION POOL
e C O N D E N S A T E STORAGE
I N I T I A L SOURCE
DRYWELL
3 SPRAY SPARGER
DHVWELL
3 RPW
Figure D.6.
7 S Y S T E M PUMP
4 RPV
6 HEAT EXCHANGERS
Residual heat removal system,

Pressure coolant injection
function
D-28
Figure D.7. High pressure core spray

system.
however, that a recent study (D.12) which sampled a five-year window from 1976
through 1980 for loss-of-coolant events found that these events were dominated
by the main stream relief valves.
These events constitute a subclass of LOCA
initiators which are included in the failure frequencies by their valve seat
area equivalent pipe diameter.
These are generally in the RSS category of
"small" breaks (between 1 and 8-1/2-in. in diameter).
For the sake of completeness, we include another system associated with

the BWR called the Reactor Core Isolation Cooling (RCIC) system (see Figure
D.2).
The RCIC system maintains sufficient water in the reactor pressure
vessel to cool the core and maintain the nuclear boiler in the standby
condition in the event the vessel becomes isolated from the turbine steam
condenser and from the feed-water make-up flow.
The RCIC operates in
conjunction with the steam condensing mode of the RHR.
It is not necessary
to consider these modes of functioning of the RHR further because they are
unimportant to the LOCA sequence.
Reviewing, first, the field experience of IGSCC, some generalizations are

possible.
IGSCC has been found, primarily, in the two recirculation loops.
Until 1978, IGSCC had not been discovered in large-diameter piping.
Since
that date, IGSCC has been found in the large-diameter piping in numerous
plants. IGSCC appears to be more prevalent in small-diameter piping. However,
there appears to be considerable variation in this generality due, possibly,
to the number of variables affecting the susceptibility and progress of the
IGSCC phenomenon.
D-29
Further, no IGSCC has proceeded to a size sufficient to be classified as

a loss-of-coolant event which would activate the ECCS.
All known events were
detected by either UT inspection or very small leakages (^ 10 gpm) (D.13).

Total avoidance of IGSCC attack is preferable; however, experience would
suggest strongly that both leak detection (or conversely, leak-before-break)
and inservice inspection are effective as components of the defense-in-depth
concept.
The thrust of this discussion is neither leak-before-break nor in-
service inspection, both of which are discussed in Sections 6 and 4, respectively, but to analyze the systems implications of IGSCC.
A review of ref-
erence D.13 and subsequent IGSCC detections suggests that the attack is
centered in the recirculation and associated systems piping with no involvement in the important safety systems in Table D.5, with the exception of the
core spray system.
The reactor water cleanup system has been found to be
heavily involved (D.9), but is not considered a primary safety system.
From a risk standpoint, a crack in a pipe can, basically, contribute to

risk in one of two ways.
The crack can be a LOCA initiator or it can disable
an important safety system.* The contribution to the LOCA frequency is

adequately addressed through the data base and application to the LOCA sensitive piping as outlined in the RSS (D.l).
IGSCC has been postulated as a
phenomenon which could produce multiple LOCA events by the Advisory Committee
on Reactor Safeguards (ACRS) (D.14) and others.
Multiple pipe failures as a
result of IGSCC warrant some discussion and will be addressed later.

* Secondary concerns associated with systems interactions and common mode
failure are not being considered here.
D-30
The probability of safety system failure (unavailability) is an integral

part of risk assessment.
The initial and current licensing philosophy is
based on what is usually termed a deterministic approach.
Briefly, it relies
on a thorough mechanistic analysis and evaluation of the nuclear power plant,

as well as the application of the concepts of system redundancy, diversity,
and independence to systems important to safety to ensure their performance.
These systems and subsystems must be capable of performing their functions
even for the most severe single-component failure.
These deterministic
considerations define an envelope termed the DBA spectrum.
Engineering judg-
ment and conservatism are used in determining whether the plant design is
acceptable in meeting the general design criteria for LWRs as published in
10 CFR 50, Appendix A.
Risk assessment quantifies the probability that severe
accidents might occur and what their consequences would be to the public.
PRA attempts to analyze and quantify all aspects of the nuclear plant design,
operation, and maintenance for internal (and external) initiators of severe
accidents which would result in residual risk to the public.
PRA techniques have provided numerous useful applications (D.4), including methods which allow a quantitative measure of the system importance
to safety (risk).
Table D.6, derived from PRA allied techniques (D.15), pro-
vides a qualitative ranking of the LOCA-related ECC systems.
The reactor
protection system has been included for comparison, since it is without question the most important safety system in the plant as demonstrated by this
table.
Risk achievement ratio is defined as the factor by which core-melt
frequency would increase if the system were never operable.
Although this
ranking is for a BWR-6 which might alter the Risk Achievement Ratio values,
the relative ranking for Hatch-2 (the selected model) would not be expected
to be significantly different.
D-31
Table D.6
Risk Worth for Safety Systems* with Respect to Core Melt Frequency (D.15)
Risk Achievement Ratio
System
Reactor Protection System
2.0 X 105
Residual Heat Removal System (RHR)
1500
Standby Service Water System
1500
Low Pressure Coolant Injection (LPCI) System
23
Automatic Depressurization System (ADS)
14
Low Pressure Core Spray (LPCS) System
7.6
High Pressure Core Spray (HPCS) System
4.4
Reactor Core Isolation Cooling (RCIC) System
2.2
*Based on Grand Gulf, a BWR-6, which has HPCS rather than HPCI as in BWR-4 as
well as some minor differences, but safety systems/functions are similar.
Some systems not directly pertinent to LOCA concerns have been omitted from
this list.
D-32
Table D.6 in conjunction with the field data discussed earlier, can be
used to obtain some measure of the systems impact of IGSCC.
IGSCC has been
found in the core spray systems of some plants. Although IGSCC in the core
spray systems should not be ignored, its presence does not increase the risk
by a substantial amount because:
1) the system is redundant, 2) core flooding
may be accomplished by other systems, and 3) a crack in these systems

(assuming leak-before-break) would most likely only degrade the system
performance.
Pipe cracking or failure in a system is generally less severe than the

failure of valves to operate.
with single valve failures.
The ECCS must be capable of performing even
Table D.7 presents the results of the single-
valve-failure analysis for Hatch-2 (D.16).
On the other hand, although the RHR system has experienced little IGSCC
attack, it is extremely important as noted by its risk achievement ratio and
should continue to be inspected for IGSCC or other faults.
The Reactor Water
Cleanup (RWCU) system has had a considerable number of IGSCC events. That
part of the RWCU system that cannot be isolated from the recirculation system
has been included in the LOCA-sensitive piping.
only nominally to risk in any other respect.
D-33
The RWCU system contributes
Table D.7
ECCS Single-Valve-Failure Analysis
Consequences
of Valve Failure
Assumed Together with
Position for
Normal Plant
Operation
System
Core Spray
High pressure
coolant injection
Valve(s)
Closed
Opened
Suction
Negate use of the

core spray loop
Injection(s)
Negate use of one

core spray loop
Test return
Negate use of one

core spray loop
Condensate suction
Use suppression pool

water
Use condensate storage
tank water
Suppression pool
suction valve
Suppression pool
test return
Low pressure
coolant injection
Partial flow loss

due to flow to
suppression pool
Injection(s)
Negate HPCI
Turbine inlet(s)
Negate HPCI
Injection(s)
Negate use of LPCI
Minimum flow
Automatic depressurization system

system
Design Basis LOCA
Partial flow loss in

one loop due to flow
to suppression pool
Test return
No consequences
Pump suction
Negate one out of four

pumps
One relief valve
Vessel depressurizes
faster
D-34
D.4.4
Multiple Pipe Failures
The concern with multiple pipe failures stems principally from the possibility that IGSCC damage could be present in more than one location and in
more than one piping system.
If the plant were to experience a severe loading
event, such as an overpressure transient or an earthquake, the plant loading

event might lead to leaks or breaks at more than one location at essentially
the same time.
A reduction of safety margins of this type is clearly undesirable if it

can be prevented.
However, from a risk standpoint and for this discussion, it
must be assumed that the possibilty exists to some degree.

consider that likelihood.
ties:
The following will
The chief concern may be limited to two possibili-
a) that a LOCA larger than the DBA results from a multiple failure or
b) that a multiple failure causes not only a LOCA, but the second break
disables an ECC system.
The concurrent LOCA DBA and the disabling of an ECC system is very
similar to the single failure criteria already required of the ECCS discussed
earlier.
As pointed out in that discussion, this induced pipe failure is no
more severe than a valve or pump failure, conditions already included in both
the deterministic and PRA assessments.
Gerber and Garud (D.ll) point out that
ECCS redundancy leaves adequate system coverage for this concern and discuss
the probability of multiple pipe failures.
They note that considerable margin
exists both in ECCS design and conservatisms in peak clad temperature calculational methods.
D-35
with respect to the probability of multiple failures, Gerber and Garud

present a statistical argument which suggests that the probability of two
concurrent failures lies bounded at most by the probability of the least
probable single failure and is at least equal to the product of both single
failure probabilities.
Reference D.ll includes a lengthy discussion of the
statistical dependence and the correlations of variables involved which will

not be repeated here.
As an example of estimating the probability of two concurrent pipe

failures, assume that each break has the same single failure rate distribution [i.e. P(fi) = P(f2) = P(f)].
With this assumption, the joint
failure probability can be reasonably approximated as a lognormal distribution.
Utilizing the statistical methods and approach of the RSS (D.l) and
the above Gerber-Garud statistical argument regarding statistical correlation, the median of the joint probability for two concurrent failures can be
expressed as
Hf,,f^)
'î^
= TpTf) X p2(f) =yi^x^^ ^ ^^^^ .
^^^^
where fj and f2 represent failures 1 and 2, U is the upper bound [P(f)]

and L is the lower bound [p2(f)], F L is the fraction of large pipe which
is LOCA sensitive, and ^Lp is the large pipe failure rate.
in the RSS are FL = 0.047 and
^LP = 1 * 10~^.
The values used
Substituting these
numbers into (11) results in
Hi^,t^
= 2 X 10"^ P(f) ,
D-36
(12)
which indicates that the contribution to the core melt frequency for concurrent or multiple large pipe ruptures is about 3 orders of magnitude less than
that associated with a single large pipe rupture.
Although these arguments are approximate in nature, they indicate that

the risk contribution of multiple large failures associated with IGSCC is very
likely several orders of magnitude less than that of a single large failure.
Approximate though these arguments may be, they are probably much better than
would be obtained if one tried to extract a joint probability from the current or foreseeable data base.
However, there are statistical codes on re-
liability, such as PRAISE (D.7), with which this approximation may be

concurred.
D.5
REFERENCES
D.l
NRC.
October 1975.
Reactor Safety Study.
WASH-1400, NUREG-75/014.
D.2
T. P. Speis, to R. H. Vollmer. Memorandum dated January 6, 1984.

"Briefing on BWR Pipe Crack Issues."
D.3
P. Janzen. April 1981.

Nuclear Power Reactors.
D.4
NRC. February 1984. Probabilistic Risk Assessment (PRA):

Status
Report and Guidance for Regulatory Application.
NUREG-1050.
D.5
S. H. Bush. To be published. "Statistics of Pressure Vessel and Piping

Failures for a Decade of Progress in Pressure Vessel Technology."
D.6
S. H. Bush. October 1977. "Reliability of Piping in Light Water

Reactors," IAEA-SM-218/11, Vienna.
D.7
P. 0. Harris, E. Y. Lim, and D. D. Dehia. August 1981. "Probability of

Pipe Fracture in the Primary Coolant Loop of a PWR Plant," Vol. 5:
Probabilistic Fracture Mechanics Analysis.
NUREG/CR-2189. Lawrence
Livermore Laboratory.
A Study of Piping Failures in the U. S.

AECL-Mi8c-204.
D-37
NRC. October 1975. Investigation and Evaluation of Cracking in

Austenitic Stainless-Steel Piping of Boiling Water Reactor Plants.
NUREG-75/067.
NRC. February 1980. Investigation and Evaluation of Cracking

Incidents in Piping in Pressurized Water Reactors.
NUREG-O531.
10
NRC. September 1980. Investigation and Evaluation of Cracking

Incidents in Piping in Pressurized Water Reactors.
NUREG-0691.
11
T. L. Gerber and Y. S. Garud. December 1981. Significance of BWR

IGSCC Pipe Cracks Within the Primary Containment and Assessment of
Multiple Pipe Failures.
EPRI NP-2183-LD.
12
D. E. Baxter, et al. (EG&G/INEL). August 1982. ."Loss of Coolant Event

(LOCE) Data Base Study." Draft, EGG-EA-5989. DOE.
13
NRC. August 1980.

NUREG-0679.
14
M. Carbon (Chairman, ACRS) to J. Hendrie, (Chairman, NRC). Letter dated

August 16, 1979, regarding pipe cracking in light water reactors.
15
NRC. July 1983. Measures of Risk Importance and Their Applications.

NUREG/CR-3385, Battelle Columbus Laboratories.
16
Edwin I. Hatch Nuclear Power Plant Unit Number 2. Final Safety

Analysis Report.
Georgia Power Docket Number 50-366.
Pipe Cracking Experience in Light-Water Reactors.
D-38
APPENDIX E
This appendix describes each postulated IGSCC mitigation scenario and

associated assumptions.
It presents the unit ORE and dollar costs for the
various inspection and mitigation procedures considered.
It also discusses
the analysis of each factor analyzed.
E.l
IGSCC RESOLUTION SCENARIOS
The general assumptions as well as the activity sequences and assumptions

associated with each postulated scenario are described below.
E.1.1
General Assumptions
Several assumptions related to plant design and operations are needed to

complete the analysis.
For this purpose, it was assumed that BWR plants are
shut down for six weeks to refuel once every 18 months and that the plants
undertake every third refueling cycle a major maintenance activity such as
turbine-generator inspection/maintenance or main condenser tube replacement
which lasts three months.
It was further assumed that the average remaining
life of operating plants is 25 years (E.l).
The assumed average number of
welds in sensitive material in a typical operating BWR is shown in Table E.l.
E-1
Table E.l
Average Operating BWR Plant - 304 SS Piping Welds >_ 4 in.*
Total
Welds,
System
Av.
Recirculation
98
Recirculation
Bypass
8 @ 4 in.
Core Spray
9 @ 12 in.
No./size
42-(22-28 in.)
RHR
25
10-(20-24 in.)
RWCU
20
160
19-(4-10 in.)
No./size
56-04-12 in.)
15-(6-20 in.)
*Average number per plant and size of 304 SS pipe welds (4 in. and up) derived
by averaging information from IE Bulletins 82-03/83-02 inspection reports and
from GECo for 22 operating BWRs.
E.1.2 Scenario 1 - Pipe Replacement
1. Operating Sequence.
piping is replaced.
This scenario assumes that all sensitive
HWC was considered as an option. Pipe replacement was
assumed to require the following activities.
a.
Decontaminate and remove old pipe and replace with less sensitive
material.
b.
Stress-relieve field welds using IHSI process or the equivalent

treatment.
E-2
c.
Perform UT examination to establish a baseline for subsequent

inspections.
d.
2.
Resume operation.
Plant Outage.
Total plant outage for replacing all 304 SS piping
(greater than 4 in.) inside the primary pressure boundary is estimated to last
6-9 months.
Pipe replacement is assumed to be scheduled concurrent with a
refueling outage of 6 weeks or with other major maintenance outage such as

turbine-generator inspection/maintenance or main condenser retubing lasting 3
months.
Thus, the incremental plant outage time chargeable to pipe replace-
ment was assumed to range from 3 to 7.5 months with a best estimate of 4.5
months.
Cost for replaced power is assumed to be S300K per day.
3. Reduction of Pipe Welds.
As a result of pipe replacement, the
average number of pipe welds inside the primary pressure boundary is assumed
to be reduced from 160 to about 120 welds per plant subject to postreplacement ISI.
The basis for this assumption is:
a) As part of pipe replacement, plants plan to eliminate the Recirculation system discharge valve bypass piping which eliminates 8- to
4-in. pipe welds.
b) Use of design improvements in replacement pipe fittings will eliminate 30-35 (assume 32) recirculation system pipe welds of which
20 are 12 in. in diameter and the remainder are 22 to 28 in.
in diameter.
The improvements include features such as use of
E-3
bent 12-in. pipe instead of welded elbows; elimination of 22-in.

welded end caps on riser headers; use of integral long tangent
28-in. elbows instead of welded tangents.
4.
Post-Replacement Inservice Inspection.
Plants that replace 304 SS
pipe with acceptable IGSCC-resistant pipe material were assumed to be allowed

to reduce their inservice inspection (ISI) to normal ASME Section XI requirements.
The Code provides for inspection of 25% of pipe welds over each 10-
year period of plant operation.
On an 18-month refueling cycle, there would
be 6.67 refueling cycles each 10 years, which implies an average inspection

rate of 3.75% of the welds each refueling.
However, the Code does not re-
quire that inspections be performed each outage.
For this pipe replacement
scenario, it was assumed that at least 4% of the welds would be inspected in

each of the first two outages.
The frequency of inspection was then assumed
to follow the Code requirements for the remainder of the plant life.
In-
spections were assumed to be performed during a scheduled outage at least

once every 3-1/3 years and cover 8% to 9% of the welds in order to meet the
Code requirements.
outages.
This level of inspection was not assumed to extend these
It was assumed the pipe replacement would reduce subsequent defects
and repairs to negligible levels.
5.
Hydrogen Water Chemistry (HWC) Option.
replacement scenario is the addition of HWC.
An option in the pipe
Operation of this system with
replaced piping is expected to provide a small additional benefit in the

reduction of piping degradation.
However, it is believed to warrant con-
sideration for avoidance of potential long-term stress corrosion of core

internals and would provide an additional degree of certainty in avoiding
piping problems.
For this reason, it was assumed that if a plant added HWC
E-4
that ISI requirements would be reduced to ASME Section XI imnediately following piping replacement.
Like pipe replacement alone, this option was
assumed to reduce new defects and repair requirements to negligible levels.

The effect of increased main steam gamma activity due to increased 16N
resulting from HWC operation is discussed in Section E.3 below.
E.l.3
Scenario 2 - IHSI With HWC
1.
Operation Sequence.
This scenario assumes that all sensitive welds
have been previously inspected and found to be without defects.
The welds are
treated with IHSI or equivalent techniques to desensitize them to IGSCC

damage.
The following activities are assumed to be performed as part of this
scenario:
a.
Decontaminate the pipe as an optional but ALARA-recommended measure

prior to IHSI.
b.
Treat all welds using IHSI or an equivalent process.
c.
Inspect the piping to establish a baseline UT record for future

inspections.
d.
Install and operate HWC system.
e.
Resvime operation.
E-5
2.
Plant Outage.
last 3 to 5 months.
The total outage for this scenario was assumed to
Like Scenario 1, this activity could be conducted concurH
rently with major maintenance or during a normal refueling outage.
This re-
sults in an assumed outage chargeable to IHSI and HWC implementation of 0 to

3.5 months with a best estimate of 2.5 months.
3.
Post-IHSI ISI.
ISI requirements following implementation of this
scenario are assumed equivalent to the case of pipe replacement without HWC.
Assumed inspection requirements would include 4% of the welds for the first
two refueling outages followed by reduction to the ASME Section XI schedule.
ISI work is assumed to be conducted every other outage.
It was assumed that
subsequent defects and required repairs would be reduced to negligible levels

following this treatment.
4.
IHSI Without Hydrogen Water Chemistry.
is not to use HWC.
An option in this scenario
This was assumed to have a negligible effect on the outage
time for implementation.
However, the post-IHSI ISI requirements are revised
to assume inspection of a minimum of 8% (double the ASME XI average) of the

welds for the first four refueling cycles to obtain field data to confirm the
long term effectiveness of IHSI alone.
Subsequent inspections would reduce to
the normal ASME XI scope and schedule.
For these inspections, it was assumed
that one new defect would be found for every 20 welds inspected and that
repairs would be needed for two out of three defects found.
The assumed
defect rate of 1/20 is about one fourth that experienced through December 1983
by plants that performed IE Bulletins 82-03/83-02 inspections.
The assumed
repair rate of two-thirds repair per defect is the same as IE Bulletin
i
E-6
inspection experience through December 1983.
No incremental outage time has
been assumed for post-IHSI inspections; however, 0 to 2 weeks with a best

estimate of 1 week was added to the outage when repairs were required.
E.1.4
Scenario 3 - Augmented Inspection and Repair

without Hydrogen Water Chemistry
1.
Operations Sequence.
In this scenario current inspection and
repair practices are continued in order to mitigate the effects of IGSCC.

This is viewed primarily as a short-term option while a decision is being made
between the previous long-term options.
It is mainly based on the assumption
that an inspection method or basis for accepting weld overlay repairs for more
than one refueling outage will be established.
The following activities are
assumed to be completed as part of the scenario:
a.
Continue inspections similar to IE Bulletins 82-03 and 83-02.
b.
If defects are discovered, evaluate the possibility of propagation

using fracture mechanics analysis.
If repair is not required, record
for future monitoring.
c.
If repairs are required, use the weld overlay technique or equivalent.
2.
Repaired welds would be inspected before resuming operations.
Inspection Scope and Frequency.
From the following, it is assumed
that 50% of the average number of welds (160) of a typical BWR will be
examined by UT during each refueling outage:
E-7
It is assumed that the augmented ISI plan presented in Appendix C will be

applied at each refueling outage for plants that do not implement measures for|
long term mitigation of IGSCC.
This augmented ISI plan basically requires
inspection of an initial 20% sample of welds and previously uninspected and uncracked welds plus inspection of unrepaired cracks and all weld overlay
repairs on cracks measured to exceed 10% of the pipe circumference. Further,
if new cracks or significant growth of old cracks is found in the initial 20%
sample, the sample size should be expanded in accordance with IWB 2430 of ASME
Code Section XI. Experience from IE Bulletin 83-02 which specified an initial
sample of 22 welds indicates that on average the plants actually inspected
four times the number of the initial sample size. However, it is expected
that the extent of sample expansion for future inspections will be reduced.
This expectation is based on the fact that the IE Bulletins 82-03/83-02
inspections examined a large number of welds for the first time using a UT
technique that is much more sensitive in detecting IGSCC than any technique
used previously.
Also, some licensees decided to inspect 100% of the
recirculation/RHR system welds regardless of the IE Bulletin specified sample

size.
Thus, for this scenario it is estimated that future inspections will
examine on average twice the initial 20% sample plus the previous repaired
and unrepaired cracks. This could amount to about 50% the average number of
welds, for a typical plant or about 80 welds per plant during each refueling
outage.
The inspection scope was assumed to range from 40% to 60%.
E-8
3.
Defect Rate.
Experience through December 1983 from IE Bulletins
82-03/83-02 inspections indicated an average defect rate of 19% (283 defects

for 1509 welds inspected).
It is assumed that future inspection will average
a reduced defect rate of about 10% because of the high level of inspection
completed under the Bulletins and the fairly short time (18 months) between
successive inspections.
4.
Repair Rate.
The repair rate experienced under the Bulletins is
assumed to remain unchanged at about two repairs every three defects (183
repairs per 283 defects).
5.
Plant Outage.
Outage time for this scenario depends on the number
of defects that are discovered and repaired.
This level of inspection and
repair was assumed to result in three to four months of total outage.
The
outage time chargeable to inspection/repair depends on whether the work is

done concurrently with a refueling outage (six weeks) or a major plant
maintenance activity (three months).
occurs every third refueling cycle.
It was assumed that major maintenance

Thus, the assumed incremental outage
time, in weeks, is as follows:
Second
Cycle
First
Cycle
Third
Cycle
Lower Estimate
Nominal Estimate
10
10
Upper Estimate
E-9
6.
Hydrogen Water Chemistry (HWC) Option.
An option in the inspect/
repair scenario is to install and operate an HWC system.
HWC installation is
not anticipated to result in additional outage or significant ORE.

frequency would remain as indicated above.
ISI
However, it is expected that HWC
will result in the presence of fewer defects at each inspection.
Thus it was
assumed that the defect rate would reduce to one new defect for every 20
uncracked welds inspected.
This should eliminate or greatly reduce expansion
of the original 20% sample size.

of the welds are inspected:
the initial 20% sample plus 10% to reexamine
previous indications and repairs.

from 20% to 40%.
maintained.
Thus, it is assumed that on on average 30%
The inspection scope was assumed to range
The repair rate of two repairs for every three defects was
It was assumed that the outage time necessary to complete these
inspections would reduce each refueling cycle two to three months.
Using the
same assumption as above for other major maintenance activity in the plant,
the incremental outage, in weeks, is as follows:
Second
Cycle
First
Cycle
Third
Cycle
Lower Estimate
Nominal Estimate
10
Upper Estimate
E-10
^^.2
UNIT VALUES FOR ORE AND INDUSTRY COST
This section develops unit value estimates for ORE and costs for weld
inspection, weld repair, pipe replacement, IHSI, and HWC.
The unit values as
well as estimated plant outage times are based on reports and verbal communication from licensees, EPRI, the NSS vendor, and some subcontractors engaged
in various activities resulting from IE Bulletins 82-03/83-02.
They also
utilize estimates of utilities planning to replace IGSCC-sensitive pipe.

These results are used in the following sections to estimate cost impacts of
scenarios in this value-impact assessment.
E.2.1
UT Inspection
The UT experience of several plants is summarized in Table E.2.
The ORE
data are based primarily on an EPRI tabulation of UT inspection experience for

15 plants.
The cost data are based on approximate cost information provided
by three plants.
E.2.2
Weld Overlay
The weld overlay experience is summarized in Table E.3.
The ORE data are
based primarily on an EPRI tabulation of weld overlay experience for 15

plants.
The cost data are based on approximate cost information provided by
two plants.
E-U
Table E.2
UT Inspection Summary Data
No. of
Welds
UT'd
ORE
Total
man-rem
Plant
15 plants^^^
1011
(2)
Plant A
73
Plant B
100
Plant C
248
ORE -
(2)
(2)
Man-rem
weld
SK.
Cost weld
,,v
Average defect rate
Man-rem
Weld
Total
Cost, SK
UT Cost
Weld. SK
Decon
1312
0.77
101
0.72
225
2.2
no
216
0.46
501
2.3
no
82
3.0
470
5.7
Average
Low
- not initially
- yes subsequently
High
0.8
0.3>
3.0^3)
3.0
2.2
5.7
283 defects
=
= 0.19.
1509 welds inspected
(1) From EPRI summary of 1/12/84, revised to reflect specific plant updated
data.
(2) Included in EPRI summary.
(3) High-low within 15 plant data.
(4) From IE Bulletin 82-03/83-02 inspections through December 1983.
E-12
Table E.3
Weld Overlay Summary Data
Total
No. of
Welds
Overlayed
Man-rem
Overlay
2061
200
10.3
ORE
Plant
15 plants^^^
Total
Cost. SK
Cost
Overlay
SK,
Decon
Plant B
(2)
330
54
6.1
4100
76
no
Plant C
(2)
185
22
8.4
1700
77
yes
Average
ORE -
Cost
Man-rem
weld
10.3
.SK
weld
High
2.4^3)
19.5^3^
76
76
f...
Average defect rate
Low
77
183 repairs
=
= 0.65.
283 defect
(1)
(2)
(3)
(4)
From EPRI summary of 1/12/84, r e v i s e d t o r e f l e c t s p e c i f i c p l a n t updated

data.
Included i n EPRI summary.
High-low w i t h i n 15 p l a n t d a t a .
From IE B u l l e t i n 82-03/83-02 i n s p e c t i o n s through December 1983.
E-13
E.2.3
Pipe Replacement
Pipe replacement ORE and cost estimates were obtained from summaries of
NRC meetings with utilities planning to replace pipe and follow-up telephone
contact with those utilities.
The results are shown in Table E.4.
Table E.4
Pipe Replacement Summary Data
Plant
Estimate Ore for

Pipe Replacement
Man-rem
Plant D
1850
1500 - 1750
57
Plant F
# 2000
40
Average
Cost, SM
Replacement scope may

be less than Plant E & F
30-35
Plant E
ORE, man-rem
Comments
Cost. SM
Low
High
1750
1500
2000
44
35
67
Estimated Outage Time - 6 to 9 months.

E.2.4
Induction Heating Stress Improvement (IHSI)
IHSI ORE and cost estimates were obtained by telephone contact with
utilities that have used this technique in their plant. The results are shown
in Table E.5 below:
E-14
Table E.5
IHSI Summary Data
oii
Total
Man-rem
Plant
No. of
Welds
IHSI'd
Weld IHSI
Man-rem
Total
Cost. SK
IHSI Cost
Weld. SK
Decon
Plant H
240
87
2.8
1600
18.4
yes
Plant G
420
91
4.6
6200
68.1
no
ORE
Man-rem
Weld IHSI
Cost,
E.2.5
SK
Weld
Average
Low
High
3.7
2.8
4.6
43.8
18.4
68.1
Hydrogen Water Chemistry (HWC) Option
The estimated costs to install and operate an HWC system were obtained
primarily from meetings with General Electric Co. and are summarized in
Table E.6 below:
Initial installation costs range from SI to 3 million depending on the

type of system installed to generate the hydrogen injected into the reactor
feed water, with an average installation cost of $2M.
Similarly, annual operating costs range from SIOOK to S500K per year with
an assumed average of S300K per year depending on the type of hydrogen
generating system.
Installation and operating costs are combined as follows:
E-15
Table E.6
HWC Summary Data
Average
Low
300
500
Installation
Cost, SM
Operating
Cost, SK
High
3
100
It is assumed that HWC installation will incur 20 man-rem.
ORE effects
of HWC operation are discussed later in Section E.3 below.
E.3
DISCUSSION OF VALUE-IMPACT ASSESSMENT
This section develops the quantitative results of the value-impact assessment.
The strategy used in the development of low, best, and high
estimates was to assume that all parameters were at their lowest, highest, or
best levels.
This assumption is not particularly realistic but does maximize
the range that is covered by the analysis.
E.3.1
IGSCC Public Health Impacts
Public health impacts from IGSCC may occur if loss-of-coolant-accident

(LOCA) frequencies are increased.
A rigorous understanding of the impacts of
IGSCC on LOCA frequency is not known.
It is currently the subject of an NRC
E-16
investigation being conducted by Lawrence Livermore National Laboratory.

Until results of this study are available, it was decided to use results of
sensitivity studies to bound the impacts of piping failures.
Two studies have
addressed this question (E.2, E.3).
A comparison of the studies indicated that the results are similar in

scope and magnitude.
Both studies cover results of risk assessments for
Limerick, Shoreham, Millstone 1, Browns Ferry, and Peach Bottom (RSS).
In
addition, Gaertner (E.2) includes Grand Gulf and estimated that the contribution to LOCA frequencies from IGSCC could be up to 30% based on a review of
existing data.
Section 7 of this report indicates that the contribution may
range from 30%-50%.
Sensitivity studies are presented that postulate LOCA
frequency increases of up to 10 times the values used in the published risk

assessments.
This corresponds to IGSCC-induced failure rates 20 to 33 times
those supported by historical data.
Gaertner also reports a case with LOCA
frequencies three times the published values as an indication of the effect of

IGSCC failures at 10 times their historical rates.
For the purpose of this study, it was assumed that increasing the LOCA
frequency by a factor of 10 was not sufficiently supported by the Section 7
analyses to be used in this value-impact analysis.
Statistical arguments
in NUREG-1050 (E.4) suggest that differences in magnitude by a factor of 3

from the historical data for LOCA are the threshold of discriminating
between two parameters with large uncertainty.
observed as a result of IGSCC in BWRs.
Such differences have not been
Engineering judgement can be used to
conclude that the data base is sufficiently large to have included IGSCC and
that only the availability of new methods in failure analysis allows the
E-17
understanding of the specific mechanism involved in IGSCC.
Thus, the sensi-
tivity case of the factor of 3 increase was assumed for the high estimate i n ^
this analysis.
The Gaertner report lists an average core melt increase for
this case of 3.6 E-6 events per reactor-year excluding Big Rock Point.
Big
Rock Point was excluded because of differences in age, design, and siting when
compared to the "typical" plant.
This assumption is believed to have no
significant effect on the plant average risk used in this analysis.
Inclusion
of Big Rock Point would distort results of this report because of the simplified approach to risk assessment that was used.
The best estimate and low
estimates used in this analysis assumed that the historical data are a good
predictor of IGSCC contributions to LOCA and that the contribution is between
25% and 50% of the published values.
This corresponds to reductions in core
melt frequency of 9.0 E-7 per reactor year for the best estimate and 4.5 E-7
per reactor year for the low estimate.
To calculate the public risk impacts of the core melt frequency changes,
they were multiplied by average dose factors generated using the CRAC code and
assuming the quantities of radioactive isotopes and guidelines in WASH-1400,
the meteorology at a typical midwestern site (Byron-Braidwood), a uniform
population density of 340 people per square mile (an average of all U. S.
nuclear power plant sites), and no evacuation of population.
on a 50-mile release radius model.
They are based
These calculations result in 6.7 E-i-6 man-
rem per core melt accident based on the Grand Gulf accident frequencies (E.l).
An assessment of the validity of this assumption was made by determining the
LOCA contribution to each release category in the WASH-1400 Peach Bottom risk
assessment.
The contributions ranged from 3.8% to 4.7%.
This suggests that
use of an average dose factor does not induce large distortions in the risk
results. Multiplying the dose factor result by the core melt frequencies
predicts the following public risk values:
IGSCC Risk Contribution (man-rem per plant)
Low Estimate
76
Best Estimate
151
High Estimate
605
No differentiation for the various mitigation scenarios was made owing to

the relatively low levels of risk value for mitigation of IGSCC.
It was
assumed that any of the various mitigation techniques would reduce LOCA frequencies due to IGSCC to negligible levels.
E.3.2
Occupational Exposure - Accidental
Reduced accidental occupational exposure could result from reducing the

core melt frequency due to IGSCC.
This value is believed to be small owing to
the relatively low core melt reduction and was not estimated.
E.3.3
Public Property
Reduced public property loss could result from reducing the core melt
frequency due to IGSCC.
This value was not estimated owing to the relatively
low core melt frequency reduction.
E-19
E.3.4
Onsite Property
Reduced onsite property loss could result from reducing the core melt
frequency due to IGSCC.
Value would result in the avoidance of core melt
accidents and the reduction of the expected loss of onsite property.
This
value is believed low and was not estimated.
E.3.5
Occupational Exposure - Operational
Occupational exposure results from the implementation and operation of

the three scenarios.
The impacts are scenario dependent and are discussed
below:
Scenario 1 - Pipe Replacement
The ORE for decontamination, replacement, and inspection of piping is

made up of two contributors.
These include the initial replacement and
inspection activities and the inspection during resumed operations.
The
initial ORE is estimated directly from Section E.2.
Initial Replacement and Inspection ORE (man-rem)
Low Estimate
1500
Best Estimate
1750
High Estimate
2000
Addition of the HWC was assumed to add 20 man-rem during installation.
E-20
Inspection during the remaining life of the plant will be performed

per the scope of ASME Section XI.
inspected over a 10 year period.
This requires that 25% of the welds be

This scenario assumes that this inspection
level will be met with or without HWC.
Thus, the dose for operation of this
resolution will be the unit value doses for inspection from Section E.2
multiplied by 75 welds.
Results are as follows:
Inspection ORE (man-rem)
Low Estimate
20
Best Estimate
60
High Estimate
225
Total ORE for this scenario with and without HWC is as follows:
Total ORE (man-rem)

With HWC
Without HWC
Low Estimate
1540
1520
Best Estimate
1830
1810
High Estimate
2245
2225
Operation of the HWC has the potential to expose personnel in the area o
the turbine during operation.
Experience to date at Dresden 2 with HWC indi-
cates that main steam gamma activity at full power increases by a factor of 5
because of increased 16N in the steam.
However, the increase in steam
E-21
activity disappears immediately (approximately 30 seconds) after hydrogen

injection to feedwater is stopped.
Therefore, shutting off the injection a
short time before entering the proximity of the turbine or steam piping for
short inspection/maintenance tasks will reduce the radiation field to negligible levels.
of HWC.
For longer tasks, plant shutdown would be required regardless
It was assumed for this analysis that the HWC system is operated to
minimize ORE.
No additional ORE was added to account for this additional
field in the plant.
Scenario 2 - IHSI
ORE in this scenario results from the implementation of the IHSI process,
inspection of welds, and repair of welds over the remaining life of the plant.
Implementation dose was estimated to range from 450 to 740 man-rem with a
best estimate of 590 man-rem.
Inspection dose was predicated on an 8% inspec-
tion rate for the first four outages after IHSI treatment if no HWC is used
and inspection at the same level as pipe replacement without HWC if HWC is
used.
A total of 120 weld inspections would be required without HWC and a
total of 100 weld inspections with HWC over the remaining 25 years of plant
operation.
Repairs were assumed to be negligible if HWC is used.

repairs were determined to be needed if HWC is not used.
A total of five
The inspection and
repair frequencies were multiplied by the unit values in Section E.2 to generate estimates of ORE in the scenario.
E-22
IHSI ORE (man-rem)
IHSI
With HWC*
Without HWC
Implement
Inspection
Total
Low Estimate
450
30
500
Best Estimate
590
80
High Estimate
740
300
Inspection
Repair
Total
36
12
498
690
96
52
738
1060
360
98
1198
Includes 20 man-rem for installation of HWC system.
Scenario 3 - Inspect and Repair
The inspect-and-repair scenario has inspection and repair dose when used
with or without HWC.
An additional 20 man-rem would be incurred during the
installation of the HWC equipment.
To estimate the number of inspections and repairs required for this

scenario over the remaining life of the plant, inspection results over all
remaining outages were simulated.
In all cases, there was an initital period
of increasing repair activity that declines as the population of weld repairs

increases.
Inspection activity shows a steady increase over the remaining
life of the plant due to repeated inspection of the overlayed welds.
For the
purposes of ORE and cost calculations that are presented in the next section,
E-23
average inspection and repair rates were established on an annual basis.
This
was calculated by summing the inspections and repairs required over the remaining life and dividing by the remaining plant life.
Results with and
without HWC are shown in the following table:
Average Annual Activity

Insp. Rate. % Inspections/yr
Years
No HWC
Low Estimate
1-9
Repairs/yr
40
40
49
62
3.6
2.7
50
50
60
72
4.4
3.0
10-25
60
60
71
81
5.2
3.3
1-25
20
26
0.9
Best Estimate
1-25
30
38
1.3
High Estimate
1-25
40
49
1.7
10-25
Best Estimate
1-9
10-25
High Estimate
With HWC
Low Estimate
1-9
These values were multiplied by their corresponding unit values from

Section E.2 and the remaining plant life (25 years) to calculate the total ORE
commitment for this scenario.
ORE for Inspect and Repair Scenario (man-rem)

With HWC
Without HWC
249
611
Best Estimate
1094
2255
High Estimate
4504
7747
Low Estimate
E-24
E.3.6
Industry Implementation Cost
Implementation costs for the three scenarios are developed in this section.
They vary by scenario and are discussed below:
Implementation costs for this scenario are due to the hardware removal
and replacement and extended outages.
The values for estimated outage time
and other costs for pipe replacement were estimated in Section E.2.
ment power was valued for this analysis at S300K/day.
Replace-
The total implementa-
tion cost for this scenario is as follows:
Pipe Replacement Implementation Cost (S millions)

Power Replacement
Other Costs
Total
Low Estimate
27
35
62
Best Estimate
41
44
85
High Estimate
68
57
125
Scenario 2 - IHSI
Implementation costs and outage times for IHSI were discussed in Section
E.2.
Conversion of the outage times at S300K/day was done as in Scenario 2.
E-25
IHSI Implementation Costs (S million)

IHSI Treatment
Power
Other*
Total
No HWC
2.9
2.9
7.0
23
30
10.9
32
43
2.9
7.0
23
32
10.9
32
46
Low Estimate
Best Estimate
High Estimate
With HWC
Low Estimate
Best Estimate
High Estimate
*Implementation of HWC would add S1-S3 m i l l i o n w i t h a b e s t e s t i m a t e of S2

million.
This scenario without HWC has no implementation costs.
HWC
implementation costs are shown above.
E.3.7
Industry Operation and Maintenance Costs
This section discusses costs incurred following the implementation of the

resolution to IGSCC.
Each scenario is different and is discussed below.
scenarios contain discounted cash flow analyses at rates of 5% and 10% to

estimate the present value of the activity.
E-26
All
Operation and maintenance costs for this scenario result from inspection
activity.
The level of inspection was calculated on the basis of the require-
ments described for mitigation of IGSCC and ASME Section XI. Utilities were
assumed to delay inspections as long as they could and still meet the code
requirements.
The inspection schedule used for this analysis was as follows:
Years
% Inspected
HWC
No HWC
No. of Welds Inspected

HWC
No HWC
1.5
10
10
10
10
10
12
10
10
16.5
10
10
19.5
10
10
22.5
10
10
Each of these inspections was costed using the unit values from Section
E.2.
The HWC system was assumed to have operating costs of SIOOK to S500K per
year with a best estimate of S300K for 25 years. Present value results are as
follows:
E-27
Present Value Operation Cost (SM)
Low Estimate
5%
10%
Best Estimate
High Estimate
5%
10%
5%
10%
0.057
0.057
0.12
0.12
0.078
0.078
0.228
0.228
0.148
0.148
4.54
4.23
2.72
1.417
0.908
1.06
0.148
Inspection
HWC
No HWC
0.088
0.088
HWC Operation 7.05

Total
HWC
7.1
4.6
4.3
2.8
1.6
No HWC
0.088
0.057
0.12
0.078
0.228
Scenario 2 - IHSI
Operating costs for IHSI with HWC were assumed to be the sum of inspection costs for pipe replacement without HWC adjusted for 160 welds and the HWC
operating costs.
Costs for IHSI without water chemistry were developed assum-
ing that repairs and inspections were conducted under the following schedule:
Years
1.5
3
4.5
6
12
16.5
19.5
22.5
No. of
Welds Inspected
No. of
Welds Repaired
13
13
13
13
13
13
13
13
1
1
1
1
1
The costs for inspection, repair, outage time, and HWC operating costs
were estimated using the unit values from Section E.2. Present value
operating costs for this scenario are as follows:
E-28
Present Value Operating cost (SM) - IHSI

5%
10%
IHSI with HWC
Low Estimate
7.3
4.7
Best Estimate
4.4
2.8
High Estimate
1.7
1.1
IHSI without HWC

Low Estimate
0.39
.27
Best Estimate
6.8
4.8
High Estimate
13.3
9.3

This scenario has operating costs due to inspection, repair, extended
outage time and HWC operating costs.
Estimates for outage time, inspection
cost, repair cost, and HWC operating costs were developed in Section E.2.
Outage time was costed at S300K per day.
The estimates for required repair
and inspection levels are time dependent.
Averages on an annual basis were
developed in Section E.3.5 to calculate the operational ORE.
These values
were multiplied by the Section E.2 unit cost values, added to the outage and
HWC (if used) costs and discounted at 5% and 10% to generate the following
present value estimates for the inspect-repair operational costs.
Present Value Inspect and Repair Operational Costs (SM)

5%
10%
No HWC
Low Estimate
85
56
Best Estimate
125
82
High Estimate
167
109
36
24
Best Estimate
63
43
High Estimate
95
63
With HWC
Low Estimate
E-29
E.3.8
NRC Implementation Support Costs
Not estimated - believed to be relatively minor.
E.3.9
NRC Operational Support Costs
Not estimated - believed to be relatively minor.
E.4
REFERENCES
E.l
W. B. Andrews et al. 1983. Guidelines for Nuclear Power Plant Safety

Issue Priorization Information Development.
NUREG/CR-2800 (PNL-4297),
Pacific Northwest Laboratory, Richland, Washington.
E.2
Gaertner et al. 1983. Sensitivity of Core-Melt Frequency to LOCA

Initiator Frequency for BWRs.
Electric Power Research Institute, Palo
Alto, California.
E.3
NRC. 1984. T. P. Speis to R. H. Vollmer. Memorandum dated

January 6, 1984. "Briefing on BWR Pipe Crack Issues."
E.4
"Probabilistic Risk Assessment (PRA): Status Report and Guidance for

Regulatory Application." Draft Report, NUREG-1050.
E-30
APPENDIX F
FRACTURE MECHANICS EVALUATION OF BWR RECIRCULATION PIPING
F.l
INTRODUCTION
Over the last ten years, intergranular stress-corrosion cracking (IGSCC)

has been found in various stainless steel piping systems at operating boiling
water reactor (BWR) facilities.
During this time a number of fracture
mechanics evaluations (F.l through F.7) have been performed to evaluate the
integrity of piping containing IGSCC and to determine the margins against
unstable crack extension in the cracked piping.
Generally, the previous piping evaluations have been performed using

either limit load analysis or tearing stability analysis based on an assumed
fully plastic condition at the crack section.
However, recent experimental
data for stainless steel weld materials indicate a reduced resistance to crack
extension compared to wrought stainless steel base metal.
These data suggest
toughness characteristics that may preclude cracked pipe sections reaching

limit load prior to significant crack extension; consequently, analysis methods based on achieving limit load may not be strictly applicable for evaluation of flaws in stainless steel piping welds.
F-1
In view of the recent crack extension resistance data obtained for stainless steel weld materials, elastic plastic fracture mechanics analysis that do
not necessarily assume limit load conditions at the cracked section were performed to assess the following:
(1) the margin against unstable crack exten-
sion for finite length, deep part-through surface cracks, (2) the margin
against unstable crack extension for throughwall cracks, and (3) the effect of
variation in material toughness and tensile properties on the margin against
unstable crack extension.
The analysis methods and the computational results are presented in the
following section of this Appendix.
F.2
METHODOLOGY
F.2.1
General Considerations
The approach used to assess the integrity of cracked piping and the margins against unstable crack extension included performing the following analyses.
First, part-through flaws, which are representative of the largest de-
tected during an inservice inspection and allowed to remain in service, were

evaluated to determine if they would remain stable for postulated faulted load
conditions.
Second, an evaluation was performed to determine the margin
against unstable crack extension at faulted load conditions for a throughwall

crack size corresponding to leak rates that can reliably be detected during
normal operation by existing, in plant leakage detection equipment.
Demon-
strating adequate margins against unstable crack extension for these flaw conditions would indicate that leak-before4)reak conditions would be maintained
and that there is little potential for pipe break.
F-2
Because there are questions about the accuracy with which part-through
flaws can be detected or sized (see Chapter 4 ) , additional throughwall flaw
analyses were performed.
These analyses included leak rate computations to
determine the flaw sizes that produce leakage rates (at faulted loads) equal
to the recirculation system makeup capacity, and stability analyses to define
the flaw sizes resulting in unstable crack extension at faulted loads.
The fracture mechanics evaluation involves the following steps:
(1) de-
fining appropriately combined Service Level A and SSE loads, (2) choosing a
suitably conservative flaw representation for the fracture mechanics analysis,
(3) defining lower bound material properties to represent material crack initiation and growth resistance, (4) performing fracture mechanics computations
to determine the potential for crack extension and the stability of subsequent
crack growth, and (5) assessing the margins against unstable crack extension
and net section plastic collapse in the cracked pipe section model.
The fol-
lowing paragraphs in this section provide a detailed description of each of

the variables used in the analysis and the analysis methods.
F.2.2 Loads
To perform the fracture mechanics and limit load analyses of the piping
systems, it is necessary to identify the axial forces and bending moments
acting on a section of pipe containing assumed circumferential cracks.
The
evaluations were made using combinations of Service Level A (SLA) and Safe
Shutdown Earthquake (SSE) loads.
The combined SLA and SSE loads were used
because they realistically model the load condition of interest to demonstrate

adequate margins against full break, i.e., an SSE occurs while the plant is at
normal full power operation.
F-3
Service Level A loads include contributions from pressure, dead weight,

and thermal expansion.
However, thermal loads can either increase or decreasfl
the resultant applied load; consequently, the thermal loads are included only
if they result in a larger combined load.
In general, this combined loading
scheme is somewhat more conservative than Service Level D loads, which do not
include thermal components.
The system thermal expansion stresses are treated
as primary stresses in the analyses.
The contribution to crack extension po-
tential from through-the-pipe-thickness stresses (resulting from thermal and

residual stresses) are relatively small and, for the purpose of the stability
analyses, were neglected in view of the conservative use of the pipe system
thermal expansion stresses.
Two load classifications were used in the analysis.
In one case SLA and
SSE loads considered typical of most welds in the BWR recirculation system
were used.
A second group of analyses was performed using bounding loads that
represent the highest loads at any of the welds in the recirculation system.
F.2.3
Flaw Characterization
F.2.3.1
Part-Through Flaws
Part-through flaws were evaluated to determine the margins against

unstable crack extension for piping containing either undetected IGSCC or
IGSCC that was found during inservice inspection and allowed to remain in
service.
The flaws used in the analyses are finite length, semi-elliptical
circumferential cracks that extend radially into the pipe wall from the inside
surface of the pipe.
F-4
The part-through circumferential flaw length was 50 percent of the pipe

circumference.
evaluated.
Flaw depths of 25, 50 and 75 percent of wall thickness were
The 50 percent flaw length and 50 percent depth are believed to
bound the largest respective flaw dimensions that were returned to service
following detection during inservice inspection.
The 25 and 75 percent of
wall thickness flaws were selected to obtain information about the effect of
flaw depth on the potential for unstable crack extension.
F.2.3.2
Throughwall Flaws
In addition to ensuring adequate margins against unstable crack extension

for part-through flaws, the evaluation of cracked piping generally includes
demonstration that leak-before-break conditions will be maintained during
postulated loading.
This is important because, as discussed in the next
section, cracks may grow throughwall by various time-dependent growth mechanisms, such as corrosion or fatigue.
In this instance, a crack large enough
to be detected by the leakage system should be able to withstand postulated

loads without unstable crack extension.
This assures that there is sufficient
time to take action to detect and repair or replace piping with throughwall
cracks.
The flaw selected to demonstrate leak-before-break is a flaw whose length

corresponds to a leak rate that can reliably detected by leakage detection
capabilities that meet the guidelines of Regulatory Guide 1.45 (F.8).
A leak
rate of 10 gpm is used to define the throughwall flaw length and to ensure detection with some margin to cover analytical and measurement uncertainties.
F-5
A correlation between crack size and flow rate through the crack was cal^
culated using a computer code previously developed for the Electric Power R e ^
search Institute (F.9).
Computational procedures in this code are based on
two-phase flow through a cracklike opening and consider crack face surface
roughness, crack length and wall thickness, pressure differential across the
crack, and crack opening area due to the loading.
The crack opening area and
associated leak rate are determined using normal operating (SLA) load conditions.
In addition to the throughwall flaw condition just described, leak rate

and fracture mechanics analyses were performed to determine two additional
throughwall crack sizes.
First, leak rate analyses were performed to find the
throughwall flaw size associated with a leak rate equal to the makeup capability (about 100 gpm) for the recirculation system at SLA plus SSE loads.
Second, fracture mechanics analyses were performed to determine the throughwall crack size that would result in unstable crack extension at SLA + SSE
loads.
These two flaw sizes serve to demonstrate the added margin for
throughwall crack size relative to the size that can be detected reliably at
SLA conditions by existing in-plant leakage detection systems.
F.2.4
Throughwall Crack Development
Because part-through flaws will not be found by leakage detection, definition of the flaw used in the evaluation also considered the effect of partthrough flaws on leak-before-break conditions to demonstrate that the through-
F-6
wall flaw is an appropriate bounding flaw for analysis purposes.
The purpose
of this section is to demonstrate that part-through cracks are likely to penetrate the pipe wall thickness and leak rather than progress around the pipe
and cause a significant break.
This crack growth sequence is verified for the two conditions that are of
major interest; namely, normal operation, and large bending loads in excess of
those postulated for faulted (SSE) loading.
For normal operating conditions, there is a large amount of service experience that demonstrates that stress-corrosion cracks progress radially
through the pipe wall and result in leak-before-break conditions.
As indica-
ted in F.2, F.IO, and F.11, several hundred significant cracking incidents
have been discovered in the United States in BWR primary and secondary piping
systems, and PWR secondary piping systems.
These statistics represent
experience with a wide range of crack sizes and piping systems.
These cracks
result from various initiation and propagation mechanisms, and exposure to

different combinations of stress states, i.e., bending and tension.
For all
the different conditions that actually occur in service, operating service experience indicates that the dominant crack growth trend for intermediate- and
larger-diameter piping is for the crack to grow radially through the wall and
leak.
F-7
Because large earthquake loadings occur very rarely, service experience

is not available to define the character of crack growth for these loads.
However, results from previously performed analyses do provide an indication of part-through crack characteristics for large bending loads.
The
studies described in F.5 defined the ratio of the crack driving force (in
terms of the J integral) in the circumferential-to-radial direction for a partthrough crack in a pipe.
The analytical results indicate that the value of
the crack driving potential J in the radial direction, J^, is always

greater than the value of the J in the circumferential direction, J , for all
combinations of percentage of crack through the wall, a/t, and circumferential
distance around the pipe.
taken from F.5.
This variation is shown in Figure F.l, which is
The variation of J , with J

(j)
a
shown in Figure F.l demonstrates
that part-through cracks will grow through the pipe wall and leak when the
pipe is subjected to large postulated bending (SSE) loads.
From service experience and analysis, it can be concluded that it is

likely cracks will grow radially through the pipe wall before they progress
around the pipe and cause a significant break.
This phenomenon has been
demonstrated for operating conditions ranging from normal operation to large

accident loads, and for a wide range of tensile and bending loading
combinations.
This ensures that analyses based on large circumferentially
oriented, throughwall flaws are an appropriately conservative basis for

evaluating the margin against full break and ensuring leak before break
conditions are maintained.
F-8
^ ^
F.2.5
Fracture Mechanics Evaluation
F.2.5.1
General Considerations
Elastic-plastic fracture mechanics analyses were performed for SLA and

loads in excess of SSE loads to evaluate the potential for unstable crack
extension of through and part-through flaws.
analyses are illustrated in Figure F.2.
The flaw geometries used in the
The methods used appropriately
account for material yielding and strain hardening that may occur in the
cracked section on application of large loads, and/or the presence of ductile
crack growth, which is the expected crack extension mechanism for ductile nuclear reactor pipe materials.
The analyses are based on a parameter called the J integral (F.12), which
is a measure of the intensity of the stress-strain field around the crack tip
(F. 13, F.14).
The two important aspects should be considered in general when
evaluating crack growth; namely, initiation or first extension of the existing

flaw and stability of a growing flaw subsequent to initiation.
The material
value of J associated with initiation of additional crack extension is denoted

as Jic.
If the applied value of J is less than Jj^,, no crack growth will
occur and stability of the existing crack is ensured automatically.
When ex-
tension of the existing crack is predicted, the crack extension must be evaluated to determine if it occurs in a stable manner, or if the crack will grow
unstably and result in a predicted full break.
F-9
Crack instability is evaluated using the tearing stability concept and

the associated tearing modulus instability criterion (F.15).
Tearing m o d u l u s ^ B
is defined as
T = ^
da
where
dJ/da
-^
0^2
(1)
indicates the increment of J needed to produce a

specified increment of crack extension at any
specified load and crack state,
is the material elastic modulus, and
a.f
is the material flow stress defined as one half

the sum of material yield and ultimate strengths.
To determine the margin against fracture, the values of J and T are

determined for the structure using the applied loads and crack geometry.
values are denoted as J
The
. . . or J
, and T
,. , or T
applied
app
applied
app
The material resistance to unstable crack extension is determined

experimentally from test data that relate J and crack extension.
Using these
data the tearing modulus can be determined from the slope of the crack growth
curve using Eq. (1). The tearing modulus obtained from the materials
data is denoted as T
. At any specified J level, where J
is greater than
Jj_, stable crack extension will occur when
T ^ > T
mat
app
(2
F-10
A convenient means now commonly used to illustrate the margin against

instability involves plotting J as a function of T for the applied loads and
material resistance as shown schematically in Figure F.3.
In this figure
material crack growth resistance is plotted along with the crack growth
potential associated with the load and crack in the component.
The
intersection of the curves defines the instability point.
F.2.5.2
Computation of J
and T
'^
app
app
The computation of J follows the method described in F.16 and F.17 for
through flaws. The same general formulation also was used in this study to
obtain J for part-through finite-length semi-elliptical surface flaws.
In
this generalized computation scheme the J integral is separated into elastic

and plastic components, or
Jg + Jp .
(3)
where
is the plasticity adjusted elastic contribution to J
is the plastic contribution to J.
and
F-11
The elastic portion of J is directly related to the elastic stress intensity

factor, K^, by the relationship
= K, /E.
e
(4)
Elastic K solutions are available from F.18, F.19, and F.20.
The plastic component of J for (bend loading) is expressed in the form (F.17)i
a a e ch,(M/M^)"*^
0 0
(5)
where
and e are the reference y i e l d s t r e s s and s t r a i n and a and n are

o
^
material constants determined from the material s t r e s s s t r a i n curve f i t

Ramberg-Osgood curve.*
* c/e^ " O/OQ + oCo/a^)
F-12
to a
2c is the remaining circumferential ligament of the cracked portion of

the pipe;
h^ is a function which accounts for relative crack and component size,

and material work hardening;
is the moment required to develop an average stress of magnitude cr in

the cracked section;
is the applied moment.
The total J is obtained by adding the plasticity adjusted elastic solution and
the plastic solution given by equation 5, or
' = ê^%ff> '
(6)
'r
The applied tearing modulus can be expressed in general terms (for bending)
(F.21) as
app
(|^)H-
(|R)a t^'H <^* 'l^'.>''
F-13
(7)
where
a = crack size,
M = Moment,
<j) = total angle of rotation,
(J> = angle of rotation of cracked section, and
c
C = system compliance
The system compliance ranges between two extremes; C = 0 for fixed

displacement boundary conditions and C = " for dead weight type loading.
conservative estimate of T
can be obtained by placing C = < into Eq.
app
(8) so that it becomes
dj
app
F.2.5.3
._j5 - I da
a%
(8)
Materials Data
Analyses were performed for two weld materials including submerged arc
welds and tungsten inert gas welds.
These two welds are believed to represent
the-majority of the welds in BWR recirculation piping.
The crack extension data for the materials are expressed in the form of
applied J versus measured crack extension.
To ensure that the material crack
growth characteristics are used in a conservative manner, an attempt was made

to represent the data as reasonable lower bounds.
F-14
All data were obtained from 1-in.-thick compact tension-type (ITCT) test
specimens that had face grooves.
The use of face grooves on the CT specimen
is the configuration that appears best to simulate the cracked tip condition
associated with cracked piping.
The material test temperatures were
about 550F.
F.2.6
Net Section Plastic Collapse
The presence of the postulated circumferential
throughwall flaw will
reduce the ultimate load-carrying capacity of the pipe section.
The limit moment for the pipe section containing a throughwall crack was
determined from F.l and is expressed as
(9)
M- 4o
R^t (COST - H sin 9)
where
R is the mean pipe radius,

t is the pipe wall thickness,
9 = half-crack angle, and
, rt ^ ir Axial load
F-15
F.3
NUMERICAL RESULTS
Numerical computations were performed to assess several aspects related

to the integrity of flawed piping.
These aspects include:
(1) the loads at
instability relative to the limit load, (2) the margin against first extension
of part-through cracks, (3) the margin against unstable crack extension for
throughwall cracks, and (4) the effect of material tensile and toughness
properties on the margin against first crack extension and unstable crack
extension.
The following paragraphs of this section present the numerical
input used to complete this analyses and the computational results.
F.3.1
Pipe Dimensions
Selected pipes in the BWR "ring header" design recirculation piping

system were evaluated.
These pipes include the 28-in. pvmp suction and
discharge lines, the 22-in. header, and the 12-in. risers.
The pipe
dimensions used in the analysis are presented in Table F.l.
Table F.l
Recirculation System Pipe Dimensions
Nominal Diameter
Mean Radius
Wall Thickness
in,
VOL.
isj
12
6.092
0.566
22
10.475
1.05
28
13.45
1.1
F-16
F.3.2
Loads
Two loading conditions were used in the analyses.
These included a
bounding loading condition representative of the largest stress for each pipe
size and a typical loading condition representative of the stresses at most
weld locations in recirculation piping.
The SLA and SSE stresses for each
loading condition and pipe size are presented in Table F.2.
Table F.2
Recirculation System Typical and Bounding Load Definition
Nominal Pipe
Size
in.
Service Level A Stresses

Axial
Bending
Typical Bounding
Typical Bounding
ksi
ksi
SSE Stresses*
Bending
Typical Bounding
ksi
12
10
10
22
10
28
10
*No additional axial stress is attributed to seismic loading.
F-17
F.3.3
Throughwall Crack Size Versus Leak Rate Correlation
The flaw sizes associated with a 10-gpm and lOO-gpm leak rate were
obtained for each pipe size at both the typical and bounding load conditions.
The 10-gpm crack size represents a leakage size crack that could reliably be
detected at normal operation and is used to demonstrate leak-before-break
margins relative to faulted loading.
The lOO-gpm crack size represents the
crack size that would approximately equal the makeup capacity for the
recirculation system at faulted loading.
The results are presented in
Table F.3.
Table F.3
Detectable Leakage and Makeup Capacity Size Throughwall Flaws

for Typical and Bounding Loading
Nominal Pipe
Size
in.
10-gpm Crack Size

Relative to Circumference
Service Level A Load
Typical , Bounding
e/ir
lOO-gpm Crack Size

Relative to Circumference
SLA + SSE Load
Typical
Bounding
e/iT
12
0.21
0.19
0.35
0.24
22
0.16
0.14
0.30
0.21
28
0.14
0.12
0.26
0.18
F-18
F.3.4
Fracture Mechanics Evaluation
F.3.4.1
Materials Data
F.3.4.1.1
Tensile Properties
The tensile properties for each of the materials are presented in

Table F.4.
Table F.4
Tensile Properties for Stainless Steel Weld

Materials, Temperature = 550F
Material
Property
Submerged
Arc Weld
Tungsten Inert
Gas Weld
Yield Strength, ksi
53.9
49
Ultimate Strength, ksi
63.4
65.6
Flow Stress, ksi
58.7
57.3
Reference Stress, ksi
53.9
35
2.83
11.8
F-19
2.52
5.0
F.3.4.1.2
Toughness Properties
The toughness properties for the two weld metals listed in Table F.4 were
obtained from experiments that generated J versus crack extension (J-R) curves
at a temperature of 550*^.
The J-R curves provide two measures of fracture
resistance; namely, resistance to first extension of the existing crack

(Jic) and the resistance to a growing crack.
The values of Jjc for each material are presented in Table F.5.
Generally, an assessment of stability conditions will be based on the resistance to crack extension of a growing crack.
As discussed in Section F.2.5.1,
it is convenient to construct a J/T plot from the J-R curve data for performing the stability analysis.
Figure F.4.
The J/T curve for each material is shown in
The J-R curves used to construct the J/T plots are presented in
Figures F.5 and F.6.
The data presented in Tables F.4 and F.5 and Figures
F.4, F.5 and F.6 were obtained from ongoing NRC and EPRI programs (F.22,
F.23).
Table F.5
JlC Values at 550F for Submerged Arc and Shielded

Manual Metal Arc Weld Metals
ÎC
Material
in.-lb/in. 2
Submerged Arc Weld (SA)
420
Tungsten Inert Gas Weld (TIG)
960
F-20
In most instances information presented in Tables F.4 and F.5 and Figure
F.3 were obtained from tensile and toughness tests from one or two specimens.
The tensile data presented in Table F.4 represent the average measured values
and are appropriate for use in the equations presented in Section F.2.5.2.
The tensile data in Table F.4 for the submerged arc weld were not measured
directly but were estimated by the experimental investigators based on properties from other welds (F.24).
The J/T plots generally are constructed using the lower bound of the
available J-R data and are appropriate for obtaining a realistically
conservative estimate of the instability conditions.
F.3.4.2
F.3.4.2.1
Numerical Results
Part-Through Surface Cracks
Values of J were computed for several loads and flaw sizes.
The load
levels included the bounding SLA, SLA + ISSE, and SLA + 2SSE loads.
The flaw
circtimferential length was assumed to be 180 around the pipe, while the
radial part-through flaw depth was taken as 25, 50, and 75 percent of the pipe
wall thickness.
Computations were performed for each of the three pipe
sizes and the submerged arc and tungsten inert gas weld materials.
F-21
The results obtained for the submerged arc weld indicate that J is less
than Jjc (420 in.-lb/in.2) for the loads, flaw sizes, and pipe sizes evalated, with the exception of a 75 percent part-through flaw in a 28-in.
pipe subjected to SLA + 2SSE loads.
700 in.-lb/in.2.
In this case the computed J is
Because J is greater than Jic, the value of applied
tearing modulus, T, was determined to assess if the predicted crack growth is

stable.
A comparison of the computed applied values of J = 700 and T = 20
with the J/T plot in Figure F.4 for the submerged arc material indicates
stable crack growth.
Similar computations were made for the tungsten inert gas weld material.
The results of these analyses indicate applied J values less than J-^Q (960
in.-lb/in.2) for all three pipe sizes with SLA + 2SSE loads and flaw depths
of 50% of the wall thickness.
At flaw depths of 75% of the wall thickness
applied, J was greater than Jjc for all three pipe sizes at SLA + 2SSE
loads.
The computed applied J values are 1450, 1370, and 1990 in.-lb/in.2
for the 12-, 22-, and 28-in.-diameter pipes, respectively.
The respective
applied T values for each pipe size were determined to be 260, 133, and 165.
Comparison of these values with the plot in Figure F.4 for the tungsten inert
gas weld indicates stable crack extension for the 180**-long, 75 percent partthrough flaws at SLA + 2SSE loads.
F-22
F.3.4.2.2
Throughwall Cracks
This section presents the results of the computation performed to

determine the instability moments for the 10-gpm leakage crack sizes
associated with the bounding SLA loads (see Table F.3), and the instability
crack sizes at SLA + ISSE loads.
The instability moments for the leakage size cracks in the 12-, 22-, and
28-in . pipes are presented in Table F.6.
Also presented in the table is the
ratio of the instability moment to the limit moment for the
two weld
materials.
Table F.6
Instability Moment for Bounding Load SLA

Leakage Size (10 gpm) Cracks
Pipe
Leakage Crack
Diameter
Size Relative
in.
Circumference
to
Instabil ity Moment
10^ in,.-lb
e/iT
SA^
SA
3
TIG^
SA
TIG
12
0.19
2.5
2.2
0.81
0.74
22
0.14
14.9
12.9
0.79
0.70
28
0.12
26.8
23.1
0.76
0.67
Ratio instability moment to limit moment.

2
I/L'
" submerged arc weld.
TIG = tungsten inert gas weld.

F-23
The results in Table F.6 indicate that the instability moment ranges from
about 70 to 80 percent of the limit moment for the materials evaluated in this'
study.
This result suggests that the use of limit load is not strictly appro-
priate for flaw evaluation.
However, the instability moments are significant-
ly larger than the SLA + ISSE moments and significant margins against unstable
crack extension exist as discussed in Section 3.5.
The throughwall crack lengths that would produce unstable crack extension
at typical and bounding SLA + ISSE loads were determined for the two weld
materials.
The results are presented in Table F.7.
Table F.7
Crack Size for Instability at SLA + ISSE Loads
Pipe
Diameter
in.
Instability Crack Size
Instability Crack Size
Relative to Circumference at
Relative to Circumference at
Typical SLA + ISSE Load
TIG
SA'
Bounding SLA + ISSE Load
TIG
SA
e/TT
e/iT
12
0.42
0.44
0.31
0.32
22
0.46
0.47
0.32
0.35
28
0.41
0.43
0.28
0.31
SAA = submerged arc weld.

TIG = tungsten inert gas weld.
F-24
The results in Table F.7 indicate that relatively large (30% around the
circumference) throughwall flaws can be tolerated at the bounding SLA *- ISSE
loads and somewhat larger sizes (40% around the circumference) for the typical
SLA + ISSE loads.
The results are about the same for the two welds.
The results in Table F.7 also serve to point up the importance of tensile
properties in the analyses.
For example, the instability crack size for the
two welds are about the same even though there is a significant difference in
the J-R curves for the materials (see Figure F.4). The difference in J-R
curves is compensated for in this case by the higher strength, flatter (lower
work hardening) stress strain curve for the submerged arc weld compared to the
tungsten inert gas arc weld.
F.3.5
Margins
F.3.5.1
Part-Through Wall Flaws
The margins against first crack extension or unstable crack extension can
be estimated from the results in Section 3.4.2.1.
Although the loads
corresponding to unstable crack extension were not obtained, the margin

against unstable crack extension can be estimated using the SLA + 2SSE load as
a reference.
At this load the flaws either did not extend or exhibit stable
growth for each of the pipe sizes and materials evaluated.
Consequently)
realistically conservative estimates of margin at SLA and SLA + ISSE load

can be obtained by the ratio of P + P. at each of these conditions (see
m
0
Table F.2) to the value of P^^^ + P^^ at SLA + 2SSE.
in Table F.8 for the bounding load case.
F-25
The results are presented
Table F.8
Estimated Margin Against Unstable Crack Extension

for Part-Through Flaws Up to 75% of Wall for
Submerged Arc and Tungsten Inert Gas Arc Welds
P, + P
Pipe Diameter
at
Margin Relative to SLA + 2SSE Bounding Loads
SLA + 2SSE
in.
ksi
SLA
SLA + ISSE
12
36
2.3
1.4
22 & 28
32
2.7
1.5
The margins shown in Table 8 generally can be considered relatively close

to the margins corresponding to instability margins for the tungsten inert gas
weld as seen from the J/T values in Section 3.4.7.1.
The margins would be
somewhat higher for the submerged arc weld where the applied J and T values
are typically lower for the 75 percent part-through flaw.
The results in this section also provide limited data that can be used to
evaluate the margins in section IWR 3640 of Section XI of the ASME Code.
Even
though limit load is not achieved for the weld materials considered, margins
equivalent to those in the code are achieved (for the loads used in this evaluation) because the part-through flaw depths allowed by IWB 3640 for a 180"
flaw are less than the 75 percent of wall depth that was shown to remain
F-26
stable by this analysis.
To ensure this conclusion can be universally true,
additional flaw depths and lengths and loading conditions representative of

the scope of IWB 3640 would have to be evaluated.
F.3.5.2
Throughwall Flaws
The margins against instability for pipes with leakage size (10-gpm leak
rate) throughwall flaws were determined using the instability results from
Table F.6 and stress conditions in Table F.2.
The results are presented in
Table F.9 for the two weld materials and both the bounding and typical SLA ->ISSE loads.
Table F.9
Margin Against Instability for Total Load

(P^ + P^) at SLA + ISSE Loads for
10 gpm Leakage Size Throughwall Flaws
Pipe
Diameter
in.
+ P, at
Instability, ksi
SA
TIG
12
42.9
22
28
Margin Against Instability at SLA + SSE

Typical Loads
Bounding Loads
SA
TIG
SA
39.3
2.7
2.5
1.7
1.5
47.2
41.6
3.9
3.5
2.1
1.9
48.9
43.
4.1
3.6
2.2
2.0
F-27
TIG
The results in Table F.9 show significant margins against unstable crack
extension for the bounding loads and large margins for the typical loads.
These results indicate that most weld joints in the recirculation system have
large margins for leak-before-break and that the highest stress weld joints
have margins typical of values normally associated with ASME Code margins for
faulted conditions.
In addition to determining margin on loads, margin on flaw size also can

be defined by comparing the leakage size flaws in Table F.3 with the instability flaw sizes in Table F.7.
From these data, the margin on the 10-gpm
leakage size flaw generally exceeds a factor of 2.0 relative to the

instability size, except for the 12-in.- diameter pipe where the margin is
about 1.6.
The data in the tables also indicate that the instability flaw
size is always larger than the flaw size corresponding to the system makeup
capacity of lOO-gpm at SLA +1SSE.
F.4
CONCLUSIONS
Part-through cracks as large as 50 percent around the circumference and

75 percent throughwall will remain stable at loads in excess of SLA + 2SSE.
This implies margins on total load of 1.4 and 2.3 relative to SLA + ISSE and
SLA, respectively, for the highest stressed weld joints in the recirculation
system.
F-28
The margin against unstable crack extension at SLA + ISSE for leakage
size (10-gpm) cracks ranges from factors of 1.5 to 2.2 on load for the highest
stressed weld joints in the recirculation system.
The instability crack size
generally exceeds the 10-gpm crack size by a factor of 2.0, except for 12-in.
piping where the factor is about 1.6.
The crack size at instability is always
larger than the crack size corresponding to the makeup capacity of the recirculation system.
The margins on crack size and load indicate that leak-before-break conditions will be maintained for BWR recirculation piping.
On the basis of the limited results obtained from these analyses, it

appears that the safety factors incorporated in paragraph IWB 3640 of Section
XI of the ASME Code are generally maintained for BWR recirculation piping,
even though crack instability is predicted prior to reaching load limit for
the pipe sizes and welds evaluated.
The tensile properties play a large role in evaluating crack stability

using elastic-plastic techniques.
The results obtained in this analysis show
that the relative difference in tensile properties between the submerged arc
and tungsten inert gas welds almost completely compensates for the significant
difference in the material J/T relationships.
F-29
F.5
REFERENCES
F.l
H. Tada et al. June 1979. Stability Analysis of Circumferential

Cracks in Reactor Piping Systems.
NUREG/CR-0838.
F.2
NRC. Pipe Crack Study Group. 1980. Investigation and Evaluation

of Cracking Incidents in Piping in Pressurized Water Reactors.
NUREG-0691.
F.3
A. Zahoor and M. F. Kanninen. July 1981. "A Plastic Fracture

Instability Analysis of Wall Breakthrough in a Circumferentially Cracked
Pipes Subjected to Bending Loads." Trans. ASME 103. 194-200.
F.4
K. H. Cotter, H.-Y. Chang, and A. Zahoor. February 1982.

Application of Tearing Modulus Stability Concepts to Nuclear Piping.
EPRI-NP-2261. Electric Power Research Institute, Palo Alto, CA.
F.5
M. F. Kanninen et al. April 1982. Instability Predictions for

Circumferentially Cracked Type 304 Stainless Steel Pipes Under Dynamic
Loading.
EPRI NP-2347, Vol. 1 and 2, Electric Power Research
Institute, Palo Alto, CA
F.6
A. Zahoor and D. M. Norris. July, 1983. "Ductile Fracture of Circumferentially Cracked Type 304 Stainless Steel Pipes in Tension." Paper
presented at ASME PVP.
F.7
P. C. Paris, H. Tada, and R. Macek. September 1983. Fracture Proof

Design and Analysis of Nuclear Piping.
NUREG/CR-3465.
F.8
NRC. May 1973. Reactor Coolant Pressure Boundary Leakage Detection

Systems.
Regulatory Guide 1.45.
F.9
D. Abdollhian and S. Levy, Inc. December 1983. Calculation of Leak

Rates Through Cracks in Pipes and Tubes.
EPRI NP-3395. Electric Power
F.IO
L. Frank et al. June 1980.

Reactors.
NUREG-0679.
F.ll
NRC. Pipe Crack Study Group. February 1979. Investigation and

Evaluation of Stress-Corrosion Cracking in Piping of Light Water Reactor
Plants.
NUREG-0531.
^
F.12
J. R. Rice.
F.13
J. W. Hutchinson. January 1968. "Singular Behavior at End of Tensile

Cracks in Hardening Material." J. Mech. Phys. Solids 16. 13-31.
1968.
Pipe Cracking Experience in Light-Water
Fracture.
F-30
Vol. 2, Academic Press, New York.
F.14
J. R. Rice and G. F. Rosengren. January 1968. Plane Strain Deformation

Near a Crack Tip in Power-Law Hardening Material. J. Mech. Phys.
Solids 16. 1-12.
F.15
P. C. Paris et al. August 1977. A Treatment of the Subject of

Tearing Instability. NUREG-0311.
F.16
V. Kumar, M. D. German, and C. F. Shih. July 1981. An Engineering

Approach for Elastic Plastic Fracture Analysis.
EPRI NP-1931, Research
Project 1237-1. Electric Power Research Institute, Palo Alto, CA.
F.17
M. D. German and V. Kumar. July 1982. Elastic-Plastic Analysis of

Crack Opening, Stable Growth and Instability Behavior in Flawed 304 SS
Piping." In Aspects of Fracture Mechanics in Pressure Vessels and
Piping.
PVP-Vol. 58, ASME.
F.18
D. M. Crutchfield (NRC) to D. P. Hoffman (Consumers Power Company).

Letter dated December 1981: Appendix 2, Docket No. 50-255, Document
L505-81-12-015.
F.19
J. L. Sanders, Jr. 1982. "Circumferential Through-Cracks in

Cylindrical Shells Under Tension." J. Appl. Mech. 49. 103-107.
F.20
J. L. Sanders, Jr. 1983. "Circumferential Through-Cracks in

Cylindrical Shell Under Combined Bending and Tension." J. Appl. Mech.
ig. No. 1, p. 221.
F.21
J. W. Hutchinson and P. C. Paris. 1979. Stability Analysis of JControlled Crack Growth.

Am. Soc. Test. Mater. Spec. Tech. Publ. 668
37-64.
F.22
EPRI Program RP 1238-2, Westinghouse.
F.23
NRC Program FIN B-6290, David Taylor Naval Ship R&D Center.
F.24
Private Communication with D. McCabe, Westinghouse.
F-31
March 13, 1984.
1.0
" ^
"
>v
iO.8 -
//\^?-T~Nsf\
f f^'^Ts^j/
\\
\\
\\
2 0.6
a:
/"SJ-T^^
I-
O
o0.4 h
V
^K^^y^
^
, = 0.,
\ \
\ \
\ \ ^ x = o.e
:0.2
<
a:
BENDING LOAD
Figure F.l Stability of a

part-through circumferential
crack under fully plastic
bending loads.
\X"'
'
\
\
1\
0.2
0.4
0.6
0.8
NORMALIZED CRACK ANGLE ( S/ir)
1.0
PART-THROUGH FLAW
THROUGHWALL FLAW
Figure F.2 Through and part-through geometries used

in fracture mechanics analyses.
F-32
INSTABILITY POINT
-MARGIN AGAINST
STABILITY
Figure F.3 Illustration

of stability assessment
diagram.
TEARING MODULUS, T
-TUNGSTEN INERT
GAS WELD
C
Figure F.4 J/T Plots for Submerged

arc and Tungsten inert gas welds,
550F.
CO
O 3
I -
50
F-33
100
150
200
TEARING MODULUS, T
250
300
14880.
8.808
0.030
8.868
8.898
8.120
8.150
8.188
8.218
8.240
CRACK EXTENSION In INCHES

F i g u r e F . 5 J - R Curve f o r TIG Weld,
550F ( R e f .
F.23)
E 83
20Z Side Grooves
8.88
8.88
8.28
8.48
8.68
SPEC.
TEMP.
ENVIR.
4USA-J2
668 F
AIR
SLOPE Y-INT. JIC
781 I.87
51 I.571
566.484
8.88 1.88 1.28

CRACK6R0WTM CIN3
1.48
Figure F.6 J-R Curve for Submerged Arc Weld, WO^F (Ref. F.22)
F-34
1.68
1.88 2.88
E-81
APPENDIX G
LONG-TERM ISSUES
ULTRASONIC INSPECTABILITY
Substantial improvement in the detection of IGSCC during ISI of BWR

piping using UT has been made.
This improvement has been accomplished through
a combination of operator training, elimination of ineffective procedures, and

the implementation of improved procedures (techniques).
The current crack
detection performance level is considered acceptable only on an interim basis,

and additional efforts are necessary to achieve further improvements on a longterm basis.
Currently, programs are under way through NRC, ASME, and the
industry to bring about these improvements.
Generally, the current capability for accurately sizing flaws, particularly flaw depths, is considered unsatisfactory.
marginally acceptable.
Length measurements are
Therefore, better procedures are needed.
Flaw sizing
needs to be examined in the context of the failure probabilities resulting

from a given flaw.
For example, a short flaw even when 100% throughwall will
not cause pipe failure.
Longer flaws require greater care in measurement of
both length and depth.
Currently, programs are under way through NRC, ASME,
and the industry to accomplish these improvements.
G-1
In the long term, both UT systems and piping systems should be designed
to permit automatic examination of welds.
modification of weldments.
This may require redesign or
Such a measure is essential to ensure as low as
reasonably achievable (ALARA) personnel radiation exposure.
Today beneficial
improvements can be achieved by using current technology which automatically

records UT signals and scanning (positional) data, although at this time this
technology is not generally being used in the field.
In new plants and for pipe replacement, attention should be given to

inspectability and accountability for inspection throughout the design stage.
An objective of 100% inspectability for all primary system welds should be
established.
Inspectability is used to denote access to both sides of the
weld, a minimization of signal-to-noise ratio, and a weldment optimized to

permit UT examination.
This consideration should apply equally to mono-,
hi-, and trimetallic welds.
LEAK DETECTION
There should be a careful review of existing leak detection systems

regarding their reliability and sensitivity now and after significant
operating time.
Also, attention should be given to the development of more
sensitive leak detection and continuous crack growth monitoring systems aimed
at detecting leakage with higher confidence and at lower leak rates.
G-2
REPLACEMENT MATERIALS
The suggested replacement material for 304 SS, namely 316NG, is a much
superior material for BWR piping systems.
The material is highly resistant to
initiation and growth of IGSCC in BWR environments.

fabricated welds as well as the base metal.
This applies to properly
Although extensive data exist,
they are not sufficient to say the 316NG material will be completely resistant
to cracking; however, the probability of cracking appears acceptably low.
It
is recognized that none of the suggested replacement materials are resistant

to cracking in severe environments such as those resulting from chloride
intrusions.
WATER CHEMISTRY
EPRI and NRC are sponsoring laboratory research and demonstration

projects at operating plants to evaluate the effectiveness of changes in BWR
water chemistry in the control of both the initiation and propagation of
stress corrosion cracks in primary piping.
These changes include tightening
the controls and limits on intrusion of impurities from the demineralizers and
the injection of hydrogen into the feedwater system to reduce the oxygen
levels.
Possible adverse effects of hydrogen injection on other parts of the
system need to be evaluated from the demonstration projects before hydrogen

injection can be recommended as a mitigating action.
These include effects on
radiation levels in the steam system, on the corrosion product transport

mechanism, and on hydriding of fuel cladding.
G-3
FRACTURE ANALYSIS
Although operating experience suggests a low probabilty of catastrophic

failure in austenitic stainless steel piping systems, certain questions remain
to be answered before the magnitude of safety margins associated with leakbefore-break can be assessed reliably for the desired range of BWR operating
conditions.
These questions generally concern material properties, flaw size,
and analysis methodology.
Work to address each of these items, both in the short and long term,
either has been initiated or will be initiated soon.
Completion of this work
will provide a comprehensive basis for the justification and use of deterministic evaluation techniques to define safety margins.
A combination of probabilistic studies, experimental testing and sophisticated analytical studies should be conducted to establish the relative
probability of leak before break compared to catastrophic failure.
These
programs should address existing materials such as 304 SS and replacement

materials such as 316NG.
The NRC Degraded Piping Program and the probability
fracture mechanics program at Lawrence Livermore National Laboratory are

evaluating this issue.
As the long-term goal, major sources of failure and
system designs should be identified to eliminate such sources of failure.

This will require the joint efforts of NRC, Code Groups, and industry.
G-4
MITIGATING MEASURES
Mitigating measures such as induction heating stress improvement (IHSI)

and heat sink welding (HSW) are aimed at reducing or reversing inner surfce
residual stresses, thus preventing or greatly extending the time for IGSCC
initiation.
Available information is considered adequate to make regulatory
decisions for their use on new pipe; however, further work is required to
establish limits to their use on pipe containing cracks.
This work should
consider a range of flaw depths to establish if long-term limitations exist

with IHSI on pipes in operating plants.
There appear to be no substantial
negative effects related to use of IHSI or HSW.
Many plants have used weld overlay reinforcement on cracked pipes.

Although this is currently considered to be an acceptble short-term (one fuel
cycle) remedy, some utilities may find it expedient to continue to operate
with overlayed welds for longer time periods.
The main questions involved in
the acceptability of longer-term operation with overlayed welds is the

possibility of further crack extension and inspectability.
These subjects
will be addressed by the NRC and new EPRI programs.
IGSCC SUSCEPTIBILITY
IGSCC has been occurring in BWR piping systems for more than 20 years.
Corrective actions have been taken over the past two decades to replace
cracked weldments with low-carbon stainless steels, and these replacements
G-5
have resisted cracking.
However, even after replacement of some of the major
piping systems, there will still remain a great deal of high-carbon sensitized
stainless steel piping; attention should be given in the long term to replacing all weldments potentially susceptible to IGSCC.
Recirculation and other BWR piping systems in older plants were designed
to less rigorous criteria with regard to loads and response to emergency or
faulted conditions.
A long-term goal tied to the replacement of systems
susceptible to IGSCC should be an upgrading in overall design criteria as well

as modification of system geometries to optimize accessibility to weldments
and to minimize the number of weldments.
These efforts should be directed to
meeting new versions of regulatory positions where considered appropriate.
RISK-ASSESSMENT - VALUE-IMPACT ANALYSIS
BWR risk assessment studies, including WASH-1400, have concluded that

large-break LOCAs are only minor contributors to the public risk.
Continuing
review of these studies and the data base are expected to show no change in
this conclusion or the data on which it is based.
Long-term BWR risk
assessment is not expected to produce any different conclusions unless the

data base changes radically with further maturation of the nuclear power plant
population.
G-6
If no radical changes in risk assessment results occur, it is anticipated

that value-impact analysis will involve evaluation of various IGSCC-mitigating
schemes or corrective actions in terms of occupational radiation exposure
(ORE) and utility costs.
SUMMARY
To ensure acceptable margins against IGSCC in BWR piping, utilization of

all three methods beneficial in preventing or delaying IGSCC should be
considered.
These methods are replacing material with more resistant
material, obtaining a beneficial stress distribution through IHSI or HSW, and

minimizing corrosion initiators in the coolant.
G-7
APPENDIX H
April 6, 1984
Mr. Joseph D. Lafleur, Jr.

Deputy Director
Office of International Programs
U. S. Nuclear Regulatory Commission
Washington, D.C. 20555
Dear Mr. Lafleur:
REF:
Your letter 2/17/84, Review of NUREG-1061 Draft Report
In accordance with your wishes, we have completed a review of

NUREG-1061, "Investigation and Evaluation of Stress-Corrosion
Cracking in Piping of Boiling Water Reactor Plants," Pipe
Crack Task Group, Draft Document. Our report is enclosed for
your consideration.
Sincerely,
\UL^C
ITÂC
-J:
Yoshio Ando, Professor Emeritus
University of Tokyo, Japan
Ferenc de Kazinczy,\ President
AB Statens Anlaggningsprovning
Stockholm, Sweden
Karl Kussmaul, Head

Materialpriifungs Anstalt
University of Stuttgart
Federal Republic of Germany
r > ^ _ _ (.
U >_î^
Brian Tomkins, Head
Structural Integrity Center
Northern Division, UKAEA
Risley, United Kingdom
APPENDIX H
REPORT OF INTERNATIONAL REVIEW TEAM

ON DRAFT OF NUREG-1061 - VOLUME I
I N V E S T I G A T I O N AND EVALUATION OF
S T R E S S - C O R R O S I O N CRACKING I N
PIPING
OF B O I L I N G WATER REACTOR PLANTS
P I P E CRACK TASK GROUP
made
for
the
U.
S.
Nuclear
Regulatory
Commission
by
Yoshio Ando, Professor Emeritus

University of Tokyo, Japan
Ferenc de Kazinczy, President
AB Statens Anlaggningsprovning
Stockholm, Sweden
Karl Kussmaul, Full Professor
Head, Materialpriifungsanstalt
University of Stuttgart
Federal Republic of Germany
Brian Tomkins, Head
Structural Integrity Center
Northern Division, UKAEA
Risley, United Kingdom
H-1
INTRODUCTION
The Pipe Crack Task Group have made an extensive appraisal
of the intergranular stress corrosion cracking (IGSCC) problem
in BWR welded Type 304 and 316 stainless steel pipework.
This
has resulted in a comprehensive report, NUREG-1061 (draft document completed March 1984), which we have reviewed in detail.
In general, we find it to be a thorough investigation of the
current status of the problem and the countermeasures which are
in hand to alleviate the damage in operating plants and prevent
its occurrence in new plants.
The report is wide ranging in
its coverage of the relevant issues, but we find that the conclusions and recommendations from individual sections have often
not been sharpened up sufficiently.
Hence, it does not quite
provide as clear a quantitative basis for detailed guidance on

future requirements as is possible on the basis of present
evidence.
In our report below we have taken this into account and
also drawn on our own experience of this phenomenon in Japan,
Germany and Sweden.
However, the views expressed in this report
are personal and do not represent the official view of the regulatory body in our respective countries.
REPORT
Our report follows the order of conclusions and recommendations made in the Executive Summary of NUREG-1061, but we have
tried to identify what we feel to be the six generic issues and
made our comments in response to these.
The issues are:
H-2
1) essence of the problem, 2) inspection, 3) flaw evaluation,

4) stress improvement techniques, 5) nuclear grade stainless
steel, and 6) water chemistry options.
1.
Essence of the Problem

We agree that the IGSCC problem is a result of a combina-
tion of factors:
sensitized material, tensile stress (predom-
inantly residual) and water environment.
However, of the three
factors, the material is the most important and a change of

material to a nonsensitizing grade of austenitic steel (316NG,
347NG, 304NG) essentially removes the HAZ problem.
We believe
that if a new, adequate material replacement (in the old plants)

or designation (for new plant) has been made, there is no need
for a stringent requirement in regard to stress improvement
measures or water chemistry treatment by hydrogen injection.
The German experience with Type 347NG steel over many years and
Japanese experience with Type 316NG steel over the past five
years supports the view that material replacement effectively
removes the problem.
However, even when unsensitized material
is used, it is prudent to adopt careful control of water chemistry by limiting conductivity to <lyS/cm and CI
2.
to <0.1 ppm.
Inspection
Three generic problems must be recognized in the inspection
of piping welds in existing plants for IGSCC:

- crack detection
- crack interpretation
- crack sizing.
All three are discussed at length in the Task Group Report.
H-3
We agree that crack detection and interpretation are considerably improved when procedures laid down in Code Case N-335 M
are followed and the operators are adequately trained on pipe
welds containing IGSCC.
It would undoubtedly be desirable to
improve the validation of personnel, procedures and equipment

even further and to extend the base of reference specimens with
a larger variety of IGSCC.
It must, however, be emphasized that
numerous welds are accessible only from one side or not at all.
For these, crack detection is extremely difficult or impossible.
Therefore, full coverage with respect to detection cannot be
obtained even under optimal conditions. A policy must be developed on a case-by-case basis for such welds in existing plants.
This should take note of the feasibility of effective countermeasures such as replacement or stress improvement (e.g., IHSI).
In new plants, the provision for a reduction in the number of
welded joints and of access for weld inspection should be an obligatory design feature and not just an improvement option as stated
in the recommendations of the Task Group.
We agree with the Task Group conclusions that crack length
can be sized with sufficient accuracy for use in evaluation.
Crack depth cannot, at present, be determined under field conditions with such degree of reliability that the results could be
,used as acceptance criteria.
(NOTE - The coiranent on a 10% flaw
depth acceptability in Section 5.2.3 is not justified in the

light of depth sizing uncertainty.)
H-A
3.
Flaw Evaluation
While agreeing with the Task Group's conclusions regarding
the ability to size crack length with sufficient accuracy, we

feel that more elaborate and stringent rules should be laid down
on flaw acceptability.
The 20% rule on tolerable crack length
is evidently based on service experience with leaks and is supported by limited testing of flawed small diameter pipes and
limited fracture mechanics evaluation.
We agree with the Task
Group Report that adherence to this rule should ensure a leak

rather than a double-ended break.
However, we are concerned about
the potentially low fracture resistance of austenitic weld metal

into which HAZ-initiated IGSCC cracks may be driven.
For such
cases on the large diameter pipes, we feel it is necessary to vali

date by full-scale pipe fracture testing the fracture mechanics
assessment.
Both static and dynamic testing should be considered.
The 20% rule should be taken as giving the maximum size of flaw
which can normally be treated without immediate removal of the
cracked section and replacement.
In order to allow continued
operation in the short term for flaws <20% circumference, countermeasures such as overlay welding or clamshells are acceptable.
In the case of multiple cracks, it is recognized that they may
readily link up.
The ASME XI flaw characterization rules for
multiple cracks should be used to define an "effective" crack

length.
4.
Stress Improvement Techniques

We recognize that the reduction of residual tensile stresses
at the inner pipe wall to near or below zero is an effective
H-5
countermeasure
cracks.
in preventing the initiation and growth of short
This is achieved by IHSI and to a lesser extent by HSW.
Residual stress can also be reduced to near zero by solution

heat treatment (SHT), which also removes weld sensitization.
Use
of the latter is restricted, however, to shop-fabricated welds.

The use of IHSI is supported as a preventative measure rather
than a repair procedure.
It should not be applied to pipes which
are known to be cracked.

5.
Nuclear Grade Stainless Steel

Type 347 modified nuclear grade material is used in Germany
with a C content of <0.04%, S<0.02% and Nb<0.65%, but with a Nb

minimum of 8X%C. The delta ferrite content (Delong) is 2-10%. With
these requirements, the material is completely resistant to IGSCC
and hot cracking on welding.
No fabrication problems have been
experienced in both shop and site fabrication.

IGSCC has been observed in this material.
For 25 years, no
Type 316NG has been
used in Japan for a number of years and more recently in the

United States and Sweden.
In Japan a level of N up to 0.12% is
allowed with a limit on C+N of <0.13% and the CR and Ni equivalents are controlled to ensure a fully austenitic microstructure
in the as-manufactured condition.
After melting and resolidifi-
cation, the material delta ferrite content is a few percent, which

prevents hot cracking.
Where a lower strength level is sufficient,
Type 304NG with similar characteristics to those noted above has

proved adequate in Japanese experience.
6.
Water Chemistry
Possible modifications to the water chemistry, in order to
inhibit IGSCC, are under consideration, in particular the injection of hydrogen to reduce oxygen level to 20 ppb.
injection is still in an experimental
Hydrogen
stage and it remains to
be verified that there are no detrimental side effects (e.g., on

fuel).
We agree with statements in the report concerning syner-
gistic effects of oxygen and anion impurities.
A limit should
be put on conductivity (say ~0.2uS/cm) to warrant any conclusion

that IGSCC will be prevented in sensitized material.
We feel
that at the present time the uncertainties related to hydrogen

injection outweigh its benefits.
However, it may be desirable
to confirm it as an effective alternative countermeasure to stress

improvement for cases where the replacement option is precluded.
Start-up deaeration in order to reduce the oxygen content to
values below 0.2 ppm is desirable.
CONCLUDING REMARKS
IGSCC in BWR piping systems deserves stringent attention
to both the initiation and growth of cracks. Although it is
anticipated that leak-before-break is inherent to tough, austenitic material, it is necessary to base the safety assessment
for replacement of cracked sections and the criterion on a
limitation of crack length.
Current ultrasonic methods, properly
applied, are capable of assuring the criterion of a limiting

crack length.
For welds which are not accessible to ultrasonic
testing, from both sides, it is necessary to consider replacement
H-7
or at least to apply adequate countermeasures against doubleended failure.
Improved nuclear grade materials should be
used for all replacements and any new construction.

ACKNOWLEDGEMENTS
The reviewers are grateful to Dr. S. H. Bush for his elucidation of the philosophy and content of the Task Group Report.
They also appreciate the efficient secretarial service of
Mrs. C. M. Bennett and the support provided by Mr. Howard Faulkner
of NRC.
H-8
APPENDIX I
CONCLUSIONS AND RECOMMENDATIONS FROM NUREG-0531
The relevant conclusions and recommendations from the 1979 review,

NUREG-0531, numbered as in the report, are given below.
2.10
Conclusions
For operating plants and plants under review for an operating license
or construction permit, the industry is implementing on a plantspecific basis many of the recommendations made by the previous Pipe
Crack Study Group.
The cracking experience in a German BWR facility demonstrates that

IGSCC can occur in large-diameter (more than 20 in.) BWR stainless
steel piping and confirms the previous study group's conclusion that,
although cracking in large-diameter pipes has not been observed in the
United States, the conditions for IGSCC do exist in these lines.
We
also have reviewed the safety aspects associated with the potential
for IGSCC in large-diameter BWR stainless steel piping.
Based on our
review, we have concluded that, although throughwall and part-throughwall IGSCC can exist in BWR stainless steel piping during plant
operation, it is unlikely that significant cracks would go undetected.
I-l
In addition, based on our review and analysis, we also conclude that

it is unlikely these cracks will cause unstable crack growth and
excessive loss of coolant.
Further, since the ECCS provides
protection should a loss-of-coolant accident occur, we conclude that

IGSCC in these lines, while generally undesirable, will not be a
hazard to public health and safety.
There have been no IGSCC incidents in large-diameter (more than

20 in.) BWR stainless steel piping in Japan.
Because significant IGSCC has occurred in recirculation-riser piping

in Japan, we conclude that similar cracking may occur in plants in the
United States.
2.11
Recommendations
We recommend that the recommendations contained in NUREG-0313 continue

to be implemented for operating plants and plants under review for an
operating license or construction permit.
Based on the incidence of IGSCC in recirculation-riser piping in

Japan, we recommend that an augmented in-service inspection program
should be developed for these lines.
It is recommended that the
augmented in-service inspection program conform to that recommended

for nonconforming, service-sensitive lines in NUREG-0313.
1-2
Conclusions
IGSCC can occur only if the proper combination of stress, enviromaent

and material conditions are present.
Such a combination exists for
austenitic stainless steels in BWRs, and in FWR systems other than the
primary systems.
Types 304 and 316 stainless steel in the sensitized conditions are
susceptible to IGSCC.
Safe ends of these materials, which were
heavily sensitized by furnace heat treatment, have been cracked in

BWRs.
Piping of Type 304 stainless steel, lightly sensitized by
welding, has been cracked in BWRs and PWRs.
These materials in the
annealed condition, or with low carbon levels, appear to be immune to

IGSCC in oxygenated water, as do weld metal and cast metal of
compositions similar to Type 304 stainless steel, but with a duplex
structure containing ferrite.
In both BWRs and PWRs, stress sources are not well defined, but
residual stressesand particularly residual stresses resulting from
weldingappear to be major contributors to IGSCC.
IGSCC is accelerated by cold working if the material is sensitized.
IGSCC is accelerated by crevices or crevice-like conditions.
1-3
4.8
Recommendations
Continue to make efforts to eliminate or minimize those conditions of

stress, environment and materials that tend to cause IGSCC in BWR
primary systems and PWR systems other than primary systems.
The use of regular grades of Types 304 and 316 stainless steel in BWR
piping systems should be avoided.
If these materials are used, steps
should be taken to ensure that IGSCC cannot occur.
Such measures may
include solutionannealing, weld cladding or other measures that have

been demonstrated to eliminate sensitization and reduce residual
stresses.
Consideration should be given to techniques now being
developed, such as IHSI, Heat Sink Welding (HSW), Solution Heat

Treatment (SHT) of weldments, and Corrosion-Resistant Cladding (CRC),
which are discussed in Chapter 10.
In addition, tests should be run
on each heat of piping to ensure that heats especially susceptible to

sensitization are not used.
Avoid cold working of austenitic stainless steels in LWR piping.

material should be in the annealed condition.
Investigations of IGSCC to assess the effects of actual operating

stress and thermal loading superimposed on constant loading in a
manner that simulates reactor thermal cycles should be continued.
1-4
All
Conclusions
Oxygen dissolved in the primary coolant is the principal environmental

cause of IGSCC in BWR primary piping.
Control of oxygen is highly desirable in BWRs.
Oxygen control during
shutdown and/or startup is considered feasible at the present time for

most operating BWRs.
The data do not exist at the present time to determine if oxygen

control during shutdown and startup will prevent IGSCC in BWR piping.
Deaeration of the condensate storage tank would have a beneficial

effect by reducing the number of instances of cracking in the ECCS and
CRC lines.
The experimental techniques for controlling oxygen during steady-state

operation, currently being developed under DOE sponsorship, may in the
long run be more successful than simple control of oxygen during
shutdown and startup alone.
Contamination of the primary coolant in both BWRs and PWRs from resin
bed intrusions and demineralizer breakthroughs in BWRs or from
inadvertent opening of the boric acid or NaOH core spray lines in PWRs
represents a definite potential cause of stress corrosion.
1-5
5.7
Recommendations
Control of oxygen during shutdown and startup of BWRs should be

practiced wherever possible.
Reduction of oxygen levels in the coolant during steady-state

operation of BWRs is desirable.
Water chemistry excursions, either by accidental intrusion of resin

beds in BWRs or of impurities from core spray coolants in PWRs,
should be reduced to the extent practicable.
6.6
Conclusions
Piping systems in BWRs are complex structures containing many welds.

From a stress analysis standpoint, each weld is itself a complex
structure.
Piping design codes provide upper limits to calculated
operation stresses but, because those stress bounds exceed the

material yield strength, the code stress limits are not appropriate
for controlling the operation stress contributions to IGSCC.
Fabrication (residual) stresses in stainless steel pipe are not
controlled by piping code rules.
1-6
Welding residual stresses (not followed by annealing) are high (e.g.,

25 to 60 ksi). These are axial tensile stresses on the inside surface
and possibly are a major contributor to IGSCC.
The available test
data suggest that residual stresses are higher in small Schedule 80

pipe than in large Schedule 80 pipe; this may explain in part why
there is relatively more cracking in small pipes than in large pipes.
The General Electric Design Stress Rule is a potentially useful tool,

particularly in focusing in-service inspection on welds which, from a
stress standpoint, would be more susceptible to IGSCC.
Recommendations
It is recommended that work be continued to quantify residual stresses

that are due to welding, both at pipe-to-pipe welds and pipe-tofitting or valve welds.
It is also recoimnended that investigation of
procedures to reduce tensile residual stresses on the inside surface,

such as those discussed in Chapter 6, be continued.
It is recommended that a review of stresses in BWR piping systems be

undertaken to establish the operating stress levels as a function of
pipe size.
The purpose would be to see if operating stress averages
help to explain why there is relatively more cracking in small pipes

than in large pipes.
1-7
It is recommended that work continue to improve and apply the Design

Stress Rule.
It is recommended that grinding of the interior surface of weld

regions be avoided under conditions where IGSCC may be present.
8.5
Conclusions
The ASME B&PV Code's Sections V and XI ultrasonic examination

procedures for ferritic weldments are not adequate for the detection
and evaluation of IGSCC in austenitic piping.
Signifcant IGSCC will be detected and evaluated reliably, provided the

improved conventional UT techniques and procedures of recent years are
followed.
Quantitative sizing of IGSCC by amplitude measurements is considered

unreliable.
Difficulty of detection and evaluation of IGSCC by UT leads to

ambiguous results; as a result, a small number of IGSCC will be either
missed or called out incorrectly.
1-8
The current leak detection systems installed in BWRs provide a

reliable complementary means by which to detect cracks.
Experience
indicates that these leak detection systems will detect a crack before
it propagates to a size that would result in a significant loss of
coolant.
Plant shutdown should be initiated for inspection and corrective

action when the leakage system indicates, within a period of four
hours or less, an increase in the rate of unidentified leakage in
excess of 2 gpm, or when the total unidentified leakage attains a rate
of 5 gpm, whichever occurs first.
8.6
Recommendations
The Pipe Crack Study Group recognizes that the following recommendations
are being pursued to a substantial degree and that definite progress is being
made:
For conventional UT testing methods, all reflectors should be mapped

with respect to weld geometry.
classified as IGSCC.
1-9
Reflectors within the HAZ should be
As a minimum, implement a recording procedure by which amplitude and

metal path are recorded automatically.
Positional data of the search
unit may be recorded manually on the amplitude and metal path data
record, or if possible, automatically.
Analysis of the UT data can
then be done by independent examiners under favorable working

conditions.
Advanced detection and evaluation methods are being developed under

several industry and NRC programs.
It is recommended that those
methods particularly useful for examining austenitic materials for

IGSCC be pursued as actively as possible.
The following areas should
be given special emphasis:
- Development and application of methods using advanced signal

processing techniques to reduce dependence on amplitude and aid in
differentiating between IGSCC and other reflectors.
These include
synthetic aperture focusing and adaptive learning network

approaches.
- Development and optimization of ultrasonic probes and acoustic

parameters to enhance through-weld inspections and interpretation of
signals, such as focused pitch-catch, phased arrays, variable
frequencies, and variable angles.
I-IO
- Development and testing of acoustic emission techniques specifically

for detecting IGSCC in austenitic materials, with special emphasis
on weldments.
Any additional work required to apply acoustic
emission to leak detection should also be pursued.
Prepare specific procedures in the ASME Code to incorporate the

improvements in detection and evaluation methods that have been
developed to date.
Issue these in the form of a Regulatory Guide
pending either a Code change or an enabling Code Case to make them

applicable.
Focus inspection efforts upon those pipe welds most likely to crack.
This could reduce the number of welds to be inspected, permitting more
time to be spent on welds most likely to crack without increasing
total personnel radiation exposure or plant down-time.
The use of
General Electric's "Stress Rule" Index to select welds for inspection

should be evaluated as to its applicability in defining the welds most
likely to crack.
Complete work on ISA Standard S 67.03, ANSI Standard N 41.21 entitled.

Standard for Light Water Power Reactor Coolant Pressure Boundary Leak
Detection, and promote serious consideration by the NRC of its
adoption.
I-ll
9.3
Conclusion
From our review of IGSCC incidents in BWRs and our evaluation of the
potential for unstable crack growth, we conclude that it is unlikely that
these cracks will cause unstable crack growth and excessive loss of coolant.
Further, since the ECCS provides protection should a loss-of-coolant accident
occur, we conclude that IGSCC, while generally undesirable, is not a hazard to
public health and safety.
The following considerations are the bases for our
conclusion:
The IGSCC occurs in a small percentage of the welds in BWR piping and
is likely to be detected before it grows to significant size.
Even considering the most severe known IGSCC, large margins against
unstable crack growth are maintained during normal operation and
anticipated transients.
Unanticipated transients that may be
postulated to cause unstable crack growth prior to detection of the

most severe flaws are low-probability events.
A lost-of-coolant accident is a design basis for nuclear facilities in

the United States.
10.6
Conclusions
From the limited information discussed above and in Appendix C, the

following tentative conclusions can be drawn:
1-12
significant variability exists in different heats of commercial

Type 304 stainless steel with respect to sensitization in the asreceived condition.
Also, a significant variability exists in the
response of different piping heats to weld sensitization.
There may be no significant difference in the sensitization behavior

of 4-, 10-, and 26-in.-diameter Schedule 80 stainless steel piping
when welded according to typical welding practices.
All pipes become
sensitized.
The degree of sensitization in the HAZ of certain weldments could

increase significantly after 10-year exposure at reactor operating
temperatures.
The grain boundaries of sensitized Type 304 stainless steel piping

contain less chromium than is required for corrosion resistance.
Also, sulfur and phosphorus concentrate at the grain boundaries where
they can affect sensitization, mechanical properties, and the
electrochemical characteristics of the grain boundary areas.
The electrochemical potentiokinetic reactivation technique can

distinguish between Type 304 stainless steel piping materials that are
susceptible and nonsusceptible to IGSCC, both in the as-received or aswelded conditions.
The technique can thus be used to verify
acceptability of as-received material, of weldments, and of thermally

treated components.
1-13
Bulk residual stresses on the inside surfaces of welded Type 304

stainless steel Schedule 80 piping appear to be lower in 26-in. pipes
than in 4- and 10-in.-diameter pipes.
However, these stresses are
considerably above the 550 F (287C) yield strength of the base

material for all pipe sizes.
Welded Type 304 stainless steel pipes are more prone to sensitization
than would be expected based on total time at sensitizing temperatures
alone because of the synergistic effect between the thermal and strain
cycles.
The effects of continuous cooling on sensitization, as from welding,

are more severe than the effects of isothermal treatments.
IHSI may prove effective in redistributing the welding residual

stresses in BWR piping, so that the pipe's inside surface is left in a
favorable state of axial compression.
For repair welds and welds on new pipes made of Type 304 stainless
steel, several techniques have been identified to reduce the potential
for IGSCC.
welds.
SHT eliminates sensitization but is restricted to shop
A CRC with duplex microstructure can be weld-deposited to
minimize the HAZ on the pipe's inside surface to move the HAZ away
from the high-stressed region next to the butt weld, and to isolate
the butt weld from the environment.
1-14
A water spray on the pipe's inner
surface during welding can produce a HSW with a compressive residual

stress on the pipe's inner surface.
Finally, tighter welding
specifications to limit the amount of post-êld grinding will reduce

the source of high residual stresses.
Recommendations
It is recommended that studies be continued to understand the heat-toheat variability of Type 304 stainless steel piping with respect to
its susceptibility to IGSCC.
Since a large variability exists, it is
recommended that sensitization tests correlated to IGSCC susceptibility in BWRs be conducted for acceptance of piping material.
Further, it is recommended that such tests be used for qualification
of welding procedures and thermal treatments for each heat of piping
material.
The tests should be performed on each heat and/or product
form.
It is recommended that, when Type 304 stainless steel piping is

removed from long-time service, it should be investigated to determine
the presence of low temperature sensitization (LTS).
It is recommended that studies be conducted to determine the

effectiveness of IHSI on pipes with pre-existing flaws and on largediameter (thick-walled) pipes, and to determine if IHSI induces or
accelerates LTS.
1-15
U S NUCLEAR R E G U L A T O R Y COMMISSION
NRC FORM 33S

12 841
NRCM 1102
3201 3202
1 REPORT NUMBER M s / J / i e r f K ' / D C Jrfd I'D/ /Vo
if anyl
NUREG-1061
Volume 1
BIBLIOGRAPHIC DATA SHEET
SEE INSTRUCTIONS ON THE REVERSE

2 T I T L E ANOSUBTLTLE. ,
3 LEAVE BLANK
Report of the U. S. Nuclear Regulatory Commission

/
|Pipinq Review Committee
^
Volume 1 : Investigation and Evaluation o f Stress
Corrosion Crackinq i n Piping o f B o i l i n g Water Reactor Plaijfs
> AUTHOR ISI
Task Group on Pipe Cracks of NRC Piping Review Committee

( S. H. Bush, Chairman)
4 D A T E REPORT C O M P L E T E D
"o^^"
August
1984
6 D A T E REPORT ISSUED
August
7 PERFORMING O R G A N I Z A T I O N NAME A N D M A I L I N G ADDRESS ttncludeZip
Coda)

10 SPONSORING O R G A N I Z A T I O N NAME A N D M A I L I N G ADDRESS (Include Zip Code!
1984
8 PROJECT/TASK/WORK U N I T NUMBER
9 FIN OR GRANT NUMBER
11a TYPE OF REPORT
Regulatory Report
b PERIOD COVERED f / m : / u s / r f a t e i ;
12 SUPPLEMENTARY NOTES
13 ABSTRACT (200 words or lessl
Severe intergranular stress corrosion cracking (IGSCC) of the recirculation piping

system in several boiling water reactors occurred during 1982-1983. A Task Group on
Pipe Cracks was established by the U. S. Nuclear Regulatory Commission with the broad
charter of developing in integrated program to deal with the entirety of the stress
corrosion cracking problem. This report presents specific conclusions and recommendations that are tied closely to relevant regulatory documents so that necessary changes
can be implemented.
This report covers aspects such as the causes and descriptions of IGSCC phenomena;
current status of pipe cracking in BWR's; nondestructive evaluations of piping welds;
inspection of piping for IGSCC; decisions and criteria for replacement, repaid of
continued operation without repair; risks related to the presence of IGSCC; and a valueimpact assessment of IGSCC.
14 DOCUMENT A N A L Y S I S - d
KEYWORDS/DESCRIPTORS
intergranular stress corrosion cracking (IGSCC)

pipe cracking
15 A V A I L A B I L I T Y
STATEMENT
unlimited
16 S B C U H I T Y C L A S S I F I C A T I O N
(This page)
b IDENTIFIERS/OPEN ENDED TERMS
unclassified
(This
rgport}
unclassified
17 NUMBER OF PAGES
NOTICE
Availability of Reference Materials Cited in NRC Publications
Most documents cited in NRC publications will be available from one of the following sources:
1. The NRC Public Document Room..1717 H Street, N.W.
2. TheNRC/GPOSalesProgram,U.S. Nuclear Regulatory Commission,
3. The National Technical information Service, Springfield, V A 22161
Although the listing that follows represents the majority of documents cited in NRC publications,
it is not intended to be exhaustive.
Referenced documents available for inspection and copying for a fee from the NRC Public Document Room include NRC correspondence and internal NRC memoranda; NRC Office of Inspection
and Enforcement bulletins, circulars, information notices, inspection and investigation notices;
Licensee Event Reports; vendor reports and correspondence; Commission papers; and applicant and
licensee documents and correspondence.
The following documents in the NUREG series are available for purchase from the NRC/GPO Sales
Program: formal NRC staff and contractor reports, NRC-sponsored conference proceedings, and
NRC booklets and brochures. Also available are Regulatory Guides, NRC regulations in the Code of
Federal Regulations, and Nuclear Regulatory Commission Issuances.
Documents available from the National Technical Information Service include NUREG series
reports and technical reports prepared by other federal agencies and reports prepared by the Atomic
Energy Commission, forerunner agency to the Nuclear Regulatory Commission.
Documents available from public and special technical libraries include all open literature items,
such as books, journal and periodica! articles, and transactions. Federal Register notices, federal and
state legislation, and congressional reports can usually be obtained from these libraries.
Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference
proceedings are available for purchase from the organization sponsoring the publication cited.
Single copies of NRC draft reports are available free upon written request to the Division of Technical Information and Document Control, U.S. Nuclear Regulatory Commission, Washington, DC
20555.
Copies of industry codes and standards used in a substantive manner in the NRC regulatory process
are maintained at the NRC Library, 7920 Norfolk Avenue. Bethesda. Maryland, and are available
there for reference use by the public. Codes and standards are usually copyrighted and may be
purchased from the originating" organization or. if they are American National Standards, from the
American National Standards Institute. 1430 Broadway, New York. NY 10013.
GPO Printed copy price: O - O C J '

NUREG-1061 Vol.1 PDF

Uploaded by

Copyright:

Available Formats

NUREG-1061 Vol.1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NUREG-1061 Vol.1 PDF

Uploaded by

Copyright:

Available Formats

"0 NOT^^^weiH

U.S. Nuclear Regulatory

DISTRIBUTION OF THIS DOCUMENT IS UNLIMITED

GPO Printed copy price:

Manuscript Completed: August 1984

U.S. Nuclear Regulatory Commission

It has been reprodaeetf from the best

PIPE CRACK TASK GROUP MEMBERS

CAUSES AND DESCRIPTIONS OF IGSCC PHENOMENA

PRESENT STATUS OF PIPE CRACKING IN OPERATING BWRs

NONDESTRUCTIVE EVALUATION OF PIPING WELDS

TABLE OF CONTENTS (Cont'd)

INSPECTION OF PIPING FOR IGSCC

DECISION AND CRITERIA FOR REPACEMENT, REPAIR, OR

IGSCC RISK STUDIES

INTERGRANULAR STRESS-CORROSION CRACKING (IGSCC)

LIST OF TASK GROUP MEMBERS

Summary of All Inspection Findings on Large

Capabilities of Leakage Monitoring Methods

Condition and Inspection Required for Weld

Results of Surface Crack Analyses for 12-inch-diameter

Recommended Measurements for Controlling IGSCC in

Sensitivity of Core Melt Frequency for BWR PRAs for

Risk Worth for Safety Systems* with Respect to Core

Value-Impact Summary Pipe Replacement

Value-Impact Summary IHSI

Value-Impact Summary Augmented Inspection and Repair

Sensitivity of Core Melt Frequency for BWR PRAs for

RSS Pipe Failure Assessed Values

Hypothetical Failure Frequency Estimates

LIST OF TABLES (Cont'd)

Fatigue and Corrosion Failures as a Function of

List of Typical Piping Systems Involved in the

Risk Worth for Safety Systems with Respect to Core

ECCS Single-Valve-Failure Analysis

Average Operating BWR Plant

UT Inspection Summary Data

Weld Overlay Summary Data

Pipe Replacement Summary Data

IHSI Summary Data

HWC Summary Data

Recirculation System Pipe Dimensions

Recirculation System Typical and Bounding Load

Tensile Properties for Stainless Steel Weld

Jjc Values at 550F for Submerged Arc and Shielded

Instability Moment for Bounding Load SLA Leakage

Crack Size for Instability at SLA + ISSE Load

Estimated Margin Against Unstable Crack Extension

Margin Against Instability for Total Load

Allowed limits on reactor coolant system leakage

Histograms based on the technical specifications for

Stress-corrosion crack growth rates

Throughwall distribution of axial residual stress in

Throughwall distribution of axial residual stress in a

Assumed throughwall welding residual stress distribution

Comparison of elastic-plastic solutions for throughwall

Comparison of elastic-plastic solutions for throughwall

Comparison of elastic-plastic solutions for throughwall

Collapse curve for a lO-inch pipe with throughwall and