NUREG/CR-6456, Forwards for Info & Use,Aeod Rept, Review of Industry Efforts to Manage PWR Feedwater Nozzle,Piping & Feedring Cracking & Wall Thinning, NUREG/CR-6456
| ML20137R394 | |
| Person / Time | |
|---|---|
| Issue date: | 04/09/1997 |
| From: | Rossi C NRC OFFICE FOR ANALYSIS & EVALUATION OF OPERATIONAL DATA (AEOD) |
| To: | Gillespie F, Thadani A, Zimmerman R NRC (Affiliation Not Assigned) |
| References | |
| RTR-NUREG-CR-6456, TASK-*****, TASK-AE AEOD-*****, AEOD-E97-01, AEOD-E97-1, NUDOCS 9704140107 | |
| Download: ML20137R394 (198) | |
Text
__
N*
A pn car p-t i
UNITED STATES g
j NUCLEAR REGULATORY COMMISSION WASHINGTON. D.C. 2065H001 April 9, 1997 AE0D/E97-01 MEMORANDUM TO: Frank P. Gillespie, Director, DISP:NRR Ashok C. Thadani, Associate Director, ADT:NRR Roy P. Zimmerman, Associate Director, ADPR:NRR Gary M. Holahan, Director, DSSA:NRR Brian Sheron, Director, DE:NRR Thomas T. Martin, Director, DRPM:NRR Bruce A. Boger, Director, DRCH:NRR Lawrence C. Shao, Director, DET:RES M. Wayne Hodges, Director, DST:RES Bill M. Morris, Director, DRA:RES James T. Wiggins, Director, DRS:RGN-1 Johns P. Jaudon, Director, DRS:RGN il Geoffrey E. Grant, Director, DRS:RGN-ill Arthur T. Howell, Director, DRS:RGN-IV FROM:
Charles E. Rossi, Dire Safety Programs Division Office for Analysis and Evaluation of Operational Data
SUBJECT:
" REVIEW OF INDUSTRY EFFORTS TO MANAGE PRESSURIZED WATER REACTOR FEEDWATER NOZZLE, PIPING, AND FEEDRING CRACKING AND WALL THINNING," NUREG/CR-6456 (INEL 96/0089)
Att' ached for your information and use is a copy of the report, " Review of Industry Efforts to Manage Pressurized Water Reactor Feedwater Nozzle, Piping, and Feedring Cracking and Wall Thinning," NUREG/CR-6456. The study objective was to provide a comprehensive overview document discussing pressurized-water reactor (PWR) feedwater nozzle, piping, and feedring cracking and wall thinning; safety aspects; and industry actions taken to manage these issues. The report should be useful in identifying, assessing, and evaluating program options to manage these issues. The time frame spans from initial discovery of feedwater nozzle cracking in 1979 through 1996.
The review and assessment effort covers relevant field experience with PWR feedwater systems, the factors causing the damage, design modifications, operating procedure changes, augmented inspection programs, and repair and replacement activities carried out because of the degradation that occurred. The effort focused on the feedwater system U
CONTACT:
Earl J. Brown, AEOD/SPD/RAB (301) 415-7572 g-Q012 Q
g own Map 9704140107 970409 i
5,.l f 0 W.b%$ a % f PDR ORO NEXD PDR
,,. 3 F. Gillespie, et al. adjacent to the feedwater nozzle where fatigue cracking and wall thinning had been reported. This includes the main and auxiliary feedwater piping adjacent to the feedwater nozzle, and the thermal sleeve, feedring, and J-tubes. The principal areas reviewed were (1) feedwater system design, (2) safety significance of feedwater line rupture, (3) fatigue cracking experience, (4) flow-accelerated corrosion-induced wall thinning experience, (5) steam generator water hammer damage experience, (6) degradation mechanisms, (7) inservice inspection methods, and (8) mitigation, monitoring, and replacement activities.
This effort emphasized understanding the technical aspects for each area in order to assess the impact on managing (establish confidence that limits for safe operation are maintained) degradation of these components. In this context, the study concentrates on causes, mechanisms, conditions (temperature, pressure, environment, etc.), inspections, procedures, and corrective actions from the perspective of capability to assess a specific aspect such as crack characterization. Thus, the report is directed toward determining whether current technology is sufficient to " manage the problem".
The operating experience review addressed feedwater nozzle cracking caused by thermal fatigue; flow-accelerated corrosion wall thinning of carbon steel J-tubes, feedrings, and thermal sleeves in top-feed steam generators, and auxiliary feedwater lines in preheat steam generators; and water hammer damage in both types of steam generators.
Feedwater nozzle cracking was detected in 18 PWRs from 1979 through 1983. These were found as a result of IE Bulletin 79-13, " Cracking in Feedwater System Piping," 1979.
There was approximately one event per year from 1983 until bulletin closure in 1991. The nozzle cracking event frequency increased to six per year for 1992 and 1993 but no additional events were detected from then through 1996. It appears that licensee action has been sufficient to minimize flow-accelerated wall thinning in J-tubes and auxiliary feedwater lines and while wall thinning in feedrings and thermal sleeves was not addressed i
by specific action, there was evidence of visual inspection and repair when needed.
Similarly, it appears that licensees have taken sufficient action, primarily design modifications and operating procedure changes, to minimize water hammer events. There were 28 events reported during the 1970s,6 during the 1980s, and none after that.
The technical findings as a whole indicate that appropriate analysis, inspection, monitoring, j
mitigatiori lheluding operational procedures), and replacement techniques have been developed so that thermal fatigue and flow-accelerated corrosion damage to feedwater nozzles, piping, and feedrings can be managed effectively. This simply means the tools are available to manage. However, this managing process requires detailed knowledge about component and system design, construction, and materials; cognizance of operating procedures (especially the potential for extended operation at startup or hot standby with j
automatic auxiliary feedwater control); in-depth understanding of factors that cause thermal fatigue and flow-accelerated corrosion; and adequate training in the use of predictive analysis methods and advanced inspection techniques. The staff at several PWR plants have been proactive in managing this type of damage.
i
o
,a i
F. Gillespie, et al. An overview observation from the study is that plant specific aspects control both the degree of susceptibility and the appropriate solution scheme to " manage the problem".
Therefore, plant specific solutions would be anticipated as the rule rather than a generic program.
The Office of Nuclear Reactor Regulation and Office of Nuclear Regulatory Research provided review comments on drafts of this report. In addition, comments were received from technical experts who provided information during plant visits or information in the technical literature. The organizatic,ns included were the Tennessee Valley Authority, Southern California Edison, Pacific Gas and Electric, Westinghouse Electric, and Electric Power Research Institute.
Attachment:
As stated cc w/att.:
V. K. Chexal, EPRI Pat Lewis, INPO Debbie W. Queener, NOAC Distribution w/att.:
Public EJordan, DEDO RSavio, ACRS TMe; Center llq DTerao, NRR JMitchell, OEDO RAB R/F JFair, NRR RJohnson, RES SPD R/F AChaffee, NRR KParczewski, NRR Dross RWessman, NRR KManoly, NRR FCongel JStrosnider, NRR DHickman MHartzman, NRR PBaranowsky MMayfield, RES KRaglin, TTC REmrit, RES 6
DOCUMENT NAME: C :WP51 \\WPDO C S\\STU DI E S\\E EPRE S S 1. EJ B To receive a copy of thle document. Indkate in the box: "C" e Copy without a t/
sure "E e Copy with attachment / enclosure V e No copy j
OFFICE RAB E
RABgj l6 C:RA7 D:SPD E
Report NAME EBrown:mmkEg Gigfrip JRosenthat CRossicf g received frun DATE 03//fl97 03g./97 03/Jj/97 03($37 ons OFFICIAL RECC'RD COPY
W NUREG/CR-6456 INEL-96/0089 AEOD/E97-01 Review.of Industry Efforts To Manage Pressurized Water Reactor Feedwater Nozzle, Piping, and
- Feedring Cracking and Wall Thinning i
)
N. b$1ah A. G. Ware, A. M. Porter Idaho National Engineering Laboratory 4
Lockheed Idaho Technologies Company Prepared for U.S. Nuclear Regulatory Commission 4
J l
e t
o t.
i AVAILABluTY NOTICE Availabihty of Reference Matenals Cited in NRC Pubhcations Most documents cited in NRC publications will be available from one of the foGrwing sources:
1.
The NRC Pubuc Document Room. 2120 L Street, NW., Lower Level. Washington, DC 20555-0001 2,
The Superintendent of Documents. U.S. Government Pnnting Office, P. O. Box 37082. Washington, DC 20402-9328 I
I 3.
The National Technical information Service, Springfield, VA 22161-0002 j
l Although the listing that fonows represents the majority of documents cited in NRC pubhcations, it is not in-tended to be exhaustive.
j Referenced documents available for inspection and cooying for a fee from the NRC Public Document Room include NRC correspondence and internal NRC memoranda; NRC bulletins, circulars, information notices, in-spection and investigation notices; heensee event reports: vendor reports and correspondence; Commission papers; and applicant and Wensee documents and correspondence.
The following documents in the NUREG series are available for purchase from the Government Printing Office:
f formal NRC staft and contractor reports, NRC-sponscred conference proceedings, international agreement i
reports, grantee reports, and NRC booklets and brochures. Also available are regulatory guides, NRC regula-1 tions in the Cocfe of Federal Regulations, and Nuclear Regulatory Commission Issuances.
j
\\
Documents ava#able from the National Technicalinformation Service include NUREG-series reports and tech-nical reports prepared by other Federal agencies and reports prepared by the Atomic Energy Commission, forerunner agency to the Nuclear Regulatory Commissirn.
Documents available from public and special technical hbraries include all open kterature items, such as books, journal articles, and transactions. Federal Regtster notices, Federal and State legislation, and congressional reports can usua0y be obtained from these libraries.
Documents such as theses dissertations. foreign reports and translations, and non-NRC conference pro-condings are avanable for purchase from the organization sponsonng the publication cited.
Single copies of NRC draft reports are available free, to the extent of supply, upon written request to the Office e8 Administration, Distribution and Mall Services Section, U.S. Nuclear Regulatory Commission. Washington, DC 20555-0001, Copies of industry codes and standards used in a substantive manner in the NRC regulatory process are main-tained at the NRC Ubrary, Two White Filnt North,11545 Rockville Pike RockvlDe, MD 20852-2738, for use by j
the public, Codes and standards are usually copyrighted and may be purchased from the originating organiza-tion or, if they are American National Standards, from the American National Standards institute,1430 Broad-way, New York, NY 10018-3308.
DISCLAIMER NOTICE This report was prepared as an account of work sponsored by an agency of the United States Government.
Neitherthe United States Government nor any agency thereof, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility for any third party's use, or the results of such use, of any information, apparatus, product, or process disclosed in this report, or represents that its use by such third party would not infringe privately owned rights.
j
NUREG/CR-6456 INEL-96/0089 AEOD/E97-01 Review of Industry Efforts To Manage Pressurized Water Reactor Feedwater Nozzle, Piping, and Feedring Cracking and Wall Thinning Manuscript Completed: March 1997 Date Published: March 1997 Prepared by V. N. Shah, A. G. Ware, A. M. Porter Idaho National Engineering Laboratory Lockheed Idaho Technologies Company Idaho Falls, ID 83415-3129 E. J. Brown, NRC Technical Monitor E. A. Trager, NRC Project Manager Prepared for Safety Programs Division 04lce for Analysis and Evaluation of Operational Data U.S. Nuclear Regulatory Commission l
W;shington, DC 20555-0001 NRC Job Code E8238
r%
t UNITED STATES g
NUCLEAR REGUL.ATORY COMMISSION I
WASHINGTON, D.C. 3000541001 TO ALL RECIPIENTS:
SUBJECT:
' REVIEW OF INDUSTRY EFFORTS TO MANAGE PRESSURIZED WATER REACTOR FEEDWATER NOZZLE, PlPING, AND FEEDRING CRACKING AND WALL THINNING," NUREG/CR-6456 (INEL 96/0089)
Attached for your information and use is a copy of the report," Review of Industry Efforts to Manage Pressurized Water Reactor Feedwater Nozzle, Piping, and Feedring Cracking and
' Wall Thinning," N'UREG/CR-6456. The study objective was to provide a comprehensive
.overvie'w document discussing pressurized-water reactor (PWR) feedwater nozzle, piping, and feedring cracking and wall thinning; safety aspects; and industry actions taken to manage these issues. The report should be useful in identifying, assessing, and evaluating program options to manage these issues. The time frame spans from initial discovery of feedwater nozzle cracking in 1979 through 1996.
The review and assessment effort covers relevant field experience with PWR feedwater systems, the factors causing the damage, design modifications, operating procedure changes, augmented inspection programs, and repair and replacement activities carried out because of the degradation that occurred. The effort focused on the feedwater system adjacent to the feedwater nozzle where fatigue cracking and wall thinning had been reported. This includes the main and auxiliary feedwater piping adjacent to the feedwater nozzle, and the thermal sleeve, feedring, and J-tubes. The principal areas reviewed were (1) feedwater system design, (2) safety significance of feedwater line rupture, (3) fatigue cracking experience, (4) flow accelerated corrosion-induced wall thinning exparience, (5) steam generator water hammer damage experience, (6) degradation mechanisms, (7) inservice inspection methods, and (8) mitigation, monitoring, and replacement activities.
This effort emphasized understanding the technical aspects for each area in order to assess the impact on managing (establish confidence that limits for safe operation are maintained) degradation of these components, in this context, the study concentrates on causes, mechanisms, conditions (temperature, pressure, environment, etc.), inspections, procedures, and corrective actions from the perspective of capability to assess a specific aspect such as crack characterization. Thus, the report is directed toward determining whether current technology is sufficient to ' manage the problem".
Ths operating experience review addressed feedwater nozzle cracking caused by thermal fatigue; flow accelerated corrosion wall thinning of carbon steel J-tubes, feedrings, and thermal sleeves in top-feed steam generators, and auxiliary feedwater lines in preheat i
steam generators; and water hammer damage in both types of steam generators.
Feedwater nozzle cracking was detected in 18 PWRs from 1979 through 1983. These were found as a result of IE Bulletin 79-13, " Cracking in Feedwater System Piping," 1979.
i i
There was approximately one event per year from 1983 until bulletin closure in 1991. The l
nozzle cracking event frequency increased to six per year fo61992 and 1993 but no additional events were detected from then through 1996. ' It appears that licensee action has been sufficient to minimize flow-accelerated wall thinning in J tubes and auxiliary
.feedwater lines and while wall thinning in feedrings and thermal sleeves was not addressed l
l by specific action, there was evidence of visual inspection and repair when needed.
j Similarly, it appears that licensees have taken sufficient action; primarily design modifications and operating procedure changes, to minimize water hammer events. There were 28 events reported daring the 1970s,6 during the 1980s, and none after that.
The technical findings as a Mole indicate that appropriate analysis, insp6ction, monitoring, mitigation (including operatioral procedures), and replacement techniques have been developed so that thermal fatigae and flow accelerated corrosion damage to feedwater nozzles, piping, and feedrings can be managed effectively. This simply means the tools are j
available to manage. However, thh managing process requires detailed knowledge about component and system design, construction, and materials; cognizance of operating procedures (especially the potential for extended operation at startup or hot standby with automatic auxiliary feedwater control); in-depth understanding of factors that cause thermal fat'gue and flow-accelerated corrosion; and edequate training in the use of i
predictive analysis methods and advanced inspectior, techniques. The staff at several PWR plants have been proactive in managing this type of damage.
i An overview observation from the study is that plant specific aspects control both the degree of susceptibility and the appropriate solution scheme to " manage the problem".
Therefore, plant specific solutions would be anticipated as the rule rather than a generic program.
i i
The Office of Nuclear Reactor Regulation and Office of Nuclear Regulatory Research 3
provided review comments on drafts of this report. In addition, comments were received from technical experts who provided information during plant visits or information in the t
technical literature. The organizations included were the Tennessee Valley Authority, Southern California Edison, Pacific Gas and Electric, Westinghouse Electric, and Electric Power Research Institute.
[f g r
ossi, irector Safety Programs Division Office for Analysis and Evaluation of Operational Data
ABSTRACT His report presents our review of the nuclear industry efforts to manage thermal fatigue, flow-accelerated corrosion, and water hammer damage to pressurized water reactor (PWR) feedwater nozzles, piping, and feedrings. De review includes an evaluation of design modifications, operating procedure changes, augmented inspection and monitoring programs, and mitigation, repair and replacement activities. Four specific actions were taken to perform the evaluation: (a) review of field experience to identify trends of operating events, (b) review of related technical literature, (c) visits to three PWR plants and a PWR vendor, and (d) solicitation ofinformation from eight other countries.
He characteristics and safety implications of the damage caused by flow stratification-induced thermal fatigue are different than those caused by flow-accelerated corrosion. Hermal fatigue cracking has generally occurred in a relatively local, safety-related portion of the feedwater piping inside the containment, whereas wall-thinning caused by flow-accelerated corrosion has typically occurred, with few exceptions, in the non-safety-related balance ofplant piping outside the containment. Field experience has shown that through-wall thermal fatigue cracks in the feedwater nozzles and piping have leaked and provided warning, whereas a component damaged by flow accelerated corrosion loses its strength and can fail catastrophically without warning during normal operation.
The factors causing thermal fatigue and flow-accelerated corrosion are well understood, inservice inspection techniques are available for reliable detection and accurate sizing of thermal fatigue cracks. Cost-elfective radiographic techniques for measuring wall thinning are being developed. Modifications in the plant operating procedures and the feedwater system designs have been made to mitigate thermal fatigue damage.
Optimized water chemistry has been developed for mitigating flow-accelerated corrosion damage to both single-and two-phase portions of the PWR secondary systems. Corrosion resistant materials are used to replace the carbon steel components damaged by flow-accelerated corrosion.
Our assessment of field experience is that the USNRC licensees have apparently taken sufficient action to minimize the feedwater nozzle cracking caused by thermal fatigue and the wall thinning of J-tubes and feedwater piping. Ilowever, we did not find specific industry actions to minimize the wall-thinning in feedrings and thermal sleeves, but we found visual inspection being performed and repair when needed.
Our assessment of field experience also indicates that the USNRC licensees have taken suflicient action to minimize steam generator water hammer in both top-feed and preheat steam generators. Ilowever, we have not evaluated the industry efforts to minimize multiple check valve failures that have allowed backflow of steam from a steam generator and have played a major role in several steam generator water hammer events.
A major finding of this review is that analysis, inspection, monitoring, mitigation, and replacement techniques have been developed for managing thermal fatigue and fiow-accelerated corrosion damage to feedwater nozzles, piping, and feedrings. Adequate training and appropriate applications of these techniques would ensure effective management of this damage. Several PWR plant operators have been proactive in managing this damage.
i i
Job Code E8238 - Specialized Technical Assistance iii NUREG/CR-6456 i
x CONTENTS A B STRACT........................................................................ iii LI ST OF FIGURES................................................................. viii LI ST OF TAB LES.................................................................. xii EXEC UTIVE
SUMMARY
........................................................... xiii A C R ONY M S................................................................... xxv ACKNOWLEDGMENTS
..................................................xxvu-
- 1. INTRODUCTI ON...............................................................
1
- 2. MAIN AND AUXILIARY FEEDWATER SYSTEM DESCRIPTIONS..................... 4 1
2.1 Main Feedwater System........................................................ 4 i
2.1.1 Startup Feedwater System...............................................
10 2.2 Auxiliary Feedwater System..................................................
1I 2.2.1 Preheater-Equipped Steam Generators.....................................
1I
- 3. DESIGN, MATERIALS, AND FABRICATION.......................................
15 3.1 Design Configuration........................................................
15 3.2 Meterials.................................................................
19 3.3 Geom etric Discontinuity..................................................... 20
- 4. OPERATING TRANSIENTS AND ENVIRONMENT.................................. 24 4.1 Auxiliary Feedwater Operation..
............................................ 24 4.2 Environm ent............................................................. 27
- 5. THE SAFETY SIGNIFICANCE OF FEEDWATER LINE RUPTURE..................... 29
- 6. FIELD EXPERIENCE RELATED TO CRACKING AND WALL THINNING OF FEEDWATER NOZZLES, PIPING, AND FEEDRINGS............................... 31 6.1 Initial Discoveries of Fatigue Cracking.......................................... 32 s
I l
2 1
1 v
i 1
1 6.2 Crack Discoveries Between Initial Inspections and Bulletin Closeout.................. 36 6.2.1 Maine Yankee......................................................... 3 6 6.2.2 Indian Point Unit 2..................................................... 3 6 1
6.3 Fatigue Cracking AAer Bulletin Closcout....................................... 41 6.3.1 Sequoyah Units I and 2................................................. 42 6.3.2 Diablo Canyon Units 1 and 2............................................. 54 6.3.3 San Onofre Unit 3...................................................... 60 6.3.4 Other PWRs.......................................................... 62 6.3.5 Non.U S Plants........................................................ 63 1
4 6.4 Flow-Accelerated Corrosion.................................................. 64
,l 6.4.1 Feedwater Piping...................................................... 64 6.4.2 Feedrings and Hermal Sleeves........................................... 69 6.4.3 Non-U.S. Plants....................................................... 74 6.5 Steam Generator Water Hammer Damage........................................ 76 6.5.1 Steam Generator Water Hammer Phenomenon............................... 76 6.5.2 Top-Feed Steam Generators............................................. 78 6.5.3 Preheat Steam Generators............................................... 82 l
- 7. DEG RADATION MECHANI SMS................................................. 86 i
7.1 Herm al Fatigue............................................................ 86 7.1.1 Thermal Stratification.................................................... 86 7.1.2 Thermal Striping and Turbulent Mixing.................................... 91 7.1.3 Thermal Cycling........................................................ 92 7.2 Flow-Accelerated Corrosion.................................................. 94 7.2.1 Flow-Accelerated Corrosion Phenomena.................................... 94 7.2.2 Factors Affecting Flow-Accelerated Corrosion............................... 96 I
- 8. INSERVICE INSPECTION OF PRESSURIZED WATER REACTOR FEEDWATER NOZZLES AND PIPING..........................................
106 8.1 Inservice Inspection Requirements..........................................
106 8.1.1 Fabrication, Preservice Inspection, and Inservice Inspection Requirements.......
106 8.1.2 Inspections in Response to IE Bulletin 79-13...............................
107 8.1.3 Activities Since Close-out ofIE Bulletin 79-13..............................
10 8 i
8.2 Advancements for Inservice Inspection of Fatigue Cracks.........................
109 8.2.1 Emerging Inservice Inspection Methods.................................... I 10 8.2.2 UT Performance Demonstration.........................................
120 8.2.3 Risk-based Inspection (ASME Section XI Code Case)........................
121 8.3 Inservice Inspection of Wall Thinning Caused by Flow-Accelerated Corrosion.........
121 8.3.1 Inspection in Response to Bulletin 8 7-01...................................
12 2 8.3.2 Emerging Inservice inspection Methods for Wall Thinning....................
123 8.3.3 Wall Thinning Inspection Criteria........................................
126 8.3.4 USNRC Audits and Inspections.........................................
127 i
NUREG/CR-64$6 vi i
t 8.4 Inservice inspections at Non-US Plants......................................... 128 8.4.1 Inservice inspections of Fatigue Cracks at Non-US Plants..................... 128 8.4.2 Inservice inspections of Wall Thinning at Non-US Plants..................... 129 9.
MITIGATION AND MONITORING OF FATIGUE, FLOW-ACCELERATED CORROSION, AND WATER HAMMER DAMAGE............................................. 131 9.1 Mitigation of Thermal Fatigue Damage......................................... 131 9.2 Fatigue Monitoring of feedwater Lines.........................................
132 93 Mitigation of Flow-Accelerated Corrosion Damage...............................
137 93.1 Optimized Feedwater Chemistry.........................................
137 9.3.2 Use of Corrosion Resistant Materials...................................... 139 9.3.3 Other Modifications...................................................
140 9.4 Mitigation of Water Hammer Damage.........................................
140
- 10. FIN DING S..................................................................
14 3 s
10.1 Maj or Findin gs.......................................................... 143 10.2 Specific Findings - Degradation Mechanisms.................................
144 10.2.1 Thermal Fatigue...................................................
144 10.2.2 Flow-Accelerated Corrosion.........................................
145 10.3 Specific Findings - Inservice Inspe:tions...................................... 146 a
10.3.1 Characterization of Fatigue Cracks 146 10.3.2 Characterization of Wall Thinning...................................
146 10.4 Specific Findings - Mitigation, Monitoring, Repair, and Replacement..............
147 1 1. RE F EREN C ES..............................................................
14 9 i
APPENDIX - Questionnaire Submitted to Foreign Countries.............................. A-1 4
f i
vii NUREG/CR-6456
1 LIST OF FIGURES f
I 1.
Sketch of a portion of the PWR feedwater system and its components
- evaluated in the report........................................................ 2 s
2.
Schematic of a typical main feedwater system for a Westinghouse 4-loop plant j
with top-feed steam generators................................................. 5 3.
Sketch of a recirculating steam generator without a preheater.......................... 6 4.
Sketch of a recirculating steam generator with a preheater............................. 6 i
5.
Top view of Diablo Canyon feedring............................................. 7 f
6.
Schematic of a typical main feedwater system for a Babcock & Wilcox plant............. 8 l
- 7.
Cross-section of a Babcock & Wilcox once-through steam generator with external fe edring (header)............................................................. 9 I
8.
Schematic of an auxiliary feedwater system for a 4 loop Westinghouse plant with j
the top-feed steam generators...............................................
12 l
I 9.
Schematic of a feedwater system for a Westinghouse plant with steam l
generators equipped with preheaters...........................................
13 10.
90 degree elbowjoining feedwater nonle and piping in the original installation at D. C. Cook plants.............
15 i
11.
Expanderjoining feedwater nonle and piping at the San Onofre I plant.................. 16 12.
Sketch of a feedwater nonle at a typical Combustion Engineering plant................. 17 1
13.
Modified feedwater nonle and piping configuration at D. C. Cook plant............... 17 14.
San Onofre Units 2 and 3 feedwater distribution piping.............................
18 15.
Typical PWR feedwater nonle-to-pipe weld with a counterbore...................... 20 l
16.
Construction drawing of transition piece for Sequoyah Units 1 and 2.................. 21 i
17.
Construction drawings for feedwater nonle and piping welds at j
Diablo Canyon Units 1 and 2................................................. 22 4
i 18.
Geometric discontinuity introduced by the field welding between the feedwater nonle and the transition piece........................................ 23 4
1
!t NUREG/CR 6456 viii
i i
I l
l 19.
Flow stratification in PWR feedwater nozzle during low flow conditions............... 25 l
l 20.
Auxiliary feedwater flow for Sequoyah Unit I showing automatic and manual controls.... 26 21.
Auxiliary feedwater flow for Sequoyah Unit 2 showing automatic and manual controls.... 27 22.
Typical fatigue crack orientation at feedwater nozzle counterbore..................... 33 l
23.
Locations of thermal fatigue cracks in an Indian Point Unit 2 feedwater nozzle and steam generator shell caused by bypass leakage of cold feedwater..................... 39 24.
Indian Point Unit 2 feedwater nozzle sealing sleeve insta!!ed to prevent bypass leakage of cold feedwater.............................................. 42 25.
Stratifled flows during operation of auxiliary feedwater system....................... 43 26.
Through-wall crack in the feedwater nozzle weld at Sequoyah Unit 1.................. 44 27.
Extent and orientation of primary cracks in feedwater nozzle-to transition field weld at Sequoyah Units 1 and 2 found in 1992............................... 46 28.
Direction of propagation of a primary crack at Sequoyah Unit 1...................... 48 29.
Loop 1 auxiliary feedwater flow rates during manual and automatic operation at Sequoyah Unit 1 on November 29,1994............................. 50 30.
New thermal liner protecting the feedwater nozzle and elbow from thermal stratification loadings................................................. 52 l
l 31.
Weld buildup to eliminate counterbore as a stress raiser............................. 53 32.
Feedwater nozzle area layout at Diablo Canyon Units 1 and 2....................... 54 33.
Temperatures and flow rate during cooldown on 9/13/92 at Diablo Canyon Unit 1, Steam Generator 1.................................................... 5 8 34.
Distribution of axial stresses in a long horizontal feedwater piping for two different elevations of hot-and-cold coolant interface and for two different boundary conditions for the pipe ends.......................................... 59 35.
San Onofre Units 2 and 3 feedwater nozzle region................................. 60 36.
Surry feedwater pipe rupture caused by flow-accelerated corrosion.................... 65 37.
Changes in the Diablo Canyon J tube design to mitigate flow-accelerated corros ion d am a ge......................................................... 70 l
i i
38.
A redesigned feedwater nozzle-to-piping connection at Diablo Canyon Units I and 2..... 72 39.
Loviisa Unit i feedwater pipe rupture caused by flow-accelerated corrosion............ 74 40.
Sequence of events leading to water hammer..................................... 77 41.
Older vintage feedring design for top-feed steam generators......................... 80 42.
Main and auxiliary feedwater systems for KRSKO................................ 84 43.
KRSKO Loop 2 auxiliary feedwater line......................................... 85 44.
'Ihermal skisses in a pipe with stratified flow..................................... 87 1
45.
Fatigue data for SA-106 Grade B carbon steel smooth base metal specimens in air at room temperature and 2 8 8 'C................................................ 90 46.
Turbulence interaction regions.................................................. 93 '
47.
A change in the penetration depth of turbulence from power variations can cause thermal stratification and cycling in a branch line................. 93 48.
The flow-accelerated corrosion model........................................... 95 49.
Flow and temperature dependence of flow-accelerated corrosion rates.................. 97 50.
Influence of fluid temperature on flow-accelerated corrosion rates for carbon steel fittings estimated using the Chexal-Horowitz model...................... 98 51.
Influence of fluid velocity on flow-accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal-Horowitz model...................... 99 52.
Influence of chromium content on flow-accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal Horowitz model............
100 53.
The calculated influence of the temperature on the ferrous ion concentration and on mass transfer of ferrous ions..........................................
101 54.
Influence of dissolved oxygen content on flow-accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal-Horowitz model........
102
]
55.
Effect of cold pH on solubility of magnetite in deoxygenated water................... 102
)
56.
Influence of cold pH on flow-accelerated corrosion rates for a 90-degree 4
carbon steel elbow, estimated using the Chexal Horowitz model.....................
103 57.
Typical PWR feedwater nozzle-to-pipe weld with geometric discontinuity (counterbore) and cracking outside the Code examination volume....................
108 NUREG/CR 6456 x
i 58.
Probability of detection versus depth of thermal fatigue cracks in clad ferritic l
and cast austenitic pipe and ofIGSCC cracks in wrought austenitic pipe - piping inspection round robin results.................................................. I i 1 59.
Depth of thermal fatigue cracks in clad ferritic material as estimated by the six inservice inspection teams with the conventional UT techniques versus that estimated by Pacific Northwest Laboratory with the enhanced UT techniques................... I11 L
60.
Scheme of ultrasonic fields of creeping wave probe in test specimen depicting front and back surface creeping waves, direct and indirect longitudinal waves, l'
and direct and indirect shear waves............................................ I 14 e
61.
Potential beam propagation paths for multimode approach using creeping wave probe..... I 15 62.
Examples of time-of-flight diffraction (TOFD) signals.............................. I16 63.
Typical phased-array transducer................................................ I 18 64.
Tangential radiography......................................................
12 5 65.
Original layout of the feedwater distribution piping inside Siemens/KWU steam generators...........................................................
13 3 66.
Modified layout of the feedwater distribution piping inside Siemens/KWU j
steam generators. The upward bend in the piping prevents flow stratification...........
134 j
67.
Antistratification device for effective mixing of hot and cold coolants in the feed water nozzle...........................................................
13 5 68.
Local on-line fatigue monitoring of feedwater nozzle to-pipe weld at Diablo Canyon Unit 1.......................................................
13 6 i
69.
Local on-line fatigue monitoring results for feedwater nozzle-to-pipe weld at Diablo Canyon Unit 1......................................................
13 6 70.
Local on-line fatigue monitoring of Beaver Valley Unit i feedwater line...............
138 l
l 1
i i
LIST OF TABLES I
1.
Typical materials and fabrication codes for feedwater piping......................... 20 2.
Typical Diablo Canyon auxiliary feedwater evolutions.............................. 26 3.
PWR feedwater piping cracking 19 7 9-1980....................................... 3 5 4.
PWR feedwater nozzle cracking 1983 to present.................................. 37 I
5.
Repair / replacement activities 1983 to present..................................... 38 l
6.
Results of Sequoyah radiographic inspection..................................... 45 k
7.
PWR plants with pipe wall thinning in the feedwater-condensate systems............... 67 8.
PWR steam generator water hammer events...................................... 79 i
I i
i i
I b
i I
t i
e r
r i
l i
i i
NUREG/CR 6456 xii
EXECUTIVE
SUMMARY
1 Pressurized water reactor (PWR) feedwater noz-modifications, operating procedure changes, j
zles, piping, and feedrings have experienced augmented inspection programs, and repair and cracking and wall thinning during operation. The replacement activities that have been carried out cracking was caused by thermal fatigue resulting because of these problems. The effort focused on i
from flow stratification, and the wall thinning was a portion of the feedwater system adjacent to the caused by flow-accelerated corrosion. In re-feedwater nozzle where fatigue cracking and wall sponse, the USNRC issued Bulletin 79-13 Crack-thinning have been reported in the field. His ing in Feedwater System Piping (Revisions 0,1,2) portion of the feedwater system includes the main in 1979 and Bulletin 87-01 Thinning ofP/pe Walls and auxiliary feedwater piping adjacent to the i
in Nuclear Power Plants in 1987, requesting the feedwater nozzle, and the thermal sleeve, feedring, nuclear plant operators in the United States and J-tubes. The primary areas addressed in this i
(USNRC licensees) inspect susceptible regions of study are as follows: 1) feedwater system design, their feedwater systems and take other specific
- 2) safety significance of feedwater line ruptures,3) actions. NRC Bulletin 79-13 addressed the fatigue fatigue cracking experience,4) flow-accelerated cracking problem, and NRC Bulletin 87-01 ad-corrosion-induced wall thinning experience, 5) dressed the wall-thinning problem. The USNRC steam generator water hammer damage experience, licensees performed the inspections requested by
- 6) degradation mechanisms, 7) inservice inspection i
the bulletins, made appropriate repairs and replace-methods, and 8) mitigation, monitoring, and ments, and established corrective programs to replacement.
manage the cracking and wall thinning issues.
USNRC Bulletin 79-13 was closed in 1991.
Main and A uxiliary Feedwater System Designs.
Each PWR has main and auxiliary feedwater l
The objective of this study was to evaluate the systems, both of which are used to remove heat efTectiveness ofindustry cfTorts to manage thermal from the reactor coolant system through the steam fatigue, flow-accelerated corrosion, and water generator tubes. The main feedwater system hammer damage in feedwater piping.
Four provides the water required to maintain the steam specific actions were taken to accomplish the generator secondary side water level during nonnal l
objective: (a) review of field experience to identify operation, whereas the auxiliary feedwater system trends of operating events, (b) review of related is used to maintain a heat sink in the steam technical literature,(c) visits to three PWR plants generators during design basis events such as a where fatigue cracking occurred since the closure reactor trip or a loss of offsite power. The of Bulletin 79-13 and a PWR vendor; and (d) auxiliary feedwater system is also used in most solicitation of information from eight other plants, to provide feedwater to the steam countries.
generators during plant startup, hot standby, and shutdown.
The report surveys the relevant field experience with PWR feedwater systems, assesses the factors The main feedwater systems include multiple causing the damage, and evaluates the design trains of feedwater piping. The feedwater passes through several feedwater heaters, pumps, common headers, control valves, isolation valves, and check vah es, and then enters the steam generator through
' in the United States, flow-accelerated corrosion is the main feedwater nozzle. The main feedwater commonly known as crosion-corrosion. We have used the systems are designed to handle large volumes of tennflow-accelerated corrarion m this report because it water [for examE e' 6.8 x 10' kg/h (l 5 x 10' lb/h) l describes the degradation process causing the obsers ed wall thinning damage to the PWR feedwater piping. The at a typical 1,000 MWe unit]. The pressure and termflow-occelerated corrosion is now also used in France temperature of that water increase from 2.0 to where the termflow-assisted corrosion was used earlier.
xiii NUREG/CR-6456
_ _ = _ _
1 t
i i
3.5 MPa (300 to 500 psia) and 160 to 204*C (320 avoid flow stratification-induced thermal fatigue 4
to 400*F) at the discharge from the condensate cracking of the feedwater nonle and piping.
j system to 5.5 to 7.9 MPa (800 to 1150 psia) and 210 to 238"C (410 to 460*F) at the steam In plants with preheater-equipped recirculating generator inlet. De auxiliary feedwater source is steam generators, the main feedwater nonle is the condensate storage tank, which is typically at located near the tubesheet on b eclo leg side and about 38 to 50*C (100 to 120*F).
(This the cold auxiliary feedwater is discharged directly temperature can be lower than 38'C for plants into the steam generator above the tube bundle. In j.
with the condensate storage tank located outdoors.)
these plants, the main feedwater is also discharged i
i The flow passes through parallel auxiliary into the steam generators through the auxiliary feedwater lines, each having an auxiliary feedwater feedwater nonle during startup (when the power a
pump, check valves, control valves, and isolation is less than about 15 to 20% of full power), which j
- valves, eliminates the possibility of a condensation-in-duced water hammer in the preheater. At the 15 to i
The thermal fatigue and flow-accelerated corrosion 20% power level, a very small amount of main
]
in PWR feedwater systems depend, in part, on the feedwater flow is directed through the main feed-arrangement (design) of the system.
The water nonle, which slowly flushes the cold feed-arrangement of the feedwater system associated water from the piping downstream of the main l
with recirculating steam generators used in feedwater isolation valve and warms up the piping.
t Westinghouse and Com bustion Engineering P WRs De main purpose of warming up the piping is to
- ~
and once-through steam generators used in prevent the cold feedwater from collapsing any Babcock & Wilcox PWRs varies. In plants with steam bubbles in the preheater section, which recirculating steam generators not equipped with could cause water hammer damage to the pre-preheaters, the cold auxiliary feedwater is typically heater. De small amount of feedwater flow used I
discharged into the main feedwater line before for flushing causes thermal stratification in the entering the steam generator. (However, there are main feedwater nonle and adjacent piping with a some exceptions. In at least one plant with the top-to-bottom temperature difference, forexample, recirculating steam generators that are not of about 164*C (295'F). The main feedwater i
equipped with preheaters, the auxiliary feedwater nonles ofthe preheater-equipped steam generators is discharged directly into the steam generators.)
have been designed for these stratified flow cycles.
His configuration introduces stratified flow in the At full power, about 14 to 18% of the main feed-1 main feedwater line and makes it susceptible to water flow is diverted to the auxiliary feedwater fatigue cracking. The main feedwater line is nonle and then into the steam generator. The connected to a nonic located above the tube purpose of splitting the flow during operation, bundle, which is in turn connected to a feedring which was not included in the original design, was and J-tubes; the J-tubes distribute the feedwater in to reduce fretting damage to the steam generator i
the steam generator.
tubes in the preheater section of some Westinghouse-designed steam generators. Ilow-During plant startup, hot standby, and shutdown, ever, the auxiliary feedwater piping and the pre-the typical source of feedwater to recirculating heater bypass line were not designed for this high steam generators not equipped with preheaters is flow during operation and, as a result, have experi-the auxiliary feedwater; main feedwater can pro-enced significant wall thinning caused by flow-vide the feedwater during these operations if a accelerated corrosion.
plant is _ equipped with electrically driven main feedwater pumps. Generally, this feedwater is in plants with once-through steam generators, the cold, but in two Westinghouse PWRs, Wolf Creek main and auxiliary feedwater pass through separate and Callaway, a special startup system is provided headers located external to the steam generator j
to supply heated feedwater to the steam generator shell. The feedwater from the headers is intro-during startup, hot standby, and shutdown. The duced into the steam generator through several main objective of the special startup system is to risers which function like J-tubes in keeping the NUREG/CR 6456 xiv i
3
,~~s. -. -.
.m-
-~
headers full of water. This configuration does not granular with a circumferential orientation, introduce stratified tiows in the main feedwater initiated at the inside surface, and located at i
line.
geometric discontinuities in the counterbore reF on or at nozzle-to-pipe welds. Metallographic studies Safety Significance offeedwater Line Rupture.
of the cracked feedwater nozzle samples Main feedwater line rupture is a design basis determined that the cracking was caused by accident; its consequences include a potential for thermal fatigue. Laboratory and field testing core damage. The rupture reduces the ability to indicated that the root cause of cracking was remove heat generated by the core from the reactor thermal stratification between cold auxiliary coolant system. In addition, the resulting loss of feedwater and hot steam generator secondary feedwater would activate and challenge coolant.
safety-related systems to cool the reactor core.
Hermal fatigue has caused through-wall cracking, Thermal fatigue cracking of feedwater piping has but no ruptures of cracked pipes have been generally not been reported at plants where the reported. However, a cracked feedwater pipe auxiliary feedwater is introduced into the steam could rupture if it were subjected to seismic or generator shell rather than into the main feedwater water hammer inads. The USNRC has not line; however, in at least one case, fatigue cracking acceptd S leak-before-break concept for PWR in an auxiliary feedwater nozzle inside radius feedwater lines. Flow-accelerated corrosion has region has been reported. This cracking occurred caused significant wall-thinning ofmain feedwater because of cold coolant leaking into the gap piping and, in a few cases, it has resulted in between the auxiliary feedwater nozzle and its rupture.
thermal sleeve.
I The consequences of a major feedwater line After the inspections conducted in response to rupture depend on the size and location of the NRC Bul!etin 79-13 were completed and licensees break and the plant operating conditions. If the had performed appropriate repairs and established break occurs between the steam generator and the other corrective actions (approximately from 1979 first check valve on the feedwater piping, through 1983), eight more PWRs experienced secondary coolant from the steam generator can be feedwater nozzle cracking (from 1983 through discharged through the break. He faulted steam 1991). Most of these fatigue cracks were similar to generator would rapidly depressurize and as a those discovered in the 1979 round ofinspections, result, the differential pressure across the faulted but there were two notable exceptions: leakage at steam generator tubes would increase and multiple Maine Yankee in 1983, and extensive cracking at mptures of degraded steam generator tubes might Indian Point 2 during the 1989 to 1991 period.
occur. If the first check valve is outside the The Maine Yankee cracking resulted in a leak containment and the break is between the when a water hammer caused fracture of an already containment wall and the check valve, tube leakage existing fatigue crack located in the counterbore would cause release of radioactive products region of the feedwater line. Fatigue cracking through the break into the environment.
occurred at several locations in the Indian Point 2 steam generators, including the pipe-to-nozzle Field Erperience Related to fatigue Cracking.
weld region, the feedwater nozzle bore region PWR feedwater nozzle cracks were first discovered under the thermal sleeve, the nozzle inner radius at the D. C. Cook Unit 2 plant ia 1979. As a section, and the heat-affected zone of the weld result, the USNRC issued Bulletin 79-13.
between the feedwater nozzle and the steam Inspections conducted in response to the Bulletin generator shell of two steam generators. The revealed instances of cracking in 18 PWRs,16 of cracking in the nozzle bore and blend radius was Westinghouse design and two of Combustion caused by coki feedwater leaking through a small Engineering design. Only the cracks found in the gap between the nozzle and the thermal sleeve. A two D. C. Cook Unit 2 feedwater nozzles were feedwater nozzle sealing sleeve was installed in through-wall cracks. Typical cracks were trans-1991 to eliminate leakage of cold feedwater into xv NUREGICR-6456
the gap between the thennal sleeve and nozzle and in September 1992, enhanced ultrasonic examina-thus mitigate the nozzle cracking. Similar cracking tion revealed circumferential indications in the has been reported in a main feedwater nozzle in a counterbore region of the feedwater nozzles at non-U.S. PWR and in an auxiliary feedwater Diablo Canyon Unit 1. Metallographic examina-nozzle in a U.S. PWR, as mentioned earlier, tions of samples removed from the counterbore region showed th4 the ultrasonic examination Cracking events in PWR feedwater nozzles have overestimated the crack size; a crack that was continued to occur since the 1991 closcout ofNRC identified by ultrasonic examination as 8.9-mm Bulletin 79-13. Leakage through a circumferential (0.35-in.) deep was found to be only about one crack was found at Sequoyah Unit 1 in March tenth that depth. Segregates found in the material 1992. The crack was located at the feedwater near the crack location may have caused the incor-nozzle-to-transition piece weld region at a geomet-rect sizing by the ultrasonic examination. The rmt ric discontinuity. Metallographic examination cause of the cracking was the same as that identi-revealed that significant cracking was present at fled for the cracking at the Sequoyah plants. The five out of eight feedwater nozzles for Units I and licensee replaced the damaged portion of piping at
- 2. Most of the major cracks were located on the all four Unit I feedwater nozzles, and the auxiliary transition piece side of the weld but two of the feedwater cycles are being monitored at both units.
major cracks were located on the nozzle side of the Thermal sleeves and feedrings have experienced weld. During earlier ultrasonic inspections, there wall thinning at these two units, as discussed later were indications recorded at the sites ofthe cracks, in this summary.
i but they were misinterpreted as geometry effects.
All the transition pieces were replaced. A subse-Ultrasonic examination of the safe ends of the San quent ultrasonic examination at Sequoyah in April Onofre Unit 3 feedwater nozzles during 1993 1993 revealed cracking in five of the eight replaced revealed circumferential cracking. It was deter-transition pieces; radiographic examination was not mined that periodic thermal stratification resulting able to confirm any of these indications.
from the earlier on-off operation of the auxiliary feedwater flow was the root cause of this cracking.
The root cause of the cracking at the Sequoyah in 1985, the licensee modified operation of the plants was attributec to thermal fatigue caused by auxiliary feedwater flow to minimize the fluctua-periodic thermal stratification when cold auxiliary tions in the flow rate. The corrective actions feedwater was injected into the main feedwater line included weld repair of the cracks and use of an and then into the steam generator during hot augmented inservice inspection program.
standby operations. The on-off operation of the auxiliary feedwater flow during the automatic Inservice inspections during 1992 and 1993 re-mode ofoperation produced about three cycles of vealed several unacceptable indications in the thermal stratification every hour during hot stand-feedwater nozzle welds for Salem Unit I and the
'oy. After the 1993 discovery of cracks, the operat-Haddsm Neck plant. The indications were re-ing procedures were modified to reduce the time moved by grinding. Circumferential cracking at for on-offoperation of the auxiliary feedwater flow feedwater nozzle welds has been reported in and to minimize the fluctuations in the flow rate.
Belgian, Swiss, French and German PWRs.
The other corrective actions included improved inservice inspection procedures, use of enhanced Field Experience Related to Flow-Accelerated ultrasonic examination techniques, and better Corrosion. Flow-accelerated corrosion has caused training of inservice inspection personnel. In significant wall thinning of the feedwater piping addition, long thermal liners were installed in two outside the containment at a number of plants, feedwater nozzles to protect the nozzle-to-transi-resulting in failures at several of these plants tion piece weld, transition piece, and transition including the failures at the Trojan plant in 1985 piece-to-elbow weld, and Surry Unit 2 in 1986. A pressure pulse ultimately caused a rupture of the damaged piping at both Trojan and Surry Unit 2. There was no NUREG/CR 5456 xvi
leak or other warning signs indicating incipient Three reasons have been reported for the Loviisa failure during either event. Eight workers were wall thinning: (1) neutral water chemistry (cold burned by flashing feedwater during the Surry pH = 7) with low dissolved oxygen, (2) a accident, four of whom subsequently died. As a geometric discontinuity introduced by a flow result of the Surry accident, the USNRC issued orifice that produced high flow velocities, and (3)
Bulletin 87-01 requesting all utilities with the low-alloy content of the piping material.
operating nuclear power plants to inspect their Nonradioactive water and steam released from the high-energy carbon steel piping.
ruptured pipe caused significant damage to nearby cables and small piping, but no important functions There has also been evidence ofdamage caused by were lost.
flow-accelerated corrosion in the safety-related portion of the feedwater piping inside the contain.
Generally, the auxiliary feedwater system is in ment. However, in one instance, it was later found standby during normal operation and, therefore, that certain damage was not caused by flow-accel-not susceptible to flow-accelerated corrosion crated corrosion. In 1987, it was initially reported damage.
However,1 for some Westinghouse that flow accelerated corrosion caused significant recirculating steam generators equipped with pre-wall thinning in the horizontal and vertical portions heaters, a small portion of the main feedwater flow of the feedwater piping inside the containment at is diverted through the auxiliary feedwater line the Trojan plant. These portions of the piping during normal operation to reduce fretting damage were not considered susceptible to flow-acceler-to the steam generator tubes in the preheater ated corrosion damage and not included in the region. This diversion of main feedwater has inspection program because they were at least resulted in significant wall thinning of the auxiliary seven pipe diameters away from an elbow or any feedwater piping in some of those Westinghouse I
other fittings that could cause turbulence. How-plants. If this piping had ruptured, the break ever, after further analysis the plant operator would not have been isolable and would have determined that the damage to these horizontal and resulted in the steam generator coolant being vertical portions of the piping was minor and released outside the containment.
Judged not caused by flow-accelerated corrosion but was rather twm an initial manufacturing The carbon steel J-tubes and feedrings in the top-defect. No other util.m has reported flow-acceler-feed recirculating steam generators have also ated corrosion damage in straight piping away experienced significant wall thinning caused by from discontinuities. Further inspection of:11 the flow-accelerated corrosion. Wall thinning and high-energy carbon steel piping at Trojan revealed through-wall penetration of the original carbon v.all thinning at about 30 additional sites,10 of steel J-tubes were reported in the late 1970s; and which were in the safety-related portion of the they were replaced with stainless steel or Alloy 600 feedwater system (portion of the system inside the J-tubes. However, the joint design between the containment), whereas the remaining sites were in new J tubes and the fcedring introduced geometric the nonsafety-related piping. Failure analysis discontinuities on the inside surfhce of the feed-indicated that flow-accelerated corrosion caused ring. The resulting turbulence has caused wall the wall thinning at these additional sites. Failure thinning of the feedrings at the Diablo Canyon, analysis also indicated that cavitation caused by Surry, and North Anna plants.
severe flow conditions at the pump discharge elbows also contributedtothewalithinningatthe Thinning of thermal sleeves has been reported at sites in the nonsafety-related piping.
the Salem Unit 1, Diablo Canyon Units I and 2, Prairie Island, Trojan, San Onofre, and Arkansas Flow-accelerated corrosion induced wall thinning Nuclear One Unit 2 plants. The damage mecha-has resulted in several feedwater piping ruptures at nism was flow accelerated corrosion. The result-non U.S. plants. In 1990, a feedwater pipe at ing increased gap between the feedwater nozzle Loviisa Unit I ruptured when the plant was and the thermal sleeve permits increased bypass operating at full power. The failure location was leakage of the colder feedwater into the hot steam immediately downstream of a flowmeter flange.
generator coolant present near the nozzle blend xvii NUREG/CR-6456
radius region. Such bypass leakage caused crack-events associated with the auxiliary feedwater line ing of the nozzle at the Indian Point 2 plant.
have also been reported.
Monitoring of the nozzle wall temperatures can aid in detecting leakage taking place because of thin-Water hammer damage was typically limited to the ning of the thermal sleeves. The repair at the plant piping support system. However, in a few Diablo Canyon plant included installation of a cases water hammer events have resulted in forging that protects the leading edge of the ther-significant damage to the steam generator internals, mal sleeve from flow-accelerated corrosion and including cracked feedrings and expanded thermal protects the nozzle from fatigue damage. Installa-sleeves, and cracking and bulging of the main tion of a thermal liner at D. C. Cook, Sequoyah, feedwater line. The most damaging event was at Beaver Valley and other PWR plants also provide Indian Point Unit 2 in 1973, which resulted in a similar protection to the thermal sleeve and the 180-degree circumferential through-wall crack in nozzle.
a 460-mm (18-in.) diameter main feedwater line at the containment penetration. To preclude the Field E.xperience Related to Steam Generator recurrence of water hammer events, the feedring Water Hammer. The portions of the main and was modified to prevent rapid draining by auxiliary feedwater piping adjacent to main and plugging the sparging holes and installing J-tubes j
auxiliary feedwater nozzles have been damaged by on the top of the ring. In addition, the elevation of steam generator water hammer, which is caused by the long horizontal run of feedwater piping outside the collapse of a steam bubble. The sequence of the steam generator was lowered about 0.4 m (16 events leading to steam generator water hammer is in.) to prevent the water in this horizontal run from i
as follows: (a) when a feedring with bottom-draining into the steam generator whenever the discharge holes is uncovered, the upper portion of water level in the steam generator dropped below the drained feedring and the horizontal length of the feedring.
the adjacent feedwater piping is filled with steam; (b) the auxiliary feedwater flow, which is resumed In 1983, a water hatnmer occurred at the Maine shortly, enters the horizontal length of the piping Yankee plant about 15 minutes after the reactor and flows under the steam blanket; (c) if the flow tripped from full power. The water hammer rate is high, the resulting turbulence seals off a caused rupture of a main feedwater line near the pocket of steam and forms a slug of cold water; (d) steam generator feedwater nozzle where a the trapped steam condenses rapidly and, as a preexisting fatigue crack caused by flow result, the slug accelerates into the void; and (e) stratification was most likely present. Each steam finally the slug traveling with a high velocity generator feedring was later modified by closing impacts an incoming water column and a pressure offall 76 bottom-mounted,25-mm (1 -in.) diameter pulse is produced. The magnitude of the pressure nozzles and installing 28 top-mounted,90-mm pulse depends on the slug velocity, which is (3.5 in.) diameter elbows.
Other preventive determined by the distance (horizontal length of actions included removing stress raisers from the the adjacent piping) it has traveled.
pipe inside surface and modifying operations instruction and guidance to reduce the probability About 37 steam generator water hammer events of thermal cycling and steam generator water associated with the main and auxiliary feedwater hammer events.
lines of 17 top feed generators have been reported.
Five of these events occurred prior to commercial in 1985, a severe water hammer occurred at San operation, whereas the others occurred afterward.
Onofre Unit 1. The events leading to water ham.
The data show that the reported occurrences of mer included loss of power to a main feedwater steam generator water hammer events dramatically pump while another one remained energized; decreased after 1979; 28 events occurred during failures ofmultiple check valves that allowed back the 1970s, whereas only 6 occurred during the flow of high-pressure feedwater which ruptured a 1980s. A few steam generator water hammer condenser tube; drainage of the feedwater through NUREG/CR-6456 xviii
a ruptured condenser tube; and filling of a long, DegradationMechanisms. The characteristics horizontal portion of the main feedwater line with of the damage caused by thermal fatigue are steam from the steam generator. Water hammer different from those caused by flow-accelerated occurred when the auxiliary feedwater flow was corrosion. Thermal fatigue cracking generally established and the isolation valves and flow occurs in a relatively local, safety-related portion of control valves were closed. As a result, the the feedwater piping inside the containment, 254-mm (10-in.) diameter pipe inside the contain-whereas wall-thinning caused by flow-accelerated ment was distorted from its original configuration, corrosion typically occurs, with few exceptions, in pipe supports were damaged, and a 2-m (80-in.)
the non-safety related balance-of-plant piping long crack was developed. He corrective actions outside the containment. A through-wall crack included replacing faulty and damaged check caused by thermal fatigue will generally leak valves, implementing logic which closes the flow-before the component ruptures. However, during control valves upon a loss of main feedwater, and an unlikely event resulting in a large overload, a installing redundant check valves in the feed water pipe with fatigue cracks might fail catastrophically piping.
with no prior leakage. A component damaged by flow-accelerated corrosion loses its strength and ne preheat ste-ra generators were not expected to can fail under normal operating pressure; a large experience water hammer events because the fitting or pipe might fail catastrophically without lessons leamed from the experience with the water warning.
hammer events associated with the top-feed steam generators were incorporated in the preheat steam Sites susceptible to thermal fatigue cracking have generator designs. However, a water hammer only been found in those horizontal poitions of the occurred in the auxiliary feedwater line of a piping and nozzles where stratified flows and preheat steam generator at KRSKO, a two-loop coolant leakage, respectively, are present; these Westinghouse plant in the former Yugoslavia, locations are generally well identified. Sites during preoperational testing of the auxiliary susceptible to flow-accelerated corrosion are found feedwater line. He water hammer was caused by throughout the feedwater system where the fluid a steam bubble collapse in an auxiliary feedwater velocities are high and very low-alloy content line during a hot functional test in July 1981. The piping materials are present. These sites are horizontal portion of the line where the water difficult to identify without predictive analysis.
hammer occurred was about 50 ft below the i
auxiliary feedwater nozzle. Damage was mainly Three different phenomena have caused thermal limited to inside the containment: the piping was fatigue damage to PWR feedwater piping: local shifted and bulged about 6 mm (0.25 in.) near the thermal stratification and thermal striping caused secondary wall (the probable location of water slug by the flow stratification, and turbulent mixing impact), and the pipe hangers were damaged. He caused by the bypass leakage through the gap piping movement was negligible in the between the thermal sleeve and feedwater nozzle.
intermediate building, but paint on the auxiliary he most severe fatigue cracking is caused by local feedwater piping was blistered back to the motor-thermal stratification.
Factors causing high driven auxiliary feedwater pumps. The sequence stresses during thermal st atification are the of events that led to the KRSKO water hammer are temperature difference between the hot steam not well understood. However, the observed paint generator coolant and the cold auxiliary feedwater, damage implies that significant back leakage of and geometric discontinuities on the inside surface steam from the steam generator through the of the feedwater piping. The stress distributions auxiliary feedwater line must have taken place, and resulting from thermal stratification are complex several check valves in the auxiliary feedwater because of the feedwater nozzle and elbow end lines must have leaked.
constraint effects and because of the geometric discontinuities at the inside surface of the piping.
xix NUREG/CR-6456
The temperature distribution in e pipe wall caused Laboratory test results and data from operating by thermal stratification is plant-specific, because plants have identified several factors that affect in addition to the differential tempe. ature between flow-accelerated corrosion rates; these factors may the hot and cold stratified fluids and the flow rate, be divided in three groups: (a) hydrodynamic it depends on the piping configuratior,. Therefore, variables - fluid velocity, pipe roughness, and the results from certain laboratory tests or from flow path geometry defined by piping configura-some plants may not be directly applicable to tion; (b) metallurgical varirsbles - weight percent anothec pint.
ofchromium, molybdenum and copper in the steel; and (c) environm ental variables - coolant temper-Thermal striping takes place when high relative ature and water chemistry incleding dissolved velocities between the hot and cold coolants are oxygen, pli, and the amine used for pli control.
present and, as a result, the gradient Richardson number is small(for example, less than 0.25). The All PWR plants in the United States use the gradient Richardson number is the ratio of the CHECWORKS cooc (or its predecessor code density gradient and horizontal velocity gradient CHECMATE) for estimating flow-: eterated across the thickness of the interface layer between corrosion rates. This code has capistis for the hot and cold coolants. Stresses produced by estimating parameters (such as local water thermal str; ping can initiate cracks on the inside chemistry and flow rate) that affect corrosion rates, surface but do not cause any significant crack and for predicting corrosion rates and helping to propagation through the thickness, because select inspection locations. This computer code through-thickness stresses attenuate rapidly. It was developed by the Electric Power Research appears that thermal striping might have contrib-Institute and is based on laboratory data from uted to initiation of some fatigue cracks found in France, England, and Germany, and a set of U.S.
the feedwater piping. As discussed above, bypass plant data. The code has been validated using leakage through the gap between the thermal another set of U.S. plant data, different from the sleeve and feedwater nozzle has caused fatigue one used to develop the model. The comparison cracking in the ncule bore, nozzle inside radius, between the predicted results and measurements and steam generator she:1. It is not known whether show that the code predicts flow-accelerated such crackir g is a common occurrence. Thermal corrosion rates within 50%. He main sources of cycling cwid cause cracking in the auxiliary uncertainties are associated with the original feedwater piping where it connects to the main thickness and thickness profile of the piping feedwater line. However, such cracking has not components, trace amounts of alloy content in the l
been reported.
piping material, actual number of hours of operation, plant chemistry
- history, and l
Flow-accelerated corrosion of carbon steel discontinuities on the inside surface of the piping.
l feedwater piping occurs through dissolution of a porous oxide layer of magnetite present on the InserviceInspection. Conventional amplitude-inside surface of piping and simultaneous based ultrasonic testing (UT) techniques can detect formation of a new oxide layer at the metal-oxide fatigue cracks in ferritic steel piping but have a interface, a process that reduces the piping wall very poor sizing capability for these cracks. This thickness over time. T1.e k.inetics ofmetal removal is illustrated by the round robin UT inspections of by an oxide dissolutior. process are linear (that is, thermal fatigue cracks in clad, ferritic steel piping the corrosion ra'.e is constant in time). The wall conducted in 1981 by the Pacific Northwest thinning rates observed in the field are essentially Laboratory. Enhanced UT techniques such as tip constant when the influencing factors do not vary.
diffraction techniques can accurately size the depth Severely corroded large-diameter piping surfaces of the fatigue cracks. Creeping wave and related produced by single-phase flow-accelerated mode conversion techniques are also used to detect corrosion are characterized by overlapping and qualitatively size fatigue cracks. Radiographic horseshoepits that give an orange peel appearance.
examinations are not adequate for detecting tight thermal fatigue cracks but can detect the cracks NUREG/CR-6456 xx
i that are open and filled with corrosion products, relatively low cost. A properly conducted UT Use of an imaging technique such as Synthetic-examination can estimate the pipe wall thickness to Aperture Focusing Technique for Ultrasonic within 5% of the actual value. Industry practice is Testing (SAFT-UT) can provide accurate s; zing of to overlay a grid on the pipe wall and then spot a crack.
measure the thickness at each grid location.
The phased array technique is used for inservice The computer code CHECWORKS (or the earlier ins 9ection of components such as feedwater CHECMATE) discussed above is used by all U.S.
nozzles which have a complex geometry, and have PWR utilities for helping to select inservice very limited access and clearance. Computer inspectionlocationsin feedwater piping. Some modeling can help in determining the best position utilities perfonn inspections at additional locations and angle for a given probe for inspecting a crack based on industry experience or because of the of given size and orientation, located in the nozzle uncertainties associated with several input inside radius region.
parameters to the code.
Discrimination of reflectors, such as geometry Mitigation, Monitoring,andReplacement. Use effect2, inclusions, and crack tips, is essential for of continuous auxiliary feedwater flow rather than reliable detection and accurate sizing of fatigue intermittent flow significantly reduces the thermal cracks. Use of more than one inspection technique fatigue damage to feedwater piping. Several can provide more reliable sizing of the cracks.
utilities have implemented continuous flow Implementation of the mandatory Appendix Vill, operation by making physical modifications to the l
Performance Demonstrations for Ultrasonic feedwater system and changes in the operating l
Examination Systems, of ASME Section XI will procedures. Also, some plants use heated main l
improve the reliability of detection and the feedwater during plant startup and hot standby accuracy of sizing of thermal fatigue cracks.
conditions to mitigate the thermal fatigue damage in the feedwater nozzle region.
Risk-based criteria for selection ofinspection sites I
arc also being developed. Use of these criteria Two design changes that reduce the thermal would result in the inspection oflocations such as stresses produced by stratified flows include a counterbores that are susceptible to thermal fatigue redesign of the counterbore to reduce the stress cracking; these locations have often been missed raiser, and installation of a thermal liner to protect by the current ASME Code selection criteria.
the counterbore and other susceptible sites from thermal fatigue damage. Several design changes Use of conventional radiographic testing for have been made by Framatome and Siemens/KWU thickness measurements is limited to small-to reduce or eliminate the occurrence of flow i
diameter piping; inspection of large-diameter stratification. These changes include, for example, l
piping requires longer exposure times resulting in installation of an antistratification device that I
higher costs and increased personnel exposure.
breaks down the stratification by mixing the hot l
The main advantage of radiography is that it does and cold coolants, and incorporation of an upward not require removal of the insulation.
Use of bend in the piping connecting the thermal sleeve to filmless radiography with phosphor plates is being the feedring inside the steam generator. This anti-evaluated in the field. This technique is expected siphon configuration prevents backflow from the to dramatically reduce the ;xposure dose and steam generator.
significantly reduce the inspection time and personnel safety concerns associated with Use of a fatigue monitoring system can assist in l
performing radiography within a plant.
detecting the presence of thennal stratification and I
in predicting temperature distributions and flow Manual UT is the most commonly used inspection rates so that the fatigue usage can be estimated method for the detection and trending of changes more accurately.
in the wall thickness because ofits accuracy and 1
xxi NUREG/CR-6456
Low-alloy steels such as 1%Cr-%Mo steel (SA-Several design modifications and operational 335, Grade Pl1) and 2%Cr-1Mo steel (SA 335, procedure changes made to the feedwater systems Grade P22), austenitic stainless steel, and Alloy in the 1970s and 1980s appear to be adequate for 600 are used to replace the existing carbon steel preventing water hammer in the top-feed steam piping when excessive wall thinning caused by generators; no water hammer events have been flow-accelerated corrosion is found. The presence reported recently. The holes in the bottom-of 0.1 wt% or more chromium provides protection discharge feedring were plugged, and J-tubes or against Dow-accelerated corrosion.
discharge elbows were installed on the top of the feedring.
This modification prevents rapid The optimum pli in the PWR secondary steam /
draining of the feedring when it is uncovered. In water system is generally achieved by the addition addition, the length of the horizontal run of of reagents such as ammonia, morpholine, and feedwater piping was reduced to less than 2 44 m ethanolamine in the demineralized water. These (8 ft) so that the magnitude of the pressure pulse amines are volatile and, therefore, maintain a resulting from the water hammer is sufficiently slightly alkaline pli in both the single-phase and reduced. In some new steam generators, the two-phase regimes ofthe secondary system, which thennal sleeve is welded to the feedwater pipe to reduces the flow-accelerated corrosion in the entire stop the drainage through the gap between the system. The selection of pil is mainly a compro-sleeve and the nozzle in the event the feedring is mise between the acceptable corrosion of piping uncovered. The operating procedure change and components made of carbon steel and copper required prompt resumption of feedwater now into alloys. Ammonia can be used for pil control in the steam generator to minimize the amount of feedwater systems with no copper alloy materials steam that enters the feedring and piping.
and no condensate polishers; however, the cold pli llowever, the feedwater now rate must be limited should be increased, at least up to 9.7, to avoid to about 5701/ min (150 gpm) so that a slug of cold flow-accelerated corrosion damage. Morpholine is water does not form.
widely used in the feedwater systems with copper alloy materials in the water heaters and condensers Findings. Our assessment of field experience to maintain the cold pil in the range of 8.8 to 9.2, related to PWR feedwater nozzle cracking is that i
provided there are no condensate polishers. An-the USNRC licensees have apparently taken l
other amine, ethanolamice, may be used instead of suflicient actions to minimize the feedwater nozzle morpholine w hen condensate polishers are present.
cracking caused by thermal fatigue. As a result of i
The nuclear industry experience indicates that the examinations conducted in response to Bulletin
(
transport of corrosion products with ethanolamine 79-13, feedwater nozzle fatigue cracking was is lower than that with ammonia, and even lower detected in 18 PWRs during 1979 to 1983. Then, than that with morpholine. A large number of U.S.
there was about one fatigue cracking event per year PWRs are currently switching or contemplating from 1983 to the bulletin closure in 1991. The switching to ethanolamine.
frequency of feedwater nozzle cracking events increased to six per year during 1992 and 1993.
Use of optimum water chemistry reduces wall But since then through 1996, we have not thinning rates but does not eliminate them; some identified any additional feedwater nozzle cracking research results show that the maximum rates are event.
still about 0.1 to 0.2 mm/yr (0.004 to 0.008 inlyr),
even when the water chemistry is optimum.
Our assessment of field experience related to flow-Therefore, comprehensive, exhaustive, and conser-accelerated corrosion damage shows four compo-vative analyses of the feedwater system need to be nents in the portion of the feedwater piping within performed to identify all sites susceptible to Cow the scope of this report that have experienced accelerated corrosion and these sites need to be significant wall thinning: carbon steel J-tubes, inspected periodically for wall thickness loss.
feedrings, and thermal sleeves in the top-feed NUREG/CR-6456 xxii
A steam generators, and auxiliary feedwater lines in accelerated corrosion loses its strength and can fail the preheat steam generators. The USNRC licens-under normal operating pressure; a large fitting or ces have taken sufficient action to minimize the piping might fail catastrophically without any war $ inning in J-tubes and auxiliary feedwater warning.
lines. liowever, we did not find specific industry actions to minimize the wall thinning in feedrings Sites susceptible to thermal fatigue cracking and thermal sleeves, but we found visual inspec-are found in those portions of the feedwater tion being performed and repair when needed.
piping and nozzles where stratified flows and coolant leakage, respectively, are present; Our assessment of field experience related to steam these locations are generally well identified.
generator water hammer damage indicates that the Sites susceptible to flow-accelerated corrosion USNRC licensees have taken sufficient action to are found throughout the feedwater system and minimize water hammer in both top-feed and are difficult to identify without predictive preheat steam generators. Ilowever, we have not analysis because several factors are involved.
evaluated the industry efforts to minimize the multiple check valve failures that have played a The factors causing thermal fatigue and flow-major role in several steam generator water accelerated corrosion damage are well hammer events.
understood. Advancedultrasonicexamination techniques can be used to reliably characterize The major technical findings are as follows.
thermal fatigue cracks.
Cost-efTective radiographic techniques for estimating wall The characteristics of the damage caused by thickness are being developed as a means to thermal fatigue are different than those caused assess flow-accelerated corrosion damage.
by flow-accelerated corrosion.
Thermal i
l fatigue cracking generally occurs in a Several effective techniques have been relatively local, safety-related portion of the developed for monitoring, mitigating, and feedwater piping inside the containment, repairing the damage caused by thermal whereas wall thinning caused by flow-fatigue and flow-accelerated corrosion.
accelerated corrosion typically occurs, with few exceptions, in the non-safety related The above findings indicate that appropriate balance-of-plant piping outside the analysis, inspection, momtoring, mitigation, and containment.
replacement techniques have been developed for managing thermal fatigue and flow-accelerated A through-wall crack caused by thermal fa-corrosion damage to feedwater nozzles, piping, tigue will generally leak long before the com-and feedrings. Adequate training and appropriate ponent ruptures. Ilowever, during an unlikely applications of these techniques will ensure event of a large overload, a pipe with fatigue effective management of the damage. Several cracks might fail catastrophically without any PWR plant operators have been proactive in prior leakage. A component damaged by flow-managing this damage.
1 I
l l
xxiii NUREG/CR-6456 1
(
ACRONYMS AFW auxiliary feedwater ANSI American National Standards Institute ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials B&W Babcock and Wilcox BWR boiling water reactor CANDU Canadian Deuterium Uranium CERL Central Electric Research Laboratories CHECWORKS Chexal Horowitz Engineering-Corrosion Workstation CT computer tomography DAC distance amplitude correction EPRI Electric Power Research Institute FATS Focused Array Transducer System IIPSI high pressure safety injection IGSCC intergranular stress corrosion cracking ISI inservice inspection JAPEIC Japan Power Engineering and Inspection Corporation l
LVDT linear variable differential transformer MFIV main feedwater isolation valve MINAC miniature linear accelerator MT magnetic particle testing l
NDE non-destructive examination NPS nominal pipe size NSP Northern States Power NSSS nuclear steam supply system OD outside diameter OECD Organization for Economic Cooperation and Development PARIS Portable Automated Remote inspection System PATT pulse arrival time techniques PASC Program for Inspection of Steel Components PNL Pacific Northwest Laboratory PSI preservice inspection PWR pressurized water reactor RCS reactor coolant system RHR residual heat removal RMS root mean square RT radiographic testing RTD Rontgen Technische Dienst RTR real time radiography SAFT-UT Synthetic Aperture Focusing Technique for Ultrasonic Testing SNUPPS Standard Nuclear Unit Power Plant System SPOT Satellite Pulse Observation Technique TEMP Transient Electromagnetic Probe TOFD time of flight difTraction USAS United States of America Standard USNRC United States Nuclear Regulatory Commission UT ultrasonic testing xxv NUREG/CR-6456
ACKNOWLEDGMENTS We acknowledge the significant role played by E. J. Brown of the United States Nuclear Regulatory Commission (USNRC) in providing technical guidance and several reviews of the report. We also teknowledge E. A. Trager and J. R. Boardman of USNRC in providing programmatic guidance.
We sincerely thank the following technical experts for providing information during the plant visits and reviewing the report: G. Wade, K. House, W. Goins, M. Lee, T. Greer, and W. C. Luawig of Tennessee Valley Authority; the late E.Regala and S.Shaw of Southem California Edition; and P. Hirschberg, H. J. Thailer, D. Gonzalez, L. F. Goyette, and K. J. Dalal of Pacific Gas and Electric. The authors also thank W. Cullen, J. Houtman, W. Bamford, D. Kurek, D. Popovich, A. Thurman, and C. Hu of Westinghouse Electric for providing technical information about the fatigue cracking in feedwater nozzles and reviewing the report.
We sincerely thank R. E. Johnson of USNRC and P. E. MacDonald of Lockheed Martin Idaho Technologies Company (LMITCO) for an in-depth, criticai review of the entire report. We also thank L. F. Goyette of Pacific Gas and Electric for several discussions on flow-accelerated corrosion damage to feedwater piping.
We sincerely acknowledge the following technical experts for reviewing several sections of the report:
D. Munson, R. Carter, S. R. Gosselin and their coworkers at the Electric Power Research Institute; K. Parczewski, M. Hartzman and D. Terao of USNRC; and J. A. Seydel, M. T. Anderson, M. B. Sattison, J. F. Cook and W. J. Galycan of LMITCO. We would like to thank D. A. Prawdzik, W. F. Steinke, and J. B. Hudson also of LMITCO for providing description of several special features of the feedwater system operation. 'Ihe authors w,uld like to acknowledge the efTorts of S. T. Khericha in preparing parts of Section 2 and 5.
We sincerely acknowledge J. H. Bryce for providing the programmatic guidance and performing an editorial review of the complete report. Our thanks to the Idaho National Engineering Laboratory technical library staff for their prompt and courteous response to our numerous requests for technical reports and papers.
Our special thanks to L. Brown for performing many searches for the documents in the Nuclear Documents System. We also thank A. J. liaroldsen and C. E. White for preparing the figures and D. R. Pack for providmg editorial guidance. Finally, we sincerely acknowledge A. M. Grimes for her countless efforts and patience in promptly incorporating many of our changes in the report and preparing the final manuscript of the report.
l l
xxvii NUREG/CR-6456
Review of Industry Efforts to Manage Pressurized Water Reactor Feedwater Nozzle, Piping, and Feedring Cracking and Wall Thinning
- 1. INTRODUCTION Several pressurized water reactor (PWR) feedwater generator water hammer events have also caused a nozzles, piping, and feedrings have experienced few ruptures of feedwater piping and feedrings.
cracking and wall thinning during operation. The cracking of feedwater nozzles and piping was In response to the cracking and wall thinning caused by thermal fatigue resulting from the use of events, the United States Nuclear Regulatory cold auxiliary feedwater during certain plant Commission (USNRC) issued Bulletins 79-132and operations such as startup and hot standby and 87-01' that requested U.S. nuclear power plant from feedwater leakage bypassing the thermal operators to perform inspections of susceptible sleeve. He specific thermal fatigue mechanisms regions and take several other specific actions.
causing cracking include thermal stratification, The USNRC licensees performed the inspections thermal striping, and turbulent mixing. He wall identified in the Bulletins, made appropriate thinning of carbon steel piping and fecdrings was repairs and replacements, and established caused by a corrosion process resulting from the corrective programs to manage the cracking and use ordeoxygenated feedwater and high local fluid wall thinning issues. The Bulletin addressing the velocities. The specific corrosion mechanism cracking problem was issued in 1979 and closed in causing wall thinning is flow-accelerated corro-1991. The Bulletin addressing the wall-thinning sion.'
problem was issued in 1987.
Main fadwater line rupture is a design basis The main objective of this report is to review accident. The rupture reduces the ability to operating experience and evaluate the effectiveness remove heat generated by the core from the reactor of the licensees' progams for managing thermal coolant system. In addition, the resulting loss of fatigue, flow-acceleratei corrosion, and water feedwater would activate and challenge safety-hammer damage in the feedwater piping. This related systems to cool the reactor core. Thermal includes an evaluation of design modifications, fatigue has caused through-wall cracking of operating procedure
- changes, augmented feedwater piping at a few PWRs but has not caused inspection programs, and repair and replacement rupture of feedwater piping, flowever, a cracked activities.
feedwater pipe could rupture ifit w ere subjected to seismic or water hammer loads. Flow-accelerated Four specific actions were taken to accomplish the corrosion has caused significant wall thinning of objective: (a) review of field experience to identify main feedwater piping and feedrings, and, in trends of operating events, (b) review of related several instances, it has resulted in rupture. Steam technical literature,(c) visits to PWR plants and to 2 USNRC Bulletin 79-13. Cracking in fredwater
' In the United States, flow-accelerated corrosion is
&# rem Piping. Revisions 0, I,2.
commonly referred to as erosion-corrosion. The termflow-3 accelerated corrosion is now also used in France where the USNRC Bulletin 87-01: Thinning ofPepe Walls in termflow-assisted corrosion uas used eatlict.
Nuclear Power Plants.
INTRODUC710N a PWR vendo, and (d) solicitation of information The main focus of the report is on the ponion of from non-US utilities.
the feedwater system adjacent to the feedwater nozzle where fatigue cracking and wall thinning The literature reviewed included Information have been reported in the field. Figure 1 is a Notices and Bulletins issued by the USNRC, doc-sketch of this portion of the feedwater system, uments submitted to the USNRC by PWR utilities which includes the main and auxiliary feedwater and vendors, reports published by the USNRC and piping adjacent to the feedwater nonle, and the j
the Electric Power Research Institute (EPRI),
thermal sleeve, feedring, and J-tubes.
journal articles, papers presented at technical meetings, and conference proceedings. We visited The overall design of the main and auxiliary three PWR plants where fatigue cracking occurred feedwater systems for top-feed steam generators is after the Bulletin 79-13 closed in 1991: Diablo described in Section 2. The relevant differences in 1
Canyon, San Onofre, and Sequoyah. We visited the feedwater system layout for the steam genera-j Westinghouse Electric Corporation, which was tors with preheaters are also identified. The de-involved from the beginning in resolving the sign, materials, and fabrication of the feedwater feedwater nonle cracking problem. PWR utilities nonles, piping, and feedring are described in from eight other countries responded to our request Section 3. The operating transients and environ-to the Organization for Economic Cooperation and ment for feedwaar piping and noules are de-Development (OECD) for plant-specific informa-scribed in Section 4. The safety significance of tion about managing feedwater line cracking and main feedwater line break events is evaluated in wall thinning. Questions requesting the informa-Section 5. The field experience related to thermal tion are listed in the Appendix.
fatigue cracking, wall thinning caused by flow-Feedwater nozzle P
Thermat sleeve Generallocation J-tube of cracks Feedring g Steam generator shell eedwater fee tr ~'
Chock
{
i valve ep I
L i
Check valve feedwater Figure 1. Sketch of a portion of the PWR feedwater system and its components evaluated in the report.
=..
i INTRODUCTION accelerated corrosion, and steam generator water thermal fatigue cracking and on wall thinning hammer damage is summarized in Section 6. His caused by flow-accelerated corrosion are discussed section presents root cause analyses for damage in Section 7..Different inspection techniques experienced in the field. It summarizes 'the developed to characterize fatigue cracks and wall-inspections performed by the licensees to thinning are evaluated in Section 8. The related
- ch-rwuize the damage, and the actions taken to field experience in the United States and other mitigate the damage. The information obtained countries is also summarized in this section, ne from non-US utilities is also summarized in this current inservice inspection requirements are also section.
summarized. Several techniques to mitigate and i
monitor thermal fatigue, flow-accelerated corro-The remaining sections of the report include sion, and water hammer damage to feedwater discussion on how the interactions between design, piping, feedwater nozzles, thermal sleeves, feed-1 materials, stresses, and environment cause fatigue rings, and J-tubes are described in Section 9.
cracking and wall thinning of the feedwater piping Finally, the specific findings related to the effec-and the other components shown in Figure 1, and tiveness of the industry efforts for managing how the damage can be characterized and miti-thermal fatigue cracking and wall thinning of PWR gated. The qualitative and quantitative effects of feedwater system components are presented in piping materials, design and layout, coolant chem-Section 10.
i istries and temperatures; and flow velocities on l
a 4
- 2. MAIN AND AUXILIARY FEEDWATER SYSTEM DESCRIPTIONS Both main and auxiliary feedwater systems are feedwater heaters to the two main feedwater used to remove heat from the reactor coolant pumps; in some plants, both main feedwater system. The main feedwater system provides pumps are turbine-driven, whereas in other plants water required to maintain steam generator water both pumps are electrically driven. Some plants level during normal operation, after startup and have three rather than two main feedwater pumps before shutdown. The auxiliary (or emergency) and both types of pumps. The feedwater pumps feedwater system is used to maintain a heat sink in discharge pressurized feedwater into a common the steam generators during design basis events
~ ader, which then passes through three parallel such as loss of main feedwater, reactor trip, and high-pressure heaters into another common header, loss of offsite power. In most plants, the auxiliary and then branches into four lines to feed the steam feedwater system is also used to provide a source generators through the main feedwater nozzle. He of feedwater to the steam generators during plant nozzle is located above the tube bundle in a startup, hot standby, and shutdown. Thus the recirculating steam generator without a preheater, auxiliary feedwater system is used when the main as shown in Figure 3, whereas in a recirculating feedwater system is not available and the pressure steam generator with a preheater it is located near in the reactor coolant system is too high to permit the tube s! eet on the cold-leg side of the steam beat removal by the residual heat removal system.
generator, as shown in Figure 4.
In a steam generator without a preheater, the feedwater enters 2.1 Main Feedwater System through the feedwater nozzle to a feedring and then to J-tubes, which distribute the feedwater in l
PWR main feedwater systems generally include the steam generator, as shown in Figure 5. The J-multiple trains of feedwater piping (usually with a tubes direct the water downward into the steam J
few large headers at selected points) to handle the generator downcomer formed by the shell and the large volume of feedwater, for example,6.8 x 106 baffle around the steam generator tubes. Each kg/h (15 x 10'Ib/h) for a typical 1100-MWe unit.
feedwater line for the Westinghouse steam A multiple-train design also provides redundancy generators has its own feed control valve, isolation in the feedwater system that is beneficial fer safety valve, and check valve, reasons. The main feedwater system receives water from the condenser hot well and supplies it Generally, in a recirculating steam generator via a sequence of heaters and pumps at a much without a preheater, an auxiliary feedwater system higher temperature and pressure to the steam discharge line connects with each main feedwater generators. (The feedwater is pressurized by the line before it enters the steam generator. In some condensate and feed pumps, and in some plants, feedwater systems this connection with the feedwater booster pumps). The downstream auxiliary feedwater line is inside the primary piping terminates at the steam generator feedwater containment, whereas in others it is outside the nozzle.
containment. Ilowever, there are exceptions; for example, in the Palisades plant, the auxiliary Figure 2 presents a simplified schematic of a feedwater line connects directly to the steam typical main feedwater system for a Westinghouse generator, which is not equipped with a preheater.
4-loop plant with top feed steam generators (steam Also, in a steam generator equipped with a generators without preheaters). His flow diagram preheater, the auxiliary feedwater line connects begins downstream of the low-pressure feedwater directly to the steam generator above the tube heaters (not shown), at the booster pumps. The bundle. Westinghouse 2,3, and 4-loop plants flow proceeds through the sets of intermediate and Combustion Engineering plants have similar feedwater system design configurations.
FEEDWATER SYSTEM DESCRIPTIONS Containment wall Foodwater isolation va!ve l
h O
h8 l
SG g
Feed U @-
7#
ON4 OP
"'S" P *"
l Bypass Mrs
=g v4-&
gg nl M@Zd
=[
10 SG Feed y
d Main SGass feed control and fili valve pumps Q
Bypass n
I 2A 2B 2C P
O-SG i.ck w
Aux.
v,
,ood
~
valve ini.nnediaie I
heaters Bypass 4A 4B 4C l
p SG
,h-goo,,,,
hFeed i
pumps control A
B C
Aux valw V
feed P Pressure F Flow M Check valve Motor operated valve From M Stop valve Air operated valve low pressure heaters s41ss 4
Figure 2. Schematic of a typical main feedwater system for a Westinghouse 4-loop plant with top-feed steam generators (USNRC Inspection and Enforcement Training Center).
i 5
FEEDWATER SYSTEM DESCRIPTIONS i
Steam outlet Steam u
K
.o.y> sm.m
/ i./ C 5
/
n.
i 3
y
~-
yWater level ter
{
"q
~
u j
armius f
)
nozzle
/D Steam %ater mixturo leaves 4
31
._ f tube bundle ilLlll p Tube
- [
Preheater
- 2 P-:kFeedwater
[
4 inlet Primary irdet Primary outlet j k.
rt.
3 w
N Buffer zone Recirculating sieam generator NO3 0022 rev dp (Model E) ug3oo4s Figure 3.
Sketch of a recirculating steam Figure 4.
Sketch of a recirculating steam generator without a preheater. The main feedwater generator with a preheater. The main feedwater i
nozzle is located above the tube bundle, nozzle is located near the tubesheet elevation.
i i
FEEDWATER SYSTEM DESCRIPTIONS I
I r
m 7 g
= Feedring 10-in. xs as pipe y
J tube (typ) is is is Hot leg Cold leg N,
A,j8
.y 1
sih)
[
12 16 in.
Ece. red
'e' Schedule 40 16 in Sch. 40 x s
tee 10 in Sch. XS
/'
%. D 7, '
s2
/
g as i
~I'~
~
84 16 in. Schedule 40 ~ \\\\w thermal sleeve C
1 h
- Feedwater nozzle g
(a) Feedring y J tube fT M95 0212 i
(b) Typical feedring cross section showing J tube arrangement Figure 5. Top view of Diablo Canyon feedring (Thailer Dalal, and Goyette 1995). Copyright American Society of Mechanical Engineers; reprinted with permission.
7 NUREG/CR-6456 l
FEEDWATER SYSTEM DESCRIPTIONS A simplified schematic of a Babcock & Wilcox (B&W) main feedwater system is shown in separate lines feeding the feedwater header ofcach Figure 6. There are two parallel trains of feed.of the two cnce-through steam generators. Then water.
In each, the flow passes through the feedwater enters each steam generator throug low-pressure heaters, the main feedwater pump, several risers spaced along the feedwater header, as and high-pressure heaters, finally disc? :.,.,ing into shown in Figure 7. Each feedwater line has its a common header. The flow then branches into on feed control valve,' main stop valve, and check valve.
l l
Atmospheric dump valves I
i Rehef valves typical typicalof 3 in each Gne E E0 of 9 in each hne i
y
- to turbines I
g'i]6--*
-d h Airoperatedcontrof valve F
l Steam d Motor operated valve
'* *i
- l generators
, f l
- Air operated valve l-iw
++ Check valve l
t 1
1 1
- Block valve l
l-J l
Reactor coolant pumps pg 1
i 1
l valve High pressure Control feed pump Low pressure
[aters heaters t
i t
I Main
\\
,f l
L Heater Main i
i l
drain Main feedwater feed hnes from Reactor l
condensate g
up o
systern f
n inside reactor building l
Outside reactor building I
U w
,Emerg u
j N93 00U Emergency feed pump (motor driven)
Emergency feed pump (turbine driven)
Figure 6. Schematic of a typical main feedwater system for a Bab cock & Wilcox plant.
_ _ _ - - - - - - - - - ~ - - - - - - - - - - -
~~~~
}
Primary irdet 7//
+-- U r
)f tu sheet 3-in. inlet pipe f
(typ of 7) 7
[
g l
h Auxiliary (l ""'"'
c inlet
[- g 6 ft - O in.
g l.
j.a Steam 6-in. auxiliary r
Jl j
outlet feedwater 7
,l header f
- g fj'g 4 3, inlet O
G j 4
header Feed -
- I :>
Aspirating annulus s'
steam (typ of 32)N
=
s
- 3: l i
- M eh
=
h 3 ft - 9 in.
support feedwater I
,[
I I
T First 14-in. main s:s:*
l' h
C.
g;
[EI C
i plate header ll f/
+- Lower N
y
- M tube N
Hg sheet Primary outlet g
M95 0334 (a) Once-through uss 0331 steam generator g
h (b) Sketch of feedrings triw x
O tT1 b
9
-e n
[
Figure 7. Cross-section of a Babcock & Wilcox once-through steam generator with external header (Han and Anderson 1982).
hz a
w a
FEEDWATER SYSTEM DESCRIPTIONS Main feedwater flow generally initiates at the end to 3.5 MPa (300 to 500 psia) and 160 to 204*C of hot standby (Mode 3 operation, see Section 4)
(320 to 400*F), respectively; while at the outlet, and the beginning of plant startup (Mode 2 they vary from 5.9 to 8.3 MPa (850 to 1200 psia) operation). Initially at low power, the main and 160 to 204*C (320 to 400*F), respectively.
feedwater pumps are driven with steam from the The typical feedwater pressure and temperature at main steam system; later on at higher power, the the steam generator inlet vary from 5.5 to 7.9 MPa pumps switch to the steam from the main turbine.
(800 to 1150 psia) and 210 to 238*C (410 to in some plants with electrically driven main 460*F). The typical bulk flow velocities at 100%
feedwater pumps, the main feedwater syem, load range from 3 to 7.6 m/s (10 to 25 ft/s).
instead of auxiliary feedwater system, can be used However, local flow velocities may be as high as j
to provide feedwater to the steam generators during 15 m/s (50 f1/s) or higher.
plant startup, hot standby, and shutdown.
The main turbine extraction steam supplies the heat to the main feedwater heaters; initially the A special startup system is part of the main amount of heat supplied is small and it increases as feedwater system in two Westinghouse plants, i
the power increases.
So initially the main Wolf Creek and Callaway. The startup system j
feedwater is cold when it is started and there is a provides heated feedwater to the steam generators large temperature difference between the main during startup, hot standby, and shutdown feedwater and the secondary coolant in the steam operations of the plant. The startup system is generator; this temperature difference decreases as provided as an option to feeding the steam the power increases. For example, Consumers generators with cold auxiliary feedwater, which is Power (1979) reported that the initial temperature the practice at several PWR plants and is discussed of the main feedwater at the Palisades plant was in the next section. The main objective of the about 127'C (260 F) at less than 1% power; then special startup system is to avoid thermal fatigue the temperature was raised to about 160*C resulting from flow stratification and cracking of (320*F) at 20% power, and to about 215*C the steam generator feedwater nozzles and piping.'
(420*F) at full power.
The temperature differences between the main feedwater and the A low flow, motor driven feedwater pump is secondary coolant in the steam generator were installed within the special startup feedwater about 130*C (250*F),105 *C (200*F), and 53 *C system to provide heated feedwater flow during (100"F) at 1%, 20%, and 100 % power startup and shutdown operations. The suction of respectively. During low power operation, the the startup pump is taken from the heated main feedwater flow rate is manually controlled condensate outlet of the steam generator blowdown and there are fluctuations in the flow rate.
> egenerative heat exchanger, which also provides Therefore, if the flow rate is suffkiently low, dditional heating of the condensate.
The stratified-flow cycles could take place during low maximum flow rate through the startup pump is power operation.
.imited to about 1.4 x 10 5 kg/h (3 x 10 5lb/hr, which is equal to 600 gpm) to prevent excessive Some PWR units have steam generator feedwater tube vibrations in the steam generator blowdown pressures much higher than the coolant saturation regenerative heat exchanger. The startup pump pressure, whereas other units have feedwater discharges to the high-pressure feedwater heaters, pressure at the suction of the feedwater pump close which use steam generator blowdown flash tank to the saturation pressure (Jonas 1988). The steam as the heat source during a plant heatup.'
operating pressure and temperature at the inlet to the main feed pumps at 100% load vary from 2.0
' USNRC Technical Training Center. " Condensate and Feedwater." Westmghouse Tecimology Systems Manual, Volume I, Chapter 7 2 NUREG/CR-6456 10
FEEDWATER SYSTEM DESCRIPTIONS i
2.2 Auxiliary Feedwater System In some PWR plants, each auxiliary feedwater line has a minimum flow orifice installed, which restricts maximum flow through the line, especially Figure 8 is a schematic of an auxiliary feedwater when the steam generator is depressurized. In system for a 4-loop Westinghouse plant with the some PWR plants, an automatic isolation of the top feed steam generators (Finn 1989). The auxiliary feedwater line is provided if the steam auxiliary feedwater system consists of two subsystems, cach capable ofsupplying 100% flow, generator is depressurized.
In addition, an auxiliary feedwater line in a typical plant can be one utilizes a turbine-driven pump and the other isolated manually or remotely from the control utilizes two motor-driven pumps. Both subsystems room.
can deliver feedwater to all four steam generators.
Some plants have only one motor-driven pump and The auxiliary (or emergency) feedwater lines in the at least one plant has a diesel-driven pump rather B&W plants introduce feedwater directly into the than a turbine-driven pump.
The auxiliary steam generators as shown in Figure 7(b).
feedwater flow initiates at the condensate storage tank and passes through three parallel, auxiliary 2.2.1 Preheater-Equipped Steam feedwater pumps, and through individual check, Generators control, and isolation valves for each of the three lines. Part of the flow from the pumps returns to i
Figure 9 is a schematic of a feedwater system for a the condensate storage tank via a recirculation line.
Westinghouse plant with preheater-equipped steam The flow to the steam generators is usually generators. A 102-mm (4-in.) diameter preheater controlled by means of an air-operated control bypass line connects the main feedwater line to the valve in each line to the steam generators. Motor auxiliary feedwater line outside the containment.
operated valves can be used should the air supply in some plants there is no check valve on the fail. Each auxiliary feedwater line connects to the auxiliary feedwater line inside the containment as j
main feedwater line somewhere between the shown in Figure 9, whereas in other plants there is feedwater nozzle and the first check valve one. The preheater bypass line serves two pur-upstream from the nozzle as shown in Figure 2.
poses. During startup, when the plant power is below approximately 15 to 20% of full power, all The auxiliary feedwater condensate storage tank is the feedwater flow to the steam generator is di-typically at a temperature ofabout 38 to S0*C (100 rected through the bypass line and then through to 120*F); however, the condensate storage tank the top-feed auxiliary feedwater nozzle. This can be at a temperature much lower than 38'C if eliminates the possibility of a condensation-in-located outdoors. The tank capacity varies, at one duced water hammer in the preheater. At about plant it is 500,000 gallons. A minimum of 15% to 20% of full power, a very small amount of 170,000 gallons are required for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> at hot feedwater flow [for example,27,000 to 36,000 standby, followed by 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br /> of cooldown. When kg/h (60,000 to 80,000 lb/h or 120 to 160 gpm)) to the condensate storage tank is exhausted, the the steam generator is directed through the main auxiliary feedwater system can be supplied from feedwater nozzle via the main feedwater isolation other sources such as the service water system.
valve (MFIV) bypass line (not shown in Figure 9).
This small amount of feedwater flow slowly flush-During normal operation the auxiliary feedwater es the cold feedwater from the piping downstream system is in standby. The motor driven pumps of the main feedwater isolation valve and warms generally start on either a steam generator low-low up the piping. This process of warming up is level signal, a safety injection signal, or a loss of generally referred to asforwardflushing. The electric power. The turbine-driven pump is started main objective of warming up the feedwater piping on either a low-low level in any two steam is to prevent the cold feedwater from collapsing generators or a complete loss of electric power.
FEEDWATER SYSTEM DESCRIPTIONS Ie
!!i
!!i j5 d
d d
d B
j m
i i
!,% 1% !%
- m%
iii i
m m
.ii s.ac D k
.l
[]
[]
[7 13 i,li 9
9 9
1h I
1 1,
1 1
1 z
s og eg %
4
=
.s d'l"l.
Nell
','i $:
i i
T' P' "
i
~
- t!,l u tl f.
Y m
- c D
D4
}
on on U on l
-I E
oX oX oX DK DK 3'il g
bi fill fil i fil!
i j
d P.
1 x
1 d
x I
.. ~
To condenser ai 14% leed flow From at 100% power W ausdliary FRPO -
-Q o
b aF sF > <l lll From main feeowater I
Sham generator AFFE FBTV equelzat nheader AU" FBTO nozzle Preheemt bypass Ene Purge line
-+
\\
/
W
=
u FPBV 1 r
+-
V C
Main k
9 g
,eedw,
m o
g
.p'
.. =
A
- - _ From rnain feedwater T
MFIV 59 6 %
equalizati n header h
MF E U
V Xt B
m M
86% tiow at
,a n
.-o e x
FCBV g
u g
M e
Listof Abbreviatons FPBV Feedwater preheater bypass valve W
AFFE Auxiliary feedwater flow element FPV Feedwater purge valve h
FBTO Feedwater bypass tempering orifice FRFC Feedwater reverse purge onfice ~
M cr ckvehe unear operused vehe g
C FBTV Feedwater bypass tempenng valve MFFE Main feedwater flow element y2 O
FCBV Feedwater control bypass valve MFFR Main feedwater flow restricting orifice g
O FCV Feedwater control valve MFIV Feedwaterisciation valve y
n
- emwines*
b
- =
O T
2:
y Figure 9. Schematic of a feedwater system for a W&..gw plant with steam generators eqinpped with y.JA Courtesy of Den Caldwell, Duke m
Power.
=
4 any steam bubbles in the preheater section, which the auxiliary feedwater line. A check valve in the could cause water hammer damage to the bypass line prevents How from the auxiliary feed-preheater. The feedwater temperature at 15%
water line to the main feedwater line. This split-power is about 129'C (265'F), whereas the corre-ting of the main feedwater flow during operation sponding temperature of the secondary coolant in was not included in the original design. The main the steam generator is about 293*C (560*F). So reason for this design change was to reduce fretting
)
the small amount of feedwater How used during damage to the steam generator tubes in the pre-j forward Dushing will cause thermal stratification in heater region of some Westinghouse-designed the feedwater nozzle and adjacent piping with a steam generators. However, the preheater bypass top-to-bottom temperature difference of about line and the portion of the auxiliary feedwater i
164*C (295*F). Therefore, about 2000 cycles of piping between the bypass line and the auxiliary stratified flow in the preheater-equipped steam feedwater nozzle were not designed for this opera-generator feedwater nozzle have been allowed for tion and they are now exposed to high-velocity in the design!
feedwater flow during normal operation and are susceptible to wall thinning. In some Combustion The second purpose for the preheater bypass line is Engineering plants with preheater-equipped steam to reduce the main feedwater flow into the pre-generators, a similar split-flow operation is carried J
heater during operation above 15 to 20% of full out and the corresponding small-diameter piping is power. This is accomplis!ied by installing a flow susceptible to wall thinning. Field experience orifice in the main feedwater line as shown in related to this wall thinning is described in Section Figure 9, which directs a fraction of the main 6.4.1.
feedwater flow (for example,14% in one plant) to
' V. N. Shah, Private conversation with J. L lloutman.
Westinghouse Electric Corporation, July 22,1996.
1 NUREG/CR-6456 14
- 3. DESIGN, MATERIALS, AND FABRICATION His section describes the design and materials of basically similar. However, the feedwater piping the secondary system components adjacent to the was designed by the plant architect engineer and, feedwater nozzle, including the feedwater as a result, the designs take several different forms distribution system inside the steam generator.
(Hu, Houtman, and White,1981). He piping may Relevant geometric characteristics ofthe feedwater have a several-foot long horizontal section nozzle are also described.
Overall design preceded by an upstream elbow, or it may have a configuration of the main and auxiliary feedwater 90-degree vertical downward elbow welded systems is described in Section 2.
directly to the nozzle. Figure 10 shows the original configuration at the D.
C.
Cook plants 3.1 Design Configuration
(' westinghouse 4-loop plants) where the feedwater nozzle is welded to a 90-degree elbow. A reducer, The feedwater nozzles are part of the steam transition piece, or safe end may be present between the nozzle and the elbow, the other end of generator shell designed by the nuclear steam which is welded to the feedwater piping.
supply system (NSSS) vendors and, therefore, are N
%h
~~
~
~
n t
16-in 00
/ b'diograp$..
therrnal sleeve
)
and feedring inlet uaem
,/
plug f
,."' p,
j f
/
I e'
y
(
/
/
N p
/
i
/
/
i
/
\\
/
.i
/
24ln.
l f
nominal radius J Nomina'l wall at L
\\
'I weld = {890 in. T l
l l o(alj I
I N
l 1'*!
l J Nomina'l well at 1 weld = Q.750 in.
Figure 10. 90-degree elbowjoining feedwater nozzle and piping in the original installation at D.C. Cook plants.
l 15 NUREG/CR-6456
DESIGN, MATERIALS, AND FABRICATION Figure 11 sbws the con 6guration of the San The J-tubes are bent so that the exiting feedwater Onofre Unit 1 plant (Westinghouse 3-loop plant) is directed away from the steam generator shell.
where an expander (sometimes referred to as a The distribution of the feedwater in the l
reducer, but in this case the flow area is downcomer is not uniform; there are more J-tubes expanding), which acts as a transition piece, is on the hot leg side of the steam generator than on placed between the piping and the nozzle.
the cold leg side. Because the J-tubes are mounted Figum 12 shows a typical Combustion Engineering on the top of the feedring, rapid draining of the feedwater nozzle welded to a safe end. As shown feedring is prevented in the event the steam in Figure 13, some modi 6ed feedwater nozzle and generator water level falls below the feedring, thus piping designs have a thermal liner protecting the reducing the potential for a water hammer event.
elbow and transition piece. Weld connections are In the early steam generators, sparging holes at the described in Section 3.3. A radiograph plug (also bottom of the feedrings were used to distribute the called a gamma plug) is installed in the feedwater feedwater, which led to rapid drainage of the piping adjacent to the nozzle to allow access for feedwater line when the feedring was uncovered.
single-wall radiography (see Figure 10).
Consequently, the feedrings were modined to prevent rapid drainage. The modi 6 cation included in the Westinghouse design, the feedwater enters plugging the sparging holes on the underside of the the steam generator through a thermal sleeve, feedring and installing J-tubes on the top of the passes through a feedring (sparger) and then J.
ring. In fact, the steam generator feedrings were tubes mounted on the top of the feedring, as shown eventually modined at most of the operating in Figure 5, and is eventually distributed into the Westinghouse units in response to the water steam generator downcomer formed by the shell hammer event at Indian Point 2 (Han and and the bame around the steam generator tubes.
Anderson 1982).
Feedwater nozzle 1
m m
e
.........N ozpe,,pjpe _,,,,,,,,,,,,,,,,,.
expander Thermal sleeve l
Normal (feedring inlet) l
\\
^
N A
1 C144 WHT 1095 02 I
Figure 11. Expanderjoining feedwater nozzle and piping at the San Onofre 1 plant.
i DESIGN, MATERIALS, AND FABRICATION Thermar si. eve leend b4m25e weld 406 mtn (16 in. )
M**O*
Steam generator Feedwater pipe b I
nozzle juncture
?my we l
Sale end SA-508 CL 1 (carbon steel)
SA 508 CL 2 (low-aMoy steel)
M9s 0336 Figure 12. Sketch of a feedwater nonle at a typical Combustion Engineering plant.
l XN ui Elbow hner Liner %
Piston rings--e
.....................9
.g 4
I NF'_
Il l Thermal N
i,
\\
\\
Feedwater nozzle
'g
[Inconel600 buf. dup i
a Main reducer
\\
J C253-WW1 1BB647 Figure 13. Modified feedwater nonle and piping configuration at D.C. Cook plant. A thin liner protecting the counterbore and elbow from thermal stratification is installed.
i 17 NUREG/CR-6456
DESIGN, MATERIALS, AND FABRICATION
! 1 m rmaisieove re4nd-to-nor.zie weld h'Obdiy b
19
>eny,, p 3
406 mm (16 in. )
foodwater Ene generator Feedwaterpipe to nozzle juncture i I M sgs&-
l
- gP"-
i P
Safe end q
,j SA-508 CL 1
[
(carbon steel)
G SA 508 CL 2 (lew-alloy steel)
M95 0336 Figure 12. Sketch of a feedwater nozzle at a typical Combustion Engineering plant.
h i'
in Liner #j Elbowliner Piston rings 4 k
's 161n.
wrmaid I
' \\(
's.'
erbow sleeve
\\
Feedwater nozzle i
I u
[Inconel600 buildup 3
Main reducer i
i i
C253 WHT.199647 the counterbore and elbow from thermal stratification is inenRhd_ Figure 13 i
DESIGN, MATERIALS, AND FABRICATION that are welded to the shell wall. A 3-in, elbow through aspirating ports, heats the feedwater to the and tee vent assembly is attached to the upper saturation temperature.
portion of the end of the distribution box toward the interior of the steam generator. Modifications In some B&W plants, the auxiliary feedwater have also been made in the feedwater distribution header (feedring) is located external to the steam system in the San Onofre steam generators and generator shell as shown in Figure 7(b). In other are described in Section 6.4.2.
Design B&W plants, the feedring for distributing the modincations have also been made at other auxiliary feedwater was originally located inside Combustion Engineering plants. For example, in the steam generator. However, the feedrings in May 1979 the Calvert Cliffs I steam generator three of these plants were found collapsed in the feedring was modified by adding thirty-six carly 1980s. So as a corrective action, these 90-degree elbows to the top of the ring and auxiliary feedwater distribution systems were plugging 72 discharge nonles on the bottom of the modified by replacing the internal feedrings with ring to minimize the possibility of water hammer external auxiliary feedwater headers similar to the events.
one shown in Figure 7(b)(Serkiz 1983).
He feedwater nonle bore, blend radius, and steam 3.2 Materials generator shell inside surface directly beneath the nonle are susceptible to thermal fatigue if there is At most U.S. plants, the majority of the main and any leakage of the feedwater through the joint auxiliary feedwater piping is made of carbon steel, between the nonle and thermal sleeve. In the typically SA-106 Grade B or SA-333 Grade 6 for original Westinghouse steam generators, slip-on straight portions of piping and SA-234 Grade type thermal sleeves were tightly Gtted in the WPB for elbows, as listed in Table 1. The feed.
feedwater nonle, whereas in the replacement ring, J-tubes, and thermal sleeve are also made of steam generators the sleeves are welded to the SA-106 Grade B carbon steel. The steam genera-nonie. He radial gap is typically 0.010- to 0.020-tor shell is made of low-alloy ferritic steel plates, in. at the slip fit and 0.25-in. over the remainder of typically SA-533 Grade A, B, C, Class 1 or 2, or the sleeve diu, ttouunm., d White 1981).
SA-302 Grade B materials. The feedwater nonle is a low-alloy ferritic steel forging, SA-508 Class Distribution of th: main feedwater into the B&W 2 material. Manganese sulfide inclusions are steam generators is different than that in the West-generally present in the carbon steel piping materi-inghouse and Combustion Engineering steam als and low-alloy steel shell materials. These generators. Water from the feedwater header inclusions reduce the crack growth resistance of enters each unit through 32 inlet pipes (risers) these materials. Because of the difference in the spaced around the steam generator shell, as shown coefficients of thermal expansion between the in Figure 7. The header functions like a feedring thermal sleeve and nonle materials, the design located external to the steam generatcr shell and includes a small gap between these two compo-the risers function like J-tubes with resm: :a 7,;,3, keeping the header full of water, that is, the feed-water inlet pipes are at a higher elevation than the Carbon steel J-tubes with very low levels of main feedwater header and, therefore, the feed-chromium (<0.01 wt%) as a trace element have water line cannot be rapidly drained when the experienced significant wall thinning because of steam generator water level falls below the eleva-now-accelerated corrosion and, in some cases, tion of the feedwater nonles. Therefore, the resulted in failure (Roarty 1986). Several utilities feedwater lines in the B&W plants are not suscep-have replaced their carbon steel J-tubes with Alloy tible to water hammer damage. The feedwater 600 J-tubes to address this problem. Some utilities entering the steam generator is sprayed into the have replaced their carbon steel J-tubes with J-downcomer. Steam, drawn into the downcomer tubes made of 2 % Cr-1 Mo material.
DESIGN, MATERIALS AND FABRICATION 0
Table 1. Typical materials and fabrication codes for feedwater piping (Florida Power & Light).
Piping Material Code Feedwater nonle to outermost ASME SA-106, Grade B ASME Section III, Class 2 containment isolation valve Balance of piping ASME SA-155, ANSI B31.1 Grade KC-65 ASME SA-106, Grade B Elbows ASTM A-234, Grade WPB ANSI B31.1 Auxiliary feedwater ASME SA-106, Grade B ASME Section III, piping Class 2 or Class 3,as applicable 3.3 Geometric Discontinuity sign and the codes of construction, the weld and its inspection varies from plant to plant.
The The butt weld that connects the steam generator geometric discontinuity at the junction of the nonle to the feedwater pipingjoins components nonle and the pipe causes stress concentration and with a large difTerence in wall thickness because of has been the location for many of the cracking the difference in the strength of respective incidents. As shown in Figure 15, a counterbore at materials. 'Ihe steam generator nonle is typically the inside diameter, and a backing ring for the made of SA-508 Class 2, which has an allowable weld were originally used. In many cases the stress of 184 MPa (26.7 ksi) at the design cracked component was replaced in kind with the temperature of 316*C (600*F). The pipe material, same design, which, as discussed in Section 6, has typically SA-333 Grade 6 or SA-106 Grade B has led to a second, and in some cases, a third-an allowable stress of only 119 MPa (17.3 ksi).
replacement. Consequently, redesigns of the Therefore, the pipe thickness is approximately 1.5 configuration have been proposed to reduce the times the nonle thickness. Depending on the de-damaging effects of the piping discontinuity.
These redesigns are discussed in Sections 6 and 9.
Pipe end Nonle end
[406 mm (16 in.) Schedule 80}
[406 mm (16 in.) Schedule 60]
J,
?
,.,. y' W r ;
17 mm
. p.
di O
P 21.4 mm
- ,, M Sc, j'
- : ', 1(0.656 in.)
(0.843 in.)
Backing ring Counterbore Figure 15. Typical PWR feedwater nonle-to-pipe weld with a counterbore.
DESIGN, MATERIALS, AND FABRICATION Thermal sleevs le+nd lo4wzzle weld
[.
mA {Q 406 mm (16 in. )
Steam generator Feedwater pips to I
nozzle juncture I
e fe.
?i '. ;
.,9 3
- a _* s.
[
SA 506 CL 1 (carbon steel) i SA 508 CL 2 (low eRoy steel)
M95 0336 Figure 12. Sketch of a feedwater noule at a typical Combustion Engineering plant.
A N
in Elbowliner Liner #,
j Piston rings--e
....._.............t 16 in.
r s===r oi
\\
's.
Thermal Sleeve N
Feedwater nozzle
- g i
[ inconel 600 bundup 4
Main reducer C2S3 WHT 199647 Figure 13. Modified feedwater nonle and piping conHguration at D.C. Cook plant. A thin liner protecting the counterbore and elbow from thermal stratification is installed.
DESIGN, MATERIALS, AND FABRICATION l
The distribution of the main feedwater into the sign of the distribution system in the San Onofre 2 Combustion Engineering steam generators is and 3 steam generators (Martin et al.1990). The l
somewhat similar to that in the Westinghouse feedring consists of 12-in. diameter piping and is steam generators. Figure 14 shows the overall de-attached by two U-bolts at each of four supports i
1 Transition pipe Discharge elbows I
Thermal inlet box
}
sleeve N I,,
Vent Feedwater
~
'""lli O
' y
\\ p,
- g vj Weds g
q g j
3 shIjfM.<
J h
I Support m
a
\\
penter support steam Support
-l Auxiliary feedwater shown) generator piping connection shell Steam generator shell l
(a) Side view Feedwater i
4 Feedwater nozzle pipe l
- 4 Distribution box Feedring 904egree support I
A m%
U bolts i
4 Discharge elbow Steam generator shell 1704egree support -
M9e 032s i
(b) Top view l
Figure 14. San Onofre Units 2 and 3 feedwater distribution piping (Martin et al.1990). Courtesy of t
E. Regala, Southern California Edition.
DESIGN, MATERIALS, AND FABRICATION that are welded to the shell wall. A 3-in. elbow through aspirating ports, heats the feedwater to the and tee vent assembly is attached to the upper saturation temperature.
portion of the end of the distribution box toward the interior of the steam generator. Modifications In some B&W plants, the auxiliary feedwater have also been made in the feedwater distributior header (feedring) is located external to the steam l
syctem in the San Onofre steam generators and generator shell as shown in Figure 7(b). In other are described in Section 6.4.2.
Design B&W piants, the feedring for distributing the modifications have also been made at other auxiliary feedwater was originally located inside Combustion Engineering plants. For example,in the steam generator. However, the feedrings in May 1979 the Calvert Clifts I steam generator three of these plants were found collapsed in the feedring was modified by adding thirty-six carly 1900s. So as a corrective action, these 90-degree elbows to the top of the ring and auxiliary feedwater distribution systems were plugging 72 discharge nozzles on the bottom of the modified by replacing the internal feedrings with ring to minimize the possibility of water hammer extemal auxiliary feedwater headers similar to the events.
one shown in Figure 7(b)(Serkiz 1983).
He feedwater nozzle bore, blend radius, and steam 3.2 Materials generator shell inside surface directly beneath the nozz!e are susceptible to thermal fatigue if there is At most U.S. plants, the majority of the main and any leakage of the feedwater through the joint auxiliary feedwater piping is made of carbon steel, between the nozzle and thermal sleeve. In the typically SA-106 Grade B or SA-333 Grade 6 for original Westinghouse steam generators, slip-on straight portions of piping and SA-234 Grade type thermal s!eeves were tightly fitted in the WPB for elbows, as listed in Table 1. The feed-feedwater nozzle, whereas in the replacement ring, J-tubes, and thermal sleeve are also made of steam generators the sleeves are welded to the SA-106 Grade B carbon steel. The steam genera-nozzle. He radial gap is typically 0.010- to 0.020-tor shell is made of low-alloy ferritic steel plates, in, at the slip fit and 0.25 in. over the remainder of typically SA-533 Grade A, B, C, Class I or 2, or the sleeve (Hu, lloutman, and White 1981).
SA-302 Grade B materials. He feedwater nozzle is a low-alloy ferritic steel forging, SA-508 Class Distribution of the main feedwater into the B&W 2 material. Manganese sulfide inclusions are steam generators is different than that in the West-generally present in the carbon steel piping materi-inghouse and Combustion Engineering steam als and low-alloy steel shell materials. These generators. Water from the feedwater header inclusions reduce the crack growth resistance of enters each unit through 32 inlet pipes (risers) these materials. Because of the difference in the j
spaced around the steam generator shell, as shown coefficients of thermal expansion between the in Figure 7. The header funcdons like a feedring thermal sleeve and nozzle materials, the design located external to the steam generator shell and includes a small gap between these two compo-the risers function like J-tubes with respect to nents.
keeping the header full of water, that is, the feed-water inlet pipes are at a higher elevation than the Carbon steel J-tubes with very low levels of j
main feedwater header and, therefore, the feed-chromium (<0.01 wt%) as a trace element have water line cannot be rapidly drained when the experienced significant wall thinning because of steam generator water level falls below the eleva-flow accelerated corrosion and, in some cases, tion of the feedwater nozzles. Therefore, the resulted in failure (Roarty 1986). Several utilities f
feedwater lines in the B&W plants are not suscep-have replaced their carbon steel J-tubes with Alloy tible to water hammer damage. He feedwater 600 J-tubes to address this problem. Some utilities entering the steam generator is sprayed into the have replaced their carbon steel J-tubes with J-downcomer. Steam, drawn into the downcomer tubes made of 2 % Cr-1 Mo material.
DESIGN, MATERIAI S, AND FABRICATION Table 1. Typical materials and fabrication codes for feedwater piping (Florida Power & Light).
Piping Material Code Feedwater nonle to outermost ASME SA-106, Grade B ASME Section Ill, Class 2 containment isolation valve Balance ofpiping ASME SA-155, ANSI B31.1
)
Grade KC-65 ASME SA-106, Grade B Elbows ASTM A-234, Grade WPB ANSI B31.1 Auxiliary feedwater ASME SA-106, Grade B ASME Section 111, piping Class 2 or Class 3,as applicable i
3.3 Geometric Discont:nuity sign and the codes ofconstruction, the weld and its in:pection varies from plant to plant.
The The butt weld that connects the steam generator geometric discontinuity at the junction of the nonle to the feedwater pipingjoins components nonle and the pipe causes stres concentration and with a large difference in wall thickness because of has been the location for many of the cracking the difference in the strength of respective incidents. As shown in Figure 15, a counterbore at materials. The steam generator noz:.!e is typically the inside diameter, and a backing ring for the made of SA-508 Class 2, which has an allowable weld were originally used. In many cases the stress of 184 MPa (26.7 ksi) at the design cracked component was replaced in kind with the temperature of 316*C (600*F). The pipe material, same design, which, as discussed in Section 6, has typically SA-333 Grade 6 or SA-106 Grade B has led to a second, and in some cases, a third an allowabic stress of only 119 MPa (17.3 ksi).
replacement. Consequently, redesigns of the i
Therefore, the pipe thickness is approximately 1.5 configuration have been proposed to reduce the times the nonle thickness. Depending on the de.
damaging effects of the piping discontinuity.
These redesigns are discussed in Sections 6 and 9.
Pipe end Nonle end
[406 mm (?? %)ihedule 60]
[406 mm (16 h) Schedule 6' ]
h i
,c4:
y o
s c.; m b
- v = : -
17 mm A \\'i (0.656 in.)
n e
y a
Backing ring Counterbore Figure 15. Typical PWR feedwater nonle-to-pipe weld with a counterbore.
DESIGN, MATERIALS, AND FABRICATION The steam generators have been designed and ing was procured to USAS B31.1.0-1967, but most constructed by the NSSS vendors according to the other piping, including the feedwater piping, was ASME Code. Before 1963, Section Vill of the procured to B31.7-1969, with no addenda.
Code was used.Section III, Nuclear Vessels, was first issued in 1963 and it and its successors have There are several configurations of the first up-been used for nuclear vessel construction ever stream weld from the steam generator in different since. The vessel rules are generally applied to the PWRs. For example, at the San Onofre 1 plant, end of the noule. Feedwater piping is designed by the first upstream weld was a nonle-to-expander the architect engineer. Before 1969, USAS B31.1 weld, and the next weld was an expander-to-pipe was used, from 1969 to 1971 USAS B31.7 was weld as shown in Figure 11. At the D. C. Cook used, and since 1971 Section 111 of the ASME plants, the pipe elbow is welded to the nonle as Code has been used. The ASME Code essentially shown in Figure 10. As shown in Figure 16, at the incorporated the piping rules of B31.7; however, Sequoyah plants, TVA chose to install a transition the requirements for Section III are more stringent piece of SA-508 Class 2 material, which was shop than those for B31.7. Plants designed before 1971 welded to the piping using a qualified P3 to P1 followed different codes and standards for different weld procedure.' The field weld between the vessel and piping components. For example,in the nonle and the transition piece, which have Sequoyah plants, designed before 1971, most of essentially uniform wall thicknesses, was made the nuclear equipment, including the steam using a qualified P3 to P3 weld procedure. If the generators, was procured by Westinghouse. Its field weld was within the limits of reinforcement code of record is ASME Section 111,1968 edition for the nonle, then the weld had to meet Section (Bressler 1994). The piping was designed and Ill requirements. If it was located outside the procured by the architect engineer, National Valve limits of reinforcement, then it was a pipe weld and
& Manufacturing Company. The main steam pip-had to meet B31.7 requirements.
1/8-in. R min Field weld Field weld 164n. transition piece N
ASME SA-508 Class 2 fk;a% *< N1/1.0 ir@ }Nu,Y kN$ ' W <
Ar g
\\,
k t
- Q j
.s h,
hb yhMk,,
f 9[
3/4 in.
4 MM j$y y,
4"*d**'""
aM Ek U4 In
~
ASME SA-508 Class 2 79 79 16 in.feedwater pipe
.d
.d ASME A-420 Grade WPL-6 Sch 80 g
g fabricated from A-333 Grade 6 5
seamless pipe s
q M95 0197 Figure 16. Construction drawing of transition piece for Sequoyah Units 1 and 2 (TVA 1992b).
' To reduce the number of welding procedure qualifications required, base metals have been assigned P-numbers. These assignments are based essentially on comparable base metal characteristics. such as composition, weldability, and mechanical properties. The P-number for SA 106 steelis 1 and for SA 533B,it is 3 (ASME 1995b).
1 21 NUREG/CR-6456
i DESIGN, MATERIALS, AND FABRICATION Figure 17 shows configurations for the Diablo (0.843 in.)] meets the Schedule 60 nonle [ nominal Canyon nonle-to-pipe and pipe-to-pipe welds thickness 16.7 mm (0.656 in.)]. The weld prepar-(Cofie et al.,1994). The 230-mm (16-in.),
tion counterbom reduces the thickness, as measur-1 Schedule 80 piping (nominal thickness 21.4 mm ed in one nonie, to as low as 13.7 mm (0.54 in.).
20i 21/2 deg.
\\
Q u.*dlAmMiB W MW @@&{%
' 19 '.-, > 'If$tendgenitath
! Schedule'80K Mf';, ',"' A 4' 't..~
,jfeedOcter nozzle?f firanMtion ploceL
-^
'dSA-SOS 4* 1^73%l class 2)T;"-'N 0.6561n.
k
..\\
- 0.843 in.S(SA4106 grade B)#--
4 e
x:
WW4 Wir /32
- 1/64 inF @
J.c 7'h*QF< ya. ? < 4.~&r, ' **iOSL..
34;t%
... 13
^!*
'~
ncy in
. d S:4 N;1
. pg.,,,..,:,,
>~
Q s; A if c
94
(
Nb I4 I"'
Consumable insert n
30' (max.)
- *I l +-
-+-
ho of eracking y
5
\\
v 0
i l
(a) Field Butt Weld Detail Feedwater transition piece-to-nozzle joint 30 deg.
$16 inid!amete.rj ' '"i^'
f:.~16 lEdiAmeten
- ySchedule 80.. :
i feedwater pipe f
.JSchedulo 802
"! ransition piece d.Jp c;;
- t(SA-106 grade B)'- 0.843 in.
j S
gde Spo.75dinf 0.843 in.
M lL.-
- .g m.c
.. o
,:m al-(W F - 1 JY
., ~...
i
, nyy o
u v
ot = 14 de9 f
E h = 0.084 in.
I h
i (b) Typical Butt Weld Detail Feedwater transition piece-to-pipe connection use os2s Figure 17. Construction drawings for feedwater nonle and piping welds at Diablo Canyon Units 1 and 2 (Cofie et al.1994). Copyright American Society of Mechanical Engineers; reprinted with permission.
DESIGN, MATERIALS, AND FABRICATION The thickness transition is significantly more between the nozzle and the transition piece which severe than for a typical piping system butt weld is smaller than the design angle a; this causes a and introduces a high stress concentration. The stress concentration greater than the design value.
field weld sometimes makes the discontinuity more This higher stress concentration factor contributed severe, resulting in a higher stress concentration.
to the initiation of a primary crack at this location For example, as shown in Figure 18, the notch (discussed in Section 6.3.1).
caused by the weld penetration makes an angle p Nozzle weld Primary Cracks i
Transition piece Secondary Cracks Nozzle E
a Detsil A Nozzle Transition Elbow Elbow weld piece weld Nozzle Detail A P,rgry Secondary y,ee377 cracks Figure 18. Geometric discontinuity introduced by the field welding between the feedwater nozzle and the transition piece (TVA 1992)(modified).
- 4. OPERATING TRANSIENTS AND ENVIRONMENT The feedwater system experiences five different been started and the condensate flow through the major transient conditions as the plant starts from idle feed pumps is returned to the condenser.
cold shutdown. These conditions, corresponding reactor coolant system (RCS) temperatures, and The residual heat removal system is secured once the modes of operation for a Westinghouse plant the reactor coolant system temperature reaches are as follows:
177'C (350 *F), and hot standby conditions (Mode 3 operation) are established. Turbine warmup Cold Shutdown RCS temperature begins when steam generator pressure reaches 7.53
< 93*C (200*F)- Mode 5 MPa (1092 psig), the automatic action of the steam dumps removes additional heat from the primary llot Shutdown RCS temperature 93 to system. As the steam dumps are opened to
+
177'C (200 to 350*F)- Mode 4 increase steam flow, auxiliary feedwater flow to the steam generators is also increased to maintain flot Standby RCS temperature 177 to the water level in the steam generators. The a
292*C (350 to 557'F)- Mode 3 reactor coolant system temperature is raised to 292*C (557'F).
- Startup, i.e.,
reactor critical RCS temperature 292*C (557'F), power The reactor is taken critical as startup (Mode 2
< 5% - Mode 2 operation) occurs. The first main feedwater pump is started when at least 2% power is reached. Once Power Operations, power > 5% -
the power is above 5%, the reactor is in Mode 1
=
Mode 1.
operation. When the power reaches 15%, the turbine is rolled and the steam dumps are closed.
The reactor coolant system is vented to remove The main feedwater regulating valves are placed gases during cold shutdown (Mode 5 operation) on automatic prior to increasing the load above and then a steam bubble is drawn 5 the 20%. The main feedwater pump speed control is pressurizer. The reactor coolant pumps are idle, placed in automatic when the main feed regulating the residual heat removal pumps are running, valves are controlling steam generator level.
decay heat is removed by the residual heat removal During loading from 20 to 100% power, additional system, and the steam generators are in wet layup, secondary pumps are placed on line, including all No secondary systems are in operation with an the main feedwater pumps.
exception of one circulating water pump. Thus there is no secondary flow or transients and, 4.1 Auxiliary Feedwater therefore, no fatigue damage to the feedwater nozzle and adjacent piping.
Operation The reactor coolant pumps are started during hot The auxiliary feedwater system is used to supply shutdown (Mode 4 operation) and the residual heat the steam generators during hot standby (Mode 3) removal pumps are turned off, since the reactor and startup (Mode 2). The relatively cold auxiliary coolant pumps will provide flow through the feedwater flow is low and unsteady. In some residual heat removal system. At approximately plants it can be operated in both manual and automat,c modes. Under low feedwater flow i
105'C (220*F), steam begins to form in the steam generators.
The condensate and circulation e nditions, the flow in horizontal sections of the IP Pe becomes stratified (flow stratificat, ion is systems are started, and the steam lines are warmed. Ilowever, the feedwater pumps have not discussed in Section 7).
OPERATING TRANSIENTS AND ENVIRONMENT Figure 19 shows such a low-flow condition during about 1514 t/ min (400 gpm) at the Sequoyah hot standby. The upper portion of the pipe fills plants, where a 757 f/ min (200 gpm) flow rate j
with hot steam or water from the steam generator leads to a stratifled flow condition with an while the bottom portion is filled with relatively interface at the center of the pipe cross section.
i cold auxiliary feedwater. The figure also shows the general location of fatigue cracks produced by Auxiliary feedwater transients vary from plant to thermal stratification. The height of the interface plant, and even from startup to startup, depending layer between colo and hot coolants in the pipe upon the time the plant is in Modes 3,4, and 5 i
depends on the auxiliary feedwater flow rate, the operations, the steam generator temperature, and temperature difference between the coolants, and whether the mode of operation for the auxiliary the piping layout. The height increases as the feedwater system is autorr.atic or manual. Typical cuxiliary feedwater flow rate increases, and it auxiliary feedwater evolutions including durations fluctuates as the flow rate is increased and and flow rates for the Diablo Canyon units are decreased to maintain the steam generator water listed in Table 2 (Peterson 1992). For the period level. While the steam dumps are cycling open covered in the table, the total time spent in Modes tnd closed, there is a corresponding on and off 2 and 3 was 339.5 days for Unit 1 and 238.6 days cycling of the auxiliary feedwater. When the for Unit 2.
cuxiliary feedwater flow rate is increased above a threshold flow rate, the cross section of the Most plants use a single-element auxiliary horizontal piping connected to the nozzle becomes feedwater flow controller (steam generator level).
completely filled with cold feedwater, and thermal For example, the automatic control for Diablo stratification is no longer present. The threshold Canyon auxiliary feedwater system was designed flow rate depends on the same parameters on for minimum flow rates on the order of 832 f/ min which the height of the interface layer depends.
(220 gpm) per steam generator, which is required i
Therefore, the threshold flow rates are different at during accident conditions (PG&E 1992). Since different plants. For example, the threshold flow the required auxiliary feedwater flow during hot rate at the Diablo Canyon plants is about 757 t/ min standby and startup is typically less than 832 t/ min, (200 gpm). The threshold flow rate is higher, on/off operation of the auxiliary feed water system occurs under automatic control.
Hot General location steam generator water of cracks g
(260*C(500*F) or more]
( Hotw e Feedwater Hot standby conditions Temperature gradients (stratification) in horizontal pipe Low flow, relatively cold water [38'C(100*F) or Cyclic temperatures less, may be intermittent) at hot / cold interface level Figure 19. Flow stratification in PWR feedwater nozzle during low flow conditions (USNRC 1980).
OPERATING TRANSIENTS AND ENVIRONMENT Table 2. Typical Diablo Canyon auxiliary feedwater evolutions.
Evolution Duration Auxiliary Feedwater Flow Reactor trip recovery 20 to 30 minutes Approximately 16911/m (400 gpm)
IIolding at Mode 3 from a forced Typically 2 to 4 days Steady outage Shutdown (cooldown) to RilR Boration (8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />)
Initially steady conditions Cooldown,286 to 149'C 0 to 851/m (20 gpm)
(547 to 300*F)(8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />) as RilR transfer nears Transfer to RIIR (8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />)
Stanup from a refueling outage lleatup to Mode 3 (20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br />)
O to 1691/m (40 gpm)
Mode 3 (72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />)
Mode 2 on AFW (48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />) llot functional testing 91 days for Unit i Similar to startup from refueling 40 days for Unit 2 outage Figure 20 shows a typical history of auxiliary al. 1994). The periods of manual auxiliary feedwater flow rates measured at Sequoyah feedwater flow exhibit relatively small fluctuations Unit I during Modes 2 and 3 operation (Cofie et al around a base flow rate of 277 to 302 t/ min (60 to 450 400 --
350 --
300 250 --
l 200 --
{
M M
l 150 -
I
\\
l
~
~
100 h
j j
0 100 200 300 400 500 800 700 600 Time (h)
Figure 20. Auxiliary feedwater flow for Sequoyah Unit I showing automatic and manual controls (Cofie et at.1994). Copyright American Society for Mechanical Engineers; reprinted with permission.
OPERATING TRANSIENTS AND ENVIRONMENT 80 gpm). However, automatic auxiliary feedwater to 1150 psia) and 210 to 238*C (410 to 460*F).
control resulted in large flow cycles from 0 to 757-However, auxiliary feedwater flow initiates from 1514 f/ min (200-400 gpm) at frequencies of about the condensate storage tank, in which the water i
2 cycles /h.
Though the automatic control may be at 38'C (100*F) or less. Feedwater j
accounted for only 13% of the total time, severe temperatures at Diablo Canyon are 21 to 30*C (70 fluctuations in the flow rates occurred during to 100*F) during Modes 2 and 3 operation when Cutomatic control that caused significant fatigue the steam generator temperature is greater than j
damage. Figure 21 shows typical periods of 177'C (350*F) (PG&E 1992). For plants in cutomatic and manual feedwater flows at colder regions, for example, Indian Point, the Sequoyah Unit 2 (Shvarts et al.1994). Obviously, auxiliary feedwater temperature could go as low as the time of automatic auxiliary feedwater use is 5'C (40*F) during winter months. Some plants important for fatigue crack growth.
have added preheaters for the auxiliary feedwater to lessen the temperature difference between the 4.2 Environment feedwater and steam generator.
A PH range of 8.5 to 9.5 measured at 25*C is The typical' values of feedwater pressure and temperature at the steam generator inlet during typical for feedwater systems. For example, the normal operation range from 5.5 to 7.9 MPa (800 pH of feedwater at Diablo Canyon is about 8.7,
^* *
^ " ' *
- t i
430 _. I -
- q an 33o 300 l
25o j
l J
p i
h zu
- I
(
\\
7 iso
.i 50 1,
I 1
h b i-U.'97 0 00 9/d/9212fD 9/S/92010 9/5/92 12.1%) 9/o/92 0.00 9/6/92 12 00 9/7/920C0 9/7/92 12 00 9/8/92000 Day, Time Figure 21. Auxiliary feedwater flow for Sequoyah Unit 2 showing automatic and manual controls (Shvarts et al.1994). Copyright American Society for Mechanical Engineers; reprinted with permission.
OPERATING TRANSIENTS AND ENVIRONMENT whereas that at Surry Unit 2 is maintained between that the flow-accelerated corrosion rates of carbon 8.8 and 9.2. Feedwater system materials determine steel are high when the water contains less than 20 the optimum pli level.
Feedwater systems ppb oxygen, but decrease rapidy with the addition containing copper require a pli level in the range of more oxygen. Higher oxygen levels lead to the of 8.8 to 9.2 to prevent excessive copper pickup.
fonnation of a stronger, less soluble iron oxide Systems with steel materials can tolerate a pli level (hematite), capable of reducing the flow-in the range of 9.3 to 9.6. If ammonia is used for accelerated corrosion rate by a factor of 3 to 10.
l pli control in the PWR feedwater system, the corresponding optimum pli level is maintained in Although low dissolved oxygen levels in the I
the range of 9.3 to 9.6. If morpholine is used feedwater accelerate flow-accelerated corrosion instead of ammonia, the pli level is maintained at damage, as discussed in the previous paragraph, 9.1 to 9.2 and, therefore, morpholine is used in the low oxygen levels inhibit corrosion-fatigue feedwater systems containing copper. Further provided that such levels are maintained at all discussion on the feedwater water chemistry may times. For example, at some plants the average be found in Section 7.2.
dissolved oxygen content is quite low in the main feedwater but high in the auxiliary feedwater, and, Flow-accelerated corrosion rates are inversely therefore, the oxygen levels are not low during all afTected by the amount of dissolved oxygen in the operating modes. Copper oxides are present in the feedwater. Nonnally, the oxygen level is kept low feedwater at plants with copper in the feedwater (typically 4 ppb) in PWR secondary systems to system. These oxides, in addition to dissolved minimize the degradation of the steam generator oxygen, also act as an oxidizing agent. Poor tubes. Such a low level of oxygen is harmful to control of feedwater chemistry (particularly the carbon steel r: ping because the oxygen fonus a oxygen content and possibly copper in the porous oxide film of magnetite on the inside feedwater system) was identified as a root cause surface of the piping and the piping becomes for cracking in the nozzle bore, knuckle region, susceptible to flow-accelerated corrosion. Tests and girth welds at two plants (Westinghouse 1990, with neutral water at 100*C (212*F) have shown Bamford et al.1992).
- 5. THE SAFETY SIGNIFICANCE OF FEEDWATER LINE RUPTURE Main feedwater line rupture is a design basis veloped to detect and size wall thinning (see accident; its consequences include a potential for Section 8). However, flow-accelerated corrosion core damage. The rupture reduces the ability to can lead to rupture if the associated wall thinning remove heat generated by the core from the reactor is not detected and repaired or replaced in a timely coolant system. In addition, the resulting loss of manner.
feedwater would activate and challenge safety-related systems to cool the reactor core. A A major feedwater line rupture can prevent the consequent transient induced steam generator tube addition of sufficient feedwater to the steam rupture could result in the release of significant generators to maintain secondary coolant inventory cmounts of radioactive material into the in the steam generators. If the break occurs environment, bypassing the containment. Failure upstream of the first check valve from the steam of high-energy piping, such as the main feedwater generator (see Figure 1), it would be similar to an j
piping, can also result in complex challenges to the accident caused by a loss of normal feedwater. If plant operating staff because of potential the reactor is not tripped during this accident and interactions of the high-energy steam and water the auxiliary feedwater is not available, more l
with other systems, such as the electrical drastic measures by the operator would be required I
distribution, fire protection, or security systems.
to prevent core damage.
Catastrophic failure of any high-energy piping can present a safety problem for plant personnel if the break occurs between the steam generator (USNRC 1989).
and the first check valve on the feedwater piping, secondary coolant from the steam generator can be Thermal fatigue and flow-accelerated corrosion discharged through the break. Also, a break at this have damaged PWR feedwater piping during location would preclude the subsequent addition of j
operation. Thermal fatigue has caused through-auxiliary feedwater to the faulted steam generator.
wall cracking, but no ruptures of cracked pipes The consequences of this accident depend on the have been reported. However, a cracked feedwater size and location of the break and the plant pipe could rupture if subjected to seismic or water operating conditions at the time. The break could hammer loads. (Water hammer is expected to be either cause a cooldown or a heatup of the reactor less likely to occur because, as discussed in Section coolant system. He cooldown would result from 9, all the U.S. PWR plants have revised and an excessive energy discharge through the break, modified the feedwater system design and similar to that which takes place during a operation to limit occurrence of steam generator postulated main steam line break (Diablo Canyon water hammer events.) The USNRC has not FSAR).
accepted the leak-before-break concept for PWR feedwater lines.
A postulated break between the steam generator and the check valve could reduce the ability to Flow-accelerated corrosion has caused significant remove heat from the reactor coolant system and wall-thinning of main feedwater piping, and in a result in the heatup of the reactor coolant system, few cases has resulted in rupture. Wall-thinning for three reasons:
has often been found at unexpected sites. Factors affecting flow accelerated corrosion have been (a) The main feedwater line rupture would better understood in the last decade. Computer reduce the feedwater flow to the steam codes have been developed to identify sites generators. Because feedwater is subcooled, susceptible to significant wall thinning and to its loss could raise the reactor coolant estimate the rate of wall thinning (see Section 7).
temperature and pressure prior to reactor trip.
Enhanced inspection methods are also being de-1 29 NUREG/CR-6456 i
SAFETY SIGNIFICANCE (b) The coolant in the faulted steam generator steam generator tubes will increase, and multiple may be discharged through the break, and tube ruptures might occur, especially if the steam this would not be available for decay heat generator tubes have a large number ofincipient or removal after reactor trip, undetected cracks (USNRC 1988c). Subsequent pressurization of the primary system would (c) The break may be large enough to prevent increase the leakage rate. If the first check valve is the addition of any main feedwater after outside the containment and the break is between reactor trip.
the containment wall and the check valve, tube leakage would cause release of radioactive l
However, the supply of auxiliary foodwater to the products through the break into the environment.
unaffected steam generators must be adequate to In addition, during the accident, the water from the remove decay heat and prevent over pressurizing refueling water storage tank is injected into the the reactor coolant system. Otherwise, more reactor coolant system through the safety injection l
drastic measures by the operator would be required system. His water leaks to the faulted steam t
to prevent core damage.
generator and is then released through the break to the environment. He safety concem is that, in an i
A faulted steam generator will rapidly depressurize unlikely event, the refueling water storage tank during a main feedwater line break between the could be drained before the shutdown cooling steam generator and the first check valve. As a mode is achieved, eventually causing core damage, result, the differential pressure across the faulted i
t t
L l
t S
i NUREG/CR 6456 30
- 6. FIELD EXPERIENCE RELATED TO CRACKING AND WALL THINNING OF FEEDWATER NOZZLES, PIPING, AND FEEDRINGS Feedwater nozzle cracks were first discovered at nozzle area, the presence of thermal stratification the D. C. Cook Unit 2 plant in 1979. (An earlier has been verified frorn temperature measurements crack in the pipe-to-nozzle weld at Diablo Canyon on the piping wall far upstream of the nozzle but Unit I was thought to have been the result of inside the containment, and also in the feedring inadequate welding procedures.) As a result, the piping downstream cf the nozzle (Bain, Collins, USNRC issued Bulletin 79-13. Further inspec-and Testa 1992). The high loads resulting from tions revealed instances ofcracking in a number of thermal stratification have bowed pipes and other plants. mainly of Westinghouse design, but damaged both feedwater piping and feedring a few were of Combustion Engineering design.
supports.
The cause of the cracking was determined to be thermal fatigue, possibly corrosion assisted, and Fatigue is not the only degradation mechanism that the root cause was generally attributed to thermal has caused damage to the PWR feedwater systems.
stratification of the coolant in the nozzle. The Flow-accelerated corrosion has caused wall inspections identified in the Bulletin were com-thinning in feedwater lines, feedrings and J-tubes.
pleted and the USNRC licensees performed appro-Also, the feedwater nozzle thermal sleeves have priate repairs and established other corrective experienced thinning at the leading edge, possibly rctions during 1979 through about 1983. Subse-because of flow-accelerated corrosion.
quently, at least eight plants experienced feedwater nozzle cracking in the 1983 through 1991 time Water hammer events can fracture piping at areas frame. Based on the responses from the USNRC degraded by fatigue or flow-accelerated corrosion, licensees implementing the actions recommended and have caused through-wall cracks and ruptures i
in Bulletin 79-13, which included augmented of piping, nozzles, and feedrings. Water hammer inspections at some plants,Bulletin 79-13 was events have also damaged piping supports.
closed in 1991 (Foley, Dean, and llennick 1991).
The objective of this section is to summarize the The basis of the Bulletin closure was that the field experience with degradation in the feedwater feedwater nozzle cracking problem was being nozzle area. Three types of damage are discussed:
adequately managed. However, afler the Bulletin (a).hermal fatigue cracking with emphasis on the was closed in 1991, feedwater nozzle cracking was more recent discoveries after the closcout of j
detected in at least 12 plants during 1992 and Bulletin 79-13,(b) wall thinning caused by flow-1993, including one through-wall crack. Thus, the accelerated corrosion, and (c) steam generator number of feedwater nozzle cracking events water hammer damage. Section 6.1 summarizes increased during 1992 and 1993 compared to the the fatigue cracking field experience during the 1983-1991 time period; there were about six period from the initial crack discoveries in 1979 to cracking events per year during 1992 and 1993 the completion of the inspections required by the compared to one event per year during the earlier Bulletin.
Section 6.2 describes the fatigue time period. However, we have not identified cracking events after the initial inspections but additional feedwater nozzle cracking events since before the Bulletin closcout, and Section 6.3 1993. Sections 6.2 and 6.3 present further details presents the field experience after the closcout of about the field experience, including the size and the Bulletin.
Degradation caused by flow-location of the cracks.
accelerated corrosion is discussed in Section 6.4, and examples of steam gnerator water hammer Although fatigue cracking from thermal damage are included in Section 6.5.
stratification has been found only in the feedwater 31 NUREG/CR-6456
i l
FIELD EXPERIENCE 6.1 initial Discoveries of the feedwater piping and piping supports inside the containment, review the adequacy of procedures a
e SC %
for responding to a feedwater line break, and repoit the sensitivity of the methods for detecting feed-nis section briefly describes the fatigue crack.mg water leaks within the containment. If cracking in the US plants from 1979 to the early 1980s, was detected, the licensee was to examine all the 1
These cracks were discovered because ofleakage feedwater piping welds up to the first piping or during subsequent mspections requested in support and at all high stress points of the piping Bulletin 7913.
within the containment to determine the extent of the cracking. Revision 1 of the Bulletin was issued i
De first discovery of feedwater nonle cracks was to designated applicants for an operating license in Diablo Canyon Unit 1, a 4-loop PWR of West-and asked that they conduct volumetric and visual inghouse design, during hot functional testmg in examinations. Revision 2 of the Bulletin reduced 1977(USNRC 1979a). Subsequent investigations the scope of the examinations based on prior with radiographic testing (RT) and ultrasome usults.
testing (UT) equipment revealed a crack approxi-mately 152-mm (6 in.) long, originating in the B&W plants were exempted from the actions
- weld heat-affected zone of the leaking nonle weld.
requested in the Bulletin because the B&W design ne cause was ascribed to either corrosion-fatigue has a separate nonle for auxiliary feedwater into or thermal fatigue initiating at small cracks proba-the steam generator, the feedwater is always pre-bly induced by welding or post-weld heat treat-heated above 180*C (360*F) for all modes of ment. The system was repaired by replacement of operation, and the feedwater velocity [ greater than the piping weld with one of identical design, 3 m/s (10 fps)] is sufficient to keep steam from commonly referred to as an in-kind replacement, backing up into the feedwater line from the steam and employmg greater controls on the welding, generator. He 3 m/s (10-fps) velocity represents including maintaming the preheat temperature until a flow rate of greater than 15,000 f/ min (4,000 the post-weld heat treatment.
gpm). Some Combustion Engineering plants, for example, Ft. Calhoun and Calvert Cliffs Units 1 About 2 years later, coolant leakage was reported and 2, as well as later vintage Westinghouse plants, at D.C. Cook Unit 2, another 4 loop PWR of also have separate auxiliary feedwater nonles.
Westinghouse design, which had operated for about 1 year (USNRC 1979a). Through-wall As a result of the examinations conducted in circumferential cracks were discovered m the response to Bulletin 79-13, cracks were discovered 405-mm (16-m.) diameter feedwater imes in the in the feedwater nonle areas of I 8 PWRs, sixteen vicmity of two of the four steam generator feed-designed by Westinghouse (including Salem Unit water nonles. The through wall cracks were in 1), and two by Combustion Engineering. Only the the upper half of the cross section. Subsequent cracks initially found in the two D. C. Cook Unit 2 radiographic examinations revealed crads or feedwater nonles were through-wall. Although crack like mdications in all eight steam generator there were no cracks found in the Yankee Rowe feedwater, let lines of both D. C. Cook uruts. He m
nonles, the welds were deemed unacceptable by most severe crack was a through-wall circumfer-present-day codes, and all unacceptable welds were ential crack, 90-mm (3.5-in.) long at the outer either repaired or replaced. No crackbg was found in 17 other plants (including Haddam Neck).
Thirteen plants were designated applicants for He USNRC issued Bulletin 79-13 because of the peradngHeensedncWgdaManyon Unhs leakage at D.C. Cook Unit 2 (USNRC 1979a).
"8
'.and Sequoyan ne Bulletin requested that the USNRC licensees of operating PWRs perform volumetric examina-yn s 1 an
), an n exammations wem identified for five B&W plants. The bulletin was tions (radiographic and ultrasonic) of the feedwater ci sed for the 13 plants that were des,gnated i
piping weld areas, perform visual inspections of NUREG/CR-6456 32 l
1
FIELD EXPERIENCE applicants for operating licenses on the basis of Typical cracks were transgranular, oriented in the having completed or committed to complete the circumferential direction and initiated at the inside inspections required by the bulletin (Foley, Dean, surface, as shown in Figure 22. In this figure, the end Ifennick 1991). Therefore, the required primary crack (deepest crack) is a circumferential inspections must have been completed at i1 of crack that has grown in the radial direction. There these plants, which have been in commercial are several shallow secondary cracks adjacent to operation for more than 10 years. The required the primary crack. Whereas the maximum crack inspection must have been partially completed at depth in some units was less than 2.5 mm (0.100 Watts Bar 1, which went n. mmmercial operation in), several cracks had propagated 6 to 19 mm in May 1996, whereas Watts Bar 2, the remaining (0.250 to 0.750 in.) through the wall. Most of the designated applicant for an operating license, is cracking occurred in the counterbore region of the still at the construction stage.
base metal (see Section 3.3), primarily at the re-Weld 0.84
.)
Flow -+ -
Pipe end
[406 mm (16 in.)
[406 mm (16 in.)
schedule 60]
Counterbore schedule 80]
Backing ring (a) Crack profile in degraded f eedwater piping.
Deepest penetration 12.7 mm (0.50 in.)
s J
M96 0336 (b) Crack location in the D.C. Cook nozzle (USNRC 1980).
Figure 22. Typical fatigue crack orientation at feedwater nozzle counterbore.
FIELD EXPERIENCE entrant corner of the counterbore at the upstream closely bisected the angle formed by the end, not in the weld or heat-affected-zone, intersection of the two surfaces of the flowever, the major cracks in a few units initiated discontinuity.
away from the counterbore. For example,in San He piping from some plants contained Onofre Unit 1 (a 3-loop Westinghouse plant), the major crack [2.54-mm (0.1-in.) deep] initiated in many shallow cracks or corrosion spikes the weld whereas in Palisades (a Combustion less than 0.25-mm (0.010-in.) deep.
Engineering plant), the major crack [4.3-mm Samples from other plants contained a (0.17-in.) deep) initiated near the weld in the heat-number ofintermediate size cracks 0.25-j affected zone (Consumers Power 1979). At to 1.0-mm (0.010- to 0.039-in.) deep, and j
Kewaunee (a 2-loop Westinghouse plant) a two plants (Palisades and San Onofre 1) crack initiated in the piping at the weld interface.
contained a number of cracks greater than Examinations of the weld areas found tool marks 1-mm (0.039-in.) deep. The remaining and machining grooves in the counterbore region.
plants had either no cracks or very few The through-wall cracks occurred at the top of the cracks in addition to the major crack.
pipe, though many other cracks were found at the The cracking was predominantly trans-3 and 9 o' clock positions. The fracture surfaces of the through-wall cracks showed striations, typical granular. Some minor branching was of fatigue crack growth, and the presence of evident in a few plants.
corrosion products. Table 3 summarizes the extent Corrosion products, mainly Fe30 and of damage at the 18 PWR plants where cracking was discovered during the inspections performed Fe2O, were present on all crack surfaces 3
in response to the Bulletin.
except on those from the deep D. C. Cook crack. Corrosion pitting, especially at the l
The cracking was attributed primarily to thermal root of machining grooves, probably ac-fatigue caused by thermal stratification and striping celerated 'he initiation ofcracking but was in the nozzle region and concentrated at the stress a minor overall contributor.
raiser at the counterbore. Corrosion fatigue apparently played a secondary role in the cracking.
Test instrumentation consisting of strain gages, Some of the cracks initiated at pits on the inside accelerometers, thermocouples, pressure transduc-surface, and corrosion products have been found ers, and displacement transducers were temporarily on crack surfaces. Although oxygen is thought to installed on the feedwater nozzles and adjacent be a factor in accelerating the fatigue, it is not piping of several plants where cracking had already considered to be the primary factor.
occurred (including two nozzles of D. C. Cook Unit 2), to monitor for the stressors causing fa-The following conclusions are drawn from a tigue. Westinghouse Electric Corporation also j
metallographic study of eracked feedwater nozzle carried out laboratory tests (Hu, Houtman, and j
samples from nine pWRs (Goldberg, Streit, and White 1981), finite element analysis (Thurman, Scott 1980):
Mahlab, and Boylstein 1981), and crack growth analyses (Bamford, nurman, and Mahlab 1981) to i
The major cracks initiated along the verify the cause and consequences of the cracking.
discontinuous section change at the All the resulting information tended to confirm that bottom of the counterbore slope and/or the root cause of the cracking was fatigue induced other geometric discontinuities, and along mainly by thermal stratification, enhanced by the weld surfaces, weld counterbore and possibly environmental effects. Augmentedinspectionswereimplemented The cracks extended circumferentially at several plants as part of the refueling outage along the inside diameter and propagated inservice inspection (ISI) program.
relatively straight in a direction that NUREG/CR-6456 34
FIELD EXPERIENCE Table 3. PWR feedwater piping cracking during 1979-1980 (USNRC 1980). (1 in. = 25.4 mm).
Maximum Vender / steam depth' Circumferential Lines Plant generator model (ie.)
Lcestion cracked
- Comments Beaver Valley I E
0.400 9 o' clock 3 of 3 N'
D.C. Cook I,2 E
Trough-Top 8 of 8 Through-wall wall cracks in two of the four Unit 2 lines Ginna E
0.107 800 o' clock 2 of 2 N
Kewaunee E
0.050 7 o' clock 2 of 2 3-in. auxiliary feed near SG inlet Millstone 2 CE 0.22 12 o' clock 2 of 2 N
Palisades CE 0.170 3 & 9 o' clock 2 of 2 Cracks also found at weld in vicinity of i
horizontal pipe Point Beach I,2 E
0.047 3 o' clock 2 of 4 3-in. auxiliary feed near SG inlet II. B. Robinson 2 E
0.750 9 o' clock 3 of 3 Shallow cracking of nozzle under thermal sleeve Salem i E
0.235 N
4 of 4 N
f San Onofre 1 E
0.100 Lower half of 3 of 3 Multiple-reducer branched cracks, fatigue Surry 1,2 E
0.080 2 & 5 o' clock 6 of 6 N
Turkey Point 3,4 E
N N
6 of 6 N
Zion 1,2 E
0.088 N
N N
- a. The typical thickness of a feedwater line pipe wall is approximately 13 to 25 mm (0.5 to 1 in.).
- b. Number of total feedwater lines into steam generators that were found to be cracked. For example, the D.C. Cook plants are 4-loop Westinghouse units, so all eight lines in the two plants were cracked.
- c. Additional informatior, was not found.
FIELD EXPERIENCE Many PWR units experienced lengthy outages in 6.2.1 Maine Yankee 1979 and 1980 while these inspections and appropriate modifications and repairs were The Maine Yankee cracking resulted in a leak performed. The repairs generally consisted ofin-when a water hammer caused fracture of an already kind replacements of the cracked sections. Some existing fatigue crack (Garrity 1983). J-tubes j
of the cracks were sufficiently shallow to simply be had not been installed on the steam generators at ground out.
On several units, the sharp this time. The nonle is connected to a safe end, i
discontinuity associated with the counterbore was which connects to a 20-degree bend. The bend is redesigned with a fillet having a 12.7-mm connected to a 0.7-m (2-fl) length of straight pipe, (0.5-in.) radius, and a 125 root mean square (RMS) which connects to a 90-degree elbow that turns surface finish was specified for the transition downward (Stoller 1983). The crack was located region. Units such as D. C. Cook installed at the bottom of the pipe near the first weld protective liners, made of carbon steel and upstream of the nozzle.
The crack was approximately 13 mm (0.5-in.) thick, inside the propagating from the inside surface of the pipe at nonles. Some utilities, such as Callaway and the reentrant corner of the counterbore. The crack Wolf Creek, have special startup feedwater was 0.9-m (35-in.) long on the inside surface and systems (described in Section 2.1) that supply the 0.28-m (11-in.) on the outside surface.
heated main feedwater.
More complete Examinations revealed that there had probably descriptions of the types of modifications been a tool mark located at the counterbore that available to mitigate damage are presented in aggravated the stress concentration. Fatigue cracks 4
Section 9.1.
were found at the same location on the other two feedwater lines, but these cracks did not become 6.2 Crack Discoveries Between through-wall because water hammers did not occur i" 'h S*II"*S ^II"'k' ****"d*d I' * 'h* II '
Initial Inspections and 7 oclock position (240 degrees around the 1
Bulletin Closeout circumference). Sections of pipe were replaced, counterbores were removed in the redesigned Feedwater nozzle cracking was found in eight pieces, small cracks were ground out, and the PWRs during the time period of 1983 to 1991, repaired welds were inspected by RT and UT to several years afler the initial inspections performed establish a baseline.
in response to Bulletin '79-13. These included i
cracks in the nozzle areas of two plants (Beaver 6.2.2 indian Point Unit 2 Valley 1 and Turkey Point 4) that had experienced cracking earlier and for which repair or Cracking Incidents in 1989. Extensive replacement had been performed. Six of these thermal fatigue cracking was found in the Indian plants were of Westinghouse design; two were of Point Unit 2, a Westinghouse 4-loop plant, feed-Combustion Engineering design. The events are water nozzles during the 1989 inservice inspection.
summarized in Table 4, and the repair and The cracking was found in the feedwater nozzle replacement activities that followed in response to blend radius or inner radius section (also called these cracking events are summarized in Table 5.
knuckle region), steam generator shell inside Most of these incidents ofcracking were similar to surface beneath the nozzle, the feedwater nozzle those discovered in the 1979 round ofinspections, bore region under the thermal sleeve, and in the but there were two notable exceptions, leakage at pipe-to-nozzle weld region as shown in Figure 23 Maine Yankee and extensive cracking at Indian (Westinghouse 1989). The cracking was caused Point 2, which are described here.
by cold feedwater leaking past the thermal sleeve, abetted by environmental effects resulting from poor water chemistry control (high dissolved oxygen). The leak path was through the small NUREG/CR-6456 36
i FIELD EXPERIENCE TsWe 4'. PWR feedwater nozzle cracking 1983 to present'(1 in. = 25.4 mm).-
- Ptset Deee VanderWese gem.
Mesimum Clreenforemelal Umes
. Comesmes erecer model depik locatieelenemme'.
cracked i
1953 to 1991 SL 1meis l' 1983 CE N*
N 2/2 safe and base metal -
Meine Yambee' 1983 CE through-wall leak at bouem of 3/3 exiarbs crack prepasssed by.
pipe; creaking Il to.
waler hammer,crecidag only in 7 o' clock pipe aide of weld Beaver Valley 2' 1983-E/5tM
'N N
N N
j 1985-j i
Twbey Point 4*
1964 Elder N
270* (A),180* (C) 2/3 base metalin pipHo.noesie around weld region.
[')
Farley l' 1984 g/51 7 37 %
11 1,7-9,4 5 (B 3/3 base metalin pig through well
. caly)o' clock weld region Trojan
- 1987 E!51A 0.533 in.
4,7,9,12 o' clock 3/4 base metal in pipe to nossie l
weld region e
i
'Besver Valley l' 1987-
' E/51 N
N 3/3 pipe-to-nozzle weld region.
Indien Point 2*
1989 -
. E/44F 0.347 in.
bottom 120' -
2/4 pipe-to-nozzle weld region;up.
l (bore); 0.388 stream piping; nozals inner bare l
in. (weld) l l
h Seever Valley I' 1992 M51 N
N N
N l
i Sequaysh I'
.1992 E/51 through-wall 3 and 9 o' clock 3/4 leak in one line; transition piece;
}
1993
- AFWauto/ manual seguoyah 2*
1992 E/51 60% through-most at 3 and 9 2/4 transition piece; AFW 1993 wall o' clock (slightly ro-auto / manual tated), one at top l
2 Sales l
1992 E/51 N
N 4/4 nozrfe-to expander and expander-to elbow weld areas Diablo Canyon I" 1992 E/S t 0.06 360* around 4/4 base metal in pipe-to-nozzle j
circumference weld region e
Prairie Island I & 2' 1992 ESI N
N N
N fladdma Neck" 1993 E/27 0.25 lower 180*
3/4 base metal in pipe to nozzle weld region e
i Italunson 2' 1993 E144F N
N N
N l
?
Turkey Point 3' 1993 E'44F N
N N
N 4
5en Ono6e 3" 1993 CE-0 02 3 and 9 o' clock 2/2 AFW on/off pnor to 1986; safe end Parley 2'-
1993 E58 N
N N-N i
s The superscripts in the first column refer to the references listed at the end of Table 5.
f
- b. The upper caec letter in the parenthesis identifies the affected steam generators. Such identification of affected stearn generators was not provided for in all the cracking events reported here.
i
- c. Infonnation was not found.
FIELD EXPERIENCE Table 5. Repair / replacement activities 1983 to present.
Plant Replacement Griad/ -
Thermal Heer/
Joint design
' Thermal weld seeHag sleeve mediScotica sleeve repairs lastellation replacement Beaver Valley 2' 1985 St.Lucie18 1983 Maine Yankee' 1983 1983 stress raiser removed abe Turkey Point 4' 1984 removed Farley l' 1984 Bloop l
w a ~s Partial i
Trojan' 1988 redesigned; A 1988 loop only,
. a.'ss moved away from weld Beaver Valley I' 1988 1992 1989 1991 Indien Point 2' two loops 1994 stress raiser Sequoyah 2' 1992 j
removed Salem l
1992 1992 planned Diablo Canyon I" 1992 1992 1994 I
Prairie Island I and 2' 1992 1992 Sequoyahl' 1992 1995 stress raiser removed lladdarn Neck" 1993 1993 l
Robinson 2' 1993 Turkey Point 3' 1993 San Onofre 3" 1993 1.
Westinghouse proprietary Class 2C handout at Westinghouse /NRC meeting 3/21/95 2.
USNkC 1983b 3.
Garrity 1983 4.
Fkwida Power A Light 1984 5.
USNRC 1984b 6.
Cockneld 1988 7.
USNRC 1987b 8.
Westinghouse 1989,1990 9.
TVA 1992a and b, Coley 1992. Cone et al.1994
- 10. PSEAO 1992. Stoller 1992b j
l 1. PGAE 1992, Cone et al.1994, Theiler et al.1995
- 12. McBreerty 1993
- 13. Mostafa and Ramsey 1994 NUREG/CR-6456 38 i
4 i
P FIELD EXPERIENCE 4
gap, typically 0.25 to 0.50-mm (0.010- to storage tank to prevent oxygen ingress. Contami-0.020-in.) wide, between the thermal sleeve and nants from condenser inleakage may have aggra-the nozzle, Whereas the dissolved oxygen in the vated the problem.
feedwater is maintained at a low level during normal operation (it averaged 2 ppb in 1989), cold, Visual examination detected linear indications on oxygenated water was drawn from the condensate the inner radius section of one of the four 4
storage tank when the plant was in hot standby and feedwater nozzles. 'Ihese indications were located shutdown conditions and injected via the auxiliary on the lower 120-degree segment of the nozzle.
feedwater system. A 203-mm (8-in.) diameter vent Liquid penetrant examination confirmed these to the atmosphere to release insoluble gas buildup indications and also found additional indications tilowed oxygen to enter the condensate storage on the two support brackets welded to the nozzle tank. It is believed that this introduction ofoxygen just below the knuckle. Visual and penetrant was a major contributor to the cracking. Some examination on the other three nozzles revealed L
plants use a nitrogen blanket in the condensate linear indications on the inner radius section of one
}
I I
Thermal sleeve pp, d
Weld Bore cracks Nozzle blend a
radius cracks l
Shell face cracks j
l Figure 23. Locations of thermal fatigue cracks in an Indian Point Unit 2 feedwater nozzle and steam generator shell caused by bypass leakage of cold feedwater (Westinghouse 1989).
l l
FIELD EXPERIENCE nozzle. No indications were found on the other welds) in all four steam generators. All indications two nozzles.
The maximum depth of the were removed by grinding. De maximum grind-indication was less than 5 mm (0.2 in.). All out depth for the indications was 11 mm (0.4 in.).
indications were removed by grinding.
The thermal sleeves were removed from all four Fiberscope examinations of the inner and outer steam generators so that visual, liquid penetrant, surfaces of the thermal sleeves and the inner and ultrasonic examinations of all four nozzle surface of the nozzle bore over the lower 90 inside surfaces (bores); visual, liquid penetrant, j
degrees did not reveal any linear indications.
and radiographic examinations of all four nozzle to Ultrasonic examinations of the lower 180 degrees feedwater piping welds; and visual examination of of all four feedwater nozzle bores and 100%
the thermal sleeves could be performed. UT i
radiographic examinations of all four feedwater revealed indications on the nozzle bore ranging in nozzle-to-pipe welds did not reveal any linear depth from 5 to 9 mm (0.198 to 0.347 in.), and indications.
indications at the nozzle-to-pipe welds ranging in depth from 5 to 10 mm (0.200 to 0.388 in.).
Westinghouse (1989) reported two other examples Cracking was also found at the 6 o' clock position of feedwater nozzle cracking. In one case, linear in the horizontal section of the feedwater piping indications were found in the nozzle bore near the upstream of the nozzle on all four steam knuckle region. Cracking was attributed to excess generators. This feedwater pipe cracking is leakage through the gap between the thermal believed to be the result of thermal stratification, sleeve and the nozzle bore. Excess leakage which occurs during the cold, low flow auxiliary occurred because of two reasons: (1) cracking of feedwater injection. These horizontal portions of the thermal sleeve caused by a water hammer, and the feedwater piping were replaced. De auxiliary (2) thinning of the leading edge of the thermal feedwater tee located outside containment, several sleeve, possibly caused by flow-accelerated feet away from the feedwater nozzle, was also corrosion. These indications were removed by inspected with a fiberscope and no indications grinding. In the second case, the cracking occurred were detected, in the knuckle region of an auxiliary feedwater nozzle of a PWR steam generator. The cracking Metallographic examinations conducted on a was a result of leakage past the slip fit joint nozzle bore sample taken from one steam generator between the nozzle and thermal sleeve. Similar showed multiple axial cracks originating from leakage has caused cracking in the knuckle region oxide-covered pits that had linked together. The of several boiling water reactor (BWR) feedwater maximum depth of the crack was approximately 2 nozzles.
mm (0.07 in.). Copper deposits were also found in the area. He presence of the oxide-covered pits, Cracking incidents in 1990. Indian Point multiple cracks which initiated from these pits, Unit 2 was shutdown in February 1990 to perform and the copper contaminants led to the conclusion a mid-cycle inspection of all four steam generator that the damage mechanism was corrosion fatigue.
feedwater nozzles and other locations that have Fractographic examinations of the fracture faces been determined to be susceptible to cracking revealed the presence of a multitude of fine based on the 1989 inspections. Visual and striations, which confirmed that the crack growth magnetic particle inspections of the shell face in was caused by corrosion fatigue. De beach marks the nozzle region, the nozzle inner radius (blend on the fracture surface suggested different periods radius), and the support bracket welds beneath the oferack extension, and that the cracking may have nozzle were performed. Linear indications were occurred over a long time such as during one fuel found on the nozzle inner radius sections of two cycle.
steam generators and in the heat-affected zone of the weld between the feedwater nozzle and the Although the primary cause of the cracking at J
steam generator shell (near the support bracket Indian Point 2 was leakage under the thermal NUREG/CR-6456 40
5 FIELD EXPERIENCE
'l sleeve and turbulent mixing of the cold feedwater location where one of th'c original support brackets and the hot steam generator water, poor oxygen was welded.
control for ~ long periods probably made the situation worse than in other plants. Whereas A feedwater noule sealing sleeve was installed in -
j cracks at the pipe-to noule weld area can' be 1991 to eliminate the leakage ofcold water into the.
i propagated by cyclic thermal stratification, cracks gap between the thermal sleeve and nouie and in the noule bore under the thermal sleeve are not thus mitigate the cracking in the nozzle bore and
.txpected to propagate because the stresses from knuckle region. ' A schematic of the seal ring is turbulent mixing rapidly attenuate through the shown in Figure 24 (Consolidated Edison 1991).
thickness. Since the noule is thicker than the -
The sealing sleeve is welded to the feedwater weld region, through-wall penetration is unlikely.
nozzle at the feedwater-pipe end, sealing the j
All cracks and indications were removed by leakage path of the incoming feedwater. A groove grinding and were repaired by welding (using the cut at the other end of the sealing sleeve contains temper bead technique) to the design coafiguration a spring loaded sealing ring. The sealing ring while the thermal sleeves were removed. Future compresses against a pad ring welded to the augmented inspections were planned. Longer term existing thermal sleeve but is free to slide to j
mitigation actions include replacing the copper accommodate differential thermal expansion i
condenser tubes with titanium tubing, conducting between the steam generator noule and the corrosion product transport studies, eliminating the feedring. Blocks are welded to the thermal sleeve bypass flow in the feedwater nozzle, and to prevent the sealing sleeve from entering the I
introducing nitrogen gas cover in the condensate steam generator if the weld should fail.
storage tank to prevent any ingress of oxygen into the auxiliary feedwater, and deaerating the 6.3 Fatigue Cracking After auxiliary feedwater.
Bulletin CLOS 90ut i
Cracking incidents in 1991.
A UT g events m. PWR feedwater nozzles have Crackm.
examination of a portion of an Indian Point Unit 2 e nt, ued to occur since the 1991 closcout of m
feedwater noule bore and knuckle region, and the Bulletin 79-13 and are summarized m Table 4.
adjacent feedwater piping was perfonned from the Feedwater nozzle cracking was detected m at least l
outside surface during the 1991 refueling outage.
12 plants after the Bulletm, closure. Eleven of De examination detected several reflectors in the these were Westinghouse design; one was of t
bore region but did not reveal any indications in Combustion Engineering design. All of the the knuckle region or piping.
Subsequent Westinghouse steam generators with feedwater l
magnetic particle testing of the nozzle bore region n zzle cracking were Model 51 or earl,er models.
i after removal of the sleeve did not reveal any Later m dels (D2 through D5 and E) have cracking. Visual examination ofthe inside surface feedwater preheaters and have separate auxiliary of the piping during installation of the thermal feedwater nozzles. A few mstances of noule sleeve also confirmed the absence ofany cracking.
crackmg have occurred m, non-U.S. PWR plants as Also, radiography of a nozzle-to pipe weld on one steam generator did not detect any cracking. And,
~
inspection of the feedring support bracket welds H.is section describes crack.mg m the feedwater with magnetic particle and liquid penetrant
" "I* areas of several plants since 1991, techniques did not reveal any cracking.
begmn.mg with Sequoyah Units 1 and 2, where a leak occurred in one nozzle in 1992. In response Liquid penetrant testing of the knuckle region in to crack, g and leakage at Sequoyah, m, spections m
all four steam generators detected one 13-mm c nducted at Salem Units 1 and 2 shortlythereafter (0.5-in.) long crack located near the 7 o' clock revealed crackmg, Unit I nozzles. Later that m
position. De crack was removed by grinding to a year, cracking was discovered at Diablo Canyon depth of 1.25 mm (0.05 in.). This crack was at a 41 NUREG/CR-6456
FIELD EXPERIENCE i
Weld l
1 4
Spring loaded seeling ring Pad ring Sealing Blocks sleeve Therrnal sleeve 1
Figure 24. Indian Point Unit 2 feedwater nozzle sealing sleeve installed to prevent bypass leakage of cold feedwater (Consolidated Edison 1991).
'T Units I and 2. Based on these cracking incidents in the transition pieces connecting the feedwater at Sequoyah and Diablo Canyon, the NRC issued nonles to the feedwater piping. Figure 16 shows j
infbrmation Notice 93-20 (USNRC 1993) for the construction drawing of the transition pieces, 3
Westinghouse and Combustion Engineering plants, which are 38-mm (1.5-in.) wide, joining the steam Cracking was also discovered at the Haddam Neck generator nozzles and the feedwater piping. Dese and San Onofre Unit 3 plants. This section transition pieces are made of the same material f describes the cracking experience at these plants (SA-508 Class 2) as the nozzles (Wilson 1992, and at other US and non-US plants. The repair and TVA 1973). The thinner end of each transition replacement activities that followed in response to piece was field welded to each nozzle. He inside these cracking events are summarized in Table 5.
surface was counterbored in this region, resulting in a reentrant corner on either side of the weld. A 6.3.1 Sequoyah Units 1 and 2 weld preparation area of base metal beneath the weld on the inside surface is shown in Figure 16.
J Sequoyah Units 1 and 2 (4-loop Westinghouse However, in Unit 1, Loop 3, the nozzle-to-PWRs) cxperienced fatigue cracking in early 1992 transition piece weld penetrated through the base I
t NUREG/CR-6456 42
FIELD EXPERIENCE metal, leaving a reentrant corner with a larger The auxiliary feedwater piping at the Sequoyah angle and, therefore, a larger stress concentration plants connects to the mein feedwater piping inside than that shown on the construction drawing (see the containments on Loops 2 and 3 at a vertical rise Figure 18). This is the general area of major of the piping upstream of the elbow at the steam cracking. A larger field weld joined the thicker generator feedwater nozzle, as shown in Figure 25, end of the transition piece and the pipe; no but outside the containments on loops I and 4.
cracking has been experienced at the weld, Eacii connection consists of an expander [102 to probably because of a lower stress concentration 152 mm (4 to 6 in.)] and a tee fitting [405 x 406 x factor, but there are some cracks at the nearby 152 mm (16 x 16 x 6 in.)] containing a thermal geometric discontinuities.
sleeve; these details are not shown in Figure 25.
t Feedwater nozzle Interf ace betwe#.:
Hot region hot and cold fluids Thermal sleeve
~
,.,/
ramum 7
w.
To steam generator,_ _ _,,,,,,
feednng D
N Transition piece Cold region j
Feedwater (not in service during mode 3)
M96 0308 Figure 25. Stratified flows during operation of auxiliary feedwater system (TVA 1992a).
FIELD EXPERIENCE The following is a brief history of the operations through-wall circumferential crack in the feedwater and inservice inapections at the two units, a nozzle-to-transition piece weld region. The crack summary of the cracking discovered in 1992, was on the transition piece side of the weld, as results of metallographic examinations of the shown in Figure 26. Although the crack was cracks, results of the root eause analysis (including estimated to be 25 mm (1 in.) in length on the a discussion of the plant operation that affects outside diameter by visual inspection, radiography feedwater nozzle cracking), a summary of the 1993 revealed that the crack was 51-mm (2-in.) long on cracking, and repair and mitigation efforts that the outside diameter and 178-mm (7-in.) long on have been undertaken at Sequoyah.
the inside diameter.
Past History. The construction radiographic examinations of the feedwater nozzles were com-l 4
pleted in 1979. Full-power licenses for the two M%ggg g
g _.
units were issued in 1980 and 1981 respectively, and normal operation began in mid-1981 for Unit L pece Ni
- W eld I and mid-1982 for Unit 2. Nozzle inspections
=FT ~
fusion M@WyI jg line were performed in accordance with Bulletin 79-13 7
M l
requirements at the first refueling outages on Unit I (1982) and Unit 2 (1983). Augmented examina-
' kg' tions initiated in 1983 included UT of the feed-
- W water nozzle and the adjacent piping of one loop a
per refueling outage. The scope of the augmented N,
' - Crack ultrasonic examinations was expanded in 1988 to i
t include four loops per refueling outage. Several f
d indications were recorded at the crack locations but
(
g4 NEgg$
were misinterpreted as being caused by geometry; 8
Nozzle weldf"4 later, cracks were discovered at the locations of
- 5 U*
these indications. The first unmistakable indica-tion of cracking was when leakage occurred at Figure 26. Through-wall crack in the feedwater Unit I in 1992.
nozzle weld at Sequoyah Unit 1 (TVA 1992b).
The feedwater chemistry during operation was as Radiographic examinations (Iridium 192 source follows: (a) oxygen was low, typically less than with Type 1 film) of all eight nozzles at the two 100 ppb, but copper oxides present in the corrosion units showed that five of these nozzles had products may contribute to corrosion fatigue; (b) significant cracking (USNRC 1993). The results the secondary water included hydrazine,2 to 3 are summarized in Table 6 (TVA 1992a). All but ppm of ethanolamine, and 5 to 100 ppm of boric one of the cracks were located on the two sides of acid; and (c) the room temperature (25*C) pH of the pipe cross section at about the 90- and 270-i the feedwater was in the range of 8.5 to 9.2.
degree (3 and 9 o' clock) locations. The crack on Steam Generator 1 of Unit 2 was located near the Cracking incidents in 1992. In March top (0-degree or 12 o' clock location). Most of the 1992, with Unit 1 in Mode 3 (hot standby) major cracks were about 127 to 178 mm(5 to 7 conditions, a high level alarm for the containment in.) in length and located on the transition piece sump went off. Personnel performing inspections side of the weld. However, two of the major in the lower ice condenser bays observed water cracks were located on the nozzle side of the weld.
j streaming trom the area of Steam Generator 3.
The destructive examination results, discussed in The plare. was cooled down to Mode 5, and an the next section, indicated that the cracking was investigation of the leak was begun. After removal caused by fatigue.
of the insulation, visual inspections revealed a NUREG/CR-6456 44
.=.
4 i
FIELD EXPERIENCE i
Table 6. Results of Sequoyah radiographic inspection (TVA 1992a)'. (1 in. - 25.4 mm).
Crackleg le Crocklegla tressielemas-messie-te.
Appresimate Unit aG ethew weld treesition weld locaties Length Comments N/A Interminent machine marks confirmed 1
i N
N on Construction Radiosrephs 2
N N
N/A N/A N/A 3
N Y
270*
6.7" Nozzle weld, transition side of root 4
N Y
90* A 270*
I. 90'5.3"
- 1. Noz. Weld, transition on side of root j
- 2. Noz. Weld, multiple circumferential
- 2. 270' 4.9" cracks, transition side of foot
- 3. Noz. Weld, nozzle side of root
- 3. 270' 5.5" j
2 1
N Y
Essent. Top 5.1" Noz. Weld, transition side of root 3
Dead Center (O')
2 N
N N/A N/A N/A 1
3 N
Y 90'& 270'
- l. 90* 0.4*
- 1. Noz. Weld, transition side of root
- 1. 90' 2.4"
- 2. 270* 6.7*
- 2. Noz. Weld, transition side of root 4
N Y
90' l.6" Noz. Weld, transition side of root
' Radiographic Inspections perfornied in 3/92 were performed using high sensitivity "M" grade film and greater source to film i
distance.
4 Figure 27 shows the extent and orientation of some explanations for these asymmetric crack profiles l
of the primary cracks (TVA 1992a). Although are (a) the elbow may be in a plane inclined to the only one of the cracks had penetrated the wall, the vertical plane instead of in the vertical plane, or (b) other five cracks shown were 50 to 80% through-the counterbore may have an asymmetric wall. Figure 18 shows the primary and secondary geometry.
cracking at the 270-degree location of Steam Generator 3 at Unit 1. The cracks on the Unit 1 All eight nozzles in Units 1 and 2 had been steam generators were at the 90iand 270-degree inspected several times prior to the time that the locations, which indicates that the elevation of the reported leakage occurred, but no cracks were hot and cold fluid interface might have been at the detected (Coley 1992).
Initial radiographic horizontal conterline of the pipe during periods of examinations were performed at the end of auxiliary feedwater operation. Thermal striping at construction in 1979.
The first inservice i
the interface may have caused crack initiation, and inspection examinations in accordance with the cracks were then propagated by cyclic thermal Bulletin 79-13 were performed during the stratification. The crack locations on Unit 2 are refueling outage in 1982, at the end of Cycle 1.
more difficult to explain. The cracks on Steam Augmented inspections of one loop per refueling Generator 3 at Unit 2 were rotated approximately outage were initiated in 1983 and expanded in 30 degrees clockwise from the 90-to-270-degree 1988 to all four loops per refueling outage (Cycles line, and the crack on Steam Generator I at Unit 2 4 and higher on Unit I and Cycles 3 and higher on was located near the top of the pipe. Two possible Unit 2). During these inspections, indications were 45 NUREG/CR 6456
FIELD EXPERIENCE UNIT 1 UNIT 2 0*
[
B o-
_ #'kN il 3
r 270*
270*
i 47.60*
44 0s' B
(A) Through wall 180*
180*
(B) 50% through wall (C) 60%through wall (D) 80% through wall (E) 60%through wall w/ sirnitar parallel 0*
crack on other side 0*
of weld C
i E
9 4 D
5.40* *
.g 0*
f
,7 k' g
,g 90*
90*
+
270*
]
270*
d 247.10
\\
180*
M95 0198 Figure 27. Extent and orientation of primary cracks in feedwater nozzle-to-transition field weld at Sequoyah Units I and 2 found in 1992 (TVA 1992a).
Nur EG/CR-6456 46
~
FIELD EXPERIENCE l
recorded at the sites where cracks were later areas will be examined each refueling outage, with confirmed, but they were misinterpreted as the exception of the auxiliary-to-feedwater line geometry effects, area, which will have a longer inspection hierval.
Loops 2 and 4 will be inspected every even-num-The licensee has stated that the inspection beted outage, and Loops 1 and 3 will be inspected weaknesses involved limitations in the UT every odd-numbered outage.
procedure and the inspector qualifications (TVA 1992a). It appears that the main weakness was the Metallographic Evaluation-A de-recording ofindications as caused by a geometry structive examination was performed on the Unit 1, effect without fu ther verification with enhanced Steam Generator 3 transition piece containing the inspection techniques. The UT was performed through-wall crack, which included the adjacent according to the American Society of Mechanical welds and the nearby adjoining sections of the j
Engineers (ASME) Code Section XI requirements feedwater nozzle and pipe. The examination and using intergranular stress corrosion cracking revealed that the through-wall crack initiated at the (IGSCC) methodology. The UT techniques and inside surface and propagated both circumfer-procedures were neither optimized nor adjusted for entially and radially, reaching the outside surface the specific configuration of the weld-counterbore and resulting in the leak. The circumferential region. The inspection personnel were EPRI extent of the crack was about 234 mm (9.2 in.) at qualified in IGSCC techniques, but since there is the inside surface and about 32 mm (1.25 in.) on no industry qualification method available for the outside surface. The crack was nonbranching identification of thermal fatigue, the inspectors and open, as shown in Figure 26 (TVA 1992b).
l were not trained to identify thermal fatigue cracks.
Oxide wedging might have kept the crack from closing. (Although no mention was made of the Several corrective actions were subsequently presence of oxides in the crack in the TVA reports,
)
instituted at the two Sequoyah plants (TVA their existence is presumed because they are 1992a). Baseline radiographic and ultrasonic reported to be present on the inside surfaces and in examinations of the replaced piping has been the secondary cracks.) The propagation of the performed. The inspection personnel have re-crack, shown in Figure 28, followed a straight path ceived enhanced training by practicing on the directly into the weld, remaining nearly normal to cracked portions of the feedwater piping that have the pipe surface instead of following a path along been replaced. The cracking event has been the weld fusion line (TVA 1992b).
reviewed with the inspection personnel who now know that the feedwater piping is susceptible to The examination of the crack surface revealed thermal fatigue cracking, that the cracks can grow striation marks, which indicate fatigue damage.
rapidly and are now familiar with the location and The licensee reported that the examination also position of the expected cracks. In addition,the revealed some evidence of thermal striping at the inspection personnel have also learned how to inside surface at about the 270-degree circumfer-differentiate between geometric reiketors and ential location. The number of cmck arrest points thermal fatigue cracks as detected by Qtrasonic foutid on the crack surface may correspond to the examination. The inspection procedures have also same number oflow power or hot standby cycles been upgraded. The current surveillance instruc-during which activation of the auxiliary feedwater tions call for 100% volumetric inspection of the system caused crack propagation. The crack transition pieces, the nozzle-to-transition piece surface was oxidized and had differem colors at i
welds,theauxiliary-to-main feedwaterlinewelds, various points along the crack front through the and base metal adjacent to each weld for a distance pipe wall, which indicates that a major portion of of two wall thicknesses. These welds will be the crack had existed for a significant period of volumetrically examined by UT, supplemented by time. 'Ihe crack surface was relatively smooth and RT when specified by TVA Engineering. And all free of extensive corrosion pitting or erosion.
1 t
i i
t I
i r
s-
{
w - r!
Y a
6 04 n
s I
f e;.
}
i I
O s
- l
'k!j rn l { ll x
=
{
i m
u*
a o
en r.
Z O
=
tu I
. u 2 *A 1.1 Crack '. ~
.rs 1
l
\\'
' (
w
' ;f..t:y ls\\*j :% '*
j m-
.. c.:
- n y;..
- 3 hjd. b i
- h ?
[
v ::
A' c.
h 1.:. 99 3
^jQ, n3,3fk
' ' +
f
=
I a
- h
- ,; y$ b #
r x.,
, + v wkz o.
. ~
=
-..- 9
^T. ;e:.
- ['7 kj*dy ; Pipe :)C-5 t
{
- i. -. --.
s e
?g:gg.
. h. m. m- _ p, m c+ z
.< q. 7 m /..,
. Q#..- h.w 5 %.
sy _ -
i i
eP%4 t
4*
usscais Figure 28. D rection of ropagation of a primary crack in Sequoyah Unit 1 (TVA 1992b).
i P
?
i FIELD EXPERIENCE 1
Pits were observed adjacent to the nozzle-to-Stratification in the feedwater piping leads to three transition piece weld bead. These pits linked in a types of cyclic loadings:
global thermal circumferential pattern. The linked pits could stratification, which results in bowing of a long, provide an area of stress concentration, more horizontal pipe; cyclic or local thermal l
intense than a single pit, that may lead to fatigue stratification, which occurs locally when the water crack initiation. A deposit of copper (49.8%
level fluctuates up and down; and thermal striping weight as oxide), nickel (3.6% weight as oxide),
(Cofie et al.1994). The portion of the feedwater and iron (balance) was found on the inside surface piping adjacent to the feedwater nozzle is mainly of the elbow. Copper oxide promotes pitting and subject to cyclic thermal stratification and thermal corrosion fatigue, striping, whereas the feedring is mainly subject to i
global stratification. Thermal striping at the The microscopic examinations also revealed interface between the hot and cold fluid layers in multiple secondary cracks adjacent to the through-the nozzle results in high-cycle fatigue, which wall primary crack in the weld heat-affected zone initiates cracking. Evidence of thermal striping has and the base metal of the transition piece been observed in laboratory tests conducted at (Figure 18).
These secondary cracks were Westinghouse on piping of similar design (Hu, orientated circumferentially, as was the primary Houtman, and White 1981).
crack, but were much shallower than the primary crack. The secondary cracks were numerous in the The stress concentrations present because of the direct vicinity of the primary crack, but diminished counterbore, and in some cases the weld, increase about 13 mm (0.5 in.) away from the primary crack the stresses at the inner surface of the transition along the transition piece. These cracks appear to piece region and make it susceptible to fatigue have hen propagating slowly through the wall, crack initiation. Figure 18 shows initiation of the beca.ae a significant stress concentration was primary crack at a stress concentration caused by absent, and the primary crack could have relieved the weld geometry and the transition piece the stresses in the vicinity of these cracks. The inclination. In this case, the weld has formed a secondary cracks were filled with a tight, dense notch with the transition piece resulting in a stress oxide, presumed to be magnetite, and extended to concentration greater than that of the transition a maximum of 10% of the wall thickness. Most of piece inclination alone. Once a crack is initiated, the cracks followed a straight path; however, some propagation results from cyclic thermal displayed multi-branching and could have resulted stratification caused by on-off auxiliary feedwater from several different causes. Fatigue caused by operation.
cyclic thermal stresses or stress-assisted corrosion, and intensified by oxide wedging are possible The auxiliary feedwater system can be operated in mechanisms.
either the automatic or the manual mode. In the automatic mode, full flow from the auxiliary feed Other minor cracks were detected in the pumps [about 760 to 830 t/ min (200 to 220 gpm)]
counterbore region; these cracks were shallow and is initiated whenever the level in the steam had no obvious branching.
This area was generator is low, and is terminated when the level essentially free of corrosion pitting, and there was has been increased to a sufliciently high level.
no observed flow-accelerated corrosion damage.
This results in on-off operation of the auxiliary feedwater flow, estimated to be approximately Root Cause Analysis-The root cause three times per hour. In the manual mode, the of the cracking is attributed to thermal fatigue operators adjust the auxiliary feedwater flow rate caused by periodic thermal stratification when cold to match the loss of water in the steam generator by auxiliary feedwater was injected into the steam recirculating a portion of the flow from the generator during Mode 3 operations. This is auxiliary feed pumps back to the condensate consistent with the root causes identified for storage tank, as shown in Figure 8.
simik. cracking that has occurred at other PWRs.
FIELD EXPERIENCE During periods of auxiliary feedwater operation, Fatigue damage progresses much more rapidly in the flow partial!y fills the horizontal portion of the the automatic mode, during which the flow nozzle region, leading to the stratified fluid fluctuates between 0 and about 760 Umin (200 condition shown in Figure 25. The cold auxiliary gpm) at a rate of approximately 3 times per hour feedwater [about 20 to 38'C (70 to 100*F)] fills and cyclic thermal stratification takes place, than in the lower portion of the horizontal portion of the mauual mode, during which the flow is the feedwater line and nozzle, and hot fluid [about relatively constant at about 300 Umin (80 gpm) and 286*C (547'F)] from the steam generator fills the large fluctuations in the flow rates are absent. The upper portion. When the auxiliary feedwater flow automrtie mode of operation, because of the is terminated, the nozzle region temperature relatively high flow rates, might also be producing returns to the hot condition, and then when flow is thermal striping at the interface between the cold reinitiated, the stratified condition is reestablished.
auxiliary feedwater and the hot fluid from the This on-off auxiliary feedwater operation leads to steam generator. Thermal striping imposes high-cyclic thermal stratification. Figure 29 shows cycle stresses on the pipe inside surface and typical auxiliary feedwater flow intes with the contributes to crack initiation. (See Section 7.1 for auxiliary feedwater cycling on and off during a description of thermal stratification, thermal automatic operation.
striping, and thermal cycling.) In the automatic 400 I
^
~
Automatic
~
operation T
Manual operation E
R aE a 200 1
l 100 a
e a
0 r
8 0
200 400 600 800 1000 1200 1400 1600 Time on 11/29/94 M950205 Figure 29. Loop 1 auxiliary feedwater flow rates during manual and automatic operation at Sequoyah Unit I on November 29,1994. Courtesy of W. Ludwig. TVA.
3 FIELD EXPERIENCE mode of operation, the auxiliary feedwater flows at tic mode. In addition, an operating procedure cbout 760 # min and fills the lower half of the review was conducted by Westinghouse, and the feedwater piping cross section as shown in following three recommendations were made to Figure 25. The pipe wall inside surface, near the reduce the impact of auxiliary feedwater operation interface, is probably subject to high-cycle fatigue on the feedwater nozzle stresses (TVA 1992a):
damage caused by thermal striping.
- 1. The operating procedures were changed so that Cracking /ncidents /n 1993. A subsequent a change from the automatic to the manual ultrasonic examination at Sequoyah in 1993 (less mode of auxiliary feedwater operation is made than one fuel cycle later) revealed circumferential as soon as possible after a reactor trip, indicatior.s in five of the eight replaced transition pieces, including Loops 2 and 3 of Unit 2 (Cofie et
- 2. He auxiliary feedwater recirculation line was cl.1994), and Loops 1,2, and 4 of Unit 1. Loop 2 modified so that during manual operation the of Unit 2 was the only loop where the UT located net flew of auxiliary feedwater to a steam an indication on the nozzle side of the nozzle-to-generator would be about 300 to 340 t/ min (80 transition piece weld; all the other indications wert to 90 gpm), and located in the transition piece within 13 mm (0.5 in.) of the weld centerline.
Radiographic
- 3. The operating conditions during steady state inspection was not able to confirm any of these Mode 3 operations were revised such that the indications. UT and RT of the auxiliary feedwater-use of the auxiliary feedwater system in the to-feedwater piping connection did not detect any automatic mode is reduced, indications.
After the 1992 cracking, all eight transition pieces After indications were discovered in Loops 2 and were replaced with the same design (in-kind 3 of Unit 2 in April 1993, time limits were placed replacement), and the plants were returned to on the auxiliary feedwater operation. Limits of 60 service. The corrective actions discussed earlier allowable hours of automatic mode auxiliary are summarized here (TVA 1992a):
feedwater operation and 600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br /> of manual Baseline RT and UT of the replaced pieces operation were established. During periods of a
operation after heatup in October 1993, the hours were conducted, of auxiliary feedwater use were recorded, and the The ISI requirements were upgraded; for remaining hours of allowable operation were determined. In addition, crack growth rates for example, a 20% DAC (distance-amplitude periods of auxiliary feedwater operation were used correction)' recording level for indications to estimate possible increases in crack depths.
has been established, inspection areas have Over the period from October 1993 to July 1994, been expanded to 100% of the weld volume calculations estimated that an initial 2.5-mm plus the base metal for two wall thicknesses (0.1-in.) long crack would have grown to 3 mm on either side of the weld, and indications are (0.123 in.).
evaluated using enhanced UT methods (such as a high-angle longitudinal wave),
Repair and Mitigation. It has been recog-nized as part of the overall PWR steam generator feedwater nozzle cracking problem that the auxil-iary feedwater operation was a root cause. Conse-quently, an evaluation of the auxiliary feedwater 1
operating time began in 1988. It became obvious The distance-amplitude correction curves j
i that the hours of auxiliary feedwater operation compensate for the attenuating efTects of the material such should be m..mimtzed, particularly. u,e automa-that the amplitude response from unknown reficctors can m
be evaluated and compared to that of the calibration reflector independent of distance.
FIELD EXPERIENCE Re event was reviewed with ISI personnel, the fatigue damage caused by thermal stratification and enhanced training was provided using and striping. It is estimated that the thermal liner the removed pieces of the damaged is capable of sustaining the thermal fatigue loads feedwater piping and nozzle, and caused by about 900 h of automatic operation of the auxiliary feedwater system. Online fatigue Additional confirmatory inspections (RT and monitoring of Loops 2 and 3 of Sequoyah Unit 2 is UT) have been implemented.
being continued and it includes monitoring of both main and auxiliary feedwater temperatures.
Long thermal liners with 356-mm (14-in.) inside diameters were installed in loops 2 and 3 of Unit 2.
There are two potential concerns about the struc-He thermal liner, shown in Figure 30, extends tural integrity of the liner: (1) wall thinriing of the over the nozzle-to-transition piece weld, transition liner by the flow-accelerated corrosion process, piece, transition piece-to-elbow weld, and a portion and (2) fatigue cracking of the liner. The first of the elbow and protects these compone.its from concern has been addressed by selecting SA-335, i
Existing feedwater nozzle All welds in the existing section of pipe are protected by the thermalliner Liner Existing thermal sleeve Thermalliner
- /
M96 0332 f.
Figure 30. New thermal liner protecting the feedwater nozzle and elbow from thermal stratification loadings (Cofie et al.1994). Copyright American Society of Mechanical Engineers; reprinted with permission.
FIELD EXPERIENCE Grade Pil as a liner material. The chromium thermal liners have also been installed at other content of this material is in the range of 1.0 to PWRs (for example, the Turkey Point units).
1.5 wt%; therefore, as discussed in Section 7.2, significant wall thinning of the liner will not take Two other steps in addition to the installation of place.' The secad concem about fatigue cracking the thermal liner have been taken to provide pro-can be addressed by monitoring the outside wall tectico against thermal fatigue damage: elimination temperature of the piping components protected by of stress raisers and improved water chemistry.
the liner, if there is any leakage of auxiliary feed The end of the feedwater nozzle was built up such i
water through the liner, it will affect the wall that there is no change in the wall thickness from temperaturer, which can be detected by the moni-the feedwater nozzle to transition piece, which toring system. There is also a concem that thermal climinates tha stress raiser where fatigue cracking fatigue might damage the liner and produce loose occurred eailier. At least one other PWR has its parts. However, the broken pieces of the liner feedwater nozzle ends built up to eliminate the would have to be of such small size that they can stress raiser. Figure 31 shows how the counterbore pass through the S I mm (2 in.) diameter J-tubes in region can be redesigned to eliminate the stress order to damage the steam generator tubes. Long raiser. The utility is also eliminating copper from Nozzle side Pipe side M
h dlM,p+< D @~
.a L
0,75n.b fh, P
9y (e) Original Design
),
nym... -e l 7 o
16.0 in. $
-1in.
2 in. --*
- 0.5 in.
) 14.32 in. $
Counterbore r
Nozzle olde New pipe
>g
- A x,.. nv N h D O N U'I%'
-1 in P'
H"+'
(b) Replacement Design
)*hrwaa I
...m-t 10 to 1 taper l f 14.32 in. $
16.0 in. $
-5.7 degrees Wald buildup on nozzle to elirninete counterbore M96 0338 Figure 31. Weld buildup to eliminate counterbore as a stress raiser (Cofie et al.1994). Cop > Tight American Society of Mechanical Engineers; reprinted with permission.
' V. N. Shah, private communication with K. House, TVA. Chattanooga. Tennessee. July 26.1996 53 NUREG/CR-6456
FIELD EXPERIENCE all the feedwater system components. The tubes in mere notched on the upstream side of the tran:ition the condenser, which are made from copper-pir;ce-to-nozzle weld than for either the bearing material, are being replaced with titanium dawnstream side of this weld or for a typical tubes.
transition piece-to-pipe weld counterbore. The transition piece-to-pipe counterbore, shown in 6.3.2 Diablo Canyon Units 1 and 2 Figure 17(b), has a reentrant angle of 166 degrees, versus 150 degrees for the transition piece-to-Diablo Canyon Units 1 and 2 are 4-loop nozzle weld. Fatigue cracking typically occurs Westinghouse PWRs. Figure 32 is a drawing of circumferentially around the pipe at the upstream the SA-508 Class 2 steam generator nozzle and the end of the transition piece-to-nozzle weld, where adjacent SA 106 Grade B feedwater piping, the stress concentration is significantly more typical for both units. A SI-mm (2-in.) long severe.
transition piece was field welded to the nozzle at the downstream end. At its upstream end, it is This section presents a brief history of operation weldedtoa203 mm(8-in.)longhorizontalsection and inservice inspection at the two units, a of pipe, which in turn is welded to a 45-degree summary of the cracking that was discovered in elbow. Figure 17 shows a construction drawing of 1992, results of the metallographic examinations of the field welds (Cofie et al.1994). Tne inside the cracks, results of the root cause analysis surfaces were counterbored in this region, resulting (including a discussion of plant operation that in a reentrant corner on each side of the weld at the affects feedwater nozzle cracking), and repair and transition to the thinnest section. Figure 17(a) mitigation efforts that have been underi:aken at shows that the thickness change is more Diablo Canyon.
pronounced, and the counterbore reentrant angle is Steam generator feedwater nozzle First weld Transition piece Second weld Third weld
\\ ermal Th Fourth weld \\
Leadirig Fifth weld edge N f
m I
\\
/
45 degree elbow f
,,j See Figure 17(a) r 45-degree elbow
~
M96 0339 Figure 32. Feedwater nozzle area layout at Diablo Canyon Units ' and 2 (Shvarts et al.1994). Copyright American Society of Mechanical Engineers; reprinted with permission.
4 FIELD EXPERIENCE 1
Past Nistory. De Unit I steam generator 1992). The average reading was 15.5 ppb, but on field welds were originally completed in 1974 in two occasions rose to 100 ppb. Copper caahaar accuidi.a with ASME Section I (Power Boilers) tubes were replaced with titanium tubes after the r:quirements, rathec than ASME Section 111. In first fuel cycle, which resulted in a smaller amount 1977, the nozzle-to-transition piece field weld on of copper oxide present in the coolant.
Steam Generator 2 of Unit I developed a through-wall crack, resulting in a leak during hot functional 1992 Cracking incident. Based on the J
t: sting. Subsequent UT and RT revealed a cracking found at Sequoyah, an enhanced inspec-154-mm (6-in.) long circumferential crack in the tion was conducted during the Unit i fiAh refuel-heat-affected-zone off the weld root (Stoller 1979).
ing outage in September 1992. PGAE conserva-The Unit 2 welds were originally completed in tively determined from ultrasonic examinations 1978, using an improved weld procedure derived that linear, circumferential indications on the from the lessons leamed from the Unit I weld feedwater piping adjacent.o three of the uteam 4
4 friture.
He improved procedure included generator feedwater nozzics of Diablo Canyon maintaining the preheat temperature until the post-Unit i exceeded the ASME Code Section XI flaw
-weld heat treatment.
acceptance criteria. Linear indications were also
[
found in the fourth steam generator but did not i
- In response to Bulletin 79-13, the first inspection exceed the ASME Code criteria. The indications of Unit I was performed after hot functional tests were near the nozzle-to-pipe welds, as shown in in 1979. Based on the inspection results, the Figure 17(a) (PG&E 1992, Cofie et al.1994).
i transition pieces on Unit I were replaced (in-kind)
Radiographs of the area did not reveal any cracks, in 1980 (Cofie et al.1994). He full power but a visual examination after cutting out the operating license was issued in 1984 after ~
section of pipe in the degraded region detected the t
successful completion of a design verification possibility af degradation. Consequently, the
- program. ' During the first refueling outage in licensee d d to replace the piping in the vicin-
]
1986, examinations were completed in response to ity of the degradation and to conduct a metallurgi-Bulletin 79-13, and the piping and welds were cal examination of some of the degraded poitions.
found to be acceptable under ASME Section XI requirements.
The second refueling outage In addition, thinning of the leading edge of the inspection was conducted in 1988, with no carbon steel thermal sleeves was found on all four reported indications. The Unit 2 Bulletin 79-13 Unit I steam generators, but the licensee con-inspections were conducted during the first cluded that the sleeves were acceptable for contin-refueling outage in 1987, and piping and welds ued operation. Although no linear indications l
were found to be acceptable under ASME Section were found at Unit 2, thinning of the leading edge XI requirem:nts, was observed on some of the thermal sleeves.
The feedwater chemistry during operation was as Magnetic pcMicle testing (MT) of the i;1 bow after follows. The 25'C pH at Diablo Canyon had it was cut out revealed a few indications.
varied in the range of 8.7 to 9.1; for the most Ultrasonic examinations estimated the deepest i
recent period, it had been in the range of 8.9 to indications on the four steam generators as follows:
9.1.' The typical oxygen in the main feedwater l
was less than 5 ppb, whereas the oxygen content of SGl-1 0.76-mm (0.030-in.) deep, the auxiliary feedwater might be as high as 100 197-mm (7.75-in.) long ppb. The dissolved oxygen content in the Diablo Canyon Unit I condensate storage tank was SG l.2 0.25 mm (0.010-in.) deep, measured over a period of about 3 years (Peterson SI-mm (2-in.) long intermittent S
' V. R Shah. private conversation with Lee Goyette, i
PGAE, March 2.1995.~
1 55 NUREG/CR-6456 I
l
-e
<+
FIELD EXPERIENCE SGl-3 2.7-mm (0.107-in.) deep; system was calibrated using a sizing notch block.8 360-degree intermittent, longest The results from the feedwater piping were not 51 mm (2 in.)
immediately conclusive because the signals were oflow amplitude. They had some crack-like signal SG l-4 1.2-mm (0.047 in.) deep,3.2-mm characteristics, and there were no geometrical (0.125-in.) long origins. A single-wall radiography was also performed by using the access provided by the Metallographic-examination results for two gamma plug. Using single wall shots with samples of pieces cut out from Unit i Steam maximum source / film distance did not reveal any 8
Generators 1 and 3, discussed in the next section, cracking on Steam Generator 4. Panoramic shots
' determined that the deepest crack was 1.5-mm were used on Steam Generators 1 and 2, but, again, (0.060-in.) deep (Aguiar 1993). The allowable no cracks were detected. An examination by video ASME Code depth is 10.2 mm (0.400 in.).
probe showed possible indications that were not conclusive. However, the UT inspections with a Based on a review of the previous radiographic tip diffraction technique revealed cracks at the records performed in response to Bulletin 79-13, counterbore discontinuity, the deepest being about the licensee determined that the records we e 8.9 mm (0.35 in.).
incomplete;thepipe-to transitionpieceweldswere radiographed, but the nozzle-to-transition piece Afstelloyaphic Evalueefon-Samples welds were not. Also the radiographic techniques of the transition pieces from two steam generators used in the Unit 1 1979,1986, and 1987 outages (1 and 3 ofUnit 1) were removed for metallurgical may not have been in full compliance with Bulletin examination. Magnetic particle testing of the 79-13 guidance, such as penetrameter thickness,'
n uples'showed that the cracking extended 360 sensitivity, and density values, but the 1985 Unit 2 degrees around the inside surface circumference, inspection was in full compliance with the and was about 6 mm (0.25 in.) upstresm of the guidance. However, based on examinations of the transition-piece-to-nozzle-weld, at the enunterbore l-removed sections from Unit 1, the licensee reentrant corner. A crack that was idemified by l
believes that the small thermal fatigue cracks UT as 8.9-mm (0.35-in.) deep was found to be would not have been detected.
only about 0.91-mm (0.036-in.) deep (Aguiar 1993).
Inclusions (aluminum oxides and Based on the experience at the Sequoyah units, manganese sulfides) found in the piping material planning for enhanced inspections of the Unit I may have caused the incorrect sizing with the tip nozzles began in May 1992. The experience in diffraction technique, which is discuned in Section previous inspections at Sequoyah and Turkey Point 8.2.1.
were reviewed, and multiple techniques were chosen for use at Diablo Canyon Unit 1, including ne major cracks had propagated perpendicularly both enhanced manual and automated UT into the wall and were nonbranching. The deepest techniques. The past experience had shown that crack was 1.5-mm (0.060-in.) deep (Aguiar 1993).
the ASME Code examinations were not adequate Evidence of microcracking was found in the to detect tight thermal fatigue cracks, and use of enhanced UT techniques was desirable. The UT 8
Saw cuts were 1, laced in a block, and the (1T results were compared with the physically measured widths for calibration.
8 The source is placed on the back side of the pipe wall. This increases the resolution.
I A strip of sheet metal is placed on the pipe wall and
' These magnetic particle tests performed in the must be seen in the radiograph. This serves as a type of laboratory were in addition to those conducted on site and online calibration.
discussed in Section 6.3.2.
FIELD EXPERIENCE transition piece. There were multiple initiation secondary coolant temperature is greater than sites, and some of the cracksjoined randomly. He 177'C (350*F)(PG&E 1992).
cracks propagated transgranularly into the metal, most penetrating only a small depth. Some of the ne Diablo Canyon auxiliary feedwater system cracks initiated at the bottom of corrosion pits, was designed for minimum flow rates on the order whereas others initiated at locations that showed no of 830 t/ min (220 gpm) per steam generator during evidence of pitting. Most of the cracks contained accident conditions (PG&E 1992), but as the traces of what appeared to be high-temperature required flow is typically less than 830 t/ min (220 oxide, and some cracks were completely filled with gpm), fluctuations occur during automatic control oxide. Some of the cracks had very blunt crack operation. Figure 33(a) shows the temperatures tips, and there were local areas where the crack during a cooldown on Unit 1 in 1992. The main surfaces were widely separated. These features are steam temperature has been reduced from about characteristic of corrosion fatigue.
288 to 177'C (550 to 350*F), and the feedwater temperature has decreased from about 120 to 40*C Root Cause Analysis-The root cause (250 to 100*F) over about 9 hours1.041667e-4 days <br />0.0025 hours <br />1.488095e-5 weeks <br />3.4245e-6 months <br />, whereas the of the cracking was attributed to thermal fatigue auxiliary feedwater temperature has remained caused by periodic thermal stratification when cold constant at 38'C (100*F). Figure 33(b) shows the cuxiliary feedwater, typically with a fluctuating fluctuations in auxiliary feedwater flow rate during flow rate, is injected into the steam generator this period; the flow rate averages about 380 f/ min during Mode 3 operations. This is the same root (100 gpm). Plant data indicate that the auxiliary cause identified at the other PWRs with similar feedwater doer not fill the nozzle until the flow cracking.
rate reaches about 760 t/ min (200 gpm) or more (Shvarts et al.1994). Cycling involving changes Oxygen in the feedwater was low, typically less in flow rates of about 380 f/ min (100 gpm) occur than 100 ppb, but the oxidizing effects of the at a rate of about 2 to 3 times per hour.
copper oxides which were found in the oxide deposits of the crack surfaces may have Figure 34 illustrates the flow in a straight i
contributed to the corrosion fatigue. The licensee feedwater pipe, showing that when the level reports that the oxygen coritrol has been very good oscillates from the bottom to the top of the pipe, over the plants' lifetimes. Although the cracks the stress at location P alternates between tension displayed some characteristics that resemble and compression. In both diagrams of Figure 34, corrosion fatigue, the cracks did not contain the the cold auxiliary feedwater is flowing under a extensive buildup of oxides and other corrosion layer of hot fluid, and as auxiliary feedwater flow products usually associated with corrosion fatigue, rate fluctuates, the cold water level increases and so the licensee concluded that the cracking appears decreases, producing cyclic stresses. The true end to be caused more by thermal fatigue than by condition is neither fixed nor free, but is restrained corrosion fatigue (Aguiar 1993).
somewhere between the two conditions. However, at point P, the stress reversal is the same.
Typical auxiliary feedwater evolutions for the Similarly, damage can occur in the upper region of Diablo Canyon units are listed in Table 2 (Peterson the pipe when the flow is higher, almost filling the 1992). These include typical durations and flow pipe. Thermal striping also may be taking place at rctes. The total times spent in Modes 2 and 3 the interface between the hot and cold layers.
operation as of 1992 were 339.47 days for Unit I cnd 239.58 for Unit 2.
Auxiliary feedwater Repair and Mitigation. The licensee de-temperatures at Diablo Canyon are in the range of cided to replace (in-kind) the horizontal piping to 20 to 38'C (70 to 100*F) during Modes 2 and 3 all four Unit I feedwater inlet nozzles. During the operation, during which time the steam generator Unit 2 seventh refueling outage, an enhanced 57 NUREG/CR-6456
FIELD EXPERIENCE 600 500 Main steam 400 300 Main feedwater 100 Auxiliary feedwater 0
0 2.
4 6
8 10 12 Time (hrs)
(a) Main steam, main feedwater, and auxiliary feedwater ternperatures j
300 250 l
200 150 I I l 1
r b'
,d
(
h
~
~
50 0
-50 100 O
2 4
6 8
10 12 1
Time (hrs)
- 0207 (b) Auxiliary feedwater flow rates Figure 33. Temperatures and flow rate during cooldown on 9/13/f 2 at Diablo Canyon Unit 1, Steam Generator 1 (Peterson 1992).
FIELD EXPERIEWCE Stress profiles Fixed end Free and condition condition
(-)
(-)
-)
\\
cay
>p e
r-
~
o o
1 p
p C
(+)
_=
A
(+)
(-)
(a) High stratification level 1
(-)
(+)
(-)
0
~
Hot
~
op p
p j
o O
iiii W ]
C-(+)
(+)
(b) Low stratification level Figure 34. Distribution of axial stresses in a long horizontal feedwater pipe for two difTerent elevations of hot-and-cold coolant interface and for two different boundary conditions for the pipe ends (Shvarts et al.1994).
Copyright American Society of Mechanical Engineers; reprinted with permission.
inspection was conducted on one nozzle region; no tions that may inhibit crack development, as well evidence of fatigue cracking was found.' Similar as inhibiting the thinning of the thermal sleeves, inspections are planned for future refueling out-which are discussed in Section 6.4.2. Auxiliary ages. 'Ihe licensee is reviewing several modifica-feedwater cycles are being moaitored at both units by online fatigue monitoring.
' V. N. Shah, Private conversation with Dave Gonzalez, Pacific Gas and Electric, January 23,1997.
FIELD EXPERIENCE 6.3.3 San Onofre Unit 3 Although no instances of feedwater nozzle cracking had been reported on Combustion San Onofre Units 2 and 3 are 2 loop, System 80, Engineering plants during the previous 10 years Combustion Engineering PWRs. In this design, a (since the Maine Yankee cracking), the nozzles safe end is welded to the nozzle in the shop, and a and adjacent piping were inspected at San Onofre field weld connects the safe end to a 45-& ree Units 2 and 3 in response to Information Notice 3
elbow, as shown in Figure 35. This elbow is the 93-20 (Mostafa and Ramsey 1994). A review of j
terminus of the feedwater piping, and it is welded the inspection data shows that no indications had i
to another 45-degree elbow at its other end. A been recorded, though there were evaluations of counterbore has been machined into the safe-end-areas characterized as geometry or geometric to-elbow weld area. The auxiliary feedwater inlet configuration, which is allowed by ASME Section connects to the main feedwater system upstream of XI.
the two 45-degree elbows.
i$?$,'ea:::.
hkkk$ikda Shop weld Field weld
$1W DI h!M]$h[hIIhgfMfM
, W8 Qsafe.-endf) c 45-degree elbow I
Counterbore M96 0328 Figure 35. San Onofre Units 2 and 3 feedwater nozzle region (Mostafa and Ramsey 1994). Courtesy of E. Regata, Southern California Edison.
FIELD EXPERIENCE Past History. Unit 2 began hot functional X-ray spectroscopy was performed on the oxide-testing in 1982 and began commercial operation in filled cracks and confirmed that there were numer-
)
g
' 1983. The corresponding dates for Unit 3 were 1 ous manganese sulfide inclusions in the areas j
i year later. These two plants were classified by associated with the cracks. Fluorescent magnetic i
Bulletin 79-13 as designated applicants for particle testing of the other steam generator nozzle operating ' licenses, and inspections of their safe end revealed numerous linear indications of
)
j
. feedwater nozzles and piping were requested in similar shallow depths. These were removed by j
Revision I to the Bulletin. No indications were buffing and did not reqdire grinding.
found.-
~
A magnetic particle test of the weld preparation 1993 Cracking incident.
The steam area of the elbow revealed a 216-mm (8.5-in.) long generators were inspected at San Onofre Units 2 linear indication in the counterbore region at the 9 i
)
cnd 3 in response to Information Notice 93-20.
o' clock position, outside the weld zone. Further The inspections found no indications on Unit 2.
examination with the fluorescent magnetic particle However, during the Unit 3, Cycle 7, refueling test revealed that the indication extended 360 outage, several indications were found. The UT degrees around the inside diameter of the elbow.
- i examination detected indications at approximately Metallographic examination revealed an oxidized the 3 and 9 o' clock positions on the safe end ofone transgranular crack 0.56-mm (0.022-in.) deep.
l of the nozzles. These indications were initially Striations on the crack surface indicate that the r
characterized as approximately 3.8-mm (0.150-in.)
crack had experienced 2,235 cycles. Previous UT deep and 137-mm (5.4-in.) or 104-mm (4.1 in.)
examination of this area showed an indication that j
i long.
had been evaluated as a geometric discontinuity.
Examination of the other steam geneator elbow The inspection was conducted using an automated revealed a similar 360-degree linear indication l
l UT device (introspect/98 Volumetric Inspection with a maximum depth of3.8 mm (0.150 in.) at the j
System, a 45-degree shear wave transducer), whi:n 9 o' clock position. A metallographic examination -
scans the test region from two different plane:.. A revealed multiple cracking, with large sulfide computer processed the data, providing a hree-(manganese and iron) inclusions in the base metal dimensional evaluation of the area, enhancing the matrix, with significant high-tem perature oxidation 3
ability to determine the size and nature of any indicative of an old crack. The depth was approxi-defects (Mostafa and Ramsey 1994). This method mately 0.61-mm (0.024-in.). A few aluminum is specifically tailored to examine crack-like oxide inclusions were also found, j
reflectors originating on the inside nozzle surface.
Root Cause Analysis-Based on the p
Metallographic Evaluation of metallographic evaluation, the root cause of the
~
Cracking-The affected areas of both nozzle safe cracking was determined to be stratified flow from ends were cut out for examination. Magnetic on-off auxiliary feedwater operation during Modes particle tests on the safe end ofone nozzle revealed 2 and 3 operations, similar to the other incidents of numerous surface indications approximately 0.025-cracking. The cracking may have initiated at to 0.051 mm (0.001-to 0.002-in.) deep, extending sulfide inclusions in the material, i
360 degrees around the safe end inside surface over h band about 16-mm (0.625-in.) wide. Opti.
During the first several years of plant operation, cal micrographs of the affected crea revealed very the auxiliary feedwater system's control valves i
l shallow cracking about 0.038-mm (0.0015-in.)
vibrated at low flow rates, so the plant operators deep, which is about 100 times smaller than was were required to feed at 950 t/ min (250 gpm) or indicated by the initial UT examination, greater flow rates, or not at all, resulting in on-off l
i 61 NUREG/CR-6456
-- +
c n -
r-
FIELD EXPERIENCE cycles of auxiliary feedwater addition during elbows. The ability to scan the flaw area and Modes 2 and 3 operation. The cycle began with differentiate between flaws and the reflection from raising the steam generator water level by adding the counterbore is inhibited by the crown of the auxiliary feedwater. While the incoming cold safe-end-to-elbow weld when scanning upstream.
feedwater was being heated by the hot primary The counterbore itself gives such a large reflection coolant, the steam flow from the steam generator that it masks small indications when scanned was almost completely stopped because of the downstream. The weld crowns were ground down lower secondary pressure. The auxiliary feedwater to enhance UT detection in this area. In addition, flow was stopped once the steam generator water the process for disposing indications as geometric level reached the required elevation. Secondary reflectors has been revised to require a more pressure built up as the water in the steam genera-detailedjustification and review.
tor h mi up and r teaming took place. Steam was released thre";;h the aimspheric dump valves, the The auxiliary feedwater system draws from the steam bypass control valves, or da main steam line condensate storage tank, on which a nitrogen gas drains, so that the seconda y pressure was main-blanket system has been installed to prevent tained at 1000 psi. As a result ti..: steam generator oxygen introduction. However, concerns with water level decreased and me cycle was repeated respect to the operation of this nitrogen system in with another addition of cold auxiliary feedwater, a confined environment (there was a personnel The plant staff has estimated ths.t 1,494 cycles (9 safety hazard from leaking valves) have precluded cycles / dcy for 166 days) for Unit 2 and 1,089 its routine use during recent fuel cycles.
cycles (9 cycles / day for 121 days) for Unit 3 occurred before modifications were made, respec-6.3.4 Other PWRs tively, in 1986 and 1985 to allow throttling of the auxiliary feedwater control valves (Martin et al.
In June 1992, prompted by the Sequoyah incident, 1990).
radiographic examinations of a feedwater inlet nozzle were performed at Salem Unit 1, a 4-loop The modifications consisted of altering the control Westinghouse PWR. The examinations showed logic to allow adjustments in the auxiliary that the nozzle weld had a number of unacceptable feedwater flow to match the makeup requirements.
linear indications (PSE&G 1992). Radiographic The fluctuations in the flow rate were reduced by exammation of the second weld, a pipe-to-pipe introducing the cold feedwater at a uniform rate weld, revealed no unacceptable indications.
between about 380 to 950 t/ min (100 to 250 gpm)
Results of volumetric examinations of two other to maintair. the steam generator water level.
steam generator feedwater nozzle welds indicated that the expeder fittings had linear indications on Stratified flow will develop in the feedwater nozzle both the nonie-to-expander and expander-to-and feedring if the flow rate to each steam elbow welds. No unacceptable indications were generator is less than about 1500 t/ min (400 gpm).
found in the other welds. There was visual Thus, there will be only one such stratification observction of wall thinning in the thermal sleeve, cycle upon going to hot star.dby with the present resulting in larger-than-design gaps between the operation, whereas there were approximately 9 sleeve and the nozzle on two steam generators.
cycles / day during the operation before the Minor pitting and metal loss in the area modifications were made in Units 2 and 3.
immediately upstream of the thermal sleeve were found in another nozzle. No degradation had been Repair and Mitigation. The cracks were previously observed in the area since the 1979-80 repaired by excavation of the flaws and weld repair repairs.
of both 45-degree elbow areas. The inservice inspection programs for both Units 2 and 3 were The area was examined by UT, visual inspection, augmented to require automatic UT of the nozzle and liquid penetrant methods. New reducing spool areas, including the safe ends and the 45-degree pieces were installed in all four steam generators, NUREG/CR-6456 62
FIELD EXPERIENCE l
the indications were removed by excavation and foodwater nozzle and an antistratification device in blending with the surrounding area. The deepest the main feedwater nozzle.
I excavation was 2.1 mni(0.083 in.). He licensee planned to examine the repair area at the next Indications have been found in the steam generator refueling outage. Similar examinations on Salem feedwater nozzles of a Belgian PWR since 1985 Unit 2 revealed no unacceptable indications.
(Westinghouse Model 44). Despite repair by grinding [ crack depth less than 0.5-mm (0.02-in.)]
One steam generator nozzle at the Haddam Neck the cracking recurred in subsequent annual plant (Westinghouse 4-loop) was inspected during inspections.' He ultrasonic inservice inspection the 1993 refueling outage. RT showed two method in accord.r ;e with ASME Section XI was circumferential cracks 648-mm (25.5-in.) and used. Water hammer damage to thermal sleeves 44.5-mm (1.75 in.) long,38 mm (1.5 in.) from the was also found, and the thermal sleeves were weld centerline in the nozzle base metal replaced.
(McBrearty 1993).
The nozzles had been examined in 1979-80 in response to Bulletin 79-After Bulletin 79-13 was issued, cracking was
- 13. The weld areas of the other three steam found in the feedwater nozzle-to pipe weld area of generators were subsequently examined.
two Swiss PWRs.'
All four welds showed Circumferential cracks were detected in two other extensive circumferential fatigue cracking, nozzles, but the fourth one appeared to be free of originating at the inside and generally less than defects. All defects were located in the lower 180-2-mm (0.08-in.) deep. However, at one location degree portion of the nozzles. The radiography the crack depth was two-thirds through the 20-mm results from the 1979-80 inspection were reviewed,
(-0.8-in.) thick wall. Small fatigue cracks were and it was determined that the 648-mm (25.5-in.)
also found in the base metal. All four nozzles were long crack appeared previously, but had grown replaced, and auxiliary feedwater was routed longer by about i14 mm (4.5 in.). To gain access directly to the steam generators through separate to the defects, the three elbows were removed.
nozzles. At least some of these steam generators Using the tip diffraction UT method, the estimated have been replaced with new steam generators crack depths for the three affected steam generators designed by Framatome.
were as follows: 6.4,4.8, and 3.2 mm (1/4,3/16, 1/8 in.). The cracks were removed by grinding, in some German steam generators designed by and their removal was verified by radiography.
Siemens/KWU, cracks have been detected both in the pipe-to-nozzle welds and the thermal sleeve-to-6.3.5 Non-US Plants nozzle welds at the upstream end of the thermal sleeve. The damaged sections were replaced with No leakage has occurred nor cracking detected by a new design that includes a transition piece, so radiography in the French steam generators that there are now two welds instead ofone. Other 2
(Westinghouse-type PWRs).8 However, magnetic mitigation actions taken by Siemens/KWU are:
particle inspections and crack micrography using a scanning electron microscope were carried out on Installation of a destratification pot on the s:mples removed from several steam generators feedwater inlet inside the steam generator, that were being replaced. Small fatigue cracks were observed in the weld bead to heat-affected Distribution of auxiliary feedwater through a zone vicinity. UT is currently being evaluated as spraying device, located in the feedwater pipe an alternate inspection method. New and replace-upstream of the feedwater nozzle, which mixes ment steam generators have a separate auxiliary the cold feedwater with the hot water present in the horizontal pipe,
' information provided by the corre,.ending country
- Information received from P.4. Meyer, via OECD Nuclear Energy Agency.
Siemens/KWU.
l FIELD EXPERIENCE Installation of a recirculation loop to mix Surry Unit 2 in 1986 (Virginia Power 1987). A blowdown flow back into the feedwater line pressure pulse caused the ultimate rupture of upstream of the feedwater nozzle, feedwater piping already significantly degraded by flow-accelerated corrosion at both plants. In Procedural changes to allow continuous neither the Trojan nor Surry case was there a leak feeding to reduce the number of auxiliary or any other warning signs indicating incipient feedwater actuations, and failure As there had been little or no inservice inspection of the majority of the feedwater system New steam generators designed with a piping at these plants, because such inspections destratification loop (see Section 9.1).
were not required by Section XI of the ASME Code, the extent of wall thinning was not known Bere have been no reports of cracking in Japanese before the ruptures occurred. This situation has PWR main feedwater piping since a single incident changed, and all U.S. utilities are now inspecting in 1974 because heated feedwater is supplied to their feedwater piping. He Trojan and Surry Unit steam generators at plant startup time. The 2 failure events are discussed in the next few feedwater is heated by the feedwater heaters in the paragraphs.
deaerator. No cracking has been found at the feedwater nozzles of Finnish or Dutch PWRs.
A main feedwater isolation following a turbine trip at the Trojan plant produced a pressure pulse that 6.4 Flow-Accelerated Corrosion reached a maximum total pressure of approxi-mately 6.0 MPa (875 psig) in the heater drain and Flow-accelerated corrosion has damaged both feedwater system (Stoller 1985). He pressure large and small-diameter feedwater piping, and surge ruptured a 368 mm (14.5-in.) diameter feedrings and J-tubes in the ster.m generators.
section of SA 106 Grade B carbon steel pipe in the Some of the events associated with this damage are feedwater heater drain pump discharge piping and released a steam-water mixture into the turbine discussed here.
Flow-accelerated corrosion depends on several parameters, including water building. He system flow velocity was 6.1 to 7.3 m/s (20 to 24 fVs), and the normal operating chemistry (pH), flow velocity, geometry, temperature, and materials.
Most of these Pressure and temperature at the time of the break parameters have been identified for each event.
were about 3.1 MPa (450 psig) and 177'C The leading edges of the thermal sleeves have (350*F), respectively. The ruptured portion of the P P ng section had been thinned from a nominal Ii experienced thinning possibly caused by flow.
accelerated corrosion and are also discussed here.
thickness of 9.5 to about 2.5 mm (0.375 to about Section 6.4.1 briefly describes the field experience 0.098 in.). Some of the thinning may have oc-related to flow-accelerated corrosion of the curred during rupture. One worker received first-
.md second-degree burns from the feedwater piping, particularly the area inside containment. He field experience involving flow.
high-temperature fluid. Before this rupture, it was accelerated corrosion of feedrings, J tubes, and believed that only piping carrying two-phase fluid thermal sleeves in Diablo Canyon Units 1 and 2, was susceptible to flow-accelerated corrosion and San Onofre Units 2 and 3, and Arkansas Nuclear was, therefore, inspected in service. Because the One Unit 2 is discussed in Section 6.4.2. Relevant ruptured drain pipe carried single-phase fluid, it experience associated with non US PWR and was not inspected. As a result of this failure, the CANDU plants is discussed in Section 6.4.3.
entire secondary system of the Trojan plant was evaluated to identify the sites susceptible to flow-8.4.1 Feedwater Piping accelerated corrosion, and then a sample of the sites was included in the inspection program and Flow-accelerated corrosion caused the rupture of subsequently inspected by ultrasonic examination.
the feedwater piping outside the containment at Repair and replacement of the damaged sections of P P ng were performed as necessary.
ii both the Trojan plant in 1985 (Stoller 1985) and NUREG/CR 6456 64
i FIELD EXPERIENCE l
A main steam isolation valve failed closed at Surry 13 mm (0.5-in.). The reactor was at full power and Unit 2 in December,1986, and the resulting the feedwater was single phase, with a flow veloc-increased pressure in the steam generator collapsed ity of about 4.3 m/s (14 ft/s), a pH levelin the the voids in the water. This caused the system range of 8.8 to 9.2, an oxygen content of about pressure to surge beyond the normal operating 4 ppb, and a coolant temperature and pressure of pressure and led to a catastrophic failure of a 90-approximately 188'C (370*F) and 3.1 MPa degree carbon steel (SA-234 Grade WPB) elbow in (450 psig), respectively. Ammonia was used for the suction line to the main feed pump, as shown the feedwater treatment. The examination of the in Figure 36. The diameter of the elbow was ruptured elbow showed that the wall thinning was 460 mm (18 in.),and the design thickness was relatively uniform except in some local areas.
i 4-4
-I m
My.
hN] ][Q EE (a) Rupture lines in intact pipe.
l l
i 1
I i
m
^
g.:
'g
- r p
(b) Pipe after rupture.
Figure 36. Surry feedwater pipe rupture caused by flow-accelerated corrosion (Jonas 1988).
4 65 NUREG/CR-6456
FIELD EXPERIENCE The wall thickness of the elbow was reduced from However, in one instance, it was later found that a nominal 13 mm (0.5 in.) to 0.38-to 1.22-mm certain damage was not caused by flow-accelerated (0.015-to 0.048-in.) in small local areas and to 2.3 corrosion. In the June 1987 outage at the Trojan mm (0.09 in.) in larger areas (Czajkowski 1987, plant, significant flow-accelerated corrosion-Virginia Power 1987). Eight workers were bumed induced wall thinning of two safety related by flashing feedwater, four of whom subsequently portions of the feedwater piping inside the died. The flashing feedwater interacted with and containment was initially reported. The pipe wall disrupted the fire protection, security, and thickness would have reached the minimum electrical distribution systems (USNRC 1988b).
thickness required by the design code during the next refueling cycle. The wall thinning was in a As a result of the Surry accident, the NRC staff horizontal and a vertical portion of the piping not asked that all utilities with operating nuclear power considered susceptible to flow-accelerated plants inspect their high-energy carbon steel piping corrosion damage and not included in the (USNRC 1987a). The degraded components, inspection program, because these portions were at fittings, and straight runs in the feed-least seven pipe diameters away from an elbow or water-condensate systems identified in that inspec-any other fittings that could cause turbulence tion and reported to the NRC are listed in Table 7 (USNRC 1987a). However, after further analysis, (USNRC 1988a). A summary of the inspection the plant operator, Portland General Electric, programs and inspection results has been compiled determined that the damage to these horizontal and by EPRI (Mattu et al.1988).
vertical portions of the piping was minor and judged not caused by flow-accelerated corrosion New piping was installed at several locations in the but rather from an initial manufacturing defect.'
Surry-2 feedwater system as a result of the pipe No other utility has reported flow-accelerated break. During the September 1988 outage, an corrosion damage in straight piping away from elbow (installed in 1987) on the suction side ofone discontinuities.
of the main feedwater pumps was found to have lost 20% ofits 13 mm (0.5-in.) thick wall in 1.2 Further inspection of all the high-energy carbon years. The NRC preliminarily concluded that this steel piping at Trojan revealed wall thinning at abnormally high rate of wall thinning may have about 30 additional sites,10 of which were in the coincided with a reduction in feedwater dissolved safety-related portion of the feedwater system, oxygen concentration (USNRC 1988b). However, whereas the remaining were in the nonsafety-Virginia Power disagrees that the oxygen content related piping. As a result, elbows and straight was a major contributor. One explanation is that portions of piping at several of these sites were the accelerated thinning could have been replaced. High flow velocity [6.9 m/s (22.6 ft/s))
aggravated by feedwater flowing into the steam combined with other operating factors [4 ppb generators that bypassed the feedwater heaters.
oxygen content, pit of 9.0, and an operating This would have reduced the water temperature temperature of 230*C (445 F)] might be the cause from its normal 190*C (370*F) to about 150*C of the thinning. Failure analysis indicated that (300*F)(Sperber 1988). Later in the September flow-accelerated corrosion coupled with cavitation 1988 outage, Virginia Power replaced a total of caused by severe flow conditions at the pump 125 piping segments with steel piping containing discharge elbows caused the wall thinning at 2.5 wt% chromium.
Trojan (Wu 1989).
'lhe mptures at Trojan in 1985 and Surry Unit 2 in 1986 were located outside the containment and upstream of the last check valve before the feedwater enters the steam generator. In contrast,
' Chexal, B.1987. "EPRI CilEC Computer there has also been evidence of flow-accelerated Program? summary Report ofthe First User Group corrosion in the safety-related pottion of the Meetmgfor she EPRI CHEC Computer Program. st Louis.
feedwater piping inside the containment.
Missouri. septer.ber 22.
FIELD EXPERIENCE Table 7. PWR plants with pipe wall thinning in the feedwater-condensate systems (USNRC 1988a).
Plant Unit Commercial Degraded components operation (fittings, straight runs)
Arkansas Nuclear One i
August 1974 Elbows, drain pump discharge piping Arkansas Nuclear One 2
December 1978 Undefined Calvert Cliffs October 1974 Elbows, reducers, straight runs Calvert Cliffs 2
November 1976 Elbows, reducers, straight runs Callaway October 1984 Recirculation line vbows Diablo Canyon i
April 1984 ElbtrM. straight runs Diablo Canyon 2
August 1985 Elbows, Y D C. Cook 2
March 1978 Elbows fort Calhoun August 1973 Elbows, straight run lladdam Neck July 1967 Recirculation line Millstone 2
October 1975 Elbows, heater vent piping North Anna l
April 1978 Elbows, straight runs North Anna 2
June 1980 Elbows, straight runs 11 it Robinson 2
September 1970 Recirculation lines September 1974 Straight runs downstream of Rancho Seco feedwater isolation valves or main feedwater pumps minimum flow valves San Onofre i
June 1967 Reducers, heater drain piping San Onofre 2
July 1982 lleater drain piping San Onofre 3
August 1983 lleater drain piping
)
Salem i
December 1976 Recirculation line Salem 2
August 1980 Recirculation line Shearon llarris October 1986 Recirculation line i
Surry 1
July 1972 Fittings Surry 2
March 1973 rittings Sequoyah i
July 1980 Elbows, straight runt Sequoyah 2
November 1981 Elbows 1rojan December 1975 Elbows, reducers, straight runs Turkey Point 3
October 1972 Feedwster pump suction line fittings 67 NUREG/CR-6456
FIELD EXPERIENCE l
A potential generic problem was discovered at corrosion has caused significant wall thinning of Catawba Unit 2 in 1991 that may affect all the the feedwater control valve bypass line at both the Westinghouse Model D4, DS, and E steam San Onofre and Diablo Canyon plants. It was generators in which a portion of the main surprising to find significant wall thinning and feedwater is diverted to the auxiliary feedwater failures of the stanup feedwater system piping at nozzle via the preheater bypass line, as shown in both the Wolf Creek and Callaway plants because Figure 9 (Stoller 1992a, USNRC 1992). The fluid these systems are used for a very short time period velocity in the 102 mm (4-in.) diameter preheater during startup. Investigation of these failure bypass line and the connecting auxiliary feedwater showed that the cause was the flow resulting from j
t line was in the range of 9 to 11 m/s (30 to 35 ft/s).
the leaking valves on the piping (Chexal et al.
j The licensee detected several locations in this 1996).
piping that were at or near the minimum required J
wall thickness. Examinations revealed that single-EPRI has developed the computer code CHEC-
]
phase flow-accelerated corrosion had reduced the WORKS (Chexal-Horwitz Engineering-Corrosion nominal 8.56-mm (0.337-in.) wall thickness to Workstation) for managing How-accelerated 4.70-mm (0.185 in.) in only four operating cycles.
corrosion of nuclear power plant piping. This This implies a flow-accelerated corrosion rate of program has capabilities for estimating paran.eters about 1.0 mm/ cycle (0.04 in/ cycle). If the (such as local water chemistry and flow rate) that preheater bypass line had ruptured, the break affect corrosion rates, and for predicting corrosion j
would not have been isolable and would have rates and helping to select inspection locations.
i resulted in the steam generator coolant being The computer code is based on data from France, released outside containment. Over 27 m (90 ft) of England, Germany, and the U.S. The code has piping was replaced at Catawba Unit 2.
been validated using other U.S. plant data. The comparison between the predicted results and 4
Recent field experience related to pipe ruptures measurements show that the code predicts the caused by wall thinning has identified additional How-accelerated corrosion rates within *50%. The sites susceptible to flow-accelerated corrosion main sources of uncertainties are associated with damage that were not included in the plants' moni-the original thickness and thickness profile of the toring programs (USNRC 1991a). This experience piping components, trace amounts of alloy content demonstrates that the flow-accelerated corrosion is in the piping material, actual number of hours of a complex, multi-parameter phenomena, and the operation, plant chemistry
- history, and susceptibility of a given site cannot be determined discontinuities on the inside surface of the piping.
by considering only a few parameters. Generally, j
the flow-accelerated corrosion monitoring pro-All PWR and BWR plants in the U.S. use the grams concentrate on inspection of pipe eroows CHECWORKS code (or its predecessor code and tee fittings, the sites where local high veloci-CHECMATE) for estimating the flow-accelerated tie.4 may be present, llowever, flow-accelerated corrosion rates. This code is also used by many corrosion has caused rupture at other feedwater fossil plants, by the U.S. Navy, and by several piping sites, such as in the flange of a flow measur-overseas utilities. The code has been used for ing device downstream of an orifice at Loviisa identifying the sites most susceptible to flow-Unit 1 in Finland (discussed in Section 6.4.3) and accelerated corrosion, prioritizing the locations that in the straight portion of a pipe, located immedi-need to be inspected, estimating remaining service ately downstream of a level control valve, at Surry life for each susceptible component, and evaluating Unit I and at Millstone Unit 3. Flow-accelerated the efrectiveness ofdifferent water chemistries and other mitigative actions.
=
FIELD EXPERIENCE 6.4.2 Feedrings and Thermal Sleeves approximately a 0.25-to 0.51-mm (0.010- to 0.020-in.) gap between the sleeve and nozzle to De carbon steel J-tubes and feedrings within the permit free thermal growth in the axial d'rection.
recirculating steam generators have also experi.
Because of the thinning of the leading edge, the enced significant flow-accelerated corrosion-gap between the nozzle and the sleeve increased.
induced wall thinning. A portion of a severely De increase in the gap permits increased bypass thinned J-tube may be separated from the feedring leakage of the colder feedwater into the hot steam and become a loose part and cause damage to the generator coolant present near the nozzle blend i
steam generator tubes, leading to leakage of the radius region. Cracking in a feedwater nozzle primary coolant. Also, spray of relatively cold caused by the bypass leakage has been reported at feedwater from a broken J-tube may be directed only a few PWR plants such as Indian Point 2 onto the steam generator shell, causing thermal shown in Figure 23, but such cracking had been a fatigue damage.
major problem in BWR feedwater nozzles (Snaider 1980). In the replacement steam generators the Carbon steel J-tubes were installed in recirculating leading edge of the thermal sleeve is welded to the steam generators in 1975 to eliminate steam feedwater nozzle.'
generator water hammer events. Several of these J-tubes failed because of flow-accelerated Diablo Canyon Units 1 and 2. Figure 5 corrosion in both the U.S. and non U.S. PWRs shows a typical feedring assembly, which consists (Roarty 1986, Chexal et al.1996). Herefore, of a 406-mm (16-in.) diameter thermal sleeve, a licensees have replaced the original J-tubes with 406 x 406 x 406-mm (16 x 16 x 16-in.) tee fitting, ones made of stainless steel or Alloy 600.
two 406 x 254 mm (16 x 10-in.) eccentric However, concerns persist that the joint design reducers, and two 254-mm (10-in.) serpentine between the new J-tubes and the feedrings may feedrings. All of these subcomponents were have introduced geometric discontinuities on the originally SA-106 Grade B or SA-105 carbon inside surface of the feedrings.
Such steel. The carbon steel J-tubes, installed before discontinuities could generate local feedwater commercial operation began and degraded by flow-turbulence at the base of the J-tube and cause flow-accelerated corrosion, were replaced by ones of the accelerated corrosion damage in the carbon steel same design, but made of Alloy 600, during the feedrings (Shah and MacDonald 1993). Flow-first refueling outage. The original Alloy 600 J-accelerated corrosion of the feedring has been tube design is shown in Figure 37(a). The inserted reported at the Diablo Canyon, Surry, and North end of the J-tube is recessed into the pipe wall to Anna (Westinghouse) units. No original carbon reduce turbulence (Thailer, Dalal, and Goyette steel feedrings have been replaced in any 1995).
Westinghouse unit.
l The feedwater treatment through the first five The discussion in Section 6.3 about feedwater cycles of operation resulted in an average pH of I
nozzle cracking in the Salem I and Diablo Canyon 8.7 to 8.9. During the 6th cycle, the amine was I and 2 plants mentioned the thinning of the changed to ethanolamine at a pH of approximately leading edge of the thermal sleeves, shown in 9.2.
Figure 32. Dermal sleeves at other Westinghouse (Prairie Island and Trojan) and Combustion Extensive examinations during the 6th refueling Engineering (San Onofre and Arkansas Nuclear outage at Unit I using visual and video probe One Unit 2) plants have also experienced thinning.
inspection of the interior of the feedring assembly Wall thinning at this location had not been revealed significant degradation of the carbon reported until recently (after 1989). The damage mechanism is believed to be flow-accelerated 8
corrosion. In the original Westinghouse steam The leading edge of the thermal sleeve is welded to generators, slip-on type thermal sleeves were the nozzle to minimize the possibility of water hammer.
The sleeve is also welded in some German-designed steam tightly fitted in the feedwater nozzle with
)
generators, 69 NUREG/CR-6456
FIELD EXPERIENCE inoonelII8et weld 2 -in. Inoonel J tube Region degraded by n,,[-
now-moo ierated
' V
/I Carbon steel feeddng (a) Original design Weld 2-in. Inconel J tube 4 x 2 -in. Inoonet reducer j
V l
Inoonel full penetration weld Carbon steel feedring l
une ooes (b) Modified design Figure 37. Changes in the Diablo Canyon J tube design to mitigate flow-accelerated corrosion damage (Thailer, Dalal, and Goyette 1995). Copyright American Society of Mechanical Engineers; reprinted with pennission.
NUREG/CR-6456 70.
FIELD EXPERIENCE steel base metal surrounding the Alloy 600 J-tubes The feedwater velocity is approximately 5 m/s in the tee fitting and the reducer sections of the (17 ft/s) through the sleeves to the feedrings; feedring assembly. (Some Diablo Canyon however, a small portion of the feedwater leaks feedrings have a J tube in the reducer section, through the gaps between the thermal sleeves and whereas others, as shown in Figure 5, do not.) The the nozzles at a velocity of 6.75 m/s (23 fl/s).
degradation was caused by flow-accelerated Inspection of the Unit I thermal sleeves showed corrosion Similar degradation was noted at the significant degradation of the leading edges and Diablo Canyon Unit 2 J-tube connections.
the outside surfaces. Similar but less severe However, the Alloy 600 J-tubes were unaffected damage was observed on the Unit 2 thermal by the corrosion mechanism. The carbon steel sleeves during the 5th refueling outage.
region of the feedring surrounding the J-tubes exhibited signifi: ant degradation if the chromium ne licensee believes that direct impingement of content of the steel was less than 0.1 wt%.
the feedwater onto the leading edges of the thermal Spectrographic analysis was used to determine the sleeves and the high fluid velocities are the major chromium content. Although the degraded region contributors to the degradation. Observations of extended around the entire periphery of the J-wear pattems indicate that the knife-edge tubes, the magnitude and extent of the damage was appearance of the leading edges have been caused greatest in regions located 90 degrees to the by erosion, whereas the thinning of the thermal direction of flow in the feedring. In some cases, sleeve walls has been caused by flow-accelerated the damaged region extended completely through corrosion damaging the outside surfaces. As a the wall to the surface of the Alloy 600 fillet weld.
result, in one steam generator, the original Ses en through-wall holes ranging in size from a 1.25-mm (0.05 in.) gap between the thermal sleeve pin hole to about 12.7 mm (0.5 in.) across were and the nozzle was increased to a maximum of 5 observed adjacent to the plugs installed on the mm (0.20 in.), and the degradation on the outside underside of the feedring, though the degradation surface of the thermal sleeve extended for at least was not as severe as that on the areas surrounding 76 mm (3 in.) from the leading edge. The degree the J-tubes. This is believed to have been caused of damage was noted to be a function of the by defective welds.
chromium content of the base metal.'
Ten J tubes in Unit I and eight in Unit 2 have A single-piece tuning fork forging (shown in recently been replaced with a modified design, Figure 38), sometimes referred to as a transition shown in Figure 37(b). In the modified design, piece, was installed in Units 1 and 2 in 1994. The 102 x 51-mm (4 x 2 in.) Schedule 160 reducers (to forging is made of SA-508 Class 2 low-alloy steel reduce flow velocity) are welded to the feedring having 0.25 to 0.45 wt% chromium, which tnd ground flush with the feedring inside diameter provides improved resistance to flow-accelerated (to reduce turbulence). Then the J-tubes are corrosion. The ends of the transition piece are welded to the reducers. The J tube length was welded to the feedwater piping and nozzle. The shortened to maintain the original height above the downstream end of the forging supports a sealing feedring and the discharge orientation relative to device consisting of a spring and a piston ring the feedring. Similar damage to the carbon steel installed in a groove to provide a tight seal between feedring at the entrance to the J-tubes was reported the forging and a thin liner installed inside the at Sequoyah Unit 1 in the early 1980s and a similar original thermal sleeve. The chromium content repair was performed (Chexal et al.1996).
Thermal Sleeves-The leading edges and the outside surfaces of the thermal sleeves are exposed to high velocity flow during operation.
Goyette, L " Steam Generator Feedring Experience at Diablo Canyon Power Plant," presented at the Eleventh EPRICHEC/CHECAfATE Users Group Ateeting (CHUG).
Jackson Hole. HT,.;une 2,199I 71 NUREG/CR-6456
FIELD EXPERIENCE s
aaung i..w.r neuw j
k a sung e.
., m g 4%8
~
e Gamma plug pad
-j
'c@
j d,e s #?@sL h
,, ' +
m i
igju.
Sgo Lin.r Single e
3,,j tuning k Existin forging sleev.g thermal MOS0013 Figure 38. A redesigned feedwater nozzle-to-piping connection at Diablo Canyon Units 1 and 2 (Cofie et al.
1994). A single piece forging protects the thermal sleeve from flow-accelerated corrosion damage and feedwater nozzle weld from thermal fatigue damage. Copyright American Society of Mechanical Engineers; reprinted with permission.
of the liner material was greater than 0.1 wt%. De original distribution box was left in place, along counterbore was eliminated, which reduces the with about a 229-mm (9-in.)section ofSchedule thermal stresses and the potential for fatigue 40 material (called a pup) welded to each side of cracking at the pipe-to-nozzlejunction. Thus the the distribution box, as shown in Figure 14. He forging protects the leading edge of the thermal diameter of the discharge elbows was increased sleeve from flow-accelerated corrosion and from 38 to 63 mm (1.5 to 3.5 in.), and tee-shaped protects the feedwater piping and nozzle from vent assemblies were installed on the distribution fatigue damage, box. On Unit 2 only, the feedring supports were modified, including slotting the U-bolt holes so the San Onofre Unit 3. During the early part of feedrings could move through the supports. This the March 1981 startup of Unit 2, a test resulted in allowed the feedrings to move more freely in Unit a partial vacuum within both halves of the feedring 2 than in Unit 3.
of one steam generator; the feedring collapsed because of the inadequate flow area of the In 1990, several pieces of carbon steel debris were discharge elbows (whose function is the same as found on the secondary side ofthe tubesheet ofone that ofJ-tubes) and the relatively thin Schedule 40 of the San Onofre Unit 3 plant steam generators.
feedring wall.
The collapsed feedring was Further inspection of the two steam generators replaced in 1982 with Schedule 120 pipe. The revealed significant material missing from the pup NUREG/CR-6456 72
FIELD EXPERIENCE on one steam generator, surface cracks in the heat.
The Schedule 40 pup material was replaced with ofrected zone at the toe of the weld that joins the Schedule 120 piping material, the distribution box-pup and distribution box, local erosion of the pups feedring weld configuration was replaced with (in one case producing a through-wall gee nion),
weld-o-let forgings to reduce the stress concentra-wall thinning of vent pipe (caused by both erosion tions at the heat-affected zone, the vents were and flow-accelerated corrosion), missing tops of removed from the design, weld buildup was used the vent tees and deformation of many of the to repair local thinning, the feedring supports were U-bolt supports (USNRC_1991b,. Martin et al.
modified to provide greater flexibility (slotted 1990). Examination of Unit 2 found no missing holes enlarged), and stronger (SA-193 Grade material from the distribution box-feed-ring B7M) U-bolts were used. Combustion Engineer-junction (pups), but all the pups had cracked ing issued CE Information Bulletin 90-04 recom-around the perimeter, both legs of one U-bolt were mending that its client utilities perform a baseline fractured, all the pups were somewhat thinned, and inspection during their next refueling outage to many of the U-bolts were stretched and defonned.
determine if wall thinning was occurring (USNRC None of this type ofdamage had been reported at 1991b).
other Combustion Engineering plants, which do not have pups in their designs, though Calvert The Unit 2 changes included a new distribution Cliffs had experienced local thinning in the box made of Alloy 690, new J-tubes made of 21/4 distribution box.
Cr-1Mo steel (P22-grade), and a pup piece made from SA-106 Grade B, Schedule 120, 305-mm The U bolts deformed because of excessive loads (12-in.) pipe, but no material change was made for resulting from expansion-contraction of the the SA-106 Grade B feedring. A monitoring feedring every time cold auxiliary feedwater (or program will be implemented to observe the metal cold main feedwater) was injected into the thinning of the distribution box and the pups after feedring. Stratified flow develops in the feedring approximately every 5 years of operation, when the flow rate into the steam generator is less than approximately 1500 t/ min (400 gpm). A Arkansas Nuclear One Unit 2. Thinning marked temperature change at the interface of the leading edge of the thermal sleeve was between the hot and cold fluids will persist at the found in Steam Generator A of the Arkansas pup for about 30 sec. Thermal analyses show that Nuclear One Unit 2 in 1991, an older vintage the fluid interface elevation would be relatively Combustion Engineering plant (USNRC 1991d).
constant at about the 4 and 8 o' clock elevation.
The thinning occurred at the 12 and 7 o' clock Calculations show that the feedrmg will bend locations, and resulted in an open gap between the downward as a result of the stratified flow, and sleeve and nozzle which extended along the will also contract and bend inward because of circumference as much as 25.4-mm (1-in.) between overall cooling.
the 12 and 7 o' clock locations where the thermal sleeve meets the feedwater distribution box (that is, The cracks first initiated at the 4 o' clock position at at the downstream end of the thermal sleeve). A the toe of the weld thatjoins the distribution box to clamp to close the gap was chosen as a temporary the pup. The cracks then propagated through the repair, similar to the arrangement used at the i
wall to produce very high local flow velocities as Calvert Cliffs Station in 1989.
the water flowed through the crack, eroding the wall. Finally, the wall became so thin that rupture occurred and pieces were torn away by feedwater exiting the feedring.
i 73 NUREG/CR-6456
FIELD EXPERIENCE 6.4.3 Non US Plants containment with a very high quality piping. Japan and Switzerland have reported an absence of Finland, Belgiunt, and Spain have reported flow-single-phase flow-accelerated corrosion damage in accelerated corrcsion of feedwater lines in their their PWR plants.
PWRs, and Canada has reported similar corrosion in its CANDU reactors. The Netherlands has not in May 1990, a feedwater pipe at Loviisa Unit I reported flow-accelerated corrosion in its PWR, ruptured when the plant was operating at full possibly because its feedwater piping material is power. The failure location was on the pressure low-alloy steel. However, to address the problem side of a feedwater pump, immediately down-of postulated pipe rupture, the Netherlands is stream of the flowmeter flange, as shown in planning to replace the feedwater piping inside the Figure 39. The rupture was caused by a pressure l
,y Flow enfice U<g
(
a enaQ Fa ure ocatio
.l, i
~:v afkm.
/
e
- b$$hkkij kgt{hf l
Flow 4.....................................................................................................
Flow accelerated IMMjd corrosion areas
%v Piping
. ;;. pl tm qMf jr p (St 45.8 material)
$og
- x. 4, Material Composition Material C
Mn Si S
P Cr Ni Cu Type
(%)
(%)
(%)
(%)
(%)
(%)
(%)
(%)
C 22 N 0.18 0.50 0.32 0.022 0.012 St 45.8 0.20 0.89 0.26 0.018 0.029 CT 20 0.19 0.57 0.25 0.019 0.026 0.20 0.30 0.30 C174WHT 19$41 Figure 39. Loviisa Unit I feedwater pipe rupture caused by flow-accelerated corrosion (USNRC 1991a).
' Information presented in this section has been provided by the corresponding country via OECD Nuclear Energy Ageracy.
FIELD EXPERIENCE wave rest.lting from a change in the operating significant, but no important ftmetions were lost.
speed of the feedwater pump. The normal operat-ne utility had an inspection program for the ing pressure was 7 MPa (1000 psi). Investigations components exposed to single-phase flow, but only have indicated that significant wall thinning had elbows and tee fittings were inspected. After the taken place over the complete circumference of the rupture, the utility inspected the flow-orifice flange, and its wall had thinned to 1 mm flanges at both units and found that 9 of the 10 (0.039 in.) from the original thickness of more than flanges were below the minimum thickness 18 mm (0.7 in.). A 0.5 m (20-in.) long pipe requirements. Unit 2 was shut down and all the section, attached to the downstream end of the damaged flanges were repaired; both units were flange, had also experienced wall thinning along restarted in June 1990.
its entire circumference; its thickness was reduced from the original value of 18 mm (0.7 in.) to a A similar event occurred at Loviisa Unit 2 three thickness in the range of 5 to 10 mm (0.195 to years later, in May 1993, when the plant was 0.390 in.).
operating at full power. The failure location was on the pressure side of a feedwater pump, ne secondary water chemistry played an immediately behind the check valve. The original important role in causing the wall thinning; Loviisa thickness, which was bearing the normal operating Unit I used neutral water chemistry (cold pH = 7) pressure of 7 MPa (1,000 psi), was reduced by with k+ dissolved oxygen.' This water chemistry flow-accelerated corrosion to 1.5 mm (0.06 in.).
was used for protecting the austenitic stainless steel The rupture occurred when the feedwater pump steam generator tubes from stress corrosion was not in operation but was being warmed for cracking, and the cooper-alloy condenser tubes startup by recirculation. The leak through the from corrosion damage. Because of this failure, ruptured piping caused local damage to nearby the utility is planning to replace the condenser electrical equipment and instrumentation, but tubes at both Loviisa units with tubes made of again, no important functions were lost. The plant titanium or austenitic stainless steel. Then the resumed operation after necessary repairs were utility will reevaluate the water chemistry. Two made (NEA 1993).
other reasons for the significant wall thinning include:
(a) the flow orifice introduced a In September 1987, a condensate line from a discontinuity on the inside surface of the piping moisture separator-reheater at Doel Unit I in and produced high velocities downstream of the Belgium ruptured, causing a second degree burn to orifice, and (b) the alloy content in the flange and an operator. As a result, a more comprehensive the 0.5-m long section of the pipe was low. In inspection pr ogram for monitoring wall thinning of contrast, the piping downstream of the 0.5-m long feedwater, steam, and condensate system piping pipe section did not exhibit any wall thinning, was developed. This inspection program is possibly because it contained more alloying described in Section 8.4.2.
elements (0.20 wt% Cr,0.30 wt% Ni, and 0.3 wt%
Cu). This experience is similar to that at Diablo Flow-accelerated corrosion has caused several Canyon, where the components containing more failures of the feedwater piping in Spanish nuclear than 0.1 wt% Cr did not experience wall thinning power plants. One example was a rupture of the (USNRC 1991a, NEA 1990).
suction side of the main feedwater piping in the Garona nuclear power plant, a BWR plant, in The rupture at Loviisa Unit I took place in the 1989. The primary cause of this failure was a low turbine building, releasing nonradioactive water oxygen content (-10 ppb).2 In other cases, the and steam into the turbine building. The local most affected areas were located near the pump damage to nearby cables and small piping was discharge and control valves in the feedwater 2
' V. N. Shah, private communication mth V. N. Shah, private communication with D. Munson, EPRI, Palo Alto, California, June 1996.
D. Munson, EPRI, Palo Alto, California, June 1996.
FIELD EXPERIENCE system. Many carbon steel pipes and fittings that The leading edge of the sleeve attached to the were susceptible to flow-accelerated corrosion feedring could possibly experience wall thinning.
were replaced with low-alloy steel components.
It is also not known whether leakage can take place through the connection; such leakage could pose Flow accelerated corrosion has also caused signifi-potential concerns for water hammer and for cant wall thinning of redeers located downstream fatigue damage to the feedwater nozzle bore and of the main boiler level control valves in CANDU blend radius.
reactors at the Bruce B site in Canada. In 1990, significant wall thinning was detected at the small 6.5 Steam Generator Water end of the outlet reducer in Unit 5. The thickness was reduced from 96% in 1987 to 65% of the Hammer Damage nominal thickness. The code-allowed thickness was 74% of the nominal thickness. Inspection of The portion of the main and auxiliary feedwater j
P ping adj,acent to the feedwater nozzle (shown m i
all other similar reducers in Unit 5, except the R ure 1) can be affected by a steam generator 8
spare ones, revealed that the wall thickness was reduced below the minimum allowable. The wall water hammer, which is caused by the collapse of thinning was uniform along the circumference.
a steam bubble. Steam generator water hammers Inspection of the reducers in Unit 6, which was have typically occurred m several U.S. and placed in service at about the same time as Unit 5, European PWR plants with top-feed steam did not reveal any significant wall thinning.
8'"*'ators following uncoveryof the feedrings. In addit. ion, collapse of steam bubbles has caused a However, inspection of the reducers in Units 7 and 8, which were placed in service after Unit 5 and few water hammers m auxiliary feedwater lines of expected to be in better condition, revealed signifi-both top-feed and preheat steam generators. The cant wall thinning, which necessitated the replace-steam generator water hammer phenomenon,s i
described next.
Then, the field experience ment of the reducers at those units. A possible reason is the high flow velocity through the boiler associated with both the main and the auxiliary level control valve and the outlet reducer. Also,
[eedwater systems of the top-feed steam generators the laboratory analysis of the samples removed is presented and selected water hammer events are from the reducers showed that the chromium briefly described. Finally, the field experience content in the reducers from Unit 6 was about four associated with the auxiliary feedwater lines for times higher than that in the reducers in Unit 5, a both the top-feed and preheat steam generators is possible explanation for no wall thinning of the Presented.
reducers in Unit 6.
6.5.1 Steam Generator Water Hammer 1
Phenomenon No country besides the U.S. has reported any wall thinning of feedrings, J-tubes, or thermal sleeves in response to our questions. However, wall-thinning The sequence ofevents presented in Figure 40 has of carbon steel J-tubes had occurred in several non-generally been accepted as a cause for steam U.S. PWRs, and they were replaced typically with generator water hammer occurrences (Han and ones made from nickel-based materials. France Anderson 1982). This mechanism has been has reported that their inspection techniques are demonstrated with experiments reported by Block not sufliciently advanced to detect thinning of the et al. (1976). Initially, the upper portions of the thermal sleeve, whose design is somewhat different drained feedring and the horizontal length of the than the one in a typical Westinghouse plant. The adjacent feedwater piping are filled with steam.
feedwater pipe includes a monolithic thermal When auxiliary feedwater flow, which is usually sleeve that mates with a sleeve fastened to the highly subcooled, is resumed, it enters the feedring. The connection between the two sleeves horizontal run of the feedwater piping into the allows them to slide freely, relative to each cher, feedring and flows under the steam blanket as but the details of the connection are not available.
shown in Figure 40(a). If the flow rate is high, interaction forces between the steam and water can NUREG/CR-6456 76
FIELD EXPERIENCE Transient
/ hydraulic
[ Condensation of steam Feedpipe waves N
Feedring
,s S//k / / / =
~ steam ~
f,..[~ 4 t-c. y_. ~.= _--- %g- =.s
,,2 Water f NpSteam Steam flow j
~~
generator.
mixing "0"I' Subcooled water l
(a) Possible Steam Water Mixing Phenomena in the Feed System pSteam N
generator Feedpipe nonle Feedring
[ Trapped steam vold
__._ q
(,_~~ - *Subcooled watorL c; ~
.a E __
N Water Steam
~"
slug formbtlon (b) Possible Trapping of a Steam Vold l
Steam from vents in some systems Water slug Steam,
moves rapidly generator nonle into vold Feedpipe Feedring g('
' * "'*""'*1Egtt=p SIug Steam
~~ ~
~
~~:
bullds up Region near and
~
steam generator scoops up water at lead edge pressure (c) Possible Slug Acceleration into Void impact Feedpipe Feedring Steam 1
D~. ~r.
-~
8I"
Pressure NSteam Y ~
=
waves travel generator
~
through system nouie (d) Possible Water Slug impact INEL 21076' Figure 40. Sequence of events leading to water hammer (Han and Anderson 1982).
FIELD EXPERIENCE create enough turbulence to form a slug, as shown show that the reported occurrences of steam in Figure 40(b). The slug seals the main feedwater generator water hammer events dramatically line and thus prevents the normal flow of steam decreased after 1979; 28 events occurred during from the steam generator from replenishing the the 1970s, whereas only 6 occurred during the trapped steam void being condensed by the cold 1980s. The main reason for this decrease is design feedwater layer. As steam is condensed at the modifications and operating procedure changes at steamwater interface, the pressure of the trapped U.S. PWR plants to reduce the occurrence of water steam drops, and, as a result, a slug accelerates into hammer events. These mitigative actions are the void, away from the steam generator, as shown discussed in Section 9.4.
in Figure 40(c). The forces causing acceleration of the slug can be very large and include the steam The original feedrings in the top-feed steam generator pressure, which can be in excess of 1000 generators had bottom-discharge holes, as shown psi, on one side and trapped vapor pressure on the in Figure 41. When the water level in the steam other side. As the condensation continues, the generator dropped below the feedring, the feedring trapped steam pressure decreases and, as a result, became uncovered and drained in 1 to 2 minutes.
the slug achieves a very high velocity. As shown As discussed, the draining of a feedring can lead to in Figure 40(d), when a slug traveling with a high a water hammer. The design of the feedrings was velocity impacts an incoming water column, a eventually moded at most of the operating i
pressure pulse is produced. The velocity ofthe slug Westinghouse and Combustion Engineering units at the time ofimpact depends on the acceleration to mitigate water hammers. The holes in the forces acting on it and the distance it has traveled.
bottom-discharge feedring were plugged, and J-This distance is usually equal to the length of the tubes or discharge elbows were installed on the top horizontal section of the feedwater piping adjacent of the feedring, as shown in Figures 5 and 14.
to the feedwater nozzle. The pressure pulse may be as high as thousands of psi for a severe water Water hammer damage for most of the reported hammer. However, because ofthe incompressibil-events was limited to the plant piping support ity of the water, inelastic expansion of piping sptems. However, in a fe v cases water hammer reduces the amplitude of the pressure wave and a events have resulted in significant damage to the smaller value is transmitted through the system steam generator internals including cracked feed-(Izensen, Rothe, Wallis 1988). The pressure pulse rings and expanded thermal sleeves, and cracking travels throughout the system and could damage and bulging of the main feedwater line. The water the feedring, piping and supports, or valve inter-hammer-induced shocks to the steam generator nals, such as a check valve disk.
instrumentation have led to spikes in the readings (such as differential pressure) that have initiated 6.5.2 Top Feed Steam Generators safety injections in some plants.
There have been about 37 PWR steam generator Ind/an Point Unit 2. The most damaging water hammer events associated with the main and event was at Indian Point Unit 2 in 1973 (USNRC auxiliary feedwater lines of 17 top-feed steam 1980), which resulted in a 180-degree generators, as listed in Table 8.'
Five of these circumferential through-wall crack in a 460-mm events occurred prior to commercial operation, (18 in.) diameter main feedwater line at the whereas the others occurred afterward. The data containment penetration. The water that sprayed from the ruptured pipe caused gross thermal deformation of the metal containment liner near
' Table 8 includes the events reported to the USNRC.
thisjuncture. The water hammer also produced a flow ever, many morc events may not have been reported large bulge in the horizontal run of the main because they involved no safety considerations. Some nuclear industry participants at the 1986 EPRI Water feedwater pipe inside the containment. The crack llammer Workshop estimated that unreported events may was caused by excessive bending stresses, possibly occur at least an order of magnitude more frequently than resulting from dynamic reaction forces at the es ents reported in the databases (Uffer 1987).
support points along the line.
FIELD EXPERIENCE l
Table 8. PWR steam generator water hammer events.'
NSSS Commercial Number 4
Plant Vendor Operation Event Date(s) of Events
{
l Beaver Valley 2 W
l1/87 04/89,92 2
8 Calvert Cliffs I CE 05n5 08/29n4,12/3074,05/12n5' 3
l Calvert Cliffs 2 -
CE 04n7 05/18n6' 1
D.C Cook W
0755 0I/02n6',03/lOn7 2
Haddam Neck W
01/68 (date7)'
- 1 (Conn. Yankee)
Indian Point 2 W
09n3 11/13n3,01/29/74 2
l Maine Yankiec CE 12n2 01/25/83 1
Millstone CE 12n5 05/09n58 1
Palisades CE 12/71 Spring 1981 1-5 San Onofre 1 W
01/68 04/29/72,01/14 7 4,11/21/85 3
4 San Onofre 2 CE 08/83 03/30/81 1
l a
Surry 1 W
12n2 10/0172 1
l Turkey Point 4 W
01/73 06/20n3,01/05/74 2
Turkey Point 3 W
07n2 01/1403 1
i Yankee Rowe W
06/61 Between unit startup(1/61)and i
d 1965 Zion 2 W
12n3 08/2904,12/300 4,05/25S 6, 6
06/20n6,07/10n7,09/03/80 Zion 1 W
06n3 09/2606,07/08n7),07/10/77.
8 09/1408,12/1508,03/02n9, j
03/16n9,06/0809
~
37 Events TOTAL: 17 Plants
' References for water hammer events prior to May 1981 are USNRC (1979c) and Chapman et al. (1982).
8 Duquesne Light (1993).
8 Reported as nondamaging, it is likely that the event took place before 1/1/69 because it is not reported by Chapman et al. (1982), which reviewed the events since 01/01/69.
' USNRC(1986a,1986b).
' RELD EXPERIENCE ~
Main foodwater pipe -
(16 in., och 80 pipe)-
Thermal sleeve i
(16 in., sch 80 pipe)
Feed ring S/G wall (10 in., och 40 pipe) i
~ intomal clearance I
gap = 0.020
- 0.015 in.
16 in x 10 in j
b k'
reducing i
lee l
M i
f M3 3/4 in. diameter flow holes (total 261 holes roughly evenly distributed about periphery) lNEL 210s3 Figure 41. Older vintage feedring design for top feed steam generators (Han add Anderson 1982).
^
t i
The Indian Point Unit 2 water hammer resulted run from draining into the steam generator l
i from a steam / cold-water interaction in the main whenever the water level dropped below the feedwater piping adjacent to the steam generator.
feedring. He modification was not necessary in Of the four steam generators at Indian Point Unit the other three steam generator inlet lines because l
2, this lino had the longest (3.25-m (10-ft 8-in.)]
the horizontal runs were shorter, less than half the horizontal run immediately upstiram of the steam length of the horizontal run in the affected line.
generator nozzle. When the level in the steam New restraints were added both inside and outside i
generator dropped below the feedring, the sparging the containment near the feedwater line penetration holes on the underside of this ring allowed the area. The feedring was modified to prevent rapid l
water to drain rapidly from the ring and from the draining by plugging the sparging holes and l
horizontal feedwater piping outside the steam installing J-tubes on the top of the ring.
l gs.evi. De feedring and line then refilled with steam from the steam generator. Subsequent San Onohe Unit 2. On March 30,1981, a l
introduction of cold auxiliary feedwater to restore steam generator water hammer event occurred the steam generator level led to a water hammer while operators were performing the feedline water that caused the damage.
To preclude the hammer test during hot functional testing at San recurrence of water hammer events, the elevation Onofre Unit 2. The feedring of the affected steam
'of the long horizontal run of feedwater piping generator was deliberately uncovered for two hours outside the steam generator was lowered about 0.4 prior to performing the water hammer test. This m (16 in.) to prevent the water in this horizontal was a severe water hammer test because 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> 4
~
~
~ _. _ __ _
FIELD EXPERIENCE '
L
- duration is not operationally watative. Even removing stress risers from the pipe inside surface
{
though the feedring was equipped with discharge 1 and modifying operations instruction and guidance i
elbows, the 2-hour duration was enough time to to reduce the probability of thermal cycling and drain the feedring, which then filled with steam.
steam generator water hammer events. Maine When full flow of auxiliary feedwater rapidly
. Yankee also considered expanding the thermal I
initiated, the steam in the feedring got trapped sleeve into the nozzle to reduce the rate ofdraining h-==, the flow area of the discharge elbows was from the feedring during a loss-of-feedwater event.
- inadequate.
De cold auxiliary feedwater Damaged supports were also repaired (Garrity condensed the trapped steam, which created a 1983).
i partial vacuum in the-relatively thin-walled (Schedule 40) feedring and caused it to collapse.
San Onoke Unit 1. On November 21, ne U-bolts at the feedring support brackets were 1985, a main feedwater pump at San Onofre l
deformed beyond the elastic limit, and some U-Unit 1 (an older 3-loop Westinghouse plant)-
bolts were failed at the threads. The corrective experienced a loss of power, whereas another main.
- actions included replacing the collapsed feedring feedwater pump remained energized. The heater-with a Schedule 120 pipe, increasing the diameter condenser train of the portion of the. main of discharge elbows from 38 to 89 mm (1.5 to 3.5 feedwater system that experienced the power loss
. in.), and installing a vent assembly on the was exposed to the high pressure of the main distribution box, as shown in Figure 14. - Also, feedwater because multiple check valves failed,
[
actions such as much slower auxiliary feedwater which allowed back flow of higispressure l
initiation were taken to prevent the formation of a feedwater. This high-pressure exposure i esulted in j
vacuum in either the feedring or distribution box a ruptured tube and a damaged shed in the i
upon the addition of cold water (Scherer 1981, condenser. The check valve failures allowed the Stoller 1991).
water in the long, horizontal piping to drain through the condenser rupture. Then steam from -
l Maine Yankee. In 1983, the Maine Yankee the steam generator filled extensive portions of the plant was tripped from full load. The auxiliary feedwater pipiag (Bamford 1987).
j feedwater pumps started automatically as designed, i
to restore water level in the steam generators. A Auxiliary feedwater flow was established and the water hammer occurred about 15 minutes after the isolation valves and flow control valves were j
reactor trip and caused rupture of the Number 2 closed after the power was restored. Once the j
main feedwater line near the steam generator isolation valves were closed, the draining of the j
feedwater nozzle where, as discussed in Section feedwater through the condenser rupture stopped.
6.2.1, a preexisting crack caused by flow stratifica-Refilling of the long, steam-filled, horizontal l
tion may have been present. He crack was at the feedwater lines with cold auxiliary feedwater led to l
bottom of the pipe, right at an intemal stress riser.
a severe water hammer, producing a traveling j
Subsequent inspection found a similar crack at the pressure wave that generated very high loads. He same location in the No. 3 main feedwater line feedwater piping in the B steam generator loop near the steam generator feedwater nozzle. This experienced the worst loads because of a long supports the postulation that the crack had initiated horizontal run of piping. The 254-mm (10-in.)
l by thermal fatigue (Garrity 1983).
diameter pipe inside the containment was distorted from its original configuration; pipe supports were Each steam generator feedring was later modified damaged; and a 2-m (80 in.) long crack was by closing off all 76 bottom-mounted,25 mm generated (Bamford et al.1987).
A full (1-in.) diameter nozzles and installing 28 description of this event may be found in a report.
top-mounted, 90-mm (3.5-in.) diameter elbows prepared by the USNRC (1986b). The corrective (Stoller 1987). Other preventive actions included actions included replacing faulty and damaged 81 NUREG/CR-6456
-... ~
J~
-. ~. _. _., -,.... -
s_~
~,,,
..,.,,-.i
FIELD EXPERIENCE check valves, implementing logic which closes the ne location of the inside containment check valve flow-control valve upon a loss of main feedwater, on the steam generator C auxiliary feedwater line and installing redundant check valves in the feed and the piping layout at Beaver Valley Unit 2 have water piping (Bamford et al.1987).
the potential to causc water hammer. The inside containment check valve is near the main feed-Palisades. Condensation of trapped steam water line and at a highe' r elevation than the anil-and the resulting collapse of the steam bubbles iary feedwater supply tank minimum water level.
have caused water hammers in auxiliary feedwater Because of its proximity to the main feedwater lines. During an inspection at Palisades (an older line, several feet of piping upstream of the check vintage Combustion Engineering plant), the valve is likely to be at a higher temperature than following auxiliary feedwater sy. tem damage was the typical auxiliary feedwater temperature; mea-l confirmed (USNRC 1984a):
sured temperatures for the top and bottom of the I
pipe were 131*C (268*F) and 93*C (200*F),
A thermal sleeve was cracked respectively, at about 0.6 m (2 ft) upstream of the 4
check valve. De higher temperatures are condu-2 One of the three hold-down clamps for the cive to steam voiding ifleakage through check i
sparger was missing valves is present. Herefore, there is potential for i
a water hammer caused by steam bubble collapse A weld at an elbow was broken if flow of cold auxiliary feedwater is rapidly initiated to the voided auxiliary feedwater line for A clamp on a riser pipe to a sparger was steam generator C (Duquesne Light 1993).
{
broken.
The corrective actions included replacing the
]
An inspection conducted before the discovery of mechanical snubbers with hydraulic snubbers i
the steam generator internal damage revealed that having a higher load rating to withstand the loads j
eight hangers on the auxiliary feedwater piping generated by a water hammer event. Other actions were also loose or damaged. These hangers were include frequent monitoring of temperatures and inspected before the beginning of the fuel cycle pressures in the auxiliary feedwater lines and I
and were considered in good condition at that time.
inservice inspection of the lines and their supports The damage described above was consistent with in the containment following an unplanned start of j
the occurrence of a water hammer during the fuel the auxiliary feedwater pumps that results in rapid
- cycle, initiation of auxiliary feedwater flow.
Beaver Valley Unit 2. Based on the results 6.5.3 Preheat Steam Generators i
of the extensive engineering analysis of previous events at Beaver Valley Unit 2, it was The preheat steam generator is a somewhat newer y
hypothesized that the auxiliary feedwater line for design, shown in Figure 4. Lessons learned from Steam Generator C had been subjected to two the experience with the water hammer events water hammers caused by steam bubble format on associated with the top-feed steam generators were i
and its collapse since the plant went operational in incorporated in the preheat steam generator de-j 1987, one event in April 1989 and another in 1992.
signs. Thus, it was believed that the preheat steam
)-
The mechanical snubbers on the auxiliary generators were not susceptible to water hammer feedwater lines were mechanically frozen during events that had occurred in the top-feed steam both events. The cause of these snubber failures generators. However, a water hammer occurred in 1
was excessive compressive or impact loads the auxiliary feedwater line of a preheat steam apparently produced by water hammer events generator at KRSKO, a 2-loop Westinghouse (Duquesne Light 1993).
plant in the former Yugoslavia,during preoper-t NUREG/CR-6456 82
~
1 FIELD EXPERIENCE l
1 stional testing of the auxiliary feedwater line. De here. %e steam generator water level w:s below
- schematic for the main and auxiliary feedwater the elevation of the auxiliary feedwater nozzle systems for KRSKO is shown in Figure 42.
internal extension discharge. Then the significant back leakage of steam from the steam generator
' through the auxiliary feedwater line took place and l
The water hammer at KRSKO was caused by a steam bubble collapse in an auxiliary feedwater caused the observed paint damage back to the i
line during a hot functional test in July,1981. %e motor-driven auxiliary feedwater pumps. De horizontal portion of the line where the water check valves in the discharge and suction lines of I
f hammer occurred was about 16 m (52.5 ft) below the auxiliary feedwater pumps had to have leaked the auxiliary feedwater nozzle, as shown in Figure to allow the back leakage of steam. %e significant
- 43. He auxiliary feedwater pumps were being run back leakage also msulted in the presence of steam l
2 i
intermittenti; ' ringthetest. Damagewasmainly in the horizontal portionpf the auxiliary feedwater j
limited to ins.e the containment: the piping was line.
Then, when the cold feedwater was shifted and bulged about 6 mm (0.25 in.) near the introduced into this portion of the line, the water l
secondary wall (the probable location ofwater slug hammer caused by steam bubble collapse occurred impact), and the pipe hangers were damaged. De (Sexton, Kasahara, and Uffer 1982).
piping movement was negligible in the i
intermediate building but paint on the auxiliary De KRSKO event shows that multiple component feedwater piping was blistered back to the motor-failures can lead to a water hammer in a preheat driven auxiliary feedwater pumps (Sexton, steam generator notwithstanding the fact that the Kasahara, and Uffer 1982).
preheat steam generator designs incorporate j
several features that have been successful in i
The sequence of events that led to the KRSKO essentially eliminating steam generator water l
water hammer are not well understood. One hammers. As discussed, failare of multiple check i
sequence of events, though unlikely, that could valves was responsible for the water hammer at have caused the observed damage is hypothesized San Onofre 1 in 1985 and Beaver Valley Unit 2 in l
1989 and 1992.
t I
l 4
4 83 NUREG/CR-6456 y
~
- 8 95r9-1f3/03 BON l
l 5f1, /l,
~ X 4'X x;
x; l
1 l aB iD fas iD l
s
+ tX f
L s
v tX t
a p
tX\\{l 3 tX \\{IT c
x rt t_
x rl t-k l
is Pa y
e TB a
b Y-"4-8 %-
r s
s s
s X I4DBitX i
i i
F i
I m
1
,m,
f',
e s
s s
t t
e bIkEh j
j j
m m,, a m
m,, +
p x,,
-x >,
I s
l>
ir ir 33NHIHHdX3 07313
1.0 ft Steam Generekt( >
6-in. diameter pipe 52.5 R -
kWde Contenment O
Normodsfa BIAking 29.5 R 116 14.1 R j
I5 006 3.6 A 114
\\
10.5 R 110 113 6h diamotor @
5.6 ft 108 111 9R 1.6 ft 5.9 R 5.9 ft
.2 R shield weg j
6.6 ft 15.1 R feedwater 11.8 R Containment From modsery ~ \\
.emo,e, G
g m
Q m
n 2
?
Figure 43. KRSKO Loop 2 auxilary feedwater systems for KRSKO (Sexton, Kasahara, and Uffcr 1982). (1 in = 25.4 nun) m T
=
m
- 7. DEGRADATION MECHANISMS As discussed in Section 6, the main degradation nozzles subject to these design transients have mechanisms affecting the PWR feedwater lines are shown acceptable fatigue usage factors for the life thermal fatigue and flow-accelerated corrosion.
of the plant. Because the thermal stresses in the Thermal fatigue has caused through-wall cracking feedwater piping near the piping-to-nozzlejunction and leakage in sections of PWR feedwater piping are lower than those in nozzles, these transients do where stratified flow conditions were present not contribute significantly to fatigue cracking in during stanup and hot standby conditions; there the piping (Thurman, Mahlab, and Boylstein have been no reported ruptures attributed to fatigue 1981). However, the original design of the piping damage. The affected piping sections typically did not consider the effects of stratified flows, include a portion of the piping adjacent to the which are observed during hot standby and at low feedwater nozzle and, in some cases, the long power during heatup, and these conditions can horizontal portions of the piping inside the impose stresses of high magnitudes. Therefore, the containment, which is discussed later in Section fatigue damage in the PWR feedwater piping has 9.2. Flow-accelerated corrosion has caused wall been much greater than originally expected. The thinning at several susceptible locations in the reported cracking incidents have been found to be piping and, in some instances, has led to associated with the presence of stratified flows.
catastrophic rupture. Factors contributing to the flow-accelerated corrosion damage are feedwater StratificafMn Phenomena.
Stratified temperature, chemistry, and v.locity, and piping flows take place in horizontal sections of piping layout and materials. This section describes the and consist of hot and cold fluids separated into phenomena associated with these two mechanisms two layers because of their density difference. The and discusses the factors affecting the damage propensity for stratification of a fluid in horizontal caused by these mechanisms.
piping can be correlated to its Froud number, which is the ratio of the inertial force (velocity 7.1 Thermal Fatigue head) to the force of gravity (buoyancy head) acting on the fluid.
The velocity head is PWR feedwater lines and nozzles are subject to determined by the relative velocities between the both low-and high-cycle thermal fatigue. Low-two fluids. The buoyancy head is caused by the cycle thermal fatigue is caused by plant heatups, density difference between the hot (top) and cold cooldowns, thermal stratification, water hammer (bottom) regions of the fluid in the pipe, and its events, and thermal shocks. High cycle thermal magnitude is determined by the difference in fatigue is caused by thermal striping (defined temperature between the two regions. Low flow below), which is associated with thermal rates and high temperature differences (small stratification, and turbulent mixing. The auxiliary Froud number) promote thermal stratification in feedwater lines, ifconnected to the main feedwater horizontal piping. In the case of the feedwater ii P P ng, thermal stratification occurs when the cold line, are potentially susceptible to thermal cycling during normal operation.
auxiliary feedwater, introduced at low flow rates into the main feedwater line during plant startup 7.1.1 Thermal Stratification and hot standby conditions, flows under the warmer coolant present in the piping.
The thermal expansion loads imposed on the feedwater piping during plant heatup and An interface layer (mixing zone) is developed cooldown cycles and feedwater injection follow ing between the hot and cold fluid layers, as shown in a reactor trip were included in the stress analyses Figure 44(a)(Miksch et al.1985). The elevation during the design of the piping but fatigue analyses of the interface layer (height of the cold fluid layer) were not required. Some analyses of the feedwater and its thickness depend primarily on the mass NUREG/CR 6456 86
DEGRADATION MECHANISMS Hot water IH Mxing zone i
i
\\
_f Y$*
T (a) Fluid temperature detribution.
Membrane Bending Total stress stress stress Y
p sot -
Tw (280*C) lA
+
/-
Cold
-~~f Tc (120*C)
F =Fc M.Fel H
(b) Theoro6 cal axial stress distribution. Mixing zone at 60'.
t80' 180*
/
\\
gg. EXIOO
\\
Tc 20M Envelope &s j
g i
/
l o.
I
/
g (d) Mxing zone at 90' hg' l
Envelopes for stress distribudon 1 i Stress distribution for mixing zone
,I j
- -- - Theoredcol I
-- -- Theoretical Actual I
/
I Actual f
I l
j
\\
\\
j
\\
\\
0
-300-200 100 0 100 200 300 Stress (N/mm2)
(c) EnWopes for theoredcol and realisde stress detribudons Figues 44. Thermal stresses in a pipe with stratified flow (Miksch et al.1985). Copyright Elsevier Science Publishers; reprinted with permission.
DEGRADATION MECHANISMS flow and density ratio. Instabilities occur in the nitude than the axial ones (Woodward 1983,Talja interface layer because ofinduced wave formation and Hansjosten 1990). For example,in some flow-as the relative velocity between the hot and cold stratification experiments performed at the HDR fluids increases. The wavy motion of the coolants (Heissdampfreaktor) test facility, the measured in the interface layer introduces substantial temper-maximum stress on the outside surface of the ature fluctuations at a fixed position in the layer piping was 36 MPa (5,200 psi) in the circumfer-and introduces small-amplitude cyclic stresses in ential and 390 MPa (56,550 psi) in the axial the inside surface of the adjacent piping in contact.
direction. The much larger magnitude of the axial These stresses penetrate only a short distance into stress explains why the fatigue cracks found in the the wall, causing high-cycle fatigue damage. This feedwater piping generally have a circumferential high-cycle fatigue phenomenon is also known as orientation.
thermalstriping and is discussed in Section 7.1.2.
Flow rate variations cause the interface layer to be Stratified flow introduces through-wall axial and raised or lowered. In the feedwater system, these circumferential bending stresses whose magnitudes changes are caused by the auxiliary feedwater are determined by the top-to-bottom temperature being tumed on or off to adjust the steam generator difference (TrTc), and the elevation and thickness water level. As the feedwater flow rate changes, j
of the interface layer. Figure 44(b) shows the the layer moves between the upper and lower theoretical distribution of the axial bending stresses portion of the pipe cross section and changes the caused by stratified flow when the interface layer through-wall bending stress distribution introduced is at 60 degrees from vertical. This distribution by the stratified flow, which causes low-cycle assumes that the ends of the horizontal section of fatigue damage. Figure 44(c) shows the predicted the piping are constrained from rotation and the stress distributions for a mixing layer at 90 degrees thickness of the mixing layer is zero. This stress as shown in Figure 44(d), and presents the distribution is comparable to the stress distribution envelope for the distribution as a function of the in a bimetallic strip subjected to a uniform height of the mixing layer. The dashed line temperature change. The resultant axial stresses represents the theoretical axial stresses, and the are presented as the sum of the membrane and solid line represents an actual stress distribution, bending stresses. The portion of pipe below the which accounts for the thickness of the mixing mixing layer experiences tensile axial stresses, layer and the heat transfer taking place in the whereas the portion above the mixing layer piping material (Miksch et al.1985). Any given experiences predominantly compressive axial point on the pipe wall will experience a maximum stresses. The theoretical maxi.num axial stresses change in its axial stress when the interface layer are near the mixing layer. The axial stresses need passes by. These cyclic stresses are through wall to be multiplied by a stress concentration factor and cause both crack initiation and growth, introduced by any geometric discontinuity such as the counterbore typically present in the feedwater The piping inside surface temperatures, under piping.
Distribution of the axial stresses is steady state conditions, are about equal to the symmetric with respect to the vertical diameter of temperatures of the fluid that is in contact with the the pipe cross section, provided the end constraints piping at that location. Because the fluctuations in and the geometric discontinuity, if any, are also the feedwater flow rate take place at a relatively symmetric.
Iow frequency, the piping inside surface temperatures are also about the same as the fluid Circumferential bending stresses are produced temperatures (Talja and Hansjosten 1990, because of the temperature difference between the Thurman, Mahlab, and Boylstein 1981). The pipe sections above and below the interface layer.
outside surface temperatures follow the inside However, these stresses are of much smaller mag-surface temperatures with some time delay.
DEGRADATION MECHANISMS De resulting piping-wall temperature distribution elbow. Plant operational data indicate that the is plant specific because, in addition to the hot and period for the cyclic thermal stratification is gener-cold coolant temperature and flow rate, it depends ally longer than several minutes and is typically l
on piping layout. Dese low-frequency fluctus-associated with the automatic cycling of the auxil-tions in the piping wall temperatures can be esti-iary feedwater flow. At the Sequoyah plant, for mated by monitormg the outside surface tempera-example, the auxiliary feedwater system in the tures. However, as discussed in the next section, automatic control mode cycles about three times the high-frequency temperature fluctuations caused per hour.
by thermal striping cannot be estimated by moni-Major fat gue cracks have been reported at the t
toring the outside surface temperature.
geometric discontinuity. In addition, the down-Global Thermal Strew #csWon. If the stream end of the elbow experiences thermal stratification persists over a long section of pipe, stratification and resulting stresses are higher than and the temperature difference between the hot and for the straight pipe. For example, in one analysis, cold fluids is large, the pipe will bow (macroscopic the peak stresses for the elbow configuration displacement), resulting in high loads on the (elbowdirectlyco-wat dtothefeedwaternozzle) supports along its length and on the elbows or were about twice those for the straight pipe config-
. nozzles at its ends. De stress distribution will be uration (nurman, Mahlab, and Boylstein 1981).
similar to the one shown in Figure 44, except it is l
modified by the constraints exerted by the sup.
Construction methods in conjunction with design ports. His type of stratification is termed global and material considerations usually introduce two thermalstratification (Su 1990). Such stratifica-different geometric discontinuities that create tion occurs in feedwater piping under two condi-significant stress concentrations in the piping: the tions: (1) existence of long horizontal runs of weld between the nozzle and transition piece pipes between the auxiliary feedwater connection connecting to an elbow shown in Figure 16, and a and the steam generator nozzle, and (2) in-counterbore, examples of which are shown in troduction of cold water at a low flow rate into a Figures 15 and 17. He stress concentration factor hot feedwater system. The feedwater piping at introduced by the counterbore depends on its angle Beaver Valley 1, for example, satisfies these a and depth h defined in Figure 17(b). As angle a conditions and has experienced global thermal and depth h increase, the stress concentration stratification (Van Duyne et al.1991).
factor increases (Cofie et al.1994).
Local Thermal Strew #csWon. If the ne distribution of cyclic stresses introduced by stratification persists over a short section of pipe, local thermal stratification in a p!y cross section it is termed local thermal strati /ication. Local may be symmetric or asymmetric weh respect to thermal stratification produces a complex stress the vertical diameter. Symmetric distributions will state in the piping because of its short length, the lead to fatigue cracking that is also symmetric, as end constraints exerted by the feedwater nozzle shown in Figures 22(b) and 27 (Loop 4).
i and elbow, and the geometric discontinuities Otherwise, fatigue cracking will be asymmetric, as (discussed below) present in the piping. Finite shown in Figure 27 (Loop 1). One possible reason element analyses are generally performed to calcu-for an asymmetric stress distribution is that the late the stresses.
midplane of the elbow (first elbow from the feedwater nozzle) is not vertical. It is also possible Local thermal stratification occurs more often in to have some asymmetry in the geometric PWR feedwater piping than global stratification.
discontinuities.
At several PWR plants, local thermal stratification has been found in the short horizontal section of Corrosion Fs#gue.
High-amplitude, i
the feedwater piping connected to the feedwater low-frequency cyclic stresses imposed by thermal nozzle. De other end of the horizontal section is stratification have combined with coolant generally attached to a 90-degree or a 45-degree containing copper oxides, chlorides, and possibly i
1 DEGRADATION MECHANISMS oxygen in the auxiliary feedwater (if nitrogen ature of small polished specimens and do not fully cover is not provided to the coolant in the account for the LWR environmental effects. These condensate storage tank) and caused curves were developed by applying factors of 2 on corrosion-fatigue damage to the carbon steel stres: or 20 on cycies, whichever is greater, to the feedwater piping and low-alloy steel feedwater mean tailure curve for the small polished speci-nozzles. These piping and nozzle rnaterials are mens. A portion of these factors accounts for the susceptible to corrosion-fatigue if they contain effects of the environment; however, recent re-sulfur inclusions such as manganese sulfides search results show that the effects of temperature (Bamford et al.1987). Corrosion products have and oxygenated coolant on carbon steel fatigue been found on the crack surfaces in these strength is not fully accounted for in these curves components. Corrosion products, for example, (Terrell 1988, Higuchi and lida 1992).' The have been found in the secondary cracks in the low-cycle fatigue strength of smooth carbon steel Sequoyah feedwater nozzle and in the nozzle bore specimens tested in air at 288'C (550*F) is lower cracks in the Indian Point Unit 2 feedwater nozzle.
than the mean failure curve, but it does not lie below the design curve, as shown in Figure 45.
The ASME Code design curves for carbon steel However, the low-cycle fatigue strength of carbon may be used to estimate the low-cycle fatigue steel in high temperature [~300*C (572*F)] ox3u damage to the feedwater nozzle and piping.
genated pure water decreases with a decrease in flowever, these curves are based entirely on data strain rate, and, at lower strain rates (s0.01%/s),
obtained from in-air tests mainly at room temper-the fatigue strength lies below the design curve.
104 3
ASME mean data curve ASME Section 111 curvas and for carbon steel ASME SA 106-B ste' w
~
/
smooth specimern N i "lu m
/
3 104 103
)
g q
q Mean data curve for Q-
~N smooth specinwns,'
~
E f}102 N
j 288*C air i _ 103 s
g-N.,,,,,,
=
~
ASME Section lil Style safety 102 N 101 -- factors as applied to the 288'C
/
E g
mean data curve
/
j Cunem ASME
~
design curve 100 1 I i11111! I I I till!! I I I11111! I I i11111l 1 I I11111l l l l IITil 101 101 1 02 1 03 1 04 jos 10s 107 Cycles to failure N92 0221 Figure 45. Fatigue data for SA-106B carbon steel smooth base metal specimens in air at room temperature and 288'C (Terrell 1988).
' lida, K, et al. 1988. " Abstract of DBA Committee Report,1988 Survey of Fatigue Strength Data of Nuclear Structural Material in Japan."(K. lida, Professor Emeritus, Unis ersity of Tokyo, Tokyo).
DEGRADATION MECHANISMS He need for incorporation of environmental in the HDR tests was between about 0.01 and 2 effects in the ASME fatigue
- sign curves for Hz, and the amplitude of the metal temperature Class I carbon steel components is being evaluated oscillations was less than 50% of the difference in by the Pressure Vessel Research Committee the hot and cold coolant layer temperatures. The Working Group on Evaluation Methods.
amplitude of the metal temperature oscillations was smaller because of the finite heat transfer 7.1.2 Thermal Striping and Turbulent coefficient and thermal inertia of the pipe wall.
Mixing The magnitude of the thermal striping stresses is highest on the inside-surface and reduces rapidly As discussed above, thermal stratification can through the thickness. Therefore, the high-cycle induce local, small-amplitude, cyclic stresses in the fatigue damage caused by these stresses is limited portion of the inside pipe surface that is in contact to the pipe inside surface adjacent to the interface.
with the interface layer, if the flow rates are suffi-ciently high. Rese stresses are caused by oscilla.
There are several differences between fatigue tions of the fluid temperature, resulting from the damage caused by thermal stratification and strip-interfacial mixing of the hot and cold fluids in the ing. Stresses produced by thermal stratification are interface layer. As mentioned above, such interfa-through-thickness stresses and are mainly along the cial mixing results in a process called thermal axial direction, whereas those produced by striping striping. The onset ofinterfacial mixing that leads are mainly at the inside surface with little penetra-to thermal striping can approximately be correlated tion through the thickness, and have a random with the initiation of a Kelvin-IIelmholtz instabil-orientation. The stresses produced by thermal ity, which occurs when inertial forces overcome stratification have large magnitudes and low cyclic stratifying density differences between the fluid frequencies, whereas the stresses produced by layers (llafner and Spurk 1990).
thermal striping are small and have high cyclic frequencies. So the fatigue damage caused by Wolf et al. (1987) conducted thermal stratification thermal stratification is low-cycle fatigue damage experiments in horizontal feedwater lines at the that affects the entire cross section of the piping, llDR test facility in Germany. The experiments whereas the damage caused by thermal striping is were performed at several different flow rates with a high-cycle fatigue damage that affects only the thermocouples mounted on the inside pipe surface inside surface of the piping in the vicinity of the to measure metal temperatures. Thermal striping mixing layer and attenuates rather rapidly toward was observed only at relatively high flow rates the pipe outside surface. Thus, thermal stratifica-(Deardorffet al.1990). The test results and theory tion contributes to both crack initiation and growth, indicate that thermal striping is present when the whereas thermal striping contributes only to crack gradient Richardson number, R,, through the initiation. Cracks initiated by thermal stratification interface is less than 0.25 (llafner and Spurk 1990, have a circumferential orientation, whereas cracks Turner 1973). The gradient Richardson number is initiated by thermal striping have random orienta-the ratio of the density gradient and horizontal tions.
velocity gradient across the thickness of the interface layer and is given by Turbulent mixing of cold feedwater leaking through the thermal sleeve-nozzlejoint and the hot R, = -g(Sp/dy)/p(du/dy)2, coolant in the steam generator has caused high-cycle fatigue crack initiation at the feedwater w here g is the acceleration due to gravity, p is the nozzle bore and blend radius. Rese two sites are density, u is the fluid velocity, and y is the shown in Figure 23.
Similarly, in a steam coordinate along the thickness of the interface generator where the auxiliary feedwater is layer (Tumer 1973).
discharged directly into the steam generator, leaking coolant has caused cracking at the auxiliary The typical frequency content of the metal feedwater nozzle blend radius (Westinghouse temperature oscillations caused by thermal striping 1989). Turbulent mixing has also caused similar 91 NUREG/CR-6456
DEGRADATION MECHANISMS crack initiations at the BWR feedwater nozzle leakage. Figure 46 illustrates how the turbulence blend radii at several BWR plants. liowever, there in the main piping, such as main reactor coolant is one difference between the designs of these two piping or feedwater piping, penetrates into a nozzles; the PWR steam generator feedwater connecting branch line such as a safety injection nozzle does not have cladding on the inside surface line or an auxiliasy feedwater line, ne turbulence whereas the BWR feedwater nozzle had stainless intensity decays exponentially from the header pipe steel cladding.
The presence of cladding into the branch line, but the temperature remains contributes to the fatigue cracking because thennal fairly constant over the length of several diameters stresses from the turbulent mixing are higher in the and then decays.
The length of turbulent stainless steel than they would be in unciad base penetration changes with time, and it is greater for metal. In addition, the low-cycle temperature a higher flow velocity in the main pipe and smaller changes contribute to fatigue of BWR nozzles for an increasing stratified flow in the branch line.'
i because the amplitude of metal temperature fluctuations caused by the turbulent mixing is Cyclic changes in the length and intensity of larger in the cladding than in the unclad base turbulent penetration produces corresponding metal. The larger fluctuations produces higher changes in the length of the stratified fluid layers.
stresses in the cladding. Herefore, the stainless As a result, the pipe in contact with the interface steel cladding on the feedwater nozzle has been experiences cyclic stresses causing fatigue damage.
removed at several BWR plants (Snuider 1980, Thermal cycling caused fatigue cracking in the Gordon et al.1987).
safety injection lines (Farley Unit 2, Tihange Unit
- 2) and in a residual heat removal line (Genkai, a Leaking cold feedwater has also caused Japanese PWR) connected to the PWR primary circumferential cracking on the steam generator coolant piping (Kim et al.1993).
shell inside surface directly beneath the nozzle as shown in Figure 23. The mechanism causing the It appears that turbulent penetration alone, under circumferential cracking is not known. Earlier certain operating conditions and with susceptible experience related to BWR control rod drive return piping layouts, could produce thermal cycling in a line nozzle cracking is reported here because ofits branch line, even in the absence of any valve similarity to the cracking on the steam generator leakage. liowever, the effects of plant operating shell. Circumferential cracking was discovered on conditions on turbulent penetration are not well the inside surfaces of the BWR reactor pressure understood. The presence of thermally stratified vessel beneath the control rod drive return line fluids, in the absence of any valve leakage, has nozzles at several plants, and it is believed that the been detected in a branch line of a PWR plant, as cracking resulted from the cyclic thermsl stresses illustrated in Figure 47. A plausible explanation related to stratified flow of cold water along the for this presence of stratified fluids is as follows.
bottom of the nozzle and down the vessel wall as The branch line travels a certain distance vertically it mixes with the downflow of the rector vessel from the main coolant loop and then rtms (Snaider 1980).
horizontally, and it contains stagnant fluid. The turbulent penetration initially developed in the 7.1.3 Thermal Cycling vertical section of the branch line, as shown in Cyclic axial movement of an interface between hot and cold fluids is called thermal cycling. Such
' The nuclear industry has performed this research cycling takes place when a column of hot turbulent and evaluation of the presence of the. mal cycling, and also thermal stratificati n and striping, m unis table reactor fluid from the main piping penetrates into a coolant system lines in response to the USNRC Bulletin connecting branch line and interacts w.th a cooler gg.os, 73,rmai s,,,,,,, in Aning Connected to Reactor i
stratified flow, which could be produced by valve Coolant Systems. and its three supplements.
l l
DEGRADATION MECHANISMS Figure 47(a). Then, an operational transient such in power, the length of the turbulent penetration as a power change caused the turbulence to receded and the stratified layer was no longer penetrate the full length of the vertical rection and present. Thus, the base metal and welds of the produce stratification in the elbow at the end and in elbow and the horizontal portion of the branch line j
the adjacent horizontal section of the branch line, experienced cyclic thermal stresses and fatigue j
as shown in Figure 47(b). With further changes damage.
Turbulent penetration Stratified coolants
/
/
L
- g g,
- 1..e. : < ;w.
Auxiliary Main feedwater
/
pipeline Potential cycling zone 3
Turbulent decay correlation Distance from header pipe (t/D)
Temperature distribution i
a Distance from hehder pipe (t./D)
Figure 46. Turbulence interaction regions (Roarty 1993). Copyright Electric Power Research Institute; r: printed with permission.
b/
6 pi C %
J*M n
Q I
loop (r.
(
s t
t~
bi ie a
e O
M*
Initial penetrations Deeper penetration'6 (a) Initial penetration of turbulence (b) Deeper penetration causes stratification is confined to vertical section.
In the horizontal section.
Figure 47. A change in the penetration depth of turbulence from power variations can cause thermal stratification and cycling in a branch line (Martin 1993).
DEGRADATION MECHANISMS Dermal cycling could cause similar cracking in cal dissolution and mass transfer, rather than a the auxiliary feedwater piping where it connects to mechanical process involving removal of the oxide the main feedwater line. However, such cracking layer by erosion or cavitation. No evidence of the of PWR auxiliary feedwater lines has not been removal of surface oxide by mechanical shear reported. It is also not certain that similar thermal forces has been found during any macroscopic or cycling could take place in the auxiliary feedwater microscopic examination of the damaged inside line in the absence of any valve leakage.
surfaces of the feedwater piping. Hence, the term Monitoring of the temperatures at the outside flow-acceleratedcorrosion is more appropriate for surface of the auxiliary feedwater piping would the observed wall thinning of carbon steel piping probably indicate whether thermal stratification is exposed to deoxygenated feedwater (Chexal et al.
taking place and thermal cycling is occurring.
1996). Laboratory results show that the fluid velocities associated with the removal of the oxide 7.2 Flow-Accelerated Corrosion layer by mechanical processes are higher than those associated with the dissolution of an oxide As discussed in Section 6, flow-accelerated layer. Also, the kinetics of metal removal by a
]
corrosion has caused wall thinning in carbon steel mechanical process are either quasi-linear or nonlinear, whereas the kinetics of metal removal piping leading to costly repair and replacement and, in some cases, catastrophic ruptures. Flow-by an oxide dissolution process are linear (that is, corrosion rate is constant in time). De corrosion accelerated corrosion also causes damage to the steam generator tubes, not directly but indirectly.
rates (wall thinning rates) observed in the field are De corrosion products are deposited as sludge on essentially constant when the influencing factors the tubesheet and contribute to denting and crevice do not vary, corrosion of the tubes.
The flow-accelerated corrosion process is an 7.2.1 Flow Accelerated Corrosion extension of the generalized carbon steel corrosion Phenomena Process in stagnant water. In stagnant water, the carbon steel corrosion rate is low and decreases A thin layer of porus iron oxide [mostly magnetite Parabolically with time due to the formation of a (Fe30 )J forms on the inside surface ofcarbon steel Protective oxide film at the surface. Flow-4 feedwater piping exposed to deoxygenated water in accelerated corrosion takes place at low flow velocities and, as mentioned earlier, the the temperature range of about 95 to 260*C (200 to 500*F). Generally, this layer protects the corresponding corrosion rate is constant. The underlying piping from the corrosive environment difference between generalized corrosion and flow-and limits further corrosion.
However, the accelerated corrosion is the effect of water flow at the oxide feedwater interface.
i magnetite layer may be dissolved at the oxide.
water interface and be replaced by new iron oxide formed at the metal-oxide interface, resulting in The flow-accelerated corrosion process may be exP ained as follows. First, iron hydroxides are i
l material removal and thinning of the piping. He corrosion process is strongly influenced by the generated at the metal-water interface according to the reaction:
fluid velocity, chemistry and temperature, piping configuration, and alloy content of the steel as Fe + 2H 0 - Fe(OH)3 + H -
~
discussed below. This process is called flow.
2 2
accelerated corrosion.
Then, a porous iron oxide layer is formed Flow-accelerated corrosion is often called erosion, according to the reaction:
corrosion in the United States. But, this process is primarily a corrosion process enhanced by chemi.
3Fe(OH)2 - Fe30 + H + 2H O.
4 2
2 NUREG/CR-6456 94
DEGRADATION MECHANISMS ne flow-accelerated corrosion phenomena then sentatthe oxide-water interface can diffusevery follows a simple two step process which is illus-rapidly into the solution. If the ferrous ion produc-trated in Figure 48 (Remy and Bouchacourt 1992).
tion is faster than the mass transfer rate, the flow-The first step consists of production of soluble accelerated corrosion process is a convective ferrous ions and their accumulation at the oxide-diffusion-controlled process, and the corrosion rate water interface and the second step consists of increases wLh the increase in the fluid velocity.-
mass transfer of these ions into the bulk coolant.
Each factor that affects the flow-accelerated corro-1
%c ferrous ions are produced by two different sion process influences the proces:: through one of
]
reactions: (a) base metal oxidation at the metal-these two steps. These factors are discussed later oxide interface and diffusion of the ferrous ions in the section.
through the porous oxide layer to the oxide-water l'
interface, and (b) reduction of the magnetite layer A similar corrosion process causes wall thinning of at the oxide-water interface. Ferrous ion produc-carbon steel piping exposed to wet steam; this
}
tion is a steady state corrosion process which process is called two-phase flow-accelerated
' depends on the pH of the water, hydrogen pres-corrosion. However, if the piping is exposed to sure, temperature, and the oxide layer thickness dry or superheated steam, no flow-accelerated (inversely. proportional to the thickness). The corrosion takes place; a liquid phase must be oxide layer thickness, usually less than lym, present for flow-accelerated corrosion damage to remains constant during a steady-state flow-accel-occur. Corroded surfaces produced by single-Grated corrosion process.
phase flow-accelerated corrosion have a different -
appearance than those formed by the two-phase i
in the second step, the flowing water removes the flow-accelerated corrosion. When the single-phase j
soluble ferrous ions by a convective mass transfer flow-accelerated corrosion rate for a large-diameter
[
mechanism, which is a diffusion gradient driven piping is high, the corroded surface is character-process. Generally, the concentration of ferrous ized by overlapping horseshoe pits that give an ions in the bulk water is very low compared to orange peel appearance. The corroded surface of their concentration at the oxide-water interface.
alargediameterpipingexposedtotwo-phase flow j
Therefore, it is assumed that the ferrous ions pre-has a well known tiger striping appearance. Sur-l
'I Q^
hh!k h.; feef h $.
Magnetite /g E
~ g ~n,
Q
[
h5h.hf;gy lp Y
f
{
yhtv"g$
b {2 t
ca.
a j2% W@RMugh M W,, Nj W%
St WMyggygSem
~
y-
- s Fe,30, Z
Fe ++
l Fe Oxidation Reduction j
Figure 48. The flow-accelerated corrosion model(Remy and Bouchacourt 1992). Copyright Elsevier Science
- Publishers; reprinted with permission.
i 95 NUREG/CR-6456
DEGRADATION MECHANISMS face examinations of severely corroded carbon data for development of empirical models to steel components have shown that the thickness of estimate the rates (Chexal and Jones 1988, Chexal the oxide film is not uniform. Scanning electron and florowitz 1995). Evaluation of test results and microscope examinations of corroded surfaces data from operating plants have identified several show a selective attack on the pearlite constituent factors that affect the flow-accelerated corrosion of the carbon steel microstructure. This selective rates. These factors may be divided in three attack produces micropitting which has been groups: (a) hydrodynamic variables - fluid considered to be the initial step in the formation of velocity, pipe roughness, and piping configuration horseshoe pits and tiger striping (Chexal et al.
(geometry of the flow path); (b) metallurgical 1996). This report discusses only single phase variables - chemical composition including weight flow-accelerated corrosion, which is simply percentage ofchromium, molybdenum and copper referred to as flow-accelerated corrosion.
in the steel; and (c) environmental variables -
coolant temperature and water chemistry including The characteristics of the piping damage caused by dissolved oxygen, ferrous ion concentration, and flow-accelerated corrosion are different than that metallic impurities in water, pli, and the amines caused by thermal fatigue. Wall-thinning caused used for pli control. Each factor is discussed in by flow-accelerated corrosion typically occurs over the following subsections, and several calculated a broad area of the inside piping surface, whereas results using the Chexal-florowitz model are a thermal fatigue crack affects a relatively local presented. The calculated results are for the area of the surface. A component damaged by following parameters, with some variations:
flow-accelerated corrosion will eventually become 102-mm (4-in.) diameter carbon steel elbow,7 ppb too thin to support normal loads and may fail under oxygen content,6.1-m/s (20-ft/s) flow velocity, normal operating pressure; a large fitting might fail room temperature [25 *C (77 F)] pli of 7, and 0.03 catastrophically without any warning. A through-wt% of chromium, molybdenum, and copper wall crack caused by thermal fatigue will generally content; the variations are noted as appropriate.
leak long before the component ruptures.
The pli level at room temperature is also referred liowever, a large overload may cause a pipe with to as coldpH.
fatigue cracks to fail catastrophically with no prior leakage.
Hydrodynarnic Variables. The hydro-dynamic variables include fluid velocity, pipe Sites susceptible to flow-accelerated corrosion are roughness, and piping configuration (geometry of found throughout the feedwater system, whereas flow path). These variables affect the rate of mass sites susceptible to thermal fatigue cracking are transfer of the iron ions and other corrosion found in those portions of the piping where products to the bulk coolant and thus affect the stratified flows are present and are generally well flow-accelerated corrosien rate. Fluid velocity identified.
Several factors affect the flow-affects the mass transfer. At a relatively low flow accelerated corrosion rates and, therefore, it was velocity, the corrosion rate is controlled by the rate difficult to identify the sites susceptible to this of mass transfer, whereas at higher velocity (still corrosion process. Since the Surry incident in lower than the critical velocity above which metal 1986, significant efforts have been spent in removal by mechanical process takes place) the developing models for identifying these sites and mass transfer rate is higher and the corrosion rate estimating flow-accelerated corrosion rates.
is controlled by the chemical reactions at the oxide-coolant and metal-oxide interfaces.
Flow-7.2.2 Factors Affecting Flow Accelerated ucelerated corrosion is less frequently observed in Corrosion straight lengths of pipe free from hydrodynamic disturbances unless the bulk fluid velocity is high.
Flow-accelerated corrosion tests have been Laboratory studies of the effect of bulk flow conducted at several British, French, and German velocities, which varied from 2 to 18 m/s (6.6 to laboratories to identify the factors affecting the 59 ft/s), on the corrosion of carbon steel in 150*C corrosion rates (rate of metal loss) and to provide (300'F) circulating water show that the corrosion NUREG/CR-6456 96
DEGRADATION MECHANISMS rate ir creases with an increase in the flow rate and, of the manufacturing process or a result of plant for s' given flow rate, the corrosion rate is about operation. A rough surface produced by the flow-cor.stant.
Figure 49 shows how the flow-accelerated corrosion process can-be very I
aaelerated corrosion rates for. straight tube damaging. The micropits formed by the initial samples placed downstream of an orifice varied selective attack on the carbon steel microstructure with temperature and flow rate (Delp et al.1985).
grow until they touch, and thus the surface The samples were tested at Central Electric becomes rough. The dependence of mass transfer Research Laboratories (CERL) in the United on the velocity is greater for a rough surface than Kingdom. He pH level was in the range of 8.5 to for a smooth surface.
9.2. The flow rate ranged from 227 to 983 kg/h (1 to 5 gpm) through the 8.33-mm (0.33-in.) diameter The variable piping configuration takes into 1
tube, with flow velocities of 1.2 to 5.8 m/s (4 to 19 account the hydrodynamic disturbances (elbows, ft/s). He corresponding maximum corrosion rates tees, branch connections, reducers, valves, flow ranged from about 0.4 to 4.1 mm/yr (0.016 to control orifices, etc.) that produce high local fluid 0.16 inlyr).
velocities and result in a further increase in mass transfer. Experiments have shown that local-flow he flow accelerated corrosion rates are influenced velochies in elbows can be two to three times the i
by two other factors that affect the local flow bulk. flow velocities (Bosnak 1987, USNRC l
velocity: pipe roughness and piping configuration.
1937c). De results ofthe Chexal-Horowitz model Roughnen in commercial pipes is a consequence showing the effect of different piping configura-I l
4.5 Flow (kg hrd)
' Single-phase flow i
i, i
i i
i
.i i
4.0 L o 983 8.33 mm (0.328 in.) tube ID ;
j x 907 2.72 mm (0.107 in.) orifice ID -
l 3.5 ;
e 756 j
- t. 605 x
i v 491
/
\\
L 3.0 :
o 378
/
-i i
+ 227
/
Increasing 2.5 7 flow rate g 2.0 x
7 9+ 7 /rx N. x V
N N
0.5 p*
-k 0.0 1
90 100 110 120 130 140 150 160 170 180 Temperature ('C)
_.,,3.,
Figure 49. Flow and temperature dependence of flow-accelerated corrosion rates (Delp et al.1985).
i 97 NUREG/CR-6456 i
DEGRADATION MECHANISMS tions on flow-accelerated corrosion rates are to 5.0 mm/yr (0.065 to 0.195 inlyr). The presented in Figure 50. He maximum corrosion maximum corrosion rate occurs at about 150*C rate for a 90-degree elbow configuration is about (300*F). When the fluid velocity is zero, the flow-2.5 times greater than that for a straight pipe; the accelerated corrosion rate is zero or very small.
rate is 3.8 mm/yr (0.15 in/yr) for an elbow, and Derefore, flow-accelerated corrosion damage has 1.5 mm/yr (0.06 in/yr) for a straight pipe.
not been reported in the auxiliary feedwater piping Geometric discontinuities at thejoints between the in which the fluid is typically stagnant during J-tubes and feedring piping have caused significant normal operation (Mode 1) and cold when flowing flow-accelerated corrosion damage, as shown in during plant startup, hot standby, and shutdown, Figure 37(a).
and during design basis events. However, in some plants with preheaters, a small percentage of the i
Figure 51 illustrates the influence of fluid velocity main feedwater is bypassed through the auxiliary on the flow-accelerated corrosion rates on the feedwater line during normal operation as shown inside surfaces of a 102-mm (4-in.) diameter, in Figure 9. Flow-accelerated corrosion damage 90-degree carbon eteel elbow, as predicted by the has been reported in these portions of the auxiliary Chexal Horowitzmodel. Asthevelocityincreases feedwater line exposed to flowing high-i from 1.50 to 9.0 m/s (5 to 30 ft/s), the corrosion temperature fluid during normal operation (see rate increases by a factor of three, from about 1.65 Section 6.4.1).
0.25 U " 4 I"'
7 180' retum b
J$"g", PP
-- -- 90' elbow
. - - 45 elbow ts pH = 7 at 77'F
,g 0.20 cr Mo = Cu - 0.03%
- - - - Pipe coW R
1
$d
=y
,/ A 0.15 g
's i
,. - ~.
b, 0.10
' f r8
/./
\\,
s f
N cu f
/,!
\\s
\\
CD
//
N 0.05. -[/
,.'.-..sN,' - - s \\,\\
i y-y%. O - -.-
N., %..
0.00 100 150 200 250 275 '300 225 350 400 450 500
~~
Temperature (*F)
Figure 50. Influence of fluid temperature on flow-accelerated corrosion rates for carbon steel fittings estimated using the Chexal-Horowitz model(Chexal and Horowitz 1995). Copyright American Society of Mechanical Engineers; reprinted with permission. [1 in. = 25.4 mm; I ft/s = 0.3048 m/s; 0.5556 (*F - 32) =
- C].
DEGRADATION MECHANISMS l
i
~ 0.20 D = 4 in.
30 ft/s 0l18 oxygen - 7 ppb
- - - - 20 ft/s 1
pH = 7 at 77'F
-- - 10 ft/s I
Cr - Mo = Cu = 0.03%
- - 5 ft/s
~
3 0.16
- 90' elbow s
m
/
\\
\\
h 0.14
,/
s
\\
8 'c
/
m 0.12
/
\\
e
/
\\
h 5 l
/
-w
\\
increasing S ii 0.10
,/
/
'N
\\
flow rate
\\.
h 0.08
,/
\\s Rt:
\\
8 0.06
/
..'"N.
'\\.\\
e
\\
/
E
_. -~ ~ '
- s. w\\
\\
'~
m 0 04 NN S
0.02
~..w-. ;
0.00 100 150 200 s50 300 350, 400 450 500
~,-
Temperature (*F)
Figure 51. Influence of fluid velocity on flow-accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal-florowitz model(Chexal and Ilorowitz 1995). Copyright American Society of Mechanical Engineers; reprinted with permission. [1 in. = 25.4 mm; I ft/s = 0.3048 m/s; 0.5556 (* F - 32) =
- C).
Metallurgical Verlables. Trace amounts of other field experience data (Chexal et al.1996).
chromium, molybdenum, and copper in carbon The presence of chromium increases the stability steel provide resistance to flow-accelerated corro-of the oxide layer and reduces its solubility in sion. The corrosion rate is most sensitive to the deoxygenated feedwater, j
weight percent (wt%) of the chromium in the steel.
Figure 52 illustrates flow-accelerated corrosion Industry experience indicates that lightly alloyed rate dependance on chromium content of the base steel such as 1 1/4 Cr-1/2Mo steel (SA-335, Grade metal as predicted by the Chexal-llorowitz model Pl1),21/4Cr-lMo steel (SA-335, Grade P22),
for a 90-degree carbon steel elbow subject to Type 304 stainless steel and Alloy 600 will provide specific hydrodynamic and water chemistry vari-full protection against flow-accelerated corrosion, cbles. The maximum rate is equal to about 3.9 and are now being used in replacing the carbon mm/yr (0.155 in/yr) for 0.03 wt% Cr and equal to steel components. Flow-accelerated corrosion about 0.4 mm/yr (0.016 inlyr) for 0.50 wt% Cr.
rates for these steels are significantly lower than Hus a small amount of chromium significantly that for carbon steel with a nominal composition (0 reduces the flow-accelerated corrosion rate. Wall wt% Cr and 0 wt% Mo). The results of the thinning measurements for carbon steel feedrings Chexal-Horowitz model show that the corrosion and thermal sleeves at the four steam generators of rate for Grade Pl 1 steel is 39 times lower than that Diablo Canyon Unit 2 indicate that no wall thin-for the unalloyed carbon steel, for Grade P22 steel ning was found in components that had a chro-it is 76 times lower, and for Type 304 stainless mium content greater than 0.1 wt% (nailer, Dalal, steel it is more than 250 times lower (Chexal et al.
r.nd Goyette 1995), which is also supponed by the 1996).
1 i
DEGRADATION MECHANISMS 0.16' D = 41n E
0.14 -vY
= 7 at 77'F
-- - Cr = 0.20%
m
- - cr = 0.50%
O.12 J. 0.10
-l 0.08 chromium n Decreasing l
content
'l 0.06
. / ~'s
/
's fy 0.04
/
s
\\
g
,/
,.: ~ ~ ~s. N s
s
,..-(,..--~~..'
N ' s 0.02 4
.,,,' ~~,
0.00 * ~ '
100
.150 200 250 300 350 - 400 450 500 Temperature ('F)
]
4 4
Figure 52. Influence of chromium content on flow-accelerated corrosion rates for a 90-degree carbon steel elbow,' estimated using the Chexal-Horowitz model(Chexal and Horowitz 1995). Copyright American Society a
of Mechanical Engineers; reprinted with permission. [1 in. = 25.4 mm; I ft/s = 0.3048 m/s; 0.5556 (*F - 32)
< C).
4 i
i Environmental Variables. Two main As the temperature increases, the ferrous ion environmental variables. that affect flow-concentration at the oxide-water interface accelerated corrosion rate are coolant temperature decreases almost linearly. On the other hand, as and water chemistry. De water chemistry includes the temperature increases, the ferrous ion dissolved oxygen, ferrous ion concentration, diffusivity into the coolant increases, resulting in a metallic impurities, and cold pH level. His mass transfer coefficient which increases about i
section discusses the effect of these parameters on linearly. He resulting corrosion rate variation the flow accelerated corrosion of carbon steel with temperature is a bell-shaped curve as shown i
components in the single phase feedwater system.
in Figures 53 and several other figures in this Den it discusses how the optimum water section; the temperature at which the maximum i
chemistry for the PWR secondary water-steam corrosion rate occurs depends upon the other system is achieved.
environmental conditions. For most feedwater piping conditions, the maximum corrosion rate De./fuWtenqperatwe influences both the ferrous occurs at about 150*C (300*F) (Chexal and ion production and the mass transfer of these ions Horowitz 1995).
into the bulk water (Remy and Bouchacourt 1992).
NUREG/CR-6456 100
~
.n
DEGRADATION MECHANISMS 1.0 s,%.
/
\\.
f
\\.
Resultant influence f
\\
of fluid temperature
/
/
0.8 Arnrnonia treated Cold pH = 0.
\\
f Single phase flow f
\\,
y
/
E 3
\\
/
._$ 0.6 N'
h
/ N, a-
/
N N
/
0.4
/
'\\
,/
Influence on fenous q.N lon concentrations f
N
/
Influence on mass N
transfer coefficient j
O.2 50 100 150 200 250 300
~~~
Temperature ("C)
Figure 53. He calculated influence of fluid temperature on the ferrous ion concentration and on mass transfer of ferrous ions (Remy and Bouchacourt 1992). Copyright Elsevier Science Publishers; reprinted with permission.
The flow-accelerated corrosion rate varies Ferrous ion concentration and metallic impurities inversely with the level of dissolved oxygen in the in the water affects the flow-accelerated corrosion fluid. As the level of oxygen increases above a rate. The increase in theferrous ion concentration threshold value, a less porus oxide layer of in the bulk fluid reduces the mass transfer of hematite,instead ofmagnetite,is formed. Because ferrous ions from the oxide-coolant interface to the the solubility of hematite in the feedwater is bulk coolant.
An increased ferrous ion several orders of magnitude lower than that of concentration can reduce or suppress flow-magnetite, the flow accelerated corrosion rate accelerated corrosion when the corrosion process decreases significantly. Some laboratory test is controlled by mass transfer. The metallic results show that the threshold value for dissolved impurities such as metallic copper, precipitate into oxygen is less than 15 ppb (Remy and the pores of the oxide and reduce the flow-Bouchacourt 1992). The corresponding results accelerated corrosion rate for a short time period, using the Chexal-Horowitz model are presented in then the corrosion rates return to the original value.
Figure 54, which shows a reduction in the The presence of copper in the feedwater can be a maximum flow-accelerated corrosion rate from result of corrosion of copper alloys used in obout 3.2 to 0.9 mm/yr (0.125 to 0.035 inlyr) as feedwater system components such as condenser dissolved oxygen content increased from 10 to tubes and feedwater heater tubes (Chexal et al.
30 ppb.
1996).
101 NUREG/CR-6456
DEGRADATION MECHANISMS 0.16 D = 4 in.
V = M Ws O pb 0.14 pH = 7 at 77'F
- - -- 10 pob
- - 30 pbb Cr a Mo a Cu = 0.03%
3
- - 50 ppb 90' elbow
-s
.E 0.12
/
s SC
/
\\
84
/
\\
/
\\
g 5, 0.10 f
Decreasing t
/
s 3
E 0.08
/
\\
s owen
.g
,f s
content g$h
/
\\
0.06
,/
s
\\
0
\\
So g
0.04 N
i3 N.s s s 0.02 N
-.,,..s,,,...
.. y-Q j
~
0.00
.250
,300 350 400 450 500 100 150 200
~
. Temperature ('F)
Figure 54. Influence of dissolved oxygen content on flow-accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal-Horowitz model(Chexal and Horowitz 1995). Copyright American Society of Mechanical Engineers; reprinted with permission. [1 in. = 25.4 mm; I ft/s = 0.348 m/s; 0.5556 (*F
- 32) =*C].
i Flow-accalerated corrosion rates vary by an order perature, the flow-accelerated corrosion rate in-ofmagnitude over the coldpHrange of 8.5 to 9.5, creases rather dramatically as the pH is reduced which is typical for feedwater systems (Shack and below or increased above about i1.0. Figure 56 Jonas 1988). Figure 55 shows the effect of cold shows the dependence of the flow-accelerated pH on the solubility of magnetite (Fe30 )in deoxy-corrosion rate on the cold pH as predicted using 4
genated water over the temperature range of 25 to the Chexal-Horowitz model. The maximum flow-250*C (77 to 482'F)(Tremaine et al.1977). The accelerated corrosion rate decreases from 1.7 to solubility of magnetite directly correlates with the 0.38 mm/yr (0.0675 to 0.015 inlyr) as the cold flow-accelerated corrosion rate. First the solubility pH increases from 8.7 to 9.4.
decreases as the cold pH increases from 3 to a certain value, which depends on the temperature, The optimum pH and water chemistry in the PWR in the range of 10 to 12, then the solubility in-secondary steam water system are generally creases as the cold pH increases. For example, at achieved by the addition of reagents such as 150'C (300'F), the solubility decreases from ammonia, morpholine, and ethanolamine in the 30 ppm of ferrous ion concentration to 0.001 ppm demineralized water. These amines are volatile as the cold pH increases from 3 to 11. At this tem-and, therefore, maintain a slightly alkaline pH in d
NUREG/CR-6456 102
DEGRADATION MECHANISMS
'Os,
i 101
$ '" :r 250*C 100 200'C 108 [
}
150*C jp, 100*C 108 [
25'C M
5 100%
gna g 150%
g 107 [-
.6 3
200*C 2
250%
3 10s 1ga (
l I
i f 1
~
1 1
i I
jgg i
3 5
7 9
11 13 pH m 255 m ao2s7 Figure 55. Effect of cold pH on solubility of magnetite in deoxygenated water (Tresaine et al.1977).
C( pyright Elsevier Science Publishers; reprinted with permission.
0.07 D = 4 in.
8.7 @ 77*'/
O en = 7 ppb
- - - - 8.9 @ 77F
.O cu = 0.03%
_ -..- Q7M~
0.06 celiic 0.05 Tv4 U C ge s
s 0.04 f
b
~
g3 o8 s
Decreasing 5:
/
s cold pH
- e 0.03
/.
's
@.98
.# x
'x
- o. t:
s s
's. ' s 's S8 0.02
./
\\
s
.. s.
s, W
,/
s.
s O.01 i
0.00 100 150 200 250
,300 350 400 450 500 Temperature (*F)
Figure 56. Influence of cold pH on flow accelerated corrosion rates for a 90-degree carbon steel elbow, estimated using the Chexal-Horowitz model(Chexal and Horowitz 1995). CopyriP t American Society of h
Mechanical Engineers; reprinted with permission. [1 in. - 25.4 mm; I ft/s = 0.348 m/s; 0.5556 (*F - 32) =
- C).
j 103 NUREG/CR-6456
V DEGRADATION MECHANISMS both the single-phase and two-phase regimes of the alloy materials. Along with the use of morpholine, secondary system and reduce the flow accelerated hydrazine is used to lower the dissolved oxygen corrosion damage in the entire system. However, content in the feedwater.
However, the the amount of partitioning into the liquid and concentration of hydrazine is kept low because steam phases is different for the various amines.
thermal decomposition of hydrazine generates Therefore, different concentrations of these amines ammonia (Bouchacourt et al.1994),
are required to obtain a desired pH at the operating temperature. Different concentrations of amines As compared to a:nmonia, a smaller molar correspond to different cold pH (Chexal et al.
concentration of morpholine and a lower cold pH 1996). In addition to the amines, hydrazine is provide the same desired pH at the operating added as a reducing agent.
temperature. For example, to obtain a pH of 6.82 in the two-phase flow at an operating temperature The selection of the pH and water chemistry is of 175'C (347'F), use of ammonia will require a mainly a compromise between the acceptable cold pH of 10.0, whereas use of morpholine will corrosion of piping and components made of require e cold pH of 9.3. In addition, for the same carbon steel and copper alloys. Corrosion of cold pH, the flow-accelerated corrosion rate carbon steel is undesirable for three reasons: (1) associated with morpholine is lower than with the wall thinning of the piping and vessels by flow-ammonia (Chexal et al.1996).
accelerated corrosion, (2) the corrosion products transported to the steam generator, and (3) the A further increase in cold pH can provide addi-fouling of condensate polishers. Corrosion of tional resistance to flow-accelerated corrosion.
copper alloys is undesirable for three similar However, the pH level is limited by the use ofion reasons: (1) the shortened life of the copper based exchanger resins for condensate polishirg. Higher components, (2) the transport of copper and pH levels require higher amine concentrations, oxygen to the steam generator, and (3) the fouling which in tum requires more frequent changing of of the condensate polishers.
these resins, resulting in an increase in the amount of waste and in the operating costs of the ion Systems with an abrence of copper alloy materials exchangers. Therefore, different amines are being can maintain a higher cold pH level to reduce the tested to find one that is effective for the entire flow-accelerated corrosion of carbon steel compo-secondary system, including both single-phase and nents. If ammonia is used for pH control in these two-phase flow regimes. One such amine is PWR feedwater systems, the cold pH should be ethanolamine. The main advantage of ethanol-increased, at least up to 9.7, to avoid flow acceler-amine, as compared to morpholine, is the lower ated corrosion (Nordmann and Fiquet 1996).
molar concentration needed to get the desired pH llowever, if the condensate polishers are present at the operating temperature. Consequently, the and continuously in operation, the regeneration load on the condensate polishers will be lower and frequency will be high. Therefore, use of ammonia a reduced frequency of resin regeneration will be with a cold pH of 9.7 may be acceptable only for required. Therefore, the use of ethanolamine is plants without copper alloys and without conden-beneficial when condensate polishers are used. In sate polishers being continuously in operation.
addition, general industry experience indicates that corrosion products transport is lower with For feedwater system ; with copper alloy materials ethanolamine than that with ammonia' and even in the condensers and cernin water heaters, a cold lower than that with morpholine. A large number pli level in the ransc of 6.6 to 9.2 should be of plants are currently switching or contemplating maintained to prevent excessive copper pickup.
switching to ethanolamine. At this time, this i
Morpholine is widely used in these systems (all i
French PWRs and several U.S. PWRs) to maintain a cold pH in the range of 9.1 to 9.3, provided the i
condensate polishers are absent. Ammonia is not V. N. Shah. private communication with L. Goyette, used because it promotes corrosion of the copper Pacific Gas and Electric, Avila Beach. California. August 16,1996.
NUREG/CR-6456 104
1 DEGRADATION MECHANISMS j
amine or a combination of several amines are the The Chexal-Horowitz model has been validated by preferred choice for a large number of U.S. plants comparing the results with available laboratory (Chexal et al.1996).
data from EDF in France and CERL in the United i
Kingdom and data from 20 nuclear power plants.
The flow-accelerated corrosion rate is a function of The comparison showed that predicted flow-
)
amine type and temperature. According to the accelerated corrosion rates are within *50% of Chexal-florowitz model, the flow-accelerated measured rates. This model is incorporated in the corrosion rate in single-phase flow is highest for CllECWORKS, a computer code developed by the ammonia, lowest for morpholine, and in between EPRI. EPRI has developed several computer for ethanolamine. For example, for a 102-mm codes for estimating flow-accelerated corrosion (4-in.) diameter, 90-degree carbon steel elbow rates at power plants. All these codes are now l
(Cr-Mo=Cu=0.03 wt%) exposed to water having incorporated in CIIECWORKS. These codes are a cold pil of 9 and flowing at 6.1-m/s (20-ft/s) currently being used by all U.S. utilities. Field velocity, the flow-accelerated corrosion rates are applications of the CllECWORKS code is cbout 0.5,0.75, and 0.88 mm/yr (0.02,0.03, and discussed in Section 6.4.1.
0.035 inlyr), respectively, for ammonia, ethanol-amine, and morpholine.
- 8. INSERVICE INSPECTION OF PRESSURIZED WATER REACTOR FEEDWATER NOZZLES AND PIPING The ASME Code,Section XI, provides preservice spection results. Finally, ISI is performed at inspection (PSI) and inservice inspection (ISI) periodic intervals to detect service-induced requirements for Class I,2, and 3 components. As degradation such as cracking and corrosion in the a Class 2 component, the feedwater piping inside welds and adjacent base metal.
the containment is subject to surface and volumet-ric examination in accordance with the Code.
The fabrication examinations of Class 2 piping These examinations focus specifically on the welds welds are performed in accordance with the appli-and the base metal immediately adjacent to the cable construction code, that is, USAS B31.1, welds and are not sufYicient for detecting damage USAS B31.7, or ASME Section III. These exami-caused by flow-accelerated corrosion or thermal nations typically include a 100% radiographic fatigue in the base metal away from the welds examination.
(flow-accelerated corrosion and fatigue caused by thermal stratification were not considered when the Preservice and inservice examinations are Code rules were developed). This section exam-perfonned in accordance with appropriate editions ines past and present ISI requirements and activi-of the ASME Code,Section XI, specifying a ties related to the feedwater system. Also, relevant volumetric examination of the inner 1/3 volume of and emerging technologies for examination of the weld and adjacent base metal for a distance of piping systems are discussed, and the inservice 6 mm (0.25 in.) from the edge of the weld crown, inspection experiences at non-US PWR plants are and surface examination of the outside diameter summarized.
(OD) surface of the weld and 13 mm (0.5 in.) of the adjacent base metal. These requirements are 8.1 Inservice inspection 8Pplied to piping welds including those at the structural disc ntinuities such as vessel nozzle-to-Re9uirements piping welds. Early editions of the ASME Code required a volumetric examination of the entire The nondestructive examination requirements weld volume. In the Summer 1976 Addenda, the during fabrication of the feedwater piping welds, Code was revised to the current inner 1/3 the preservice and inservice inspection require-volumetric examination and surface examination, ments, and the revisions to these requirements are because it was determined that degradation discussed in this section. We also summarize the initiating on the inside or outside surfaces was the mscruce mspections performed in response to IE Bulletm 79-13 and the inspection related activities primary concem, rather than imbedded fabrication since close-out of the bulletm.
defects that do not tend to propagate (Bush 1980).
Appendix III of the ASME Code,Section XI, 8.1.1 Fabrication, Preservice inspection, specifies the oasic technique and contains and inservice Inspection Requirements calibration and examination requirements. This includes specifications for the maximum Current industry inspection practice ic!!es on three transducer element sizes and beam angles. For levels of nondestructive examinations (NDE) t most pipe weld inspections, a beam angle of 45 estahlish confidence m the pipmg system mtegrity.
degrees is used. Other angles are allowed if the These levels melude examinations of the ulds at thickness or geometry impedes the effective use of the time of fabrication, prior to service, and at a 45-degree angle beam examination. Appendix til also requires that scanning be performed at a certam mtervals during service.
Fabrication examinations are performed to establish the imtial minimum of 6dB (a factor of two) above the quality of the weld. This is followed by a PSI t primary reference level established during calibration' establish a baseline to compare future inservice in.
NUREG/CR-6456 106
INSERVICE INSPECTION ISI examinations are required at 10-year intervals, Codes andStandards rule currently endorses by and are performed on a minimum of 7.5% of the reference the inservice inspection rules in the 1989 carbon steel and low-alloy steel welds in non-Edition of Section XI of the ASME Code. Efforts cxempt' Class 2 piping, but not less than 28 welds.
are underway to revise this regulation.
Welds selected for examination must be distributed among terminal ends and structural discontinuities 8.1.2 Inspections in Response to IE and prorated among systems. Typically, each plant Bulletin 79-13 h:s several hundred to 1000 nonexempt, Class 2 welds; therefore,28 to 75 welds per unit are As stated in previous sections, the cracking discov-examined. Welds examined during the initial 10-cred in all eight feedwater lines at D. C. Cook year interval are re-examined during successive Units I and 2 (revealed by leakage and subsequent intervals. In addition to the piping welds, the Code radiographic examinations) prompted the NRC to requis es a volumetric examination of the nonle-to-issue a letter (USNRC 1979b) to all PWR owners, j
v:ssel welds and the nonle inside blend radii for advising them of the subject of feedwater piping Class 2 vessels. For multiple vessels of similar cracking and requesting information on the fabrica-design, such as the steam generators, examinations tion and operating history, and preservice and i
may be limited to nonles of one vessel or the inservice inspections performed for the feedwater l
equivalent of one vessel distributed among the system. Later, the NRC issued IE Bulletin 79-13 vessels.
(USNRC 1979a), requesting volumetric examina-j tion of the feedwater nonle-to-piping welds at all Although PSI and ISI of Class 2 components such PWR facilities with steam generators fabricated by as feedwater piping are currently required across Westinghouse and Combustion Engineering that the nuclear industry, they were not part of Section had not received volumetric examination since XI prior to 'he Winter 1972 Addenda of the 1971 May 1979. Affected licensees had 90 days to Edition. Thus, when cracking was discovered in conduct the examinations. The basis of the bulle-the D. C. Cook nonles, many plants had not tin was, "the identified degradation of these joints performed preservice or inservice examinations of in the absence of a routine inservice inspection the feedwater system. The lack of preservice requirement." The bulletin recommended a radio-examination data is significant as many plants did graphic examination, supplemented by an ultra-not have baseline UT data for the UT examiners to sonic examination of all feedwater nonle-to-use to identify the geometric reflectors, such as the piping welds and adjacent base metal.' If cracking counterbore and weld root, and discriminate them was found in these areas, additional examinations from the service-induced defects. This is espe-were specified for the remaining welds in the cially important in the case of the feedwater piping feedwater system piping up to the first piping cracking, wnich has generally initiated at geomet-support or snubber and at high stress points located ric discontinuities such as the counterbore corner.
inside the containment. In addition, the feedwater Also, the examination volume at plants that are system piping supports and snubbers in contain-performing inspections of welds in accordance ment were to be visually inspected to verify opera-with later editions of ASME Section XI, may not bility and conformance to design.
extend far enough to include the discontinuity at the counterbore corner, llowever, some utilities The bulletin also specified that all licensees with cre including an examination of the counterbore PWR facilities perform additional volumetric comers in the ISI of the feedwater system. The examinations during the next outage of sufficient Code of Federal Regulations,10 CFR 50.55a, duration or at the next refueling outage afIer the inspections mentioned in the preceding paragraph are performed. Some requirements for these
' Catain piping systems are exempt from examination based additionalexaminations follow: for steam genera-on size and function. With the exception of the high pressure tor designs with common main and auxiliary safety injection system (IIPSI), piping 4 NPS (nominal pipe feedwater noules, insE#ction of all feedwater size) and smaller is not subject to exammation. l'or the IIPSI, nonle-to-prping weld areas and all feedwater piping 1 1/2 NPS and smaller is exempt from examination.
107 NUREG/CR-6456
INSERVICE INSPECTION piping weld areas inside containmert was required; 8.1.3 Activities Since Close-out of IE also, if the anxiliary feedwater line was connected Bulletin 79-13 to the main feedwater line outside the containment, volumetric examination of weld areas connecting IE Bulletin 79-13 was closed in February 1991, these two lines was required; and for steam genera-based on actions implemented by licensees as a tors with separate auxiliary and main feedwater result of the bulletin. As part of those actions, a nozzles, volumetric examinations of all welds number of utility ISI programs incorporated inside containment and upstream of the feedwater augmented volumetric examinations consisting of nozzle for each steam generator were required.
radicgraphic and/or ultrasonic examination of feed-water piping weld areas and visual examination of As a result of these inspections, cracking was feedwater piping and piping supports inside detected in 18 operating PWRs. The cracks were containment. As a result, degradation was detected generally oriented circumferentially and located at several facilities and is listed in Appendix B of i
either at the weld root location or at the reentrant the bulletin closeout document (Foley, Dean, and corner of the counterbore on the upstream (pipe)
Hennick 1991).
side of the nozzle weld, as shown in Figure 57.
The cracking was determined to be caused by Events since the 1991 closcout of the bulletin have thermal fatigue, possibly assisted by the corrosive illustrated the potential weaknesses in ISI tech-environment of the secondary system. As shown niques and practices. In March of 1992, through-
)
in the Ogure, cracking occurred at some plants at wall cracking was discovered in a Sequoyah Unit the counterbore comer outside the Code examina-I feedwater nozzle-to-transition piece (pipe) weld.
tion volume and, therefore, could have been over-Subsequent RT examinations revealed that Eve of looked during previous periodic inservice inspec-the eight nozzles (at Units 1 and 2) contained tions. Several plants also noted pitting, which was signi6 cant cracking. All these welds had been revealed by destructive analysis. Code ultrasonic previously examined ultrasonically.
Further examination methods employ angle beam tech-investigation by NRC inspectors revealed that the niques (i.e., 45-degree shear waves) designed to UT examinations were conducted using the mini-detect cracks but not pits. Pit detection requires mal Code requirements and the indications had use of specialized, small-diameter, focused 0-been incorrectly identiGed as root geometry. No degree probes.
supplemental or enhanced tecianiques were used to Weld Examination volume Dsepest cra'ckj E 6
17 mm 21.4 mm 3)'
(0.656 in.)
(0.643 in.)
i i y 1
J.-.
Nozzle end Pipe end
[406 mm (16 in.) Schedule 60]
[406 mm (16 in.) Schedule 80] - Counterbore m.= mm Figure 57. Typical PWR feedwater nozzle-to-pipe weld with geometric discontinuity (counterbore) and cracking outside the Code examination volume.
' A distance of at least two wall thicknesses.
NUREG/CR-6456 108
INSERVICE INSPECTION verify the indications. As a result of this cracking, this report, a through-wall crack at Sequoyah Unit the licensee upgraded its procedures for 1 was not detected during UT examinations using conducting ultrasonic examinations and expanded IGSCC-qualified techniques and procedures.
the examination volume for welds having the Unoptimized IGSCC-qualified techniques and the potential for being subjected to thermal inspectors' lack of experience with thernal fatigue stratification. The expanded volume includes the cracks were cited as the cause.
welds plus adjacent base metal for a distance of two wall thicknesses.' In addition, the procedure Some optimization of the IGSCC-qualified requires 100% examination of the steam generator techniques is necessary to accommodate material nozzle transition piece. The licensee has also differences and to ensure reliable detection of incorporated into their procedures a number of cracking outside the heat-affected zone. First, enhanced ultrasonic techniques to aid in the carbon steel components exhibit less attenuation to svaluation of detected indications.
ultrasound than stainless steel components. Thus, higher frequencies which provide greater Another incident involving misinterpretation of sensitivity for detecting small defects, can be used crack indications occurred at Diablo Canyon, Unit for carbon steel. Second, an optimum scanning 1.
These cracks were oversized by ultrasonic sensitivity can be established for thermal fatigue exaniination, which resulted 'in removal of the cracks in carbon steel to provide more reliable affected weld. Subsequent metallurgical results examination procedures. As mentioned, the Con indicated that the UT had oversized the 0.86-mm requires a scanning sensitivity of at least 6dB (0.034-in.) deep crack by a factor of ten owing to above the primary reference level. For IGSCC, inclusions in the weld area (PG&E 1992). These industry practice is to use additional instrument same cracks were not detectable with RT. In a gain. Such scanning is performed with the follow-up meeting with the NRC, the licensee material grain noise amplitude set at 5 to 10%
concluded that Code examinations were not (EPRI 1983). This approach was intended to adequate for small thermal fatigue cracks and maximize the low amplitude responses from the enhanced ultrasonic techniques, including IGSCC. Third, since cracking often occurs at automated scanning, were necessary to improve geometric discontinuities outside the heat-affected reliabilly, accuracy, and repeatability (Peterson zone, inspectors can be trained to look for and 1992),
expect cracking in areas away from the weld.
Several licensees are now using enhanced The concept of using enhanced UT techniques for inspection techniques, such as tip-diffraction and thermal fatigue is not new. As a result of through-creeping wave techniques, and providing training wall cracking in stainless steel piping, caused by to their inspectors using the removed portions of thermal stratification, the NRC recommended in the damaged feedwater piping.
NRC Bulletin No. 88-08, Supplement 2, the use of enhanced UT techniques and procedures, and 8.2 Advancements for Inservice stated that UT procedures that qualified for the Inspectlod of Fatigue Cracks detection ofIGSCC in BWR recirculation pipmg hase demonstrated effectiveness for detecting The different m. service mspection techniques for thermalfatiguecracks(USNRC 1988d). Although detection and sizmg thermal fatigue cracks m the use ofIGSCC-qualified techniques can provide feedwater piping are discussed in this section. We improvement over basic Code-prescribed also discuss the selection of the inservice techniques, these enhanced techniques may not be inspection locations entical to plant safety or with effective "as is." As stated in Section 6.3.1.2 of high failure probabilities.
8 The extended examination of the adjacent base metal may not include the reentrant corner of the counterbore at some PWRs.
109 NUREG/CR-6456
INSERVICE INSPECTION 8.2.1 Emerging inservice inspection used to construct a distance-amplitude correction Methods (DAC) curve used to compensate for the attenuating effects of the material. This DAC Nuclear plant operators have the choice of two curve is considered the primary reference level.
volumetric examination methods, RT and UT, for Reflectors that produce a response greater than the the detection of cracks which initiate on the pipe specified percent of DAC are investigated and inside surfaces in a weld or adjacent base metal.
recorded. In the 1986 Edition of Section XI, this Each method has advantages and disadvantages, recording level was lowered from 50% to the but the information provided by each method can current 20% DAC, largely due to the work complement the information provided by the other performed in the Program for Inspection of Steel method. In particular, radiography can provide Components (PISC) project that indicated that the reliable information about the geometric 50% recording level was unreliable and that characteristics of the inside surface, and enhanced recording levels lower than 20% DAC produced an ultrasonic examinations can provide reliable excessive number of false calls (PISC-II 1985). In detection and sizing of fatigue cracks.
addition, a requirement was added to record any indication of a suspected flaw regardless of The inherent advantage of radiography is that the amplitude. In accordance with the Code, the testing can be performed through the insulation endpoints or length of the flaw indication are and a permanent record is obtained, which can be estimated by moving the ultrasonic transducer compared with future examinations. The resulting parallel to the reflector until the signal amplitude is image can also be used to characterize the weld reduced 1o 50% of the maximum signal amplitude.
geometry.
Many disadvantages stem from As will be discussed later, a similar amplitude-convenience factors, including:
radiological based approach was used in the past to determine controls that may interfere with critical path the through-wall extent of the flaw indication, but activities, interference from contaminated and has been abandoned for more reliable techniques.
irradiated components, and access to the inside surface (which is available only at plants with The conventional amplitude-based UT techniques gamma plugs installed in the pipe wall adjacent to can detect fatigue cracks in ferritic steel piping but the feedwater nozzle, otherwise, double wall have a very poor sizing capability for these cracks.
techniques with reduced sensitivity must be used).
This is illustrated by the round robin UT j
The final consideration is that RT is arguably a less inspections of thermal fatigue cracks in LWR sensitive method for crack detection compared to primary coolant piping conducted in 1981 using other NDE methods and is generally not capable of the 1977 Edition of the ASME Code,Section XI.
determining the through-wall extent of the cracks.
He results for crack detection reliability are shown Although RT is sensitive to defects that are in Figure 58 and those for crack sizing accuracy volumetric in nature (e.g., wall thinning, slag shown in Figure 59 (Muscara 1990).
The inclusions, etc.), the density difference caused by inspection results presented in Figure 58 show that aight crack may be insufficient for detection if the conventional UT can reliably detect thermal oriutation of the crack is not parallel to the fatigue cracks in clad ferritic steel piping whereas gamma or x-ray. However, the fatigue cracks in it usually can not detect such cracks in the feedwater piping are often open and filled with centrifugally cast stainless steel piping. This figure oxides and can be detected with radiography.
also shows that conventional UT can not detect IGSCC cracks in wrought stainless steel piping as UT is the primary volumetric method used for reliably as thermal fatigue cracks in clad ferritic inservice inspection, and, in accordance with steel piping. For example, the probability of ASME Section XI Code (1989), flaw detection is detecting a 20% through-wall crack was about achieved by comparing the signal amplitude of a 80% for clad ferritic steel, 55% for 250-mm flaw indication to that of a known reflector, such as (10-in.) wrought austenitic steel, and 20% for cast a notch or a hole in a calibration block. He austenitic piping, l
responses from known calibration reflectors are NUREG/CR-6456 110
I INSERVICE INSPECTION i i i i i i i i i I ' i ' i
^
1.0 ii,iiiiiiie i
Clad Ferritic f
t
~
0.8
~
~
0.6 Austenitic Pipe (tGSCC)
~
~
O N
~
2 2
0.4 1
a.
2
,x-
=
0.2 Cast Austenitic Pipe
,i,,,i1 i i i i I i iiil i i i '
,,, i,,,
O 10 20 30 40 50 60 0.0 '
Through-wall crack depth (%)
ca > win. mon Figure 58. Probability of detection versus depth of thmnal fatigue cracks in clad ferritic and cast austenitic pipe and ofIGSCC cracks in wrought austenitic pipe - piping inspection round robin results (Muscara 1990).
Copyright ASM International; reprinted with permission.
1.0 i i i i i i i i i, i i i i i i i i i,,3 i ii j'
,/
0.8
,f'
.i
/
Tenm 2
/
a 6 0.6
,/
e 1
- 4
/
ox 0.4
,/
5 ~
i s
v>
/
2.
s'
~
0.2
/
3
/,/
4 e
s' 0.0 '0
.'6 O.
0.2 0.4 0
0.8 1.0 1.2 PNL depth estimate (in.)
wo Figure 59. Depth of thermal fatigue cracks in clad ferritic material as estimated by the six inservice inspection teams with the conventional UT techniques versus that estimated by Pacific Northwest Laboratory with the enhanced UT techniques. 'Ihree destructive analysis results (D) are also plotted (Muscara 1990). Copyright ASM International; reprinted with permission. (1 in. = 25.4 mm) 111 NUREG/CR-6456
INSERVICE INSPECTION However, it is difficult to accurately size thermal developed Appendix VIII, which is discussed later fatigue cracks in ferritic steel piping using conven-in this section. Several enhanced inspection tional UT techniques as shown in Figure 59 techniques have been developed for reliable detec-(Muscara 1990, Doctor 1984). The figure shows tion and sizing oflGSCC.
a comparison between the crack depths estimated by the six ISI teams using the conventional UT Knowledge gained from inspecting for IGSCC has techniques and those estimated by the Pacific provided improvements in three areas that apply to Northwest Laboratary (PNL) staff using enhanced all examinations ofpiping welds. These areas are:
UT techniques such as tip diffraction techniques.
(1) improved search units employing a variety of Figure 59 also shows that the PNL crack depth angles, modes, and configurations; (2) proactive estimates compare well with three destructive procedures relying on numerous techniques to evaluation results. Accurate crack sizing is needed verify and discriminate inservice degradation from for evaluation of the structural integrity of flawed weld geometry; and (3) the use of automated piping. The national and international studies inspection equipment to improve reliability and coordinated by the Pressure Vessel Research repeatability, and to provide a permanent record of j
Committee and the PISC, respectively, have also the examination. Carbon steel is less attenuative 1
revealed the poor capability of conventional UT for acoustically than stainless steel; therefore, the same sizing of flaws in pressure vessels and nozzles, angles, modes, and configurations can be used to These studies indicate that there are large variabili-search for thermal fatigue cracks in carbon steel, ties between different UT procedures.
with small adjustments in frequency to optimize for the material di:Terences.
Although UT is capable of detecting many types, sizes, and orientations of cracking, manual UT Improvements in search unit technology have examinations have two inherent disadvantages:
moved far beyond the 45-degree shear wave reliance on the inspectors' ability and judgement, approach prescribed by the Codm Manufacturers and the lack of a permanent record. For the now build numerous types of search units with feedwater piping, these problems resulted in different wave modes (shear, longitudinal, and inconsistent results, miscalls of both cracks and multimodes), angles, and configurations (single, weld geometry (often owing to the lack of baseline dual element, tandem, phased arrays), designed to data), and a general lack of confidence in UT.
enhance the sensitivity of their equipment to Because of these inconsistencies and miscalls, the cracks. Supplementing the Code techniques with limitations of the Code-prescribed techniques are these enhanced inspection techniques and using a well recognized. These failures illustrate the need proactive approach to examining components is far to move beyond the minimum requirements of the more effective than use of the Code requirements Code to provide a more reliable examination, alone.
driving industry to develop enhanced inspection techniques.
Another improvement to consider is the use of automated inspection equipment to collect and Inservice inspections ofIGSCC in BWR primary store ultrasonic data. Automated scanning of the piping have also revealed some of the shortcom-feedwater nozzles was performed at San Onofre ings of the conventional ultrasonic examination Unit 3 using the introspect/98 volumetric inspec-methods. The inability of ultrasonic inspection tion system (Mostafa and Ramsey 1994). Using methods to detect IGSCC in the thickwall, large-computer processing, a 3-dimensional evaluation diameter, stainless steel recirculation piping at the of the data was performed to enhance the ability of Nine Mile Point Unit I resulted in the USNRC an analyst to characterize flaws and discriminate requiring specific performance capability demon-geometrical reflectors on the inside surface. Such strations for this inspection technique (USNRC enhanced evaluations are becoming routine and are 1982,1983a). In response to this need, the ASME possible with many modern-day scanning systems.
Section XI Subcommittee on Inservice Inspection NUREG/CR 6456 112
INSERVICE INSPECTION Creeping Wave and Related Mode L, and L) as a qualitative sizing tool, can provide Conversion Techniques.
An approach inspectors with additional confidence when evalu-developed for IGSCC inspection that can be ating crack indications and help avoid miscalls.
effective for the inspection of feedwater piping For example, as described in a previous section, with thermal fatigue is the use oflD creeping wave shallow cracks detected at Diablo Canyon, Unit 1, and related mode conversion techniques to detect were oversized owing to the presence ofinclusions and qualitatively size the fatigue cracks. This that were mistaken for the crack tip. Use of the family of techniques has gained wide acceptance in multimode evaluation approach can help avoid this the nuclear industry because ofits high sensitivity type ofinterpretation by allowing the inspectors to to ID connected flaws. The ID creeping wave is confirm the crack depth in a qualitative, but reli-a generated as a result of a high-angle refracted able, manner. The disadvantages of this approach longitudinal wave. When an angle ofincident, a, are the complexity of signal interpretation and for a lor,gitudinal wave is slightly less than the first potential interference from nonparallel ID and OD critical angle,4 our modes of ultrasonic waves omponent surfaces that may cause redirection of f
ne generated below the front surface (see the reflected and mode converted waves.
Figure 60): shaar wave S at the refracted angle p,,
j longitudinal wave L at the refracted angle pi, Tip-Diffraction Techniques. A widely j
creeping wave Cr along the front surface (also used method for crack depth sizing is the use of a called an OD creeping wave), and an indirect shear diffracted wave from the crack tip. Techniques wave, S,, radiating from the creeping wave Cr that use the tip-diffracted signal are generally (Brook 1986, Davis 1991). Mode conversion takes known as tip-diffraction techniques. These tech-place at the back surface because it is parallel to niques measure the time-of-flight to the crack tip to the front surface, and the indirect longitudinal estimate the crack depth. As a result of the im-wave, L,, along with the shear wave, S, are pinging sound beam, a diffracted wave is produced reflected from the back surface. In addition, the from the crack tip that radiates like a point source indirect shear wave mode converts at the back in all directions. Thus, the diffracted wave can be surface to produce a back surface creeping wave, detected as it scatters backward or forward. An C, which is also called an ID creeping wave example of a backscatter technique is shown in (Brook 1986). Creeping waves can be used to Figure 61(d), using the direct L-wave from the detect very shallow OD or ID defects, as shown in creeping wave probe. He sound beam is transmit-Figure 61(a) and (b) respectively, provided that the ted and received from the same side of the crack.
OD and ID surfaces are parallel.
Other examples of backscatter tip diffraction techniques are the Satellite Pulse Observation In addition to the creeping wave, the multiple Technique (SPOT), which measures the difference modes and high angle longitudinal wave make it in time-of-flight between the tip and a corner possible to characterize and size cracks with the signal, and PATT (pulse arrival time techniques) i same search unit, as shown in Figure 61(c)(Brook which measures the direct time to the tip signal.
1986, Davis 1991). An indirect longitudinal wave (L) is generated at the ID surface, which, if re-An example of a forward scattering tip-diffraction flected off the crack face, returns to the transducer technique is the time-of flight diffraction (TOFD) as a longitudinal wave. However, a shallow crack method (Pers-Anderson 1993). His is a crack-tip would not cause reflection of the indirect longitu-diffraction method that employs opposing ele-dinal wave. Thus, the indirect longitudinal wave ments, as shown in Figure 62(a). Two signals are can indicate the presence of deeper ID cracks (for present in the absence of a crack: a direct lateral example, deeper than 25% through-wall), which wave signal and a back wall reflection signal from may be confirmed by the presence of a tip-dif-the ID surface. Diffraction occurs when the in fract :d signal from the direct L wave, as shown in coming sound beam impinges upon a finite planar 11gure 61(d). Using this multimode approach, the reflector such as a crack. The diffracted sound presence or absence of c enain signals (i.e., Cn C.,
energy from the crack tip acts as a point source and 113 NUREG/CR-6456
INSERVICE INSPECTION l
85G p
er
\\
Front surface (OD)
Cf S/
i p,
L S
S Es e
s l
Ob Back surf e (10)
M95 0217 Cr Front surface creeping wave Cb Back surface creeping wave L
Direct longitudinal wave Li Indirect longitudinal wave l
S Shear wave Si indirect shear wave a
incident angle Criticalincident angle acr pg Refraction angle for longitudinal wave, L s
incident angle at back surface Figure 60. Scheme of ultrasonic fields of creeping wave probe in test specimen depicting front and back surface creeping waves, direct and indirect longitudinal waves, and direct and indirect shear waves (Brook 1986). Copyright ASM International: reprinted with permission.
NUREG/CR-6456 114
I I
INSERVICE INSPECTION -
)
i i
I l
Front surface F Cf (a) Detection of 00 crack using front l
surfacecreeping wave (Cg)
Back surface j
I l
(b) Detection of ID crack using back surfacecreeping wave (C )
i b
S l
L l
(c) Crack depth classification using S
indirect longitudinal wave (L;)
Li M
4 l
(d) Crack sizing using tip-diffracted direct longitudinal wave (L)
M95 0218 l
Figure 61. Potential beam propagation paths for multimode approach using creeping wave probe (Brook 1986). Copyright ASM International; reprinted with permission.
9 4
115 NUREG/CR-6456
INSERVICEINSPECTION Probe center separation,
l' i
_1 4
r 1
(a)
^^
No crack I
4 w
2 Signal 1 Lateral wave i
Signal 4 Backwall
_1 1
2 (b) inside-diameter y
crack Signal 1 Lateral wave Signal 2 Upper crack tip r
(c)
Through-wall crack No signal 1
-1 4
l 2
1 2
3 (d)
Embedded "y'
^
v I
crack Signal 1 Lateral wave Si nel 2 Upper crack tip Si nel 3 Lower crack tip Si nel4 Backwall 4
Figure 62. Examples of time-of-flight (TOFD) diffraction signals (Pers-Anderson 1993). Copyright TRC; reprinted with permission.
NUREG/CR-6456 116
INSERVICE INSPECTION radiates a sound wave to the receiving transducer.
use of an imaging technique such as SAFT-UT can The time of arrival of this signal can then be used provide accurate sizing of a crack.
to pinpoint the tip of the crack and determine crack depth. Figure 62(b) illustrates such a Phased Array Technique. Ultrasonic diffracted signal produced by the tip of an inside-examination methods based on the phased array diameter crack; note the absence of a back wall technique have been developed for inservice reDection signal and the presence of a lateral wave inspection of components such as feedwater noz-signal. As shown in Figure 62(c), a through-wall zies which have a complex geometry, and have a crack would of course eliminate all signals, and all very limited access and clearance. One such signals would be present for an embedded crack, as technique developed by Siemens has been used for shown in Figure 62(d). This approach provides a inspection of the BWR feedwater nozzle inner means of both sizing and verifying the extent of radius region, nozzle bore, and nozzle to vessel the crack, but can be hindered by weld geometry weld; the BWR bottom head ligaments; and the on the OD and ID surface which can cause a loss PWR closure head ligaments (Rathgeb et al.1992).
of surface contact and/or the loss of back wall and This technique has also been used to inspect the lateral wave signals for reasons other than the inner radius regions of PWR feedwater nozzles.
presence of a crack.
A phased array transducer consists of multiple The improvement in sizing technology using the elements that can be controlled individually to time-of flight techniques is demonstrated by the create a variety of beam patterns. The use of data from the PISC 11 trials, which, as discussed multiple elements with a computer controlled above, showed that amplitude-based methods were pulsing sequence results in the ability to steer unreliable for determining the through wall extent and/or focus the sound beam, as shown in of a crack (Cowfer 1989). Since that time, numer-Figure 63 (Bray and Stanley 1989). With an ous transducer con 6gurations (e.g., tandem, dual, appropriate phase-shifting of the transducer ele-multimode) have been developed to optimize the ments, the focal length of the transducer can be response from the crack tip and other portions of changed and the specimen can be scanned in the crack (i.e., the tip, face, and base) to aid in depth. The transducer design can be tailored to the crack sizing. Back scattering and time of Dight-needs of the specine examination. For example, diffraction techniques are just two approaches to the examination of a nozzle inner radius region
- tipddfraction techniques developed in recent employs a fixed incident angle with a variable yem.
skew angle whereas the vessel shell welds require a fixed skew angle with a variable incident angle.
The tip-diffraction techniques and other UT tech-niques can overestimate the through-wall extent of Data collected during the inspection are digitized i
a crack in carbon steel feedwater piping ifinclu-and then stored on removable computer hard disks.
sions such as MnS are present near the plane of a These data are analyzed using a personal computer crack. For example, a tip-diffraction technique program. The program provides several different overestimated the depth of crack in the carbon steel displays to facilitate accurate analysis of reDectors:
feedwater piping at the Diablo Canyon Unit 1 A-scan, B-scan, C scan, and UT Echo Tomogra-because of inclusions present near the crack tip.'
phy, which requires that the echoes received in Therefore, discrimination of reflectors (for exam-many cross-sectional directions are stored during ple, inclusion, geometry effect, and crack tip) and inspection. The echo tomograph utilizes the spatial use of more than one inspection technique for relationships of the signals in order to enhance the confirming the inspection results is advised. Also, signal to noise ratio. The combination of these display modes allows a rapid and accurate analysis of the reDectors. Flaw sizing is typically done with V. N. Shah, Private consersation wi6h Dase a tip diffraction method (Fischer et al.1992).
Gonzalez Pacific Gas and Electric, June 19.1996, 117 NUREG/CR-6456
i INSERVICE INSPECTION -
io m fgy Controller
/8/l Sh\\
Pulsers --+ A B
C D
E F
G H
I J
/ robes P
w +-
-+
~ CAIl
/
A B
C D
E F
G H
I J
e = Inclination angle -
= Sin *1 E w
Wave front Where C = Sound speedin the material
%e At = Phase delay in pulsing each probe w = Probe center-line spacing M98 0337 Figure 63. Typical phased-array transducer (Bray and Stanley 1989). Copyright McGraw-Hill; reprinted with permission.
Another example of the application of the phased to the inaccuracies of as-built drawings, weld array technique is the TestPro/ FATS system profiles of each weld examined are a necessity.
(Bisbee 1994). FATS (Focused Array Transducer One method of obtaining this is by plotting thick-System) is a phased array technique, which allows nesses obtained by ultrasonic measurements. An the beam to be focused electronically to the area of alternate method used in Japan to visualize the interest. This method reduces beam spread and weld geometry is the use ofcomputed tomography t
allows the beam to focus on the crack opening to (CT) imaging of radiographic data (Maeda and j
enhance detection, or focus on the crack-tip to Yagawa 1991). The advantage of this approach is improve sizing accuracy. The focused array that an accurate cross section of any section of the transducer system, combined with the TestPro weld, including the reentrant corner of the counter-ultrasonic acquisition system, results in a complete bore, can be obtained. The disadvantage is the system for acquisition, analysis, imaging of expense and time required for performing such an ultrasonic data, and interfacing with scanning examination.
devices.
Synthetic Aperture Focusing Tech.
Computed Tomography. One of the nique for Ultrasonic Testing (SAFT-UT).
problems with performing UT examinations on SAFT-UT is an effective technique for characteriz-piping welds is discriminating the weld geometry ing flaw size, shape and orientation. This tech-firom the cracks. Therefore, an accurate representa-nique uses reflected ultrasonic waves to form a tion of the weld geometry is important. Owing to high quality image of a detected discontinuity that variation in weld geometry from weld to weld and can be easily interpreted. The ultrasonic pulses NUREG/CR-6456 l18
INSERVICE INSPECTION l
J returned from a test' sample contain much more (0.17 to 0.44 in.), were 95% and 18%. SAFT team useful information (both amplitude and phase performance was better than the average of the information) about the interaction between the manual or automated inspection teams participat-ultrasound and the material than is commonly ing in the Miniature Round Robin.1 i
extracted from conventional ultrasonic displays.
j The image formed by SAFT-UT contains this ne blind tests cWM at the EPRI NDE Center
~ information, which can be used to characterize were for evaluation of SAFT-UTs capabilities for j
many material discontinuities including thermal IGSCC detection and sizing. SAFT-UT detected fatigue cracks.
seven out of10 defects. De detection tests identi-fied several weaknesses of the system and where De SAFT-UT method is based on the concept of improvements were needed. Dese improvements '
i collecting data from scanning with a small, highly -
have been made. The sizing tests indicated that divergent (unfocused)transducerelement. De use SAFT-UT has a tendency to undersize the deeper of a large divergence is based on the idea that the cracks.
l more often a given discontinuity is seen by a l
transducer, the more information can be gathered Inspection of the PISC-III full scale reactor pres-(bout the discontinuity, and more precisely the sure vessel was selected as a blind test for evaluat-discontinuity can be characterized. The first step ing SAFT-UT. De test involved characterization in synthetic-aperture data processing is to choose of twelve defects in the vessel. The SAFT-UT a set of transducer element positions to be pro-system was used to size 8 of the 12 defects. There 4
cessed as a unit. The data are then processed elec-was a reasonable agreement for the size of six of tronically by introducing appropriate time shifts the eight defects. The comparison cf the SAFT.
l into the inspection data from each position, so as to UT system with the other inspection systems will simulate a larger. transducer with much better be made after the detailed analysis of the inspec-resolution and signal-to-noise ratio. De electronic tion results, which is currently in progress, is
[
processing of the data results in focusing of the completed. He SAFT technology has been inte-l collected data. Thus the processing of the data in grated into the ASME Code Section V, Article 4, effect synthesizes the physical processes occurring Appendix E, which addresses computerized UT in a large, but unobtainable, transducer (aperture),
imaging systems which also includes UT-phased Therefore, this method is referred to as the syn-arrays (Doctor et al.1995).
j thetic-aperture focusing technique (Seydel 1982).
)
Field application of the SAFT technology has Three different tests have been performed since included an inspection of a plate in the PISC II 1989 to validate SAFT-UT: (1) participation in program, detection and sizing of IGSCC in the the Miniature Round Robin conducted at the BWR recirculation piping at Dresden Unit 3 and Pacific Northwest Labore ry, (2) tests at the EPRI Vermont Yankee, characterization of an indication NDE Center, and (3) test conducted as part of the in Indian Point 2 reactor pressure vessel, and a Programme for the inspection of Steel Compo-special ultrasonic examination of a statically cast nents, Phase 3 (PISC-Ill). The participation in the stainless steel hot-leg elbow at Trojan, ne SAFT Miniature Round Robin was limited to an evalua-analysis of the inspection data for the Indian Point tion of the detection capability of SAFT-UT; no 2 reactor pressure vessel showed that the defect cvaluation ofits sizing capability was performed.
was volumetric and was not connected to the 4
For small cracks, average length of 7.4 mm surface. The application at Trojan showed that (0.29 in.) and depth of 1.1 mm (0.043 in.), which SAFT can be successfully coupled with commer-is equivalent to 4% through wall, the probability of cial UT equipment and scanner systems in an on-detection and false call probability were, respec-or off-line mode. This application also showed tively,67% and 14%. The corresponding values that the SAFT results are easy to understand and for large cracks, average length of 187.7 mm provide images that others can readily interpret (7.4 in.) and depth in the range of 4.3 to 11.2 mm (Doctor et al.1995).
e i
119 NUREG/CR 6456 i
INSERVICE INSPECTION 8.2.2 UT Performance Demonstration uncenain, the U.S. industry is in the process of preparing the necessary test blocks for the perfor-Nondestructive examinations are now recognized mance demonstrations (Spanner et al.1992, Beck-as an essential element in the safe operation of er et al.1992). In order to minimize the number of nuclear power plants. They can be used for dem-samples needed and eliminate the need for onstrating that defects of a size sufDcient to threat-site-specific qualifications, test blocks are being en the structural integrity of the plant are absent prepared to cover a range of pipe sizes and thick-both prior to operation and at intervals during nesses rather than one block for each specific size.
service. However, a number of events, such as Performance demonstrations are to be carried out inspection of the thermal fatigue cracks in the on test blocks with weld preparation, geometry, ferritic steel feedwater piping and the IGSCC in and access conditions representing those encoun-the wrought stainless steel piping during the 1970s tered during inservice inspection. The geometric and 80s, have diminished confidence in the ability conditions include those that normally require of the nondestructive methods to fulfil this role.
discrimination from flaws. One example of such These events have prompted the regulatory and a geometric condition is a counterbore at the Code bodies to adopt different approaches. In the feedwater nozzle. The test blocks must also con-past, the ASME Code used prescriptive standards tain a certain number of realistic defects to confirm for the nondestructive examinations. These stan-inspection technique reliability. For ferritic piping dards described in detail how the nondestructive welds, at least 75% of the defects must be either examination was to be performed. However, these mechanical or thermal fatigue cracks.
prescriptive standards had several disadvantages.
In particular, it was very difficult to introduce any Acceptance criteria for detection and sizing of improvements in these standards and they inhibited flaws in the test specimens are provided in the introduction of new nondestructive examina-Appendix VIII. The detection specimens include tion methods. The ASME Code has addressed this both flawed and unflawed test specimens. A problem by replacing the prescriptive standards minimum of % of the flaws are required to have with performance requirements. This change in depths between 5% and 30% of the nominal pipe the Code approach to inservice inspection leaves thickness. At least % of the flaws are required to the responsibility for the nature and detail of the have depths greater than 30% of the nominal pipe inspection entirely with the plant owner who is free wall thickness. There are also requirements about to use whatever inspection method isjudged to be the orientation ofthese flaws in the test specimens.
appropriate. However, there must then be a dem-At least one and a maximum of 10% of the flaws onstration that the method chosen is capable of must be oriented axially. The remainder of the meeting the requirements placed on it by the Code.
flaws must be oriented circumferentially. Service-induced flaws shall be included when available.
The 1989 edition of ASME Section XI(ASME For sizing samples, a minimum of 20% of the 1989a) includes a man 6 tory Appendix VIII, flaws in each of the following three groups of flaw Performance Demonstrations for Ultrasonic depths is required: 5 to 30%,31 to 60%, and 61 to Examination Systems, which provides require-100% of the wall thickness, ments for the performance demonstration of the ultrasonic examination personnel, procedures, and The costs associated with compliance to Appendix equipment used to detect and size flaws. The VIII are high, especially for the feedwater nozzle intent of Appendix Vill is to establish a minimum inside radius section, because it requires a large level of skill and effectiveness for ultrasonic nun'ber of mockups. Use of computer modeling of inspection systems. It will be a requirement for the the ultrasonic examination process can result in a inservice inspection of both austenitic and ferritic reduction of the resources needed to comply with piping welds, vessels, vessel nozzles including the Appendix VIII. By modeling the change in nozzle inside radius sections, and bolts and studs.
ultrasonic beam angle with changing test part Although the implementation schedule is still geometry, the number of specimens required for NUREG/CR-6456 120
INSERVICE INSPECTION performance demonstrations can be rei 1
plant, is expected to be completed in the first Computer modeling can also help in determin..
quarter of 1996 (Balkey 1995). Several other pilot the best position and angle for a given probe foi studies are also underway and are expected to be inspecting a crack of given size and orientation, completed in the latter part of 1996 or early 1997.
located in the nozz!c inside radius section. ASME Section XI Code Case N-552 Alternative Methods The basis for the Code Case is contained in an
- Quahpcationfor Nozzle Inside Radius Section ASME Research White Paper entitled Risk-Based frorn outside Surface permits such modeling to Alternative Selection Process for Inservice reduce the number of specimens for demonstration Inspection of I,WR Nuc* ear Power Plant of the ultrasonic systems used for nozzle Components (ASME 1995a), which describes the examinations.' A non-mandatory Appendix to risk-based selection process as an alternative to the Section XI of the ASME Code, covering validation selection sites specified by ASME Section XI. A of computer models is in the Code-approval primary concern with risk-based inspection has
- process, been how to implement it in a consistent manner across the nuclear industry. Appendix ! Piping Silk (1996) discusses the use of computer Risk-BasedSelection Procedure of the white paper processing of ultrasonic time-of-flight diffraction contains one approach for implementing a risk-data to vary the level of noise in the signal and based selection process at a nuclear power plant.
provide estimates of the probability of detection of This appendix has been incorporated into the draft flaws as a function of the noise level. This concept Code Case to help alleviate concerns regarding the could be applied to other ultrasonic data. It should implementation of the risk-based inspection, be possible to modify the data to add the effects of geometric signals and then evaluate the capability The intent of risk-based inspection criteria is to of the UT system to discriminate flaws from allow ISI examinations to be focused on critical geometry. It should also be possible to move the components. Components would be categorized 2
locations of the recorded flaws and geometry as either high risk-significant or low risk-signals to make new simulated qualification significant, and would receive examinations specimens.
commensurate with their risk-significance and the expected failure mechanisms. The result would 8.2.3 Risk-based Inspection (ASME probably be an overall reduction in the quantity of Section XI Code Case) examinations, but would concentrate activities on components critical to plant safety and areas prone A proposed alternative to the weld selection to cracking, such as the counterbore areas in criteria of ASME Section XI is the use of risk-feedwater nozzle welds that currently fall outside based selection rules. A draft code case (Code the inspection volume required by Section XI.
Case N-XXX, Risk-Based Selection Rules for Class 1, 2, and 3 Piping.Section XI, Division 1) 8,3 lnSerVlCO lnSpection of for this alternative is being developed by the Wall ThinninU Caused bY Section XI working group on implementation of risk-based examination and is being supported by Flow-Accelerated CorroSlon ASME research. In addition, pilot studies are underway to demonstrate suitability of risk-based This section summarizes the inservice inspections technology for inservice inspection. One such performed subsequent to the pipe wall thinning study, being performed by the Westinghouse (including feedwater pipe rupture) described in IE Owners Group at Northeast Utilities' Millstone 3 Bulletin 87-01 (USNRC 1987d). Emerging in-
' The Code Case N 552 has been adopted by the For the purposes of this section, critical components 2
ASME Boiler and Pressure Vessel Cornmittee and is in the are those that have a high failure probability or severe process of publication.
consequences associated with the failure.
121 NUREG/CR-6456
INSERVICE INSPECTION service inspection methods for more reliable and main feedwater loop isolation valves (Wu 1989).
accurate estimation of wall thinning are discussed.
These components are listed in Table 7.
He criteria and procedures developed for evaluat-ing wall-thinning inspection results are summa-The review concluded that limited inspections of rized.
single-phase feedwater-condensate systems were conducted at most plants aaer the incident at Surry 8.3.1 Inspection in Response to Bulletin Unit 2. The review also found that inspection 87-01 programs to detect wali thinning in two-phase, high-energy carbon steel piping systems had On December 9,1986, a main feedwater pipe at existed at all the plants. In general, inspection Surry Unit 2 experienced a catastrophic failure as locations were established based on EPRI a result of wall thinning. In a 1987 refueling Document NP-3944, Erosion / Corrosion in outage, wall thinning was also discovered in two Nuclear Plant Steam Piping: Causes and straight sections of main feedwater piping at Inspection Program Guidelines, published in Trojan. Because of these incidents and concerns 1985. However, because the document guidelines about high-energy, carbon steel piping systems, the were nonmandatory, there was significant variation NRC issued Bulletin 87-01, Thinning of Pipe in the piping systems included in the program and Walls in Nuclear Power Plants, which requested the number of sites selected for inspection among the following information regarding high-energy, the plant inspection programs.
carbon steel piping:
The primary inspection methods in response to a) The Codes and standards to which the NRC Bulletin 87-01 were reported to be manual piping was designed and fabricated, UT, supplemented by visual examination of the inner surfaces of piping components. There were l
b) ne scope, extent, and sampling criteria of several cases where radiography was also used.
existing inspection programs to monitor Visual inspection is performed by direct pipe wall thinning of safety-related and observation through an opening in the piping nonsafety related systems, system with borescopes and fiberscopes, and with crawlers that normally have television cameras.
c) Results ofinspections performed to identify Use of crawlers generally requires that the piping pipe wall thinning, and system is drained of water. The advantage of performing visual examinations is that larger d) Plans for revising existing pipe monitoring surface areas can be inspected rapidly if an access procedures or developing new or additional to the interior of the piping system is available; for procedures.
example, through a disassembled valve or an inspection port. Visual examination can reveal the The USNRC review of the licensees' responses to presence of worn areas but is not well suited for Bulletin 87-01 showed that wall thinning in single observing the wall thinning that occurs in single-phase systems (feedwater and condensate systems) phase flow. Visual examination is typically limited is more widespread in PWRs than in BWRs; 26 to the inside surface of the large diameter piping PWRs and 6 BWRs had identified various degrees (Chexal et al.1996).
of wall thinning in the feedwater piping and fittings.
The PWR components where wall Manual UT has been, and is the most commonly thinning was reported included elbows, reducers, used inspection method for measuring wall thin-straight piping runs, drain pump discharge piping, ning because it is relatively inexpensive and accu-recirculation line, heater vent piping, fittings rate. A properly conducted UT examination can including feedwater pump suction line fittings, and estimate the pipe wall thickness within 5% of the straight runs including those downstream of the actual va!ue (Chexal et al.1996). The industry NUREG/CR-6456 122
INSERVICEINSPECTION inspection method for the detection of wall' thin.
practice for using UT to detect wall thinning ~~
ning because ofits accuracy and relatively low caused by flow-accelerated corrosion is to draw or overlay.a grid and then spot measure the wall cost. It can be adapted to many complex geome-
-thickness at each grid location.. The grid size tries, and it posses no hazard to inspection person-
' depends upon the pipe diameter; larger grid sizes nel(as opposed to radiography). The testing can tre used for larger diameter pipes. The grid size be accomplished using either a digital thickness varies from about 25 x 25 mm to 150 x 150 mm (I gage or a standard flaw detector with an A-scan
- x l'in. to 6 x 6 in.). If thinning is found, the presentation. Digital thickness gages provide an common practice is to completely scan each grid instantaneous thickness measurement without space with UT and record the minimum thickness.
relying on operator interpretaticn and are effective Use of grids provides data that can be used for on clean, uncoated surfaces. However, digital estimating long-term trends and also for evaluating readout devices can have problems in heavily the structural integrity of a thinned component, cwroded areas and may misrepresent small thinned areas or the edges of larger thinned areas (Fay The UT examinations are usually performed on 1987). This is primarily the result of the roughness components that are at an ambient temperature.
of the corroded surface, which tends to scatter the Occasionally, a UT examination is performed on a ultrasonic beam and interfere with the automatic component that is not taken out of service and the sating and measurement scheme of the digital wall temperatures are higher. This type of thickness gage. Consequently, an A-scan presenta-examination is referred to as hot UT. Special UT tion is still preferred by many inspectors because transducers and couplant are used-' for this trained operators can more reliably interpret signal examination. The wall thickness measurements patterns that may be misinterpreted by digital made with hot UT are less accurate than those instruments. Fonunately, many equipment vendors made on components that are at ambient conditions have developed compact equipment with both an because of the difficulties associated with working A-scan display and digital readout, so thickness in a hazardous environment and the added measurements can be verified using the A-scan -
precautions necessary to make accurate presentation. For corroded surfaces such as those measurements at elevated temperatures (Chexal et with pitting, focused transducers can provide a al.1996).
more accurate and reliable thickness measurement.
i Because of their smaller beam size, focused trans-Most often, the examination frequency is based on ducers examine smaller areas and are less prone to previous inspection data used to trend flow-loss from scatter.
l recelerated corrosion rates. In systems susceptible to wall thinning, the areas selected for inspection Although ultrasonic thickness measurements are L
varied from plant to plant but were usually areas of widely accepted and used, there are several abrupt changes in flow direction, immediately disadvantages to the ultrasonic method. UT downstream of significant pressure drops (e.g., at requires direct access to the examination surface.
orifices and control valves) and at other fittings Thus, removal ofinsulation is required, which is that cause penurbations in flow (i.e., at reducers costly and time consuming. Another disadvantage 4
[
End branch connections). In some cases, the plant is that manual methods do not provide a permanent operators used EPRI's CllECMATE program to inspection record, only a vast number of thickness renk the pipe locations prone to flow-accelerated measurements. in addition, manual UT is operator corrosion.
dependent and prone to error because of varying conditions and operator interpretation. In an 8.3.2 Emerging inservice inspection independent study performed by Northern States Methods for Wall Thinning Power (NSP), six experienced operators made thickness measurements on new clean components Ultrasonic Examinations.
As stated and corroded components that were removed fmm j
above, manual UT is the most commonly used service. The results indicated an average error 123 NUREG/CR-6456
INSERVICE INSPECTION spread of 036 mm (0.014 in.) for the new raphy. The first is a direct double-wall shot along components and average error spread of 0.91 mm with the use of a calibration curve of thickness (0.036 in.) for the corroded components (Shankar versus film density as an indication of wall thick-and Bridgeman 1990).
ness. This method can be used for inspecting larger components and also geometrically compli-Automated ultrasonic scanning has also been used cated regions, which are difficult to inspect with to map corrosion. It is less dependent on the UT He second is tangential beam radiography, as operator and provides a permanent record of the shown in Figure 64. A technical limitation of inspection (Edelmann and Gribi 1990). Another using either of these RT techniques is that multiple advantage is that the entire area can be scanned shots are required to make measurements around and presented in a C-scan or plan view. His type the entire circumference. Therefore, achieving of presentation displays UT data so that those not 100% coverage is a time-consuming process.
familiar with ultrasonic inspection techniques can llowever, for small diameter piping (for example, view, interpret, and visualize trends. The primary diameter less than 150 mm (6 in.)], tangential disadvantages of automated UT is the higher cost, beam radiography has been found more economi-compared to manual techniques, and more cal than ultrasonic examinations.
clearance needed for the scanning devices. Field experience has found that automated UT Radiography oflarger size components is limited examinations for thickness measurements are too by the strength of the source. Linear accelerators cumbersome and time-consuming for effective use such as the miniature linear accelerator (MINAC)
(Chexal et el.1996).
developed by EPRI can be used as a source for the inspection of larger size components.
For Another variation of UT is a mdtiple element example, the tangential beam radiography with an array called PARIS (Portable Autcmated Remote Iridium 192 source has been used for inspecting up Inspection System). PARIS is a flexible array that to 200-mm (8-in.) diameter pipes, and with a linear was adapted for characterization of flow-accelerator it has been used for inspecting larger up accelerated corrosion damage in sti.els (llarring-to 450-mm (18-in.) diameter pipes (Chexal et al.
ton 1988). His array of transducers can conform 1996).
and acoustically couple to complex geometries (i.e., elbows and tees) automatically with 100%
The most accurate method of displaying coverage. Field demonstrations are still needed to radiographic data is computed tomography, with validate this approach.
which a complete cross-section of a component can be obtained. The Japan Power Engineering and Radiographic Examinations. An alterna-Inspection Corporation (JAPEIC) is developing tive to UT thickness measurements is RT. RT is computed tomography to reliably look inside advantageous because it provides a permanent components to obtain dimensions and characterize record of the inspection and can be performed complex geometries (Maeda 1991). Additional while the system is on-line and without any work has been performed to make the system more removal of insulation. Radiographic examina-compact and to enhance image quality, so that
' ions can detect wear, particularly in components flaws can be easily discriminated (Miyoshi, et al.
with complex geometries, but the examination 1992). This information is used as an aide in results are less useful in trending wall thickness evaluating indications detected during inservice over time. The other disadvantages of using RT inspections. His same methodology could also be are the higher cost of implementation for large used to characterize wall-thinning damage in the bore piping and the radiation hazards it presents to feedwater piping and thermal sleeves.
As plant personnel.
mentioned in Section 6.4.2, thinning of the thermal sleeves at Diablo Canyon has occurred. Since There are generally two methods of performing conventional ultrasonic techniques are currently thickness measurements with conventional radiog-not effective, owing to the gap between the pipe NUREG/CR-6456 124
INSERVICE INSPECTION N
I y
Film Pipe f
/g g/ l
.h.-
Source Iridium 192 Figure 64. Tangential radiography.
wall and the thermal sleeve, computed tomography nique is that access to the far side or inside of the may offer a solution to this inspection problem.
pipe is not required. in addition, removal of The disadvantage of computed tomography is the insulation can be avoided. This technology has high cost and potentially long processing time to been used to examine aircraft components (Lawson create an image. Ilowever, advances in computer-1995), and at least one commercially available ized systems are reducing processing time and cost system (COMSCAN)' has been developed which end making tomography a more viable approach is used primarily for lightweight airframe materi-for nuclear power plant applications, als. For steels, the use of this technique presents practical problems due to greater absorption losses Another radiographic technique suggested for use and reduced resolution. Thus, because of the in wall thinning inspection is compton backscatter difficulties working with steel and high implemen-imaging (Lee and Kenney 1992). Compton scat-tation cost, compton backscatter has not yet gained tering is a direct interaction of the incident radia-widespread acceptance as a reliable field inspec-tior, with orbital electrons of the material under tion tool. 110 wever, because of the potential i
t:st. The resulting photon energy is scattered in all advantages and benefits, additional investigation is j
directions at a lesser energy and wavelength and is warranted, j
monitored with detectors on the same side of the i
component as the source. The degree of scattering varies directly with atomic number (Metals 11and-
' conscAx is Philips' trade name for its compton book 1989). The obvious advantage of this tech-
- x. ray nackscatter Inspection system.
125 NUREG/CR-6456 l
INSERVICE INSPECTION The EPRI NDE Center is currently looking for R6ntgen Technische Dienst B.V. (RTD), for technology improvements that will make flow-commercialization of the technology. TEMP is no accelerated corrosion inspection better, faster, and longer being evaluated by EPRI.
chegper.' Because radiographic examination has a potential to provide these advantages, EPRI is 8.3.3 Wall Thinning inspection Criteria looking at technologies employing this examination method that will reduce inspection Once the inservice inspection is completed, an times, safety hazards and costs, while improving evaluation must be made as to whether the piping the qtality ofinspections. One approach is the use is acceptable for service for the next full duty of a real time radiography (RTR) system that cycle. This evaluation must consider the measured employs multiple detectors to perform double-wall thickness, the flow-accelerated corrosion rate, the thickness measurements through insulation. The factors affecting the corrosion rate, the projected current effort is aimed at adapting the system to operating time before the next inspection, and the work on complex geometries such as elbows and requirements for minimum wall thickness.
tees, and to add the capability ofmaking tangential measurements. A second approach is the use of EPRI has developed a three-step process for evalu-phosphor plates in lieu of film. Phosphor plates ating flow accelerated corrosion damage that are expected to reduce exposure time or radiation includes: (a) screening to determine if further intensity by a factor of 20, which can significantly evaluation is warranted, (b) comparison of any reduce inspection time and/or personnel safety general wall thinning with the design minimum concerns associated with performing RT within a wall thickness requirements, and (c) evaluation of plant. Use of phosphor plates is now being any local wall thicknesses less than the design evaluated in the field.
minimum based on three alternate acceptance criteria (Gerber et al.1989). The acceptability of l
Other inspection Methods for Wall any local wall thinning is based on the axial extent Thinning. The Transient Electro-Magnetic Probe of the corrosion-induced thinning along the pipe l
(TEMP) was developed by ARCO for monitoring length, branch reinforcing requirements, and local general wall thinning in piping and tanks in the membrane stress rules.
petro-chemical industry (Walker and Martinez 1994). TEMP monitors the decay of an eddy Hoemann and Berak (1989) developed a current pulse within the wall and has been used three-stage evaluation procedure, based partly on successfully by ARCO as a high speed survey the EPRI method, using a combination of UT tool.2 An earlier EPRI evaluation of this probe results and analysis. In the first stage, ultrasonic concluded that TEMP could reliably measure wall measurements are taken using a coarse grid net-thickness to within 5% for uniform thinning work extending approximately one pipe diameter through insulation, but could not detect localized upstream and two diameters downstream from the thinning of an area as large as 130 mm (5 in.) in component ofinterest. The component isjudged diameter. Detection of local thinning would to be acceptable if the lowest ultrasonic measure-require the probe head to be redesigned and the ment shows that wall thickness will remain above measurement algorithm to be modified. TEMP has the minimum required thickness by the next in-been licensed to an inspection equipment vendor, spection period. If the Stage I criterion is not met, areas around low readings are reexamined, using a reduced grid size, and a combination of ultrasonic measurements and analysis is used to evaluate the area. If the component cannot then be qualified,
' A. M. Porter, private communication with Steve fracture mechanics and limit load analyses are Kenefick at the EPRI NDE Center, December 12,1995.
performed. Ofthe piping components evaluated as part of the analysis,47% were found to be accept-2 A. M. Porter, private communication with James E.
able during initial evaluation,41 % required further Mitchell of ARCO, November 23,1995.
NUREG/CR-6456 126
l i
1 INSERVICE INSPECTION i
i l:
analysis to satisfy acceptance criteria, and 12%
PWRs to assess the licensees' efforts toward -
t required repair or replacement.
implementing their flow-accelerated corrosion monitoring programs. The results of those visits j
ASME Code Case N-480 also presents acceptance are summarized here (Taylor 1995).
standards based on the EPRI evaluation process for pipe wall thinning caused by flow-accelerated De USNRC staff audited 10 plants during 1988, l
corrosion (ASME 1990). If the pipe thickness is 7 PWRs and 3 BWRs. The major finding of these less than 87.5% of the nominal pipe wall thickness, audits was that all the licensees had performed l
the pipe must be repaired or replaced unless an initial inspections for estimating the wall thinning i
analytical evaluation shows that an acceptable caused by flow-accelerated corrosion and had j
j safety margin.- exists.
Also, if the piping programs that meet the intent of the industry component predicted wall thickness at the next guidelines for monitoring this damage. However, inspection is less than 87.5% of the nominal wall these were short-term programs and not all the
[
thickness, pipe repair, seplacement, or acceptance licensees were committed to implementing i
by analytical evaluation is required for continued administrative controls and formalized procedures i
l service.
The Code-Case iequires that the needed for long-term monitoring. For example, in
{
component be repaired or replaced if the predicted many instances, the record keeping that would
.j wall thickness of a piping component at the next allow for future reproducibility of grids identifying i
inspection is calculated to be less than 30% of the the inspection locations and trending of wall nominal wall thickness. The analytical evaluation thickness reductions was nonexistent. Also, in t
procedure has three alternate requirements for several instances where outside contractors eccyt-ice of any local wall thinning which are a performed the inspections, the contractor's function of depth and extent of the area thinned by guidelines rather than licensees' guidelines were flow-accelerated corrosion.
If an analytical followed; this may lead to inconsistency in the evaluation is required for areas damaged by flow-future (Wu 1989).
cecelerated corrosion and the requirements for i
acceptance of local wall thinning are satisfied, the in response to this finding, the USNRC issued i
r.ffected areas should be examined during the next Generic Letter 89-08, Erosion / Corrosion-Induced three inservice examinations, at a frequency Wall 7hinning, specifically requesting the licensees determined by a flow-accelerated corrosion rate to affirm that they have implemented, or will estimated from the inservice examination results.
implement by a specified date, long term flow-i accelerated corrosion monitoring programs to 8.3.4 USNRC Audits and Inspections ensure that flow-accelerated corrosion will not lead to unacceptable wall thinning of high-energy As discussed in Section 8.3.1, the USNRC issued carbon steel piping in either single or two-phase Bulletin 87-01 because of the catastrophic failure systems. The Generic Letter did not specify the of the main feedwater piping at Surry Unit 2 and scope and content of these programs.
the wall thinning of the straight sections of feedwater piping inside the containment at Trojan, The USNRC staff conducted audits and both caused by flow-accelerated corrosion.' The inspections at 3 PWRs and 5 BWRs at 5 randomly inspections performed by the NRC licensees in selected nuclear plant sites during the first half of response to the bulletin are also summarized in 1992. The main objective of these plant visits was Section 8.3.1. Since publishing the bulletin, the to assess the general response of the nuclear USNRC staff have made several visits to selected industry to Generic Letter 89-08. The second objective was to evaluate the procedure that provides guidance to the NRC inspectors for 4
inspecting components susceptible to flow-8 The term crosiodcorrosion instead offlow-accelerated corrotton is generally used in the USNRC accelerated corrosion in addition to these audits j
information notices, tiulletin, generic letter, letters, and and inspections, the Region I staff conducted reports. -
127 NUREG/CR-6456
INSERVICE INSPECTION supplemental inspections at 10 plants,5 PWRs and licensee repaired a worn Class 1 feedwater 5 BWRs at 7 nuclear plant sites. The main finding component inside the containment with a weld was that all the licensees that were audited had buildup on the exterior of the pipe.
implemented flow-accelerated corrosion programs Most licensees did not perform baseline in accordance with Generic Letter 89-08.
However, the following specific findings imply thickness measurements on new or replaced that a number of improvements could be made piping prior to placing the piping in service.
(Richardson 1992).
The USNRC staff were interested in evaluating the j
The flow-accelerated corrosion programs use of engineeringjudgement by some licensees to varied widely.
Some licensees selected choose locations for inspections.
It started, ultrasonic mmination locations using therefore, a program where predictions were made analytical models which ranked the systems using the CHECMATE code and then compared to and components for their susceptibility to the licensee's predictions using engineering flow-accelerated corrosion damage, whereas judgement.
The evaluation at one plant is others used engineeringjudgement.
completed and good agreement between the components selected by the NRC staff with the Several licensees had excluded the high en-CHECMATE code and those selected by the plant ergy, carbon steel components in the safety-operator with engineeringjudgement. However, related portion of the feedwater and steam the main concern remains that the component generator blowdown systems inside contain-selection based on engineering judgement largely ment from the scope of their flow accelerated depends on the plant operators experience and, corrosion programs. The components may be therefore, the selection may vary significantly from located in non-isolable portions of the coolant plant to plant. Initially NRC planned to continue systems and they should be included in the this program at one more plant. However, this flow-eccelerated corrosion programs.
further evaluation may not be needed because all the plants now use the predictive codes for Several errors were found in either inputting selecting the components for inspection.
the proper parameters into a predictive model i
such as CilECMATE or in calculating Code 8.4 InSerVICO Inspections minimum wall thickness acceptance criteria.
at Non-US Plants Some inconsistencies were found in reproduc-As discussed.m Section 7, fatigue caused by ing the inspection grids during subsequent thermal stratificat,on and flow-accelerated corro-i ultrasonic examination ofpreviously inspected si n have damaged feedwater pipmg in non-US components.
plants. Conventional and enhanced mspection techniques have been used to characterize the All of the NRC licensees audited had damage. However, these inspections are not performed repairs or replacement of the carbon n a Periodic bas,s in the responding j
required i
rteel components which failed to meet the e untries. Several non-US countries have pro-licensee's minimum wall thickness or alternate vided varying amounts of information about acceptance criteria. The replacement materials relevant inspection activities carned out at their varied from carbon steels, to chromium-molybdenum steels, to austenitic stainless PWR plants.
steels.
8.4.1 Inservice inspections of Fatigue Cracks at Non-US Plants Some repairs of the safety-related piping wom by flow-accelerated corrosion may not meet Several countries use the ASME Section XI Code code repair requirements. For example, a for detecting and sizmg thermal fatigue cracks.
NUREG/CR 6456 128
i L
INSERVICEINSPECTION i
Some countries have performed inspections based plants; however, this inspection method has never on NRC Bulletin 79-13. Generally, conventional detected indications in the feedwater systems.
or enhanced manual ultrasonic inspection techniques are used._ However, according to the In Switzerland, inservice inspection of the Finnish response, these techniques are not feedwater systems includes weld surface and l
qualified for detection of thermal fatigue cracks, volumetric examinations performed by magnetic i
which is the present situation in the whole of particle testing and UT, respectively.
The Europe. His is also the present situation in other inspection zone includes the entire volume of the i
countries, including the United States. Finland has weld and 13 mm (0.5 in.) of the adjacent base performed volumetric and surface examinations of metal on each side of the weld. De crack-tip-
+
the complete piping between the feedwater nozzle diffraction technique is used for crack sizing; cnd the first check valve upstream of the nozzle, however, this technique is not yet qualified. The including both base metal and welds. These amplitude-drop method is considered to be i
ext.minations were performed at each of the six unreliable for crack sizing. De discrimination steam generators at one plant. No fatigue cracking between geometrical and flaw indications is hts been detec%i outside the inspection areas performed by localization of an indication with required by the ASME Code, which include welds scanning from both sides of the weld and using i
and the adjacent base metal. De amplitude drop additional scanning angles.
i technique is never used alone for sizing a fatigue crack.
Supplemental ultrasonic examination Ontario Hydro Nuclear of Canada does not inspect techniques such as PATT or SPOT are always used the feedwater lines in its CANDU units for fatigue j
in parallel.
cracks, possibly because such cracking is considered unlikely. However, if such inspections In Belgium, ultrasonic examination with shear were to be performed, Ontario Hydro Nuclear wave probes is used for detecting cracks in would use baselinc inspection msults, radiography 9
feedwater piping welds and the adjacent base results when applicable, and geometry data for metal. The crack-tip-diffraction technique with an weldjoints along with ultrasonic inspections, using cngle beam or a straight beam is used to size special tools such as tip-diffraction techniques f:tigue cracks in the feedwater piping. If the crack coupled with creeping wave probes. Dese probes 4
tip diffraction technique is not applicable and the are currently being developed and used during the i-flaw apparently has a small through-wall extent, inspections required by the code Periodic then the amplitude comparison method is used. In Inspection of CANDU Nuclear Power Plant all other cases, the 6 dB drop method is used. De Components to look for surface breaking flaws.
operator discriminates flaw indications from geometry by using a full-scale sketch of the weld 8.4.2 Inservice Inspections of Wall and repositioning the probes and ultrasonic beams.
Thinning at Non-US Plants Advanced ultrasonic methods are being developed Apparently, wall thinning caused by flow-in France for detecting and sizing fatigue cracks accelerated corrosion is more wide spread in the produced by thermal stratification. Mockup non-US plants than fatigue cracking caused by samples and samples from replaced steam thermal stratificatica. Each responding country i
generators are used to qualify these methods. Two has inservice inspetion activities associated with i
mein difficulties faced in application of these wall thinning. Generdly, these inspections are j
methods are limited access to the zones that have performed with ultrasonic techniques using a grid-
~
to be inspected and rough surface conditions. It based system. In Finland, a0 se feedwater piping l
wts expected that development and field testing is included in the inservice inspection program.
would be completed by the end of 1995. Gamma The WATHEC computer code developed by radiography inspection is also used at French PWR Siemens is used to analyze the system and select 129 NUREG/CR-6456
.-.e
, ~., - -.,
, + -,
INSERVICE INSPECTION i
locations for inser6 'spections. In addition to indications of wall thinning in feedwater lines have r
ultrasonic examinatwas, visual examinations of been found until recently.
discontinuities on the inside surfaces, especially the inside surfaces of welds, are performed using Ontario Hydro Nuclear uses manual ultrasonic a video camera. Supplemental locations, in examination techniques for estimating wall addition to those recommended by the WATHEC thinning at the Bruce B units. Extensive use of code, are also inspected.
nberglass templets is made for accurately and efficiently reproducing the grid at each inspection in Belgium, the feedwater system locations to be location. The inspection locations are selected and inspected are selected after taking into account ranked by taking into account the various factors various factors affecting flow-accelerated affecting the flow-accelerated corrosion damage corrosion: water temperature and pH, flow including fluid temperature and velocity and velocity and diameter of pipe, piping material piping geometry. Additional factors such as composition, and local geometry. The following dissolved oxygen concentration and pH level are components are usually inspected in the selected considered to prioritize closely ranked components.
piping: all elements downstream ofcontrol valves, The Atomic Energy Control Board of Canada including valve bodies; all tee fittings and pipe reports that the accuracy of the wall-thickness expansion pieces; and 10% of the elbows.
measurement using ultrasonic examinations is on including all elbows in series. Elbows with a large the order of 75 pm (0.003 in.).
angle are selected over ones with a small angle.
Ultrasonic examinations with the use of a No significant wall thinning has been found in the transmitter-receiver probe are performed to Dutch PWR plant because the feedwater system measure the extent of the wall thinning. Both was built with low-alloy steels, and the flow minimum and average thicknesses are estimated.
velocities are moderate. In Japan, wall thinning inspections are performed by each utility but they Manual ultrasonic examinations are typically are not required to report the inspection results to performed to detect wall thinning in Swiss PWRs.
the regulatory body. Ultrasonic examinations are Selection of inspection areas is made by the used for wall thi+.aess measurements in Spanish utilities, considering recommendations from plant PWRs. A typical surveillance program includes manufacturers and their own experience. No inspection of about 75 areas in the feedwater enhanced examinations are performed because no system.
1 NUREG/CR-6456 130
- 9. MITIGATION AND MONITORING OF FATIGUE, FLOW-ACCELERATED CORROSION, AND WATER HAMMER DAMAGE Methods employed to prevent or mitigate thermal manual operation of the auxiliary feedwater system f tigue damage are discussed in this section. The instead of automatic operation as discussed in methods include modifications to the operating Section 6.3.1. Manual operation at the Sequoyah i
procedures for the feedwater systems and structural plant, for example, supplies a continuous flow of design changes of the feedwater piping and other auxiliary feedwater at a nearly con.; tant flow rate of components. We then brieHy review the on-line about 300 to 340 f/ min (80 to 90 gpm), whereas fitigue monitoring of feedwater lines. We follow the automatic operation supplied auxiliary this with a discussion on the methods employed to feedwater with a flow rate cycling between 0 and prevent or mitigate wall thinning caused by flow-833 t/ min (220 gpm) at a frequency of about 3 accelerated corrosion. Finally, mitigation methods cycles per hour. Thus, the fatigue damage is for water hammer are discussed.
significantly smaller using the manual mode of operation.
9.1 Mitigation of Thermal An ther effective modification to the plant Fatigue Damage operatmg procedure is to use heated main feedwater during plant startup and hot standby Asd.iscussed in the previous sections of tlu.s report, conditions. As discussed in Section 2.1, in two flow st.~tification m PWR feedwater piping can Westinohouse plants, Wolf Creek and Callaway, cause significant thermal fatigue damage. Flow the main feedwater system includes a special j
stratification takes place when the incoming startup system that provides heated feedwater to j
.cedwater flow rate is low and there is a large the steam generators during startup, hot standby, temperature difference between the incoming and shutdown operations of the plant. Heated feedwater and the steam generator coolant.
feedwater is also used in Japanese PWRs during Fluctuations m the elevat,on of the,nterface startup and hot standby. This may be the reason i
i between the hot and cold coolants causes thermal that fatigue cracks have not been found in the fatigue damage. This damage generally takes place feedwater nozzle in the Japanese PWRs since at plants where the cold auxihary feedwater is j974' injected directly into the main feedwater line.
Therefore, several plants have modified their Three modifications in the feedwater piping design operatmg procedures and made design change o or layout have also been made to reduce or their feedwater systems to reduce the stre
- eliminate the thermal stresses produced by i
prodeced by stratified flows. In addition, new stratified flows. These modifications include a design featares have been mcorporated in redesign of the counterbore to remove the stress replacement steam generators and in steam raiser, installation of a thermal liner to protect the generators for new plants so that flow stratification counterbore and other susceptible sites from does not take place.
nermal fatigue damage, and a modified piping layout to introduce the auxiliary feedwater directly An effect.ive modification to the operat.mg into the steam generator instead ofinto the main procedure is to use a continuous flow of auxihary feedwater line. The end of the feedwater nozzle feedwater mstead of intermittent flow. This can be built up with weld metal so there is no modification does not eliminate stratification or change in wall thickness from the feedwater nozzle reduce the thermal stresses produced by the to the transition piece; this modification eliminates stratification but does sigmficantly reduce the the counterbore, a stress raiser, where fatigue o imber of fluctuations m these stresses and' cracking has occurred at several PWR plants. An lerefore, the resulting fatigue damage. This example of this type of modification is shown in cisange in the operatmg procedure may require 131 NUREG/CR-6456
MITIGATION AND MONITORING Figure 31. Installation of a thermal liner that Two other design changes in the Siemens/KWU protects the counterbore region, feedwater nozzle, steam generators include: (a) distribution of the and adjacent piping and elbow from the flow cold auxiliary feedwater through a spraying device stratification stresses is shown in Figures 13 and located in the main feedwater pipe upstream of the 30.
feedwater nozzle, which mixes the cold auxiliary feedwater with the hot water present in the The feedwater piping layout was modified at the horizontal pipe, and (b) installation of a Palisades plant to introduce the auxiliary feedwater recirculation loop to inject a part of the steam directly into the steam generator, after finding generator blowdown flow back into the feedwater cracks in the main feedwater nozzle in 1979. The line upstream of the feedwater nozzle.
Pelisades steam generators had originally been designed with an auxiliary feedwater nozzle The flow stratification phenomenon has been located some distance above the main feedwater avoided in some new or replacement steam genera-nozzle and connected to the auxiliary feedwater tors by diverting the auxiliary feedwater flow i
distribution system including a thermal sleeve and directly into the steam generator through a separate i
a sparger inside the steam generator. However, the smaller diameter nozzle. This arrangement elimi-auxiliary feedwater nozzle was capped during nates the thermal stratification conditions in the installation and the auxiliary feedwater piping was main feedwater nozzle and adjacent piping. In instead connected to the main feedwater piping.
addition, it does not result in thermal stratification rhe piping layout modification performed after the in the auxiliary nozzle because of its smaller 1979 cracking event included connecting the diameter.
auxiliary feedwater piping to the auxiliary feedwater nozzle as originally planned. Howevar, The flow stratification phenomena was also this layout modification led to a water hammer avoided in some new or replacement steam event in the auxiliary feedwater system and various generators by installing an antistratification device components were damaged as discussed in Section that breaks down the stratification by mixing the 6.5. Another concern with this modification is the hot and cold coolants. Framatome developed such an antistratification device, called a helix, possible leakage of the auxiliary feedwater into the gap between the auxiliary feedwater nozzle and consisting of a central hub and four helical blades, thermal sleeve and the resulting fatigue damage to extending from the hub to the sleeve inner wall, as the nozzle. Such damage to an auxiliary feedwater shown in Figure 67 (Slama 1994). Each blade nozzle has been reported at one plant makes 1.35 revolutions and is welded to the inner (Westinghouse 1989).
wall of the thermal sleeve and to the hub. A helix was installed in the replacement steam generator Several design changes have been made in the for Beznau Unit 1. Field test results showed that steam generators designed by Siemens/KWU to stratification does not take place even in the reduce or eliminate the possibility of flow presence of cold feedwater [40*C (104 F)]
stratification taking place. Flow stratification flowing at less than about 600 t/ min (159 gpm).
occurred in the horizontal portion of the main feedwater piping attached to the original Siemens 9.2 Fatigue Monitoring steam generators when the flow rate of the main of Feedwater Lines feedwater was low, as shown in Figure 65. This caused cracking in the heat-affected zone and in the base metal near the weld between the feedwater The flow rate and temperature of the feedwater are nozzle and the piping (Braschel, Miksch, and
I^tively constant through the feedwater system dun.ng power operation. However, during startup Schucktanz 1984). To prevent flow stratification from taking place, an upward berid has been and h t standby operations (Modes 2 and 3, incorporated in the piping connecting the thermal respectively) when auxiliary feedwater is injected, sleeve to the feedring inside the steam generator as the flow rate and temperatures may fluctuate.
shown in Figure 66.
Reasonable estimates of these two parameters can NUREG/CR-6456 132
MITIGATION AND MONITORING lllilli'
N w
N LEEN N
l N
Ns Feedwater Nozzle l
s.
/
\\' -\\'/
l
\\
\\
[L
/
wr//w Hot
/,,,
,.__l a
. s
._= _ m -
r.
f
.=
y V
.d~-
Cold l
(
Layer of Cold 1__. "
~
y Feedwater h
?
/
i l
P5 EM i
i i1
{8 l l
/
~
/
l l
l il
/
I, I
I l
l 4
j//
/
t
__ - g i\\
3
\\
\\
, %j (
i N
N
\\
[/
3 Feedring N
=
I hk
/
\\1 Figure 65. Original layout of the feedwater distribution piping inside Siemens/KWU steam generators.
Courtesy of P..J. Meyer, Siemens/KWU, 133 NUREG/CR-6456
MITIGATION AND MONITORING Upward Bend
~
eedwater Nozzle N~
J - Tube
-5~. l l
I N\\s i
i p
/
./.
J-_.-
l l
/
Thermal Sleeve k
N C!
I I
i 1
1 1
1 1
? Normal Water Level Mh8" Y-
-Vessel Wall
- _ q3 1
+
'q
.i.i i. i i i i i l i i e i i i t..e, iei etirriin ii TE'.
merge
( (
am i..1.....
i i
it iie' Feedring g'
~
Nozzle 1
Figure 66. Modified layout of the feedwater distribution piping inside Siemens/KWU steam generators. The upward bend in the piping prevents flow stratification. Courtesy of P.-J. Meyer, Siemens/KWU.
only be obtained by on-line monitoring. A cir-Local fatigue monitoring of feedwater piping has cumferential distribution of the feedwater pipe wall been performed at several U.S. and foreign nuclear temperatures must be measured to determine the power plants and the information used to estimate presence and extent of the thermal stratification; the auxiliary feedwater flow rates and temperatures such monitoring of a local temperature distribution during startup and hot standby operations. The on-is sometimes referred to as local fatigue monitor-line fatigue monitoring system installed at Diablo ing. The measured temperatures can be used to Canyon Unit 1, Steam Generator 2, provides a calculate the thermal stresses caused by stratifica-typical example. RTDs were attached to the tion and, then, to estimate the fatigue usage for the outside surface of the feedwater nonle-to-pipe piping and noules. This information can also be weld underneath the insulation, as shown in used as part of an overall plant on-line fatigue Figure 68 (Shvarts et al.,1994). Data were monitoring program. The calculated thermal acquired once per minute for a twelve day period stresses can also be used for estimating the growth during a plant heatup, in addition, data for five of existing eracks.
plant process parameters (feedwater pressure, NUREG/CR 6456 134
MITIGATION AND MONITORING gi
,A b
i Thermal sleeve Ii!
!li i
Iii t
2 Foodwater nozzle Figure 67. Antistratification device for effective mixing of hot and cold coolants in the feedwater nozzle (Slama 1994). Copyright Canadian Nuclear Society; reprinted with permission.
temperature, and flow rate; auxiliary feedwater coefficients for the hot and cold coo! ants, and the flow rate; and reactor coolant loop 2 temperature) height and width of the thermal stratification were collected for the same time period at the same interface. Thus a correlation between the auxiliary time intervals. The nozzle region was subjected to feedwater flow rates and the height of the interface thermal stratification with a maximum top-to-layer was established. However, the correlation bottom temperature difference of approximately will vary from plant to plant because ofdifferences 232*C (450*F) for about 3% days. The steam in the piping layouts including feedring arrange-generator was fed exclusively with auxiliary ments and in the coolant temperatures and flow feedwater during this time period. Then the main rates. The data gathered from the fatigue monitor-feedwater flow was established and the auxiliary ing system were used for the design of new transi-feedwater flow was terminated during Mode 2 tion pieces for the Diablo Canyon feedwater operations and there was no stratification.
nozzle.
l Figure 69 shows typical data from five of the nine Similar fatigue monitoring systems have been in RTDs and also the auxiliary feedwater flow rate intermittent operation at the Sequoyah plants since over a 33-hour period. A finite element analysis 1991 (Cofie et al.1994). The data were used to was used to determine the height of the interface evaluate the flaws found in the Sequoyah plant in layer of the stratified coolant that would match the 1993. The data gathered from these monitoring measured temperature distribution at the outside systems can also be used by plant operations surface. Four parameters were varied for this personnel to limit auxiliary feedwater cycling and analysis: the inside surface convective heat transfer thereby reduce crack growth.
135 NUREG/CR-6456
MITIGATION AND MONITORING RTo9 Approximate location of RW temperature sensors 7
60 RTD6 RTD5 RTD4 I2 RTD3 V
Bottom D2
/
\\
i Second-off Firstoff
?'Y_....
..".*.Y......................
......q._._
i !
,n..,,
/
\\
cownr.ime Figure 68. Local on-line fatigue monitoring of feedwater nozzle-to-pipe weld at Diablo Canyon Unit I (Shvarts et al.1994). Copyright American Society of Mechanical Engineers; reprinted with permission.
600 RTD9]
5*
- M. Oh k.[
Yf f
RTD I
l
)
h JoO i
k/
)I.
y 1
RfD2 i
8 c
200 I
h Y l
l t
h
/
f ve-RTDI i
100 k
[
b 0
. @t_.-
"7 h
- L
-100 14:24 l 19.12 l 00!OO l 04548 l 09536 l 14:24 l 19512 l 00!OO 16:48 21:36 02:24 07:12 12:00 16:48 21:36 11-6-92 to 11-7-92 RfD 1 RfD 2 RTO 3 RTD 4 RfD 9 FLOW Figure 69. Local on line fatigue monitoring results for feedwater nozzle-to-pipe weld at Diablo Canyon Unit 1 (Shvarts et al.1994). Copyright American Society ofMechanical Engineers; reprinted with permission.
NUREG/CR-6456 136
MITIGATION AND MONITORING Local monitoring of a long horizontal section of timized feedwater chemistry, use of replacement the feedwater piping at Beaver Valley Unit 1, a materials with a chromium content greater than 0.1 Westindouse 3-loop plant, was performed wt%, and other modifications such as reducing the following a reactor trip at 29% power (USNRC flow velocity or protecting the susceptible material 1991c; Van Duyne et al.1991). The reactor trip from being exposed to large flow velocities. As event was followed by the injection of cold discussed In Se: tion 7.2.2, two other reasons for auxiliary feedwater at 10*C (50*F) into the main reducing the flow-accelerated corrosion rates are to feedwater, which was at 168'C (335'F). The reduce the amount of corrosion products auxiliary feedwater flow rate was 1,400 f/ min (370 transported to steam generators and to reduce the gpm) per steam generator. He feedwater piping is fouling of the condensate polishers.
particularly susceptible to global thermal stratification because it contains a long horizontal 9.3.1 Optimized Feedwater Chemistry run inside the containment, and its connection to the auxiliary feedwater line is located outside the Flow-accelerated corrosion rates vary by an order containment. Ilowever, it had not previously been of magnitude over the cold pH range of 8.5 to 9.5, identified as subject to thermal stratification. The which is typical of feedwater systems. The rate instrumentation included 12 thermocouples, 5 increases rather dramatically when the cold pH is strain gages,8 lanyard potentiometers (to measure below about 9.5. Therefore, a relatively high displacement), I pressure transducer, and I linear feedwater cold pH is beneficial in reducing the variable differential transformer (LVDT).
extent of flow-accelerated corrosion damage.
Locations of these instruments are shown in llowever, as discussed in Section 7.2.2, high cold Figure 70. ne six thermocouples mounted around pH levels are not compatible if copper alloy the circumference of the pipe at two locations materials are present in the condensers and detected a top-to-bottom temperature difference feedwater heaters because of corrosion of copper that varied by as much as 110*C (200*F). Flow alloy materials. Also, high cold pli levels are not stratification was estimated to occur when cold desirable if the condensate polishers are present fluid, such as auxiliary feedwater, is injected into and continuously in operation because the the system at flow rates of 3,800 f/ min (1,000 regeneration frequency will be high.
gpm) or less per steam generator. Stratification has also been observed during power reductions to An optimum pli and water chemistry in a PWR levels below 30% and a subsequent reduction in secondary steamwater system is generally achieved the feedwater flow rate.
by the addition of amines such as ammonia, morpholine, and ethanolamine in the 9.3 Mitigation of Flow.
demineralized water. For feedwater systems with C PPerall ymaterialS.ac Id illevelin therange P
Accelerated Corrosion Damage of 8.8 to 9.2 should be maintamed to prevent excessive copper pickup. Morpholine is widely Several carbon steel components m the PWR used in these systems (all French PWRs and feedwater systems have experienced material loss several U.S. PWRs) to maintain a cold pH in the due to flow-accelerated corrosion, as discussed in range of 9.1 to 9.3, provided the condensate Section 6.4. The components which have been polishers are absent. Along with the use of damaged by flow-accelerated corrosion include morpholine, a low concentration of hydrazine is elbows, tees, branch connections, reducers, valves, used to lower the dissolved oxygen content in the flow contro' orifices, etc. in the feedwater pipmg feedwater.
outside the steam generator, and J tubes, feedrings, and thermal sleeves in the feedwater distribution Systems with an absence of copper alloy materials pipmg mside the steam generator. The damage can maintain a higher cold pH level to reduce the caused by flow-accelerated corrosion can be flow-accelerated corrosion of carbon steel reduced m a number of ways, including use of op-components. If ammonia is used for pH control in 137 NUREG/CR 6456
MITIGATION AND MONITORING o
- i f Spring hanger.
. Vertical restraint l
c Horizontal snubber
'L 4
Steam 0 Rupture restraint.
generator O
Containment penetration b
k f
Crane wall l
i L8 l
Steam 6
E" i
generator J-y
/
- 4/efline,g, L4 j Q
r Thermocouples Thermocouples T1 -T6 T7 T12 and strain gages S1 - S4 and AC1 r,
and AC2 gg4 T1 T7 T2 T8 Nomenclature T = Thermocouple T6 TJ T12 T9 S = Strain gage T4 T10 AC= Accelerometer 5
T11 L = anyad Thermocouole Locations C152-WHT it964e Figure 70. Local on-line fatigue monitoring of Beaver Valley Unit I feedwater line (USNRC 1991c).
138
MITIGATION AND MONITORING these PWR feedwater systems, the cold pH should ne availability of the material, allowable stresses be increased, at least up to 9.7, to avoid flow-and thermal expansion coeflicient of the makrtal, accelerated corrosion provided that the condensate and required heat treatments play a roic in polishers are not in continuous operation.
selecting the replacement material. The P1I steel tends to be less available than P22 steel. The As discussed in Section 7.2, test results have allowable stress and the thermal expansion shown that ethanolamine is effective for the entire coefficients for the SA-335, Grade Pil and P22 secondary system, including both the single-phase steels are similar to those for carbon steels, and a and two phase flow regimes. Use ofethanolamine new stress analysis is, therefore, not normally is also compatible with the continuous operation of required. However, pre-and post-weld heat the condensate polishers.
In addition, the treatment is usually required (Chexal et al.1996).
corrosion product transport with ethanolamine is lower than that with ammonia and even lower than Type 304 stainless steel is widely available but its that with morpholine. A large number of U.S.
allowable stress is lower than that of carbon steel plants are currently switching or contemplating and its thermal expansion coefficient is greater.
switching to this amine or a combination of several Therefore, a new stress analysis is required when amines.
an appreciable length of carbon steel piping is l
replaced with stainless steel piping. Another Despite these feedwater conditioning enhance-difficulty with the use of stainless steel is the ments, use of the most effective reagent or combi-special precautions required for making the nation of reagents for feedwater conditioning will bimetallic welds between the stainless steel and not completely eliminate the risk of wall thinning carbon steel sections of the piping. In addition, the during long-term operation, such as a 40-year susceptibility of the stainless steel piping to stress operation. Research results show that the maxi-corrosion cracking needs to be considered.
mum wall thinning rates are still about 0.1 to 0.2 mm/yr (0.004 to 0.008 inlyr), even when the water Similarly, carbon steel piping with a stainless steel chemistry is optimum (Bouchacourt et al.1994).
cladding or coating on the inside surface is There fore. comprehensive, exhaustive, and conser-immune to flow-accelerated corrosion.
A vative analyses of the feedwater system need to be three-layer stainless steel coating designed to performed to identify all sites susceptible to flow-prevent flow-accelerated corrosion damage to accelerated corrosion and these sites need to be carbon steel piping was developed and successfully inspected periodically for wall thickness loss.
applied to pipes containing wet steam at 10 power plants in Europe, 6 BWRs arid 4 PWRs (NEI 9.3.2 Use of Corrosion Resistant Materials 1989). He first application was in a BWR plant in 1977. He coating is flame-sprayed on the interior The risks of wall thinning can be significantly of the piping. He top layer is Type 304 stainless reduced if the susceptible sections of the feedwater steel, which is very resistant to flow-accelerated piping are replaced with piping made from corrosion. The bottom layer is chosen to provide materials, such as SA-335 Grade Pl 1 and P22 steel a sufficient mechanical bond to the carbon steel and Type 304 stainless steel, with a higher pipe, and an intermediate layer provides bonding chromium content. The experience at the Diablo between the top and bottom layers. The total Canyon plants show that flow-accelerated thickness of the three layers is usually about 0.5 corrosion of carbon steel components took place mm (0.020 in.). In situ application requires a only if the chromium content was less than 0.1 minimum pipe diameter of about 600 mm (24 in.),
wt%, as discussed in Section 7. Several PWR whereas shop application requires a minimum pipe plants have made such replacements. For example, diameter of about 100 mm (4 in.). A similar the replacement steel piping at Surry Unit I coating can be applied to the feedwater piping, contains 2.5 wt% chromium, which carries single-phase fluid. Similarly, carbon 139 NUREG/CR-6456
MITIGATION AND MONITORING steel piping clad with stainless steel will provide the transition piece and other replacement protection against flow-accelerated corrosion.
components was kept higher than 0.1 wt% to mitigate any future flow-accelerated corrosion Several other carbon steel components in the damage. The thermal liners installed in several feedwater distribution system have also been re-feedwater nozzles (see Figure 30) for protecting placed with components made from corrosion the feedwater nozzle and elbow from thermal resistant steels. De carbon steel J-tubes have been stratification also protect the leading edge of the replaced with Alloy 600 (16 wt% Cr) J-tubes in all thermal sleeve from being exposed to flowing the Westinghouse steam generators. In some water.
Combustion Engineering plants, the carbon steel discharge elbows have been replaced with elbows 9.4 Mitigation of Steam made ofP22 steel. In some Combustion Engineer-Generator Water Hammer mg plants, the distribution box has experienced local thinning and has been repaired with weld Damage a
buildup, whereas in some other plants the damaged l
distribution boxes were replaced with ones made Over 30 steam generator water hammer events of Alloy 690, which has a chromium content of have been reported in the PWR feedwater piping, about 30 wt%.
as summarized in Section 6.5. Some events did not cause any damage, but several others distorted 9.3.3 Other Modifications the intemals of the steam generator, especially the feedrings, and damaged the piping supports. Few l
The role played by high flow velocities was not steam generator water hammer events caused recognized at the time of the J-tube failures. So, in cracking in the feedwater piping.
some instances, installation of the replacement J-tubes led to a geometric discontinuity at the inside in the older Westinghouse and Combustion surface of the feedring around the entrance to the Engineering plants with top-feed steam generators, l
J-tubes.
As shown in Figure 37(a), this the water level is normally well above the feedring.
discontinuity resulted in high flow velocities that Under certain plant transients, such as a main caused wall thinning of the feedring in one PWR feedwater pump trip, the feedwater flow decreases plant. The feedring was repaired and the J-tube rapidly, and the water level in the steam generator design and installation were modified to lower the may drop below the feedring and the feedring will flow velocitics. The modification included an be uncovered. It takes about I to 2 minutes to Alloy 600 reducer welded to the feedring to lower drain the feedring ifit is designed with bottom-the flow velocity. He reducer is ground flush with discharge holes shown in Figure 41, which was a l
the feedring inside diameter to reduce turbulence, typical design for the original feedrings in the older l
as shown in Figure 37(b).
plants. The draining of a feedring can lead to a water hammer as explained in Section 6.5.1.
Wall thinning of thermal sleeves has been reported Malfunction of check valves has also played a at some plants. He possible reason is damage to major role in some water hammer events.
the leading edge of the sleeve caused by impingement of feedwater on its square edge, The steam generator water hammer problems :re followed by flow accelerated corrosion of the not expected in some newer Westinghouse and outside surface of the sleeve caused by the high Combustion Engineering plants with preheat steam flow velocity of the bypass coolant leaking generators because, as discussed in Section 2.2.1, through the gap between the sleeve and the a forward flushing practice is followed at the start-feedwater nozzle. As shown in Figure 38,in one up of the plant. In addition, the design changes PWR plant, a redesigned transition piece was identified for prevent ng water hammer in older i
installed so that a leading edge is not exposed to top-feed steam generators were incorporated in the flowing feedwater. The chromium content la preheat steam generators. However, a steam generator water hammer did occur in the auxiliary NUREG/CR 6456 140
l MITIGATION AND MONITORING
/
Reduce the length of the horizontal run of pipe feedwater line of a preheat steam generator at a non-U.S. PWR, possibly because ofmultiple check to less than 2.44 m (8 ft) at the steam genera-salve failures, tor inlet nozzle, and Promptly resume feedwater flow into the Steam generator water hammer problems are not expected with the B&W once-through steam steam generator to minimize the amount of generator design because it has separate main and steam that enters the feedring and piping.
tuxiliary feedwater headers located external to the However, limit the feedwater flow rate to steam generator shell as shown in Figure 7. The about 570 # min (150 gpm) per steam genera-feedwater from the headers is introduced in the tor so that a slug of cold water does not form.
steam generator through several risers which function like J-tubes and keep the headers full of Design guidelines and preoperational test require-water (Han and Anderson 1982, Serkiz 1983).
ments for preventing water hammers and minimiz-ing the consequences of water hammers in top-feed Several design modifications and operationa' pro-steam generators were incorporated into cedure changes were implemented to prevent or Section 10.4.7 of the Standard Review Plan in mitigate water hammer in the top-feed steam 1975. Starting with Trojan, all licensed Westing-generators. The most efTective method for prevent-house and Combustion Engineering plants ing water hammer is to keep the feedring and equipped with top-feed steam generators have J-piping filled with water so that steam does not tubes installed on the top of the feedring as shown i
enter into them. This is generally accomplished by in Figures 5 and 14, and have short horizontal runs plugging the bottom discharge holes shown in of inlet piping. The plant operators have per-Figure 41 and installing J-tubes on the top of the formed a test to confirm the adequacy of plant feedring as shown in Figure 5. He J-tubes prevent operating proceduies to avoid damaging water rapid draining, but do not stop it. A slow leakage hammers (USNRC 1979c). In addition, several i
of the feedwater takes place through the annular operating plants have made specific modifications gap between the thermal sleeve and the feedwater in the hardware or procedures. For example, some nozzle and through the closed feedwater valve. In plants installed a separate nozzle to inject cold addition, evaporation of feedwater also takes auxiliary feedwater into the steam generator when places. As a result of these leaks and the evapora-the main feedwater line is isolated in this case, tion, it takes about 20 minutes or more to drain the cold auxiliary feedwater will not enter a steam-feedrings with J-tubes. Herefore, installation ofJ-filled feedring during steam generator refilling. In tubes is effective only if the feedwater flow is some newer steam generators, such as the Delta 75 reestablished before a significant amount of steam model designed by Westinghouse, the thermal enters the feedring and piping. However, the sleeve is welded to the feedwater pipe to stop the resumed feedwater flow rate should be limited so leakage through the gap between the thermal that a water slug does not form. The explanation sleeve and the feedwater nozzle in the event the of the water hammer phenomenon presented in feedring is uncovered. The thermal sleeve is also Section 6.5.1 indicates that a shorter horizontal welded to the feedwater nozzle in the steam gener-feedwater pipe at the feedwater nozzle would ators designed by Siemens/KWU.
l reduce the magnitude of the pressure pulse result-ing from a water hammer. These design e d c pera-Other approaches used at the design stage to tion modifications are summarized aC ollows:
reduce the impact of potential steam generator water hammer include: (a) use of vents such as the Plug the holes in the bottom-discharge feed-one shown in Figure 14,(b) inspection of check ring and install J-tubes on the top of the valves, and (c) installation of redundant check
- feedring, valves on feedwater piping. Other preventative
(
L MITIGATION AND MONITORING
- measures include increased operator awareness and I-training, and utilization of surveillance equipment i
such as temperature sensors to detect voids in systems with a history of steam generator water hammer events (Serkiz 1983).
i
)
l NUREG/CR 6456 H2
-i
- 10. FINDINGS He main objective of the study reported here was Our assessment of field experience related to steam to evaluate the effectiveness ofindustry efforts for generator water hammer damage indicates that the managing thermal fatigue, flow-accelerated corro-USNRC licensees have taken sufficient actions to sion, and water hammer damage that has occurred minimize water hammer in both top-feed and in the feedwater piping. This includes an evalua-preheat steam generators. However, we have not tion ofdesign modifications, operational procedure evaluated the industry efforts to minimize the changes, augmented inspection programs, and multiple check valve failures that have played a repair and replacement activities. Four specific major-role in several steam generator water actions were taken to accomplish the objective: (a) hammer events.
review of field experience to identify trends of operating events,(b) review of related technical 10.1 Major Findings literature, (c) visits to three PWR plants and a PWR vendor, and (d) solicitation of information The characteristics of the damage caused by from eight other countries.
thermal fatigue are different than those caused by flow-accelerated corrosion.
Thermal Our asses ;mer.t of field experience related to PWR fatigue cracking generally occurs in a feedwater nozzle cracking is that the USNRC relatively local, safety-related portion of the licensees have apparently taken sufficient action to feedwater piping inside the containment, minimize the feedwater nozzle cracking caused by whereas wall-thinning caused by flow-thermal fatigue. As a result of the examinations accelerated corrosion typically occurs, with conducted in response to Bulletin 79-13, feedwater few exceptions, in the non-safety related nozzle fatigue cracking was detected in 18 PWRs balance-of-plant piping outside the during 1979 to 1983. Then, there was about one containment.
fatigue cracking event per year from 1983 to the bulletin closure in 1991. The frequency of feed-A through-wall crack caused by thermal
=
water nozzle cracking events increased to six per fatigue will generally leak long before the year during 1992 and 1993. But since then component mptures. However,in the urdikely through 1996, we have not identified any addi-event of a large overload, a pipe with fatigue tional feedwater nozzle crackmg event.
cracks might fail catastrophically without any prior leakage. A component damaged by flow-Our assessment of field experience related to flow-accelerated corrosion loses its strength and can accelerated corrosion damage shows four compo-fail under normal operating pressure; a large nents in the portion of the feedwater piping within fitting or pipe might fail catastrophically the scope of this report that have experienced without any waming.
significant wall thinning: carbon steel J-tubes, feedrings, and thermal sleeves in the top-feed Sites susceptible to thermal fatigue cracking steam generators, and auxiliary feedwater lines in are found in those portions of the feedwater the preheat steam generators. He USNRC licens-piping and nozzles where stratified flows and ecs have taken sufficient action to minimize the coolant leakage, respectively, are present; wall thinning in J-tubes and auxiliary feedwater these locations are generally well identified.
lines. Ilowever, we did not find specific industry Sites susceptible to flow-accelerated corrosion actions to minimize the wall thinning in feedrmgs are found throughout the feedwater system and l
cnd thermal sleeves, but we focn3 visual inspec-are difficult to identify without predictive tion being performed and repair when needed.
analysis because several factors are involved.
l 143 NUREG/CR-6456 l
The factors causing thermal fatigue and flow-10.2.1 Therm:I Fctigue accelerated corrosion damage are well understood. Advancedultrasonicexamination Thermal fatigue cracking of the main techniques can be used to reliably characterize feedwater piping has not occurred at plants thermal fatigue cracks.
Cost-effective where the auxiliary feedwater is directly radiographic techniques for estimating wall introduced into the steam generator shell and thickness are being developed as a means to not into the main feedwater line. Such plants assess flow-accelerated corrosion damage.
include Babcock & Wilcox PWRs with once-through steam generators and Westinghouse Several effective techniques have been and Combustion Engineering PWRs with developed for monitoring, mitigating, and recirculating steam generators equipped with repairing the damage caused by thermal preheaters.
fatigue and flow-accelerated corrosion.
Use of cold auxiliary feedwater with cyclic The above findings indicate that appropriate on/off flow control is mainly responsible for analysis, inspection, monitoring, mitigation, and the flow stratification which causes ther-replacement techniques have been developed for mal fatigue damage.
managing thermal fatigue and flow-accelerated corrosion damage to feedwater nozzles, piping, and The major fatigue cracking is caused by local feedrings. Adequate trammg and appropriate thermal stratification. Factors which cause li APP cations of these techniques can ensure high stresses during thermal stratification are effective management of the damage. Several the temperature difference between the hot PWR plant operators have been proactive in managing this type of damage.
steam generator coolant and cold auxiliary feedwater and the geometric discontinuities on the inside surface of the feedwater piping.
The major findings are supported by the followm.g specific findings, which are grouped m three Major cracks have generally been found at categories: (a) damage caused by degradation geometric discontinuities (such as counter-mechanisms, (b) inservice inspection methods for bores and welds) where large, throughwall charactertzmg the damage, and (c) techniques used stresses develop and cause crack growth.
to mitigate and repair the damage.
The stress distribution resulting from local 10.2 Specific Findings -
thermal stratification is complex because of the short length of the piping exposed to the Degradation Mechanisms stratified flow, the feedwater nozzle and elbow end constraint effects, and the Effective management of aging damage requires geometric discontinuities at the inside surface that the degradation mechanisms causing damage of the feedwater piping.
be well understood. The nuclear industry has accumulated considerable plant experi-nce, The temperature distribution in a pipe wall is performed flow stratification and flow-acceler-plant specific, because in addition to the hot ated corrosion tests to determine the factors and cold coolant temperatures and flow rates, affecting the damage, and developed tools to it depends on piping layout. Therefore, the predict the flow-accelerated corrosion. The plant results from laboratory tests or from other L
operating experience and our review of the plants may not be directly applicable to a related literature indicate these mechanisms are given plant.
adequately understood. Our specific findings for thermal fatigue and flow-accelerated corrosion of PWR feedwater piping are as follow.
NUREO/CR-6456 144
FINDINGS Piping configuration (elbow, tee, nozzle, Thermal striping takes place in a stratified a
fluic when large relative velocities between orifice, etc.) and piping inside surface irregu-hot.md cold coolants are present. Stresses latities determine the local velocities of the produced by thermal striping can initiate feedwater. Inside surface irregularities are cracks on the inside surface but do not cause developed during the fabrication process or through-wall crack propagation because caused by design related features such as through-thickness stresses attenuate rapidly.
weld discontinuities and surface damage caused by fatigue cracks or cavitation pitting.
It appears from the available data, that the The piping configuration has a primary effect recerut fatigue cracking of feedwater piping is on the local flow velocities of the feedwater, limited to U.S. PWRs. Information obtained whereas the surface irregularities have a from foreign countries does not mention any secondary effect.
recent fatigue cracking events.
Components in carbon steel (SA-106 GrB)
Bypass leakage through the gap between the feedwater piping containing a very small thermal sleeve and feedwater nozzle has amount of chromium (less than 0.1 wt%) and caused fatigue cracking in the nozzle bore, exposed to high local velocity fluids are more nozzle inside radius, and steam generator susceptible to flow-accelerated corrosion.
shell regions. At least two main feedwater The coolant chemistry (dissolved oxygen and nozzles and one auxiliary feedwater nozzle have experienced such cracking.
pH level) also affects the flow-accelerated corrosion rates. A lower dissolved oxygen Thermal cycling could cause cracking in the content and pH level are associated with auxiliary feedwater piping where it connects higher corrosion rates. Even though the cold to the main feedwater line. However, such pH level is the same for the entire feedwater cracking has not been reported.
system, the hot pH level varies throughout the system. As a result, the susceptibility to 10.2.2 Flow-Accelerated Corrosion flow-accelerated corrosion also varies throughout the system.
Flow-accelerated corrosion is widespread; it Use of optimum water chemistry reduces has been reported at both U.S. and foreign a
PWRs.
wall thinning rates but does not eliminate them; some research results show that the The main factors affecting feedwater piping maximum rates are still about 0.1 to 0.2 flow-accelerated corrosion are fluid tempera-mm/yr (0.004 to 0.008 in./yr), even when ture, bulk and local flow velocities, material the water chemistry is optimum.
composition, and water chemistry. These Several uncertainties are present when esti-factors may vary along the feedwater piping and, therefore, play important roles in mak-mating flow-accelerated corrosion rates.
l ing some locations more susceptible to flow-These uncertainties are associated with the l
accelerated corrosion than other locations.
original piping component thickness. %
exact amounts of the trace alloys present m 1
l The maximum corrosion rate for most the material, the actual number of operating feedwater piping conditions occurs at about hours, the plant chemistry history, and the 150*C (300"F).
presence of discontinuities on the inside surface of the piping.
145 NUREG/CR-6456
FINDINGS Flow-accelerated corrosion has caused wall Advanced ultrasonic techniques s :ch as tip-thinning of steam generator feedrings and diffraction techniques can reliably size crack distribution boxes. Many utilities have mod-depths.
eled these components in CHECWORKS to Discrimination ofreflectors, such as geometry determine their susceptibility to such dam-age.
effects, inclusions, and crack tips, is essential for reliable detection and accurate sizing of It is not known whether thinning of thermal fatigue cracks. Use of more than one inspec-sleeves is a widespread phenomenon. It tion technique provides more reliable sizing of appears that flow-accelerated corrosion could cracks.
have caused this thinning. Monitoring of nozzle wall temperature can detect leakage Radiographic examinations are not adequate taking place because of thinning.
for detecting tight fatigue cracks but can detect cracks that are open and filled with corrosion Flow-accelerated corrosion generally does Products.
not cause wall thinning in auxiliary feedwater lines because the coolant in these lines is Field experience shows that counterbores at stagnant during power operations and it is the feedwater nozzle-to-pipe welds have cold when it is flowing. One exception is experienced thermal fatigue cracking. The PWRs with preheat steam generators where cracking locations have often been outside the ASME Code examination volume and, a small portion of the main feedwater is diverted through the auxiliary feedwater lines therefore, not inspected during ISI.
during normal operation. A portion of these The phased array technique can be used for lines exposed to hot main feedwater flowing inspection of the feedwater nozzle inner radius at high velocities have experienced regi n.
This technique has been used at significant wall thinning.
several BWR plants to mspect the feedwater n zzle inner radius region, which is also Valve leakage has also caused flow-susceptible to fatigue damage, and its accelerated corrosion damage in piping Performance has been demonstrated with containing stagnant fluid.
nozzle mockups.
10.3 Specific Findings -
10.3.2 Characterization of wall Thinning inservice Inspections Use of conventional radiographic testing for 10.3.1 Characterization of Fatigue Cracks thickness measurements is limited to small-diameter piping; inspection of large-diameter P ping requires longer exposure times resulting i
Conventional ultrasonic examinations are reliab!c in detecting thermal fatigue cracks in in higher costs and increased personnel carbon stee! feedwater piping, but have a very exposure. The main advantage of RT is tnat it poor sizing capability, does not require removal of the insulation.
Use of filmless radiography with phosphor Ultrasonic examinations conducted using the P ates is being evaluated in the field. This l
minimum ASME Code requirements are reliable for estimating the length of a fatigue technique is expected to dramatically reduce crack, but not the through-wall extent of the the exposure dose and significantly reduce
- crack, inspection time and personnel safety concerns associated with performing RT within a plant.
NUREG/CR-6456 146
FINDINGS Several plants have redesigned the feedwater Manual UT is the most commonly used in-spection method for the detection and trending nozzle counterbores to eliminate the stress of wall thickness changes because ofits accu-raisers and installed thermal liners to protect racy and relatively low cost. A properly con-the counterbores and other susceptible sites ducted UT examination can estimate the pipe from thermal fatigue damage. One plant has wall thickness within 5% of the actual value.
modified the feedwater piping layout to Industry practice is to overlay a grid on the introduce the auxiliary feedwater directly into pipe wall and then spot measure the thickness the steam generators instead ofinto the main at each grid location.
feedwater lines. (These steam generators originally had auxiliary feedwater nozzles that Field experience has found that use of auto-were not used.)
mated UT for thickness measurements is too l
In one plant, the feedwater nozzle-to-piping cumbersome and time consuming.
connection has been redesigned to protect the Computed tomography, an advanced radio-leading edge of the thermal sleeve from flow-graphic examination technology, can be used accelerated corrosion.
This design also to characterize wall-thinning damage in piping prevents feedwater leakage through the gap and thermal sleeves. However, this technology between the nozzle and the thermal sleeve and, is slow and expensive, and the equipment is thus, protects the nozzle bore and inner radius too large for use in the field.
region from fatigue damage. Use of a thermal liner also provides similar protection to the The computer code CHECWORKS (or its thermal sleeve and feedwater nozzle.
earlier versions) is used by all U.S. PWR Use of a fatigue monitoring system can assist utilities for estimating flow-accelerated corrosion rates and identifying inservice in detecting the presence of thermal stratifica-inspection locations in feedwater piping.
tion and in predicting temperature distributions Utilities perform inspections at additional and flow rates so that the fatigue usage and locations based on industry experience and crack growth rates can be estimated more because of the uncertainties associated with accurately.
several input parameters to the code.
Low-alloy steels such as 1%Cr-%Mo steel 10.4 Specific Findings -
(SA335, Grade Pil) and 2%Cr-lMo steel (SA335, Grade P22), austenitic stainless steel, j
Mitigation, Monitoring, and Alloy 600 provide protection against flow-Repair, and Replacement accelerated corrosion.
Ammonia may be used for the pH control in Use of continuous auxiliary feedwater flow rather than intermittent flow significantly the feedwater systems with no copper alloy reduces the thermal fatigue damage to the materials and no condensate polishers present; feedwater piping.
Several utilities have however, the cold pH should be increased, at implemented continuous auxiliary feedwater least up to 9.7, to avoid flow-accelerated l
flow by making modifications to the feedwater corrosion damage.
system and changes in the operating I
Morpholine is widely used in the feedwater procedures systems with copper alloy materials in the Some plants use heated main feedwater during water heaters and condensers to maintain the plant startup and hot standby conditions to cold pH in the range of 8.8 to 9.2, provided the mitigate thermal fatigue damage in the condensate polishers are absent. Ethanolamine feedwater nozzle region.
instead of morpholine may be used when condensate polishers are present.
147 NUREG/CR 6456
FINDINGS The most effective method for preventing I
' Nuclear industry experience indicates that the corrosion products transport with ethanol-water hammer is to keep the feedring and i
amine is lower than that with ammonia and piping filled with water so that steam does not even lower than that with morpholine.
enter into them. Therefore, the thermal sleeve in several newer or replacement steam Several piping design modifications and generators is welded to the feedwater piping to operational procedure changes made to the stop leakage through the gap between the feedwater system appear to be adequate for thermal sleeve and the feedwater nozzle, in the
)
preventing water hammer in top feed steam event the feedring is uncovered. This change generators; no water hammer events have been also protects the thermal sleeve from flow-reported recently, accelerated corrosion damage and the feedwater nozzle bore and inside radius
{
section from thermal fatigue damage.
i
\\
l NUREG/CR-6456 148
l
- 11. REFERENCES Aguiar, D. J., l993. Failure Investigation ofDiablo Canyon Power Plant Unit #1 Feedwater Nozzle Piping Adjacent to Steam Generators 1 1 and 1-3, Report 420DC-93.19, Pacific Gas & Electric.
ASME 1995 a. Risk-BasedAlternative Selection Processfor Inservice inspection ofL WR Nuclear Power Plant Components, ASME Research Task Force on Risk-Based Inspection Guidelines - LWR Nuclear Power Plant Application, November.
ASME 1995b. Section IX-Quah)1 cation Standardfor Welding and Brating Procedures, Welders, Brazers, and Welding and Brazing Operators, ASME Boiler and Pressure Vessel Committee, Subcommittee on Welding, pp. 65-89.
ASME 1990. " Case N-480: Examination Requirements for Pipe Wall Thinning Due to Single Phase Erosion Corrosion," Cases ofASME Boller and Pressure Vessel Code,Section XI, Division 1, American Society of Mechanical Engineers, New York.
Bain, R. A., S. O. Collins, and M. F. Testa 1992. "New Insights into Thermal Stratification of Feedwater Piping in PWR Plants," PVP Vol. 235, American Society ofMechanical Engineers, pp. 113-118.
Balkey, K. R.,1995. "Recent Industry, Regulatory and ASME Developments," American Society of Mechanical Engineers Research Task Force on Risk-BasedInspection, San Francisco, California, November 28.
Bamford, W. H., A. Thurman, and M. Mahlab 1981. " Fatigue Crack Growth in Pressurized Water Reactor Feedwater Lines," American Society ofMechanical Engineers 81-PVP-2.
Bamford, W. H., et al.1987. " Integrity Issues in PWR Steam Generators and Feedwater Systems," American Society ofMechanical Engineers P VP-Vol.119, pp.19-30.
Bamford, W. H., G. V. Rao, and J. L. Houtman 1992. " investigation of Service-Induced Degradation of Steam Generator Shell Materials," Proceedings of the Ftfth International Symposium on Environmental Degradation ofMaterials in Nuclear Power Systems - Water Reactors, August 25 29, 1991, Monterey, Cahfornia, pp. 588-595.
Becker, F. L., et al.1992. " Performance Demonstration Test Blocks to Meet the Requirements of the ASME Code," Proceedings ofthe lith International Conference on NDE in the Nuclear andPressure Vessel Industries, April 1992, Albuquerque, New Mexico, ASM International, pp. 213 218.
Bisbee, L H.,1994. " Enhanced Ultrasonic Examination of Feedwater Pipe-to-Nozzle Welds," Nuclear Plant Journal, March-April, pp. 42 53.
Block, J. A., et al.1976. Evaluation ofPWR Steam Generator Water Hammer, NUREG-0291, U.S. Nuclear Regulatory Commission.
Bosnak, R. J.,1987. " Meeting Minutes of 1/15/87 Technical Panel Discussion of Surry 2 Pipe Failure implications," Memorandum to T.
P.
Speis, January 15,1982, U.S. Nuclear Regulatory Commission.
149 NUREG/CR 6456
REFERENCES I
Bouchacourt, M., et al.1994. "The BRT-Cicero Code, An Exhaustive Approach to Predicting Flow Accelerated Corrosion," Proceedings ofthe InternationalSymposium on Contribution ofMaterials Investigation to the Resolution of Problems encountered in Pressurized Water Reactors (FontevraudIN),1986, Volume 2, Royal Abbey of Fontevraud, France, pp. 761-768.
Braschel, R., M. Miksch, and G. Schucktanz 1984. " Thermal Stratification in Steam Generator Feedwater L1nes,"JournalofPressure Vessel Technology, Vol.106, February, pp. 78-85.
Bray, D. E., and R. K. Stanley 1989. Nondestructive Evaluation - A Toolfor Design, Manufacturing, and Service, McGraw-liill, New York, pp. 98-99.
Bressler, M. N.,1994. An IntegratedApproach to Address Engineering ofOperating Nuclear Power Plants, Part I, Codes & Standards, PVP-16, notes from tutorial presented at the 1994 ASME Pressure Vessels and Piping Conference, Minneapolis, Minnesota, June 1994.
l Brook, M. V.,1986. " Ultrasonic Transducers Design for Detection and Characterization of Cracks in Nuclear
\\
Primary Pressure Piping," Proceedings ofthe 8th International Conference on NDE in the Nuclear Industry,17-20 November 1986, Kissimmee, Florida, American Society for Metals, Materials Park, Ohio, pp. 439-445.
Bush, S.11.,1980. " Nondestructive Examination and Component Reliability - What Should We Be Looking For," Presented at the ThirdInternational Conference on NDE, Salt Lake City.
l Chexal, B., et al.1996. Flow-Accelerated Corrosion in Power Plants, EPRI TR-106611, Electric Power Research Institute, Palo Alto.
Chexal, V. K., and J. S. llorowitz 1995. "Chexa!-Horowitz Flow-Accelerated Corrosion Model-Parameters and Influences," Current Perspectives ofInternational Pressure Vessels and Piping Codes and Standards, PVP-Vol. B, pp. 231-243. American Society of Mechanical Engineers, New York.
Chexal, V. K., and R. L. Jones 1988. " Implications of the Surry Piping Failure for Other Nuclear and Fossil Units,"InternationalJournalofPressure Vessels & Piping. 31, pp. 331-343.
Cockfield, D. W. (Portland General Electric),1988. Letter to USNRC, April 27.
Cofie, N. G., et al.1994. " Management of Steam Generator Feedwater Nozzle Cracking in PWRs," American Society of Mechanical Evineers, PVP Volume 286, pp. 31-43.
Coley, J. L.,1992. " Inspection reports 50-327/92-09 and 50-328/92-09, Sequoyah I and 2," U.S. Nuclear Regulatory Commission, Region 11.
Consolidated Edison 1991. Indian Point Unit 21991 Steam GeneratorInspection. Evaluation andRepair Activities, Consolidated Edison Company, New York, NY, pp. 8-1 to 8 11.
Consumers Power 1979. " Meeting with NRC on October 25,1979 - Feedwater Pipe Cracking Problem,"
h Consumers Power Company, Michigan.
NUREG/CR-6456 150
REFERENCES Cowfer, C. D.,1989. " Basis / Background for ASME Code Section XI Proposed Appendix VII: Ultrasonic Examination Performance Demonstration," Nondestructive Evaluation Planning andApplication, R.
D. Streit (ed), Presentedat The Pressure Vessels andPiping Conference, Honolulu, Hawaii, July 23-27,1989, NDE-Vol. 5, American Society of Mechanical Engineers, New York.
Czajkowski, C. J.,1987. Afetallurgical Evaluation of an 18-inch Feedwater Line Failure at the Surry Unit 2 Power Station, NUREG/CR-4868, BNL-NUREG-52057.
Davis, J. M.,1991. "ID Creeping Wave Techniques for Detection, Characterization, and Sizing of Planar Flaws,"NDE: Applications, AdvancedAfethods, AndCodes andStandards, PVP-Vol. 216/NDE-Vol.
9, American Society of Mechanical Engineers, New York.
DeardorfT,
A. F., et al.1990. "L%velopment of a Thermal Striping Spectrum for Use in Evaluating Pressurizer Surge Line Fatigue," Advances in Dynamics ofPiping andStructural components, PVP Vof.198, American Society of Mechanical Engineering, New York, pp. 23-28.
Delp, G. A., et al. 1985. " Erosion / Corrosion in Nuclear Plant Steam Piping: Causes and Inspection Program Guidelines, " EPRI NP-3944, Electric Power Research Institute, Palo Alto.
Diablo Canyon FSAR. " Accident Analysis," Chapter 15, DCPP Units 1 & 2 FSAR Update.
Doctor, S. R., et al.1995. Development and Validation cfa Real-Time SAFT-UTSystemfor the Inspection ofLight Water Reactor Components, NUREONR-4583 (draft), PNL-5822, Vol. 4, February.
- l Doctor, S. R.,1984. "NDE Reliability Assess lent," Non-Destructive Examination of Pressuri:ed Components, Proceedings of the ThirdIny'rnationalSeminar on Non-Destructive Examination in Relation to StructuralIntegrity, Afonterey,'Cahprnia. USA, 30-31 August,1983, EIsevier, pp. 323-335.
Duke Power 1979. "Oconee Nuclear Station, Docket Nos. 50-269,-270,-287, Information Requested on PWR Feedwater Lines," letter from W. O. Parker, Jr. (Duke Power) to H. R. Denton (U.S. Nuclear Regulatory Commission), docket Nos. 50-269,270-287, July 27.
Duquesne Light 1993. " Design Stress for the Auxiliary Feedwater System Exceeded due to Water Hammer,"
Licensee Event Report 93-01, Beaver Valley Unit 2.
Edelmann, X., and M. Gribi 1990. " Mechanized wall thickness inspection measures up to its cost," Nuclear Engineering International, pp. 54-55.
EPRl 1983. UTOperator Trainingfor Intergranular Stress Corrosion Cracking (IGSCC), Competency Area 910.
Fay, C. W.,1987. Response to IE Bulletin 87-01, Point Beach Nuclear Plant, Units 1 and 2, September 10.
Finn, S.,1989. Nuclear Power Plant System Sourcebook - Zion 1 & 2, SAIC 89/1532, prepared for U.S.
Nuclear Regulatory Commission.
151 NUREG/CR-6456
REFERENCES Fischer, E., et al.1992. "A New Approach for the Inservice Inspection of BWR Reactor Pressure Vessels,"
Proceedings of the 1Ith International Conference on NDE in the Nuclear and Pressure Vessel Industries, Albuquerque, NewMexico, USA, April 30- May 2, ASM Internstional, pp. 87-91.
Florida Power & Light. St. Lucie Final Safety Analysis Report, Docket 50 389.
Florida Power & Light 1984. " Steam Generators Feedwater Nozzles Cracking," Licensee Event Report 84-005, Turkey Point Unit 4.
Foley, W. J., R. S. Dean, and A. Hennick 1991. Closeout ofIE Bulletin 79-13: Cracking in Feedwater System Piping, NUREG/CR 5285, PARAMETER IE-176.
Garrity, J. H. (Maine Yankee Atomic Power Company),1983. Letter to the U.S. Nuclear Regulatory Commission, " Main Feedwater System," February 17.
. Gerber, T. L., et al. 1989. " Acceptance Criteria for Structural Evaluation of Erosion / Corrosion Thinning in Carbon Steel Piping," Transactions ofthe 10th International Conference on Structural Mechanics in Reactor Technology, Analwim, Cali ornia,1989, Association for Structural Mechanics in Reactor f
Technology, Los Angeles, Vol. D, pp.93-101.
Goldberg, A., R. D. Streit, and R. G. Scott I980. Evaluation ofCracking in Feedwater Piping Adjacent to the Steam Generators in Nine Pressurized Water Reactor Plants, NUREG/CR-1603, UCRL-53000.
Gordon, B. M., et al.1987. " Service Experience of BWR Pressure Vessels," Performance and Evaluation of Light Water Reactor Pressure Vessels, PVP Vol. I19, American Society of Mechanical Engineers, New York, pp. 9--17.
Hafner, W., and J. H. Spurk 1990. " Formation of Waves in Thermal Stratified Flow," presented at the International Conference on Physical Modeling of Transport and Dispersion, MIT, Cambridge, Massachusetts, August 7-10,1990.
Han, J. T., and N. Anderson 1982. Prevention andMitigation ofSteam Generator Water Hammer Events in PWR Plants, NUREG-0918.
Harrington, T. P.,1988. Adqptation ofthe PARIS Technologyfor Measuring Wall Thinning in Ferritic Steel Components, EPRI NP-5961, Electric Power Research Institute, Palo Alto.
Higuchi, M., and K. lida 1992. " Fatigue Strength Data of LWR Structural Materials in Japan (Effects of LWR Water Environment on Fatigue Strength)," TechnicalInformation)>om Workshop on Cyclic Life on EnvironmentalEfects in NuclearApplications, P VRC Volume 2, Clearwater Beach, Florida, January 20-21, 1992, Welding Research Council, Inc., New York.
Hoemann, C. S., and E. G. Berak 1989. " Engineering Evaluation of Erosion / Corrosion in Piping Systems" Dansactions ofthe 10th International Conference on StructuralMechanics in Reactor Technology, Anaheim, California, Vol. D, American Association for Structural Mechanics in Reactor Technology, Los Angeles, pp. I17122.
1 NUREG/CR-6456 152
REFERENCES Hu, M. H., J. L. Houtman, and D. H. White 1981. " Flow Model Test for the Investigation of Feedwater Line Cracking for PWR Steam Generators," American Society of Afechanical Engineers paper 81-4, American Society of Mechanical Engineers, New York.
Iida, K., et al.1988. " Abstract of DBA Committee Report,1988-Survey of Fatigue Strength Data ofNuclear Structural Materials in Japan," unpublished article sent to V. N. Shah (K. Iida is Professor Emeritus, University of Tokyo, Tokyo).
Izenson, M. G., P. H. Rothe, G. B. Wallis 1988. Diagnosis of Condensation-Induced Waterhammer, NUREG/CR-5220, Creare TM-1189, Vol.1, pp. 48-49.
Jouns, O.,1988. Erosion-Corrosion of PWR Feedwater Piping-Survey of Experience, Design, Water Chemistry, and Afaterials, NUPIG/CR-5149, ANL-88-23.
Kim, J. H., et al.1993. " Thermal Stratification and Reactor Piping Integrity," Nuclear Engineering and Des /gn,139, pp. 83-95.
Lawson, L.,1995. " Compton X-Ray Backscatter Depth Profilometry for Aircraft Corrosion Inspection,"
Afaterials Evaluation, November, pp. 936-941.
Lee, H., and E. S. Kenney 1992. "A New Pipe Wall Thinning Inspection System," Nuclear Technology, October, pp. 70-78.
Maeda, N., and G. Yagawa 1991. "Some experiences and development in Japan related to nondestructive examination and fracture mechanics in structural integrity assessment of nuclear power plant components," Nuclear Engineering and Design,131, May, pp. 329-336.
Martin, J. H., et al.1990. Safety Engineering Root Cause Report, Root Cause Report 90-001, Southern California Edison.
Martin, J.,1993. "ANO-2 Shutdown Cooling Line Stratification," presented at the EPRI Workshop on Thermal Stratification, Cycling, and Striping (TASCS) Evaluation, June 17-18, 1993, Orlando, Florida, Electric Power Research Institute, Palo Alto.
McBrearty, R. A.,1993. " Report of June 21-25,1993, Inspection at Haddam Neck, Report No. 93-11," U.S.
Nuclear Regulatory Commission Region I.
Mattu, R. K., et al. I988. Utility Response to NRC Bulletin 87-01. Thinning ofPape Walls in Nuclear Power Plants, EPRI NP-6066, Electric Power Research Institute, Palo Alto.
Metals Handbook 1989. " Nondestructive Evaluation and Quality Control". Metals Handbook, Ninth Edition, Volume 17, American Society for Metals International, Materials Paris, Ohio.
l Miksch, M., et al.1985. " Loading Conditions in Horizontal Feedwater Pipes of LWRs Influenced by Thermal Shock and Thermal Stratification Effects," Nuclear Engineering and Design. 84, pp.179-187.
153 NUREG/CR-6456
REFERENCES Miyoshi, S., H. Fukui, A. Uehara, and M. Yoshida 1992. " Development of the X-Ray CT Scanner for the Piping in Nuclear Power Stations", Proceedings ofthe lith International Conference on NDE in the Nuclear andPressure VesselIndustries, Albuquerque, New Mexico, April 30-May 2, American Society for Metals International, Materials Park, Ohio, pp. 263-265.
Mostafa, S., and M. B. Ramsey.1994. Unit 3 Steam Generator Feedwater No:zle Pipe Cracking, San Onofre Unit 3, RCE 93-024, Southern California Edison.
Muscara, J.,1990. "NDE Reliability: A Major Element in the Defence-in-Depth Concept," Proceedings of the 10th International Conference on NDE in the Nuclear andPressure VesselIndustries, Glasgow, Scotland,11-H June,1990, ASM International, pp.17-23.
NEA (Nuclear Energy Agency) l990. Feedwater Pipe Rupture at Loviisa 1 Nuclear Power Plant Unit 1, NEA #1102.02, Organization of European Community Development (OECD).
NEA (Nuclear Energy Agency) 1993. Feedwater Pipe Rupture at Loviisa 2 Nuclear Power Pla,st Unit, NEA
- 1352.02, Organization of European Community Development (OECD).
NEI 1989. "How Stainless Steel Coating Combats Erosion / Corrosion in Wet Steam," Nuclear Engineering International, May, pp.16-18.
Nordmann, F., and Fiquet, J.1996. " Selection Criteria for the Best Secondary Water Chemistry," Nuclear Engineering andDesign, pp. I93-201.
l Pers-Anderson, E.,1993. "PWR Vessel Head Penetration Inspections," Assuring Structuralintegrity ofSteel Reactor Pressure Boundary Components,12th SAfiRT-Post conference Seminar No. 2, Paris France, 23-23 August,1993.
l Peterson, S. R.,1992. Summary ofOctober 20,1992, Public Afeeting to Discuss Steam Generator Feedwater Norrie Cracking, U.S. Nuclear Regulatory Commission, November 23,1992.
PG&E l992. Indications on the Afain Feedwater Piping Near the Steam Generator Feedwater Nozzles Due to ThermalFatigue, Licensee Event Report 92-022-01, Diablo Canyon Unit 1.
PlSC 111985. A summary ofthe Programfor the Inspection ofSteelComponents-11(PISC-II) project, P1SC 11 Report No.1, June.
PSE&G 1992. " Steam Generator Feedwater Piping Linear Indications," Licensee Event Report 92-014-00, Salem Generating Station, Unit 1.
l Rathgeb, W., et al.1992. "Recent Applications of UT Phased Array Techniques for Inservice Inspection of 1
Primary Components," Proceedings ofthe lith International Conference on NDE in the Nuclear and Pressure VesselIndustries. Albuquerque, New Alexico, USA, April 30 - Afay 2, ASM International, pp.317 321.
Remy, F.N., and M. Bouchacourt 1992. " Flow-Assisted Corrosion: A Method to Avoid Damage," Nuclear Engineering and Design,133, pp. 23-30.
NUREG/CR-6456 154 1
REFERENCES Richardson, J. E.,1992. " Summary of the Erosion / Corrosion Audits: Related Events and Proposed Future Actions," USNRC Aiemorandumfor W. T. Russell, December 1,1992.
Roarty, D.,1993. " Overview of TASCS (nermal Stratification, Cycling, and Striping) Testing Program,"
presented at the EPRI Workshop on Thermal Stratification, Cycling. and Striping (TASCS)
Evaluation. June 17-18,1993, Orlando, Florida, Electric Power Research Institute, Palo Alto.
Roarty, J. D.,1986. " Corrosion-Erosion of Steam Generator J-Tubes," Proceedings of the International Afeeling on Nuclear Power Plant Maintenance, Afarch 23-27,1986, Salt Lake City, Utah, pp. 7-53 to 7-62, American Nuclear Society, LaGrange Park, Illinois.
Scherer, A. E. (Combustion Engineering),1981. Letter to Mr. Stello (U.S. Nuclear Regulatory Commission),
Steam Generator Feedring Damage," July 17.
Serkiz, A. W.,1983. Thermal-Hydraulics ofNuclear Reactors, Vol. II, papers presented at The Second International Topical Afeeting on Nuclear Reactor Thermal-Hydraulics, Santa Barbara, Cahfornia, January 11-14, American Nuclear Society, pp. 803-805.
Sexton, D. E., M. Kasahara, and R. A. Uffer 1982. Evaluation of Water Hammer Potentialin Preheat Steam Generators, NUREG/CR-3090, December.
1 Seydal, J.,1982. " Ultrasonic Synthetic-Aperture Focusing Techniques in NDT," Research Techniques in Nondestructive Testing, Volume 1, Sharpe, R. S. (editor), Academic Press, pp.1-47.
Shack, W. J., and O. Jonas 1988. " Investigation of Erosion-Corrosion in Reactor Piping," presented at the Workshop on StructuralIntegrity ofReactor Piping Systems, Afay16-19. Tokyo, Japan, U.S. Nuc1 ear Regulatory Commission and Japanese Ministry ofInternational Trade and Industry.
Shah, V. N., and P. E. MacDonald (editors) 1993. " Aging and Life Extension of Major Light Water Reactor Components," Elsevier Science Publishers, pp.535-539.
Shankar, R., and J. Bridgeman,1990. " Erosion / Corrosion Data Handling for Reliable NDE",10th International Conference on NDE in the Nuclear and Pressure VesselIndustries, Glasgow, Scotland, June Il-14, pp. 397-401.
Shvarts, S., et al.1994. " Development of Methodology for Evaluating and Monitoring Steam Generator Feedwater Nozzle Cracking in PWRs," Plant Systems / Components Aging Afanagement, PVP-Vol.
283, American Society of Mechanical Engineering, New York, pp.97-109.
Silk, M. G.,1996. " Estimates of the Probability of Detection of Flaws in TOFD Data with Varying Levels of Noise," Insight, Vol. 38, No.1, January, pp. 31 36.
Slama, G.,1994. " Steam ienerator Towards Greater Reliability," Steam Generator and Heat Exchanger Conference Proceedings, Volume 1, Canadian Nuclear Society, Toronto, pp. 1.35-1.45.
Snaider, R.,1980. BWR Feedwater No::le and Control Rod Drive Return Line Nozzle Cracking, NUREG-0619.
REFERENCES Spanner, J. C., et al.1992. " Performance Demonstrations for Inservice Inspection of Nuclear Power Plant Components," Proceedings ofthe lith International Conference on NDE in the Nuclear andPressure VesselIndustries, April 1992, Albuquerque, NewMexico, ASM International, pp. 205-212.
Sperber, M., 1988. " Inspection of Surry Feedwater Piping Uncovers New ninning Problems,"Inside NRC, October 10, p. 9.
Stoller (S. M. Stoller Corporation) 1979. " Cracked FW-to-Nozzle Pipe Elbow Welds," Nuclear Power Experience, PWR-2, VI.E.211.
Stoller (S. M. Stoller Corporation) 1983. " Reactor Trip - Water Hammer D=ing SG Level Restoration -FW Pipe Weld Cracks," Nuclear Power Experience, PWR-2, VI.E.435.
Stoller (S. M. Stoller Corporation) 1985. " Heater Drain Pump Discharge Pipe Ruptured, Personnel Burned-FW Pressure Transient During Turbine, Reactor Trips--Pipe Eroded / Corroded," Nuclear Power Experience, PWR-2, III.
Stoller (S. M. Stoller Corporation) 1986. " Addition to VI.E.561 re: Aux FW Pumps Inoperable - Steam Supply Line Hangers Damaged - Water Hammer from Condensed Steam," Nuclear Power Erperience, PWR-2, VI.E.627.
Stoller (S. M. Stoller Corporation) l987. Characterization ofthe Perfbrmance ofMajor LWR Components, EPRI NP-5001, Electric Power Research Institute, Palo Alto.
Stoller (S. M. Stoller Corporation) 1991. "SG FW Sparger Damage - Metal Debris - Feedring Weld Hennal Stress, Vent Assembly Erosion & Corrosion - Inadequate Design," Nuclear Power Experience, PWR-2, V.D.551.
Stoller (S. M. Stoller Corporation) 1992a. "IN 92-07 re: Rapid Flow-Induced Erosion / Corrosion of FW Piping
- Westinghouse SG Design Deficiency," Nuclear Power Experience, PWR-2, VI.E.1131.
l Stoller (S. M. Stoller Corporation) 1992b. " Linear Indications on Weld in SG Piping - Thermal Fatigue Suspected," Nuclear Power Experience, PWR-2, V.D.590.
l Su, N. T.,1990. SpecluiStudyReport-Review ofThermalStratifcation Operating Erperfence, AEODIS902, l
U.S. Nuclear Regulatory Commission, Office ofAnalysis and Evaluation of Operational Data, March.
Talja, A., and E. Hansjosten 1990. "Results of Thermal Stratification Tests in a Horizontal Pipe Line at the HDR Facility," Nuclear Engineering and Design,118, pp. 29-4 I.
Taylor, J. M.,1995. " Erosion / Corrosion (Flow Accelerated Corrosion). Induced Pipe Thinning in U.S.
Nuclear Plants - Update on NRC and Industry Activities," USNRC Memorandum for The Commissioners, February 7,1995, SECY-95-026, U.S. Nuclear Regulatory Commission.-
Terrel1, J. B.,1988. Fatigue Life Characterization ofSmooth andNotchedPiping SteelSpecimens in 2 ?8*C Air Environments, NUREG/CR-5013, MEA-2232.
Thailer, H. J., K. J. Dalal, and L. F. Goyette 1995. " Flow accelerated Corrosion in Steam Generators,"
presented at the 1995 ASME PVP Conference, Honolulu. Hawall, June 1995.
NUREG/CR-6456 156
REFERENCES Tremaine, P. R., et al.1977. " Calculation of Gibbs Free Energies for Ferrous Ions and the Solubility of Magnetite in H O and D 0 to 300*C," Thermochim Acta, Netherlands, p. 287.
2 2
Turner, J. S.,1973. Buoyancy Efects in Fluids, Cambridge University Press, Cambridge.
Thurman, A. L., M. S. Mahlab, and R. E. Boylstein 1981. "3 D Finite Element Analysis for the Investigation of Feedwater Line Cracking in PWR Steam Generators," 81 PVP-3, American Society ofMechanical Engineers.
TVA 1973. " Mechanical Transition - Feedwater to S. G. Nozzle," Drawing 47B17-4, Revision 1, Sequoyah Nuclear Plant.
TVA 1992a. " Cracking of Feedwater Piping," handout at TVA/NRC Management Meeting, April 3.
TVA 1992b. Feedwater Nozzle Crack Steam Generator Loop 3, Sequoyah Nuclear Plant, Unit 1, Report Number 92-7334, June 22.
UITer, R. A., et al. I982. Evaluation of Water Hammer Events in Light Water Reactor Plants, NUREG/CR-2781, QUAD-1-82-018, EGG-2203, pp. 5-1 to 5-3.
Uffer, R.,1987. " Nuclear Plant Water Hammer - Current Status," Paper No. 87-PVP-17, presented at the Pressure Vessel and Piping Conference, San Diego, California, June 28 - July 2,1987, American Society of Mechanical Engineers, New York.
USNRC,1979a. Cracking in Feedwater System Piping, IE Bulletin No. 79-13, Revisions 0,1,2, U.S. Nuclear Regulatory Commission.
USNRC 1979b. Letter from V. Stello (NRC) to all PWR licensees, Information requested on PWR Feedwater Piping pursuant to 10 CFR K50.54(f), May 25.
USNRC 1979c. Water Hammer in Nuclear Power Plants, NUREG-0582, U.S. Nuclear Regulatory Commission, Washington, D. C., July.
l USNRC 1980. USNRC Investigation and Evaluation ofCracking Incidents in Piping in Pressurized Water l
Reactors, NUREG-0691.
l USNRC 1982. " Stress Corrosion Cracking in Thick Wall, Large-Diameter, Stainless Steel Recirculation System Piping at BWR Plants," USNRC Bulletin 82-03, Revision 1.
l USNRC 1983a. " Stress Corrosion Cracking in Large-Diameter Stainless Steel Recirculation System Piping I
at BWR Plants," USNRC Bulletin 83-02.
USNRC 1983b. Inspection Report 50-348/83-08, St. Lucie Unit 1.
l USNRC 1984a. " Auxiliary Feedwater Sparger and Pipe Hanger Damage," USNRC Information Notice 84-32.
1 USNRC 1984b. Inspection Reports 84-07,84-10, and 85-22, Joseph M. Farley Unit 1.
l USNRC 1985. "Recent Water Hammer Events," USNRC Information Notice 85-76.
157 NUREG/CR-6456
REFERENCES USNRC 1986a. " Failure of Main Feedwater Check Valves Causes Loss of Feedwater System Integrity and Water-liammer Damage," USNRC Information Notice 86-01.
USNRC 1986b. Loss ofPower and Water Hammer Event at San Onofre, Unit 1, on November 21,1985, NUREG-1190, U.S. Nuclear Regulatory Commission.
USNRC 1987a. "Significant Unexpected Erosion of Feedwater Lines," USNRC Information Notice 87-36.
USNRC 1987b. Inspection Report 88-01, Beaver Valley Unit 1.
USNRC 1987c. "Feedwater Line Break," USNRC Information Notice 86-106, Supplement 1.
USNRC 1987d. " Thinning of Pipe Walls in Nuclear Power Plants," USNRC Bulletin 87-01.
USNRC 1988a. " Summary of Responses to NRC.Bulletin 87-01," USNRC Information Notice 88-17.
USNRC 1988b. "Feedwater Line Break," USNRC Information Notice 86-106, Supplement 2.
USNRC 1988c. NRCIntegratedProgramfor the Resolution of UnresolvedSafety issues A-3, A-4, and A-5 Regarding Steam Generator Tube Integrity, NUREG-0844.
USNRC 1988d. " Thermal Stresses in Piping Conne..:ed to Reactor Coolant System," USNRC Bulletin 88-08, Supplement 2.
USNRC l989. Erosion / Corrosion-Induced Pipe Wall Thinning, USNRC Generic Letter 89-08, May 2.
USNRC 1991 a. "High-Energy Piping Failures Caused by Wall Thinning," USNRC Information Notice 91 -l 8.
USNRC 1991 b. " Steam Generator Feedwater Distribution Piping Damage," US CRC Information Notice No. 91-19.
USNRC 1991c. "1hermal Stratification in Feedwater System Piping," USNRC Information Notice No. 91-38.
USNRC 1991d. Erosion ofSteam Generator FeedNo::le ThermalSleeve Update, Morning Report,-Region IV, March 15,1991.
USNRC 1992. " Rapid Flow-Induced Erosion / Corrosion of Feedwater Piping," USNRC Information Notice No. 92-07.
USNRC 1993. " Thermal Fatigue Cracking of Feedwater Piping to Steam Generators," USNRC Information Notice 93-20.
Van Duyne, D. A., et al.1991. " Global Stratification Effects in Feedwater Line to Steam Generator in PWR Power Plant," PVP-Vol. 218, pp. 9-12, American Society of Mechanical Engineers.
Virginia Power 1987. Surry Unit 2 Reactor Trip and Feedwater Pipe Failure Report, Rev. 0.
NUREG/CR-6456 158
i i
I REFERENCES i
i Walker, S. M., and E. Martinez,1994. " Evaluation of New Technology for Detection of &osion-Corrosion,"
12th International Conference on NDE in the Nuclear and Pressure VesselIndustries, Philadelphia, i
Pennsylvania, October 11-13, pp.147-150 Westinghouse (Westinghouse Electric Corporation) 1989. Indian Point Unit 2, Steam Generator Girth Weld, Feedwater Nonles R,rport, Spring,1989 Outage, Consolidated Edison of New York, WCAP 12294, Revisior 4.
Westinghouse (Wend gl,vuse Electric Corporation) 1990. Indian Point Unit 2, Steam Generator 1nspection, Repair, andRestOration Program, WCAP 12574, SG-05-022.
Wilson, J L. (Tennessee Valley Authority),1992. Lener to the U.S. Nuclear Regulatory Commission, I
"Sequoyah Nuclear Plant (SNP)- Reque.;t for Relief from the American Society of Mechanical l
Engineers (ASME),Section XI, Hydrostatic Pressure Test Requirements," March 23.
Wolf, L., et al.1987. "Results of Thermal Mixing Tests at the HDR Facility and Comparisons with Best Estimate and Simple Codes," Nuclear Engineering andDesign, 99, pp. 287-304.
Woodward, W. S.,1983. " Fatigue of LMFBR Piping Due to Flow Stratification," Paper 83-PVP-59, American Society of Mechanical Engineers, New York.
j
-l Wu, P. C.,1989. Erosion / Corrosion-Induced Pipe Wall Thinning in U.S. Nuclear Power Plants, NUREG-1344, April.
l r
l l
W s
159 NUREG/CR-6456 i
..,-._.-.,__r
.,,-.---m.
=
r
4 i
i I
I
)
APPENDIX
APPENDIX QUESTIONNAIRE SUBMITTED TO FOREIGN COUNTRIES
- 1. What are your current ISI practices / requirements regarding the examination of the feedwater (FW) system?
What documents imposed these requirements.
- 2. Has your utility experienced cracking and/or erosion / corrosion in the feedwater system? If so, provide a brief description of the problem.
- 3. Provide a brief description of current or past supplemental examinations that have been implemented for the FW system. Considering that some cracking has taken place outside the Code requimd examination volume, has any effort been made to extend examination volume; has any effort been made to extend examination coverage into the adjacent base metal.
4.
Provide a description of the examination techniques that are being used to detect fatigue cracks in the FW system. Have the inspection techniques been qualified for the detection of fatigue? If so, provide a brief description of the qualification process (e.g., flaw types, blind tests, etc.).
1
- 5. It has been shown that sizing the through-wall extent of cracks with the amplitude drop technique is unreliable. Describe the standard or supplemental ultrasonic techniques used to size cracks.
- 6. Describe the methodology used to discriminate geometry from flaw indications. What percentage of the ultrasonic examinations performed on the FW system are performed with automated systems? What system (s) are used?
- 7. Provide a brief description of the examination techniques used to detect erosion / corrosion. What area of the FW system are examined and how are they selected? Are any enhanced inspection techniques being considered?
l h
A-1 NUREG/CR-6456
NRCPORM 3as UA NucLsAR RsGULAToRY "'""*"'"
- 1. REPORT NUhWER No) posi8ned by MRc, Add Vel, Supp., Rev.,
E" BIBLIOGRAPHIC DATA SHEET
""'****"*""""'"'I NUREG/CR-6450
- 2. TITLE AND SUaTITLE INEL-96/0089 Review of industry Efforts to Manage Pressurtzed Water Reactor Feedwater Nozzle, 3.
KE REPORWauSHED Piping, and Feedring Cracking and Wall Thinning l
MONTH YEAR March 1997
- 6. AUTHOR (S)
- 6. TYPE OF REPORT V. N. Shah, A. G. Ware, A. M. Porter Technical
- 7. PERICO COVERED (hedueme Dease)
- s. PERFORMNG ORGANIZATION. NAME AND ADDRESS trMtc. povese Dvam, once erMegart, u s. Mwedeer Aapudskry t-
, and meeng entuee, s contecer, j
para nome end meane ***en) 1 idaho National Engineering Laboratory Lockheed Martin Idaho Technologies Company P.O. Box 1825 j
id:ho Fcils, ID 83415
- e. SPONSORING ORGANEzATION NAME AND ADORESS (r Mtc. Wsame as e6 ave; #oanescer, pave Mtc Dum Osa or Regen, u s Nicasar Asyudskry coms mese end meene ***ees) l S:fety Programs Division Office for Analysis and Evaluation of Operational Data U.S. Nuclear Regulatory Commission W7shington, DC 20555-0001
- 10. suPPLEME.NTARY NOTES E. J. Brown, NRC Technical Monitor: E. A. Trager, NRC Project Manager
- 11. A87 TRACT rJoo wonde ar hu)
Review of industry efforts to manage thermal fatigue, flow-accelerated corrosion, and steam generator water hammer damage to Pressurtzed Water Reactor (PWR) feedwater nozzles, piping, and feedrings is presented in this report. The review includes an evaluation of design modifications, operating procedure changes, augmented inspection and monitoring programs, and mitigation, repair cnd replacement activities. Four specific actions were taken to perform the evaluation (a) review of field experience to identify trends of operating events; (b) review of the related technical literature; (c) visits to three PWR plants and a PWR vendor, and (d) solicitation ofinformation from foreign utilities. Our assessment of field experience indicates the USNRC licensees have apparently taken sufficient adion to minimize the feedwater nozzle cracking caused by thermal fatigue, wall thinning of J-tubes and feedwater piping, and steam generator water hamrner in both top-feed and preheat steam generators. A major finding of this review is that the analysis, inspection, monitoring, mitigation, and replacement techniques have been developed for managing thermal fatigue and flow-accelerated corrosion damage to feedwater norries, piping, and feedrings. Adequate training and appropriate applications of these techniques would ensure effective management of this damage. Several PWR plant operators h:ve been proactive in managing this damage.
- 12. KEY WORDS/DESCRIFTORS (Let wave er p* sees past we samet mesonore m beetng re report) 13 AvAs.ABluTY STATEMENT uMmW Pressurtzed water reactor feedwater nozzle, piping, feedring, J-tubes, flow stratification, thermal fatigue, H SECURWCLASSEADON flow 4ccelerated corrosion, erosiorworrosion, inspedion and monitoring, field experience, water hammer rrhe Peee) unclassified Inus nepas unclassified
- 16. NUMBER OF PAGEs 1s. PRICE NRCFORM M M TNs form wee seecagrecany produced by Dhe Federal Fenns. Inc.
1 I
l Printed on recycled paper Federal Recycling Program
r--
...i,u,q
$l 3
5 E
5ll9$l, 8
e igl-
- i-g g e
h l
t i
4 1.
BR{
ll
.________-__________l