Artículos Sobre SCC en Turbinas de Vapor

EPf21
EPRI TR-108982
Final Report
September 1997
Proceedings: Steam Turbine
Stress Corrosion Workshop
Prepared by
EPRI Repair and Replacement Applications Center
Charlotte, North Carolina
Proceedings: Steam Turbine Stress
Corrosion Workshop
TR-108982
Final Report, September 1997
Prepared by
EPRI Repair and Replacement Applications Center
1300 Harris Boulevard
Charlotte, North Carolina
and
Electric Power Research Institute
3412 Hillview Avenue
Palo Alto, California
Principal Investigators
P. Sabourin
R. Viswanathan
Prepared for
Palo Alto, California 94304
EPRI Project Manager
R. Viswanathan
Strategic Research and Development
DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES
THIS REPORT WAS PREPARED BYTHE ORGANlZAllON(S) NAMED BELOWAS AN ACCOUNT OF WORK SPONSORED
OR COSPONSORED BYTHE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF
EPRI, ANY COSPONSOR, THEORGANlZAllON(S) BELOW, NORANY PERSON ACTING ON BEHALFOFANYOFTHEM:
(A) MAKES ANY WARRANTY OR REPRESENTAllON WHATSOEVER, EXPRESS OR IMPLIED, (I) WITH RESPECT TO
THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS REPORT,
INCLUDING MERCHANTABIUTY AND FITNESS FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT
INFRINGE ON OR INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARlVS INTELLECTUAL
PROPERlY, OR (Ill) THATTHIS REPORT IS SUITABLETOANY PARllCULAR USER'SCIRCUMSTANCE; OR
(B) ASSUMES RESPONSIBIUlY FOR ANY DAMAGES OR OTHER L1ABIUTY WHATSOEVER (INCLUDING ANY
CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTAllVE HAS BEEN ADVISED OF THE
POSSIBIUlY OF SUCH DAMAGES) RESULllNG FROM YOUR SELECTION OR USE OF THIS REPORT OR ANY
INFORMAllON, APPARATUS, METl-IOD, PROCESS, OR SIMILAR ITEM DISCLOSED INTHIS REPORT.
ORGANlZAllON(S) THAT PREPAREDTHIS REPORT
EPRI REPAIRAND REPLACEMENTAPPUCATJONS CENTER
ELECTRICPOWERRESEARCH INSTTTlITE
ORDERING INFORMATION
Requests for copies of this report should be directed to the EPRI Distribution Center, 207 Coggins Drive,
P.O. Box 23205, Pleasant Hill, CA94523, (510) 934-4212.
Elecbic Power Research Institute and EPRI are registered service marks of the 8ecbic Povver Research Institute, Inc.
EPRI. POWERING PROGRESS is a service mark of the Becbic Power Researdllnstitute, Inc.
Copyright1997 8ecbic Power Research Institute, Inc. All rights reserved.
REPORT SUMMARY
A recent survey of utilities commissioned by EPRI indicated that cracking of steam
turbine disk rims by stress corrosion was a pervasive problem in both fossil and nuclear
power plants. There is a clear need to document industry experience in this area so that
guidelines can be provided to utilities on managing the problem.
Background
The catastrophic failure of an LP rotor disk at Hinkley Point A Power Station in 1969
focused industry attention on the problem-the cracking at bores and keyways of disks
due to stress corrosion. Based on initial experience, reheat machines of up to 500 MW,
such as those in fossil stations, were considered to be safe. This myth was shattered by
the burst of an intermediate pressure (IP) turbine disk in the UK in 1987. After ten more
years and millions of more dollars of research, the SCC problems continue to plague
both fossil and nuclear turbines, with severe cracking found at the rim locations.
Objectives
To focus industry attention on the rim cracking problem.
To consolidate the experience and the collective wisdom of industry, so that
guidelines can be provided to utilities on how to manage the problem.
Approach
EPRI organized a workshop jointly with Baltimore Gas & Electric Company. All major
OEMs, a cross section of utilities, and consultants were invited to participate. The first
dayof the workshop was devoted to an OEM/consultant perspective of the issues. On
the second day, utility perspectives were presented, after which the conference was
concluded with a panel discussion that included several experts fielding questions. The
gist of these discussions and the papers presented were documented.
Results
The workshop brought out the following key issues/ results:
iii
The rim cracking has affected nearly one-third of the nuclear and fossil fleet.
However, there are no reported instances of catastrophic failures, indicating that
crack tolerance-based approaches can be used to avoid premature replacement of
rotors.
Remedial actions for cracked rotors include weld repairs as well as mechanical
repairs, such as flaw excavation, shot-peening, blade removal, use of steel/titanium
notch blocks, pressure plates, and long shank buckets.
Though crack growth-based life prediction methods exist, uncertainties arise from
NDE procedures for crack sizing, crack growth rates, and other unit-specific
variables. Improved NDE procedures and crack growth data based on re-inspection
of units periodically will improve life prediction capability.
Controversy exists regarding the role of oxygen, steel purity, shot-peening, and the
usefulness of probabilistic analysis.
EPRI Perspective
From an EPRI perspective, the workshop was very successful in promoting continued
research to help solve and manage the problem. It was encouraging to note that the
OEMs were actively involved in improving their NDE capabilities and developing
repair technologies and improved coatings to mitigate the problem. This is the
catalyzing effect of workshops such as these. The need for more field testing, and the
need for resolving by further research any controversies about the role of oxygen, steel
cleanliness, and shot peening, are quire apparent.
TR-108982
Interest Categories
Applied science and technology
Fossil steam plant O&M cost reduction
Turbines and generators
Keywords
Steam turbines
Rotor disks
Stress corrosion
Nuclear power plants
Fossil power plants
iv
FOREWARD
In1969, Unit 5 at Hinkley Point fA" Power Station in the UK suffered catastrophic
failure of a disked LP rotor during an overspeed trip test following desynchronizing.
Both of the first disks in each flow had burst from stress corrosion cracks (SeC) present
in the keyways. In the wake of this failure, extensive inspection programs were
undertaken of nuclear turbines by turbine manufacturers (OEMs) worldwide. see
cracking was found in many instances, and the disks at risk were identified to be those
operating in a wet steam environment, as characteristic of non-reheat machines with
high disk-bore stresses. Reheat machines of up to 500 MW output, such as those in
fossil stations, were considered to be safe.
Examinations of reheat turbine disks from fossil power stations built after 1987 started
to reveal that these disks are not immune to sec. This concern was highlighted in
February 1989 when a disk burst at speed on the IP turbine of a 200 MW machine in
UK. Subsequent to this incident, several U.S. utilities were alerted by the turbine
manufacturers regarding a potential problem of see in fossil reheat turbine disks.
Several affected utilities contacted EPRI for guidance. A workshop was organized by
EPRI in 1991 to gather and consolidate the pertinent industry experience(l). More
recently Lyle (2) reviewed this early research and combined with it the results of
international projects on the characterization of the susceptibility of rotor steels of
various composition and strength levels to environmental factors, including his own
results with simulated crevices. This recent review, combined with other observations,
underscores some significant gaps in our understanding of crack initiation and the
ability to predict the time and location of the onset of cracking.
Recent years have seen the introduction of refurbishment and repair methods to reduce
or eliminate the risk of stress corrosion cracking as well as improvements in steel
making and processing for more resistant forgings. At the same time, the industry's
efforts to reduce the risk of bore and keyway cracking through integrally forged and
welded rotor construction have shifted concern somewhat to the susceptibility of rim
attachments. In particular it is important to determine whether this cracking mode
should be expected to become more prevalent with aging of the fleet and what steps
utilities could take to mitigate the problem. These issues led EPRI to commission a
survey of US utility experience and of the efficacy of repair methods(3). This survey
showed the rim attachment cracking to have occurred in 41 out of the 109 nuclear plants
currently operating. In the case of fossil units, 29 of the 110 supercritical units (26%)
v
Foreward
and 20 of the 647 subcritical units (3%) included in the survey were reported to contain
rim cracks. In view of the pervasive nature of the problem, the current EPRI workshop,
co-hosted by Baltimore Gas & Electric Company was held in Baltimore on March 19-20,
1997. A total of 17 technical papers were presented. The first day of the conference was
dedicated to the manufacturer and consultant perspective of the issues. On the second
day, the utility perspective was presented, after which the conference was concluded
with a panel discussion with several industry experts fielding questions in a sometimes
spirited debate. Of the 10 people attending the conference, 50 people represented 23
domestic utilities. The remainder of the attendees represented 27 organizations
including 12 foreign companies.
Some of the key issues that were hotly debated are summarized below:
Dry disk cracking: The question of whether cracking can occur in stages that are
normally dry during operation was discussed. It was concluded that only wet disks crack.
So called"dry" disks are subject to wetness during shutdown transients, turning gear
operation with glancial steam, due to malfunctioning Moisture Separator Reheater
(MSR) units, or due to severe ingress of chemicals.
Effect of oxygen: Previous surveys had shown Boiler Water Reactor (BWR) turbines to
be more susceptible to cracking than Pressurized Water Reactor (PWR) turbines
suggesting a strong deleterious role for oxygen with respect to keyway cracking; with
respect to rim cracking however, there seems to be no strong with reactor
type. This ledto a discussion of the role of oxygen. Professor Speidel's results showed
that oxygen had no effect on crack growth. He was also of the opinion that oxygen had
no effect on crack initiation either. This view was strongly supported by Dr.
Holdsworth of GEC-Alsthom. The opposing view that oxygen had an adverse effect on
both crack initiation and propagation was expressed by Dr. Fred Lyle of SWRI and
J. McCracken of Westinghouse. The reasons for these differences in views are not clear.
Regarding oxygenated treatment (OT), Dr. Dooley has categorically stated in the
discussion (submitted after the conference) that there is no adverse effect of this.
treatment due to the fact that the condensate in the phase transition zone has been
shown to contain no oxygen, although the inlet steam may contain 100-200 ppb oxygen.
Since the concentrating effect of chlorides and sulfides in the condensate is actually
reduced and the liquid films forming on the blades are actually thinner due to aT, his
results suggest that OT may even be beneficial. While the merits of OT need to be
discussed separately, the controversy regarding the effect of oxygen on see needs to be
resolved.
Effect of steel cleanliness: Cleanliness clearly increases the short term and long term
toughness of the steel and is beneficial. In terms of stress corrosion (SCC) crack
propagation, cleanliness seems to make no difference. In terms of crack initiation,
cleanliness may be beneficial, but only under those conditions ofenvironment and strength levels
(e.g., yield < 850 MPa) where pitting is the crack initiating mechanism.
vi
Foreward
Effect of surface condition: The needfor more studies on the effects of surface condition,
residual stresses from fabrication and shot-peening on see initiation was very apparent. While
some OEMs believed in the beneficial effects of shot peening, other OEMs,
Westinghouse in particular, expressed the view that once the surface compressive stress
region had been penetrated by pitting, further crack growth may actually be accelerated
in a shot peened component due to residual tensile stresses below the surface. W.
Welsch of the Metal Improvement Company said that according to his data, once pits
penetrated the compressive region, there was no residual tensile stress zone left. In the
absence ofpublished data to substantiate any of these claims, further research is calledfor.
(Speidel, Ortolano) A strong interest in development of barrier type and subcritical
coatings was expressed by several OEMs, although none has been perfected at this
point.
Refurbishment: Weld repair using 12%Cr filler seems to have become commonplace
with respect to fossil LP turbines. Westinghouse indicated that they have performed
weld repairs on BWRrotors and expect to perform more such repairs. There is however a
strong needfor in-situ repair techniques.
Life assessment: Many issues pertaining to life assessment of rim-cracked disks were
extensively discussed. Among the salient points were:
The need for better detection and sizing techniques for flaws (all)
The discrepancy between laboratory findings and field experience. For instance,
laboratory data show crack initiation in hundreds of hours, whereas field experience
suggests initiation only after 50 to 100,000 h of service (D. Rosario). It was felt that
field performance is too complex to be simulated in the lab and hence lab data were
only useful for screening the effect of variables and not for life prediction (C.
DeCesare). The needfor testing samples in the field, in the actual turbine environment was
emphasized by several participants.
The USA crack growth rate data, based on field experience, show considerably more
scatter than some of the European data, based on lab studies (D. Rosario). The need
for repeat measurements and crack sizing on actual disks to calibrate crack growth
models was emphasized by several OEMs.
Differing viewpoints were expressed by participants regarding the usefulness of
probabilistic life assessment procedures. One view held that the many uncertainties
involved rendered such assessments useless and that good deterministic analysis
tempered with experience-based judgment was the better approach. OEMs
particularly expressed this view. Engineering consultants on the otherhand were of
the opinion that probabilistic analysis using refined data were quite useful (T. Lam).
The needfor improved life assessment techniques was recognized beyond any doubt.
vii
Foreward
These proceedings contain the papers presented as well as a transcript of the debate that
occurred during the panel discussion. It also includes the final agenda and an
attendance list of the participants.
Lastly, we wish to thank BG&E, particularly Dr. Mark Richter, for co-hosting this
workshop. Mr. Paul Sabourin and Sherryl Stogner assisted with the organization. The
contributions of all the participants and authors is gratefully acknowledged.
R. Viswanathan, Workshop Chairman
Strategic Research & Development
EPRI
viii
Foreward
References
1. Proceedings: Fossil Steam Turbines Disk Cracking Workshop, G5-7250, April
1991, Electric Power Research Institute, Palo Alto.
2. F. F. Lyle, UStress Corrosion Cracking in Low Pressure Steam Turbines - An
Overview", Paper No. 219, Corrosion 194, The Annual Conference of NACE
International.
3. D. Rosario, C.H. Wells and G.]. Licina, ilLow-Pressure Rotor Rim Attachment
Cracking Survey of Utility Experience", TR-I07088, March 1997, Electric Power
Research Institute, Palo Alto, CA.
IX
x
FINAL AGENDA
EPRI SteamTurbine
Stress Corrosion Cracking
Conference
March 19 - 20, 1997
Hosted By
Baltimore Gas & Electric Company
Wednesday, March 19, 1997
Foreward
7:30 Registration (BGE) Headquarters
8:15 Welcome Paul Sabourin EPRI
Dr. R. Viswanathan EPRI
Dr. Mark Richter BGE
Welcome Address David Snyder BGE
Manufacturers/Consultants Perspective
Session Chairman Dr. Mark Richter BGE
8:30 Designing LP Turbine Disks Dr. H. R. JhansaJe Siemens Power
Against Stress Corrosion Cracking Corporation
9:00 GE Dovetail Stress Corrosion Carl DeCesare, Dennis General Electric
Cracking Experience and Repair Lessard, and Jack Nolan Company
10:00 Break
10:30 Qualification of Welding Rotors with Dennis Amos Westinghouse Electric
12Cr Stainless Steel to Improve SCC Corporation
Resistance
11:00 Laboratory Stress Corrosion Cracking Dr. S. R. Holdsworth GECAlsthom
Experience in Stearn Turbine Disc Steels
11:30 Management of Rotor Disk Rim Tony Lam Stress Technology,
Stress Corrosion Cracking Inc.
12:00 Lunch
Manufactures/Consultants Perspective (cont.)
Session Chairman Harold Presson EPRI
1:00 Remaining Life and Risk Assessment Avinash Sarlashkar Stress Technology,
pf Low Pressure (LP) Rotor Dovetails Inc.
Due to Stress Corrosion Cracking (SCC)
1:30 Stress Corrosion Crack Initiation in f'red F. Lyle, Jr. Southwest Research
Steam Turbine Steels Institute
2:00 Stress Corrosion Cracking of Darryl Rosario Strucmral Integrity
Steam Thrbine Rotors Associates, Inc.
xi
Foreward
2:30 Break
3:00 Probabilistic Assessment of Failure by Darryl Rosario Structural Integrity
Stress Corrosion Cracking in h r u n k ~ n Associates. Inc.
Disks on Low Pressure Thrbines
3:30 Major Influences on the Growth Rates of Dr. Markus O. Speidel Swiss Federal
Stress Corrosion Cracks in Steam Turbine Institute of
Rotor and Blade Materials Technology
4:00 Unit-Specific Assessment of Low-Pressure Dr. Sanjeev Srivastav Failure Analysis
Steam Turbine Rotor Stress Corrosion Associates, Inc.
Cracking
4:30 Adjourn SCC Conference
Convene utility working group
Thursday, March 20, 1996
8:30
9:00
9:30
Utility Perspective
Session Chairman
Millstone Unit 1, Disk Face Cracking
Turbine Stress Corrosion Cracking
- Duke Power's Perspective
Arresting Stress Corrosion Cracks.
in Steam Turbine Rotors
Paul Klein
Jay Ely
TomAlley
Russell Chetwynd
BGE
Northeast Nuclear
Energy Company
Duke Power
Company
Southern California
Edison
10:00 Break
10:30 LPTurbine Dovetail sec Management
atBGE
Dr. Mark Richter BGE
11:00
11:30
Low Pressure Thrbine Stress Corrosion
Cracking Investigation at the Navajo
Generating Station
Design, QUalification, and Testing
of RotorlDisc Weld Repairs
12:00 Lunch
Eugene Nowak
David Gandy
Salt River Project
ERPIRRAC
xii
1:00 Panel Discussion
Chairman
Panelists
3:30 Adjourn
Dr. R. Viswanathan
TomAlley
David Gandy
Darryl Rosario
Carl DeCesare
Dennis Amos
Dr. Markus O. Speidel
Tom McCloskey
EPRI
Duke Power Company
EPRIRRAC
Structural Integrity Associates, Inc.
General Electric Company
Westinghouse Electric Corporation
Swiss Federal Institute of Technology
EPRI
CONTENTS
Designing LP Turbine Disks Against Stress Corrosion Cracking by
H. R. Jhansale 1
GE"Dovetaii Stress Corrosion Cracking Experience and Repair by
Carl DeCesare, Dennis Lessard, and Jack Noland 17
Qualification of Welding Rotors with 12Cr Stainless Steel to Improve SCC
Resistance by Dennis Amos 29
Laboratory Stress Corrosion Cracking Experience in Steam Turbine Disc Steels
by S. R. Holdsworth ; 43
Management of Rotor Disk Rim Stress Corrosion Cracking by Tony Lam 61
Remaining Life and Risk Assessment of Low Pressure (LP) Rotor Dovetails Due
to Stress Corrosion Cracking (SCC) by Avinash Sarlashkar '" 89
Stress Corrosion Crack Initiation in Steam Turbine Steels by Fred F. Lyle, Jr 105
Stress Corrosion Cracking of Steam Turbine Rotors by barryl Rosario 119
Probabilistic Assessment of Failure by Stress Corrosion Cracking in Shrunk-On
Disks on Low Pressure Turbines by Darryl Rosario 139
Major Influences on the Growth Rates of Stress Corrosion Cracks in Steam
Turbine Rotor and Blade Materials by Markus O. Speidel 155
Unit-Specific Assessment of Low-Pressure Steam Turbine Rotor Stress
Corrosion Cracking by Sanjeev Srivastav 175
Millstone Unit 1, Disk Face Cracking by Jay Ely 189
xiii
Contents
Turbine Stress Corrosion Cracking - Duke Power's Perspective by Tom Alley 213
Arresting Stress Corrosion Cracks in Steam Turbine Rotors by Russell Chetwynd .... 229
LP Turbine Dovetail SCC Management at BGE by Mark Richter 239
Low Pressure Turbine Stress Corrosion Cracking Investigation at the Navajo
Generating Station by Eugene Nowak 257
Design, Qualification and Testing of Rotor/Disc Weld Repairs by David Gandy 279
Panel Discussion 301
Attendance List 323
xiv
Designing LP Turbine Disks Against Stress Corrosion Cracking
Presented By
Dr. H. R. Jhansale
Siemens Power Corporation
1
2
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
DESIGNING LP TURBINE DISKS
AGAINST STRESS CORROSION CRACKING
H. R. Jhansale
w. E. Meade
D. R. McCann
1040 South 70
th
Street
Milwaukee, Wisconsin, U.S.A
w. David
M. Brandle
G. Roettger
Siemens/KWU,
Muelheim, Germany
3
4
I
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
. j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
j
DESIGNING LP TURBINE DISKS
AGAINST STRESS CORROSION CRACKING
H. R. Jhansale
W. E. Meade
D. R. McCann
1040 South 70
th
Street
Milwaukee, Wisconsin, U.S.A
W. David
M. Brandle
G. Roettger
Siemens/KWU,
Muelheim, Germany
Abstract
Stress Corrosion Cracking (SCC) has been experienced world-wide in shrunk-on disks of
LP steam turbines both in Nuclear and Fossil Power generation units. Over the last 30
years, Siemens has continually advanced its LP turbine design to overcome stress corrosion
cracking. To date about 406 disks from 41 rotors of the older design have been inspected or
reinspected and only three disks with indications (one with SCC) have been reported. With
additional measures to avoid stress corrosion cracking, which Siemens has used since 1987
for 29 rotors with excellent experience, the newest six disk (three per flow) per rotor design
is developed to be even more successful. This paper discusses the various design
measures incorporated in Siemens turbines to achieve this successful experience.
Introduction
World wide cracking experience with shrunk-on disks in LP turbine rotors has shown, that
the phenomenon of stress corrosion cracking can significantly reduce the life expectancy of
the turbine. Siemens has conducted extensive research studies of the stress corrosion
cracking phenomenon, and continually advanced its LP turbine design to overcome stress
corrosion cracking.
Figure 1 shows a schematic of the three design configurations of Siemens 1500/1800 RPM
LP disk type turbine rotors. The ten disk design (five perflow) was first introduced in the late
1960's. The first two disks carried the drum type blading and the last three disks carried
one free standing blade row each. In 1986, with the advent of improved forging processes,
the eight disk design (four per flow) was introduced. The first two disks were combined into
one disk to carry all rows of drum blading. In the eight disk design the overall stresses were
reduced by inducing surface and near surface compressive residual stresses through heat
treatment, shot peening and honing. In 1995 the present six disc design (three per flow) was
introduced, which combines the No.2 and NO.3 disks of the eight disk design into one (new
5
H. R. Jhansale
No.2) to carry two rows of free standing blading. This design evolution has not only
improved the life expectancy of the disks but also has improved the thermal performance
and power plant output.
Siemens manufactures both disk type and mono-block rotors. For large nuclear units,
Siemens prefers disk-type rotors over the moncrblock rotors because of the following
advantages:
High quality shaft and disk forgings are, more readily available and easier to produce
because of their smaller size.
Residual compressive stresses are induced during the heat treatment process into
the entire disk including the blade attachment area.
The design of the disk type rotor allows for lower manufacturing cost.
The replacement of disk-type rotors can be carried out without any dynamic or
alignment problems.
The Siemens design avoids closed gaps on the inlet side that could cause crevice
corrosion.
The design of the disk-type rotor allows 100% ultrasonic inspection access of the
disks.
Factors Governing Stress Corrosion Cracking (SeC) Life
Siemens has conducted extensive studies into the sce behavior of materials used for rotor
disks 1-4. Based on these studies and available data from technical literature 5-7 in general, it
is of interest to note the following information about the sce behavior .
seePhenomenon
Stress Corrosion Cracking occurs in the presence of a steady tensile stress and a aggres-
sive or corrosive environment. The phenomenon consists of an initial crack initiation period
in which pitting or cracks are formed which is followed by a crack growth period. sec initia-
tion is not understood well enough to quantify its time dependent behavior. However, SCC
crack growth behavior can be modeled using linear elastic fracture mechanics. Figure 2
shows a schematic of the sec growth rate (daldt) as a function of the applied stress
intensity factor (K), which exhibits three distinct stages. Stage I shows that daldt rapidly
decreases with decreasing K, approaching no crack growth below a threshold value of K,
defined as K
1scc
. During Stage II, daldt'is Virtually independent of the K level, until K
approaches the material fracture toughness level. Then the daldt approaches unstably high
values leading to fracture. Ktscc is typically of the order of about 20-30 MPa"m. This means
that a finite threshold crack size has to be either initiated or be present in order for crack
growth to continue. The lower the stress, the larger the threshold crack size.
6
H. R. fhansale
Factors Influencing sec Initiation
Impurities in steam, conditions promoting flow stagnation such as crevices, steam conden-
sation, ratio of stress to yield strength and level of yield strength significantly influence the
potential for SCC.
In high purity water with a conductivity of < O.2flS/cm, stress corrosion crack initiation is
influenced only by the quenching and tempering process which establish the material's yield
strength value. If the yield strength exceeds approximately 1085 MPa (157 Ksi), the material
becomes susceptible to hydrogen induced stress corrosion cracking as shown in Figure 3.
Up to this threshold no stress corrosion crack initiation occurred even when operating
stresses exceeded the yield strength in notched specimens. This result is also not affected
by the purity of steel. Materials conventionally smelted 25 years ago behaved in high purity
water as well as today's steels which are smelted in accordance with the electroslag
remelting process. Under high purity water conditions, even non-metallic inclusions (e.g.
A1
2
0
3
, MnS, etc.) do not act as crack starters at the material surface. Such inclusions do not
influence the resistance to stress corrosion cracking. This even applies to water with low
oxygen content as well as to oxygen satLirated water. Pit formation was also not found
under these water conditions.
Findings from extensive testing, power plant experience and review of literature leads to
Figure 4. For yield strengths less than 1000 MPa (145 Ksi), this figure shows at what
operating stress to yield strength ratios, stress corrosion crack initiation can be expected for
specific environment conditions. As shown in the figure, an improvement of the operating
environment permits high stress levels up to and above the yield strength level of the mate-
rial. The diagram also reveals that with stress levels below 50% of the yield strength stress
corrosion cracking has not occurred even under severe corrosion conditions;
Factors Influencing sec Growth Life
For a given initial crack size, SCC growth life is governed by two factors, namely, the SCC
growth rate and fracture toughness. see growth rate which is not only a function of the
material and the environment, but is also dependent on the yield strength, temperature, and
whether the steam is wet or dry. For a given material and environment, the sec rate
increases in general with yield strength and temperature. However, as shown in Figure 5,
Siemens laboratory studies showed that the sec growth rate is virtually independent or
slightly dependent on the yield strength for yield strengths and temperatures in the range of
650 to 1085 MPa and 80 to 150, respectively. For yield strengths exceeding about 1085
MPa, the sec growth rate is strongly dependent on the yield strength, due to hydrogen
induced sec phenomenon. Increase in sec growth rate means decrease in see growth
life. Higher the fracture toughness, longer is the SCC growth life. see growth rates have
been determined based on disk cracking experience in power plants.
7
H. R. Jhansale
Design Requirements to Combat sec
Based on the above knowledge base, the various desired design combinations which com-
bat sec and improve the life expectancy are as folJows:
1. Maintain good water chemistry.
2. Keep yield strengths below 1000 MPa (145 s i ~
3. Keep ratio of operating stress to yield strength below 0.5.
4. Keep operating temperatures as low as possible.
5. Eliminate stagnant conditions and steam condensation.
6. Maximize Fracture Toughness.
Design Features of Siemens LP Disks
Table 1 summarizes essential features incorporated in the evolution from the older ten disk
to the newest six disk design configuration. Finite element techniques have been used to
optimize stresses, temperatures, keyway geometry and shrink frt. Patented heat treatment
and manufacturing techniques have been developed to induce beneficial compressive
residual stresses to reduce the net surface/near surface stress levels below 50% of the yield
strength.
In the older ten disk design 2-3.5%-Ni-Cr-Mo-V steels similar to ASTM A471 were used.
Currently Siemens utilizes a 3.5% Ni;.1.5% Cr-Mo-V steel with much higher Charpy Impact
Fracture Appearance Transition Temperature (FAIT) requirements than for ASTM A471.
This new material, in addition to excellent hardenability has also excellent fracture tough-
ness. Specified yield strength range varies from about 780-890 MPa, which is well below the
1000 MPa value to minimize the material's stress corrosion crack initiation potential. The
specified low FAn and good hardenability, will ensure that the LP disks are operating in the
upper shelf region of the material toughness. The deep seated fracture toughness is 'greater
than 200 MPa-vm (180 Ksi'J'in) at room temperature.
Operating stress levels are reduced by eliminating keyways in Disks Nos. 2 and 3 and intro-
ducing significant compressive stresses through heat treatment of the disks, shot peening of
surfaces including blade attachment areas, and honing followed by rolling of keyways in No.
1 disks (see Figure 6). The operating stress levels at or near the surfaces are well below
50% of the yield strength thus virtually eliminating stress corrosion crack initiation.
Because of their high shrink fit, NO.2 and No. 3 disks each exceed more than ten times their
operational torque caused by friction (friction coefficient conservatively assumed to be 0.05).
Siemens disks are very robust, Le. wide, yielding large shrink fit forces. As shown in Figure
7, because of the lower temperatures on the downstream side, the in-service shrink fit
forces remain high allowing elimination of keyways. Because of the lower stresses,
8
H. R. Jhansale
elimination of keyways and low temperatures, the probability of stress corrosion cracking is
virtually reduced to zero in these discs.
No. 1 disk remains keyed as a result of the temperature profile during startup. Figure 6
illustrates the shrink fit and keyway detail configuration. Siemens keyways are located down
stream of the steam flow where the metal temperature is higher than the surrounding steam
temperature. This eliminates condensation and reduces the sec potential. The keyway
which is round, is open circumferentially by a gap and is located in a non-shrink fit area.
After honing, the key way is rolled to induce residual stresses before the round key is fit.
The first 20% of the inlet side of disk no. 1 hub is relieved by a taper while the shaft is
profiled to give a relatively large cavity eliminating the shrink fit. There is a gap between the
front side of the disc and the rotor shaft. This allows an efficient "pumping" action due to
centrifugal forces to take place which expels any moisture or impurities, thus eliminating any
stagnant condition for the steam. All these design features help minimize potential for SCC.
Siemens has used the same general blade foot attachment arrangement for both drum stage
and free standing blading for more than thirty years for both nuclear and fossil units. Some
nuclear units are approaching 200,000 operating hours without blade attachment failure.
Inspections to date have not found crack initiation in the blade attachment area of disks or
rotor forgings of either nuclear or fossil units. Because nuclear unit LP disks are operating
in conditions where stress corrosion cracking is possible, blade loading on disk surfaces is
maintained significantly less than 50% of material yield strength. For older designs, the
caulking areas or the r ~ s below the blading on the disks, were shot peened to induce
residual stresses to assure no crack initiation where propagation would be downward into
the disk. On the newest six disk design, the radii of the disk blade attachment area are also
shot peened to induce residual stresses which minimize operating stress levels and further
assure that stress corrosion cracks will not initiate. This is done for all disks which are
susceptible to SCC.
Nuclear power plants need to insure (per U.S. NRC requirements) that the probability of
external missile generated by a potential tUrbine failure be limited to l' E-4 or 1E-5 per year
respectively, depending on whether the turbine unit is favorably or unfavorably oriented
relative to the reactor. A probabilistic fracture mechanics procedure is used in conjunction
with Monte Carlo simulation technique to evaluate the disk burst and external missile prob-
abilities. A detailed discussion of this topic is beyond the scope of this paper, but is sched-
uled for presentation at the upcoming EPRI-NDE workshop in July 1997 8. Using this pro-
cedure, it has been estimated that the service interval between inspections for the eight disk
and the six disk LP rotors in a Nuclear power plant can easily exceed 100,000 hours as
compared to 50,000 hours for the older ten disk design. The probability of an external
missile for a six disk design unit is estimated to be about two orders of magnitUde lower than
the NRC requirements.
Performance Experience
To date a total of 82 Siemens/KWU, No. 1 disks and 324 latter disks from 41 ten and eight
disk LP rotors in operation have been inspected or reinspected world wide over the last 15
years. Two of the newest six disk design rotors have been in operation since September
9
H. R. fhansale
1996 ano eight more are to be installed during 1997-99. Therefore, no inspections have
been made on these six disk design rotors.
Only one No. 1 disk was found to have sec type ultrasonic indications in disk hub surface.
This finding was after 67,600 operating hours on a ten disk design unit. This design did not
have the benefit of design induced compressive residual stresses on the disk hub bore.
Subsequent inspections found crack growth rate to be 3-4 mm per year 4. An investigation
of the cause, showed that the disk hub surface was contaminated by microscopic Ni- and 8-
rich particles, which were inadvertently introduced and pressed into surface during the time
of manufacture. This probably acted as the crack starter. Manufacturing procedures were
redefined to preclude such occurrences in the future.
Small indications were also found on two of the 324 latter disks. The nature of these indi-
cations could not be ascertained, but are likely to be due to water erosion Qr sec. Details of
these findings have been reported earlier 3. These two findings on the inlet side of the TE
and GE disk 4 of the same rotor, also of a 10 disk design unit without design induced
residual stresses of the disk hub bore, were after 53,000 operating hours. Evaluation found
no limitation to designed operating life, the rotor was returned to service and additional
investigation to this time has not been possible due to the disks being in service.
Conservatively, assuming that these indications are due to sec and using standard statisti-
cal procedures, the crack initiation probabilities at 90% confidence level for the No. 1 and
latter disks in the ten disk design are estimated as 4.63E-2and 1.64E-2, respectively.
These values are 3-10 times better than industry wide experience. The crack initiation
probability for the newest six disk design should be virtually zero, since the effective surface
and near surface stress levels have been reduced well below 50% of the yield strength with
residual compressive stresses. Therefore, these disks should be much better than the
industry average in terms of see crack initiation probabilities. However, until sufficient field
experience is accumulated, the above estimated probabilities of sec initiation for the older
design are being be used in the life expectancy analysis of the six disk design configuration.
Conclusions
Considerable laboratory based research together with improved design and manufacturing
techniques and field experience, have helped in the development of shrunk-on disk type LP
turbines which have successfully combated stress corrosion cracking. This has allowed the
ultrasonic inspection interval of the disks to exceed 100,000 hours, and at the same time
increased the thermal performance and power plant output.
10
H. R. Jhansale
References
1 H. Oeynhausen, G. Roettger, J. Ewald, K. Schleithoff and H. Termuehlen, "Reliable
Disk Type Rotors for Nuclear Power Plants" - Siemens paper presented at the
American Power Conference April, 1987.
2. E. Muehle, K. Keienburg and H. Termuehlen, "Considerations to Achieve Reliable
Long-Time Turbine Operation" ASME 87-JPGC-Pwr-55.
3. H. Termuehlen, K. Schleithoff and K. Neumann, "Advanced Disk-Type LP Turbine
Rotors" - Paper presented at the EPRI Workshop -Stress Corrosion Cracking in
Steam Turbines on October 10-11, 1990.
4. W. David, G. Roettger, K, Schleithoff and H. Termuehlen, "Disk Type LP Turbine
Rotor Experience" - presented at 1993 International Joint Power Generation
Conference.
5. W. G. Clark, B. B. Sheth, and D. H. Shaffer, "Procedures for Estimating the Prob-
ability of Steam Turbine Disc Rupture from Stress Corrosion Cracking". Joint
ASMElIEEE Power Generation Conference, October 4-8, 1981.
6. A. McMinn, F. Lyle Jr., G. Leverant, "Stress Corrosion Crack Growth in NiCrMoV
Turbine Disc Steels", NACE, Vol. 41, No.9. Sept 1985.
7. F. Lyle, "Low Pressure Rotor Disc Cracking and Remaining Life Analysis", SwRI
Report.
8. J. Zhang, H. Jhansale, and D. McCann, "Application of Probabilistic Techniques in
Condition Assessment of Rotors and Disks", 5th EPRI Steam Turbine/Generator
Workshop, July 29-August 1, 1997.
11
12
TABLE 1: DESIGN EVOLUTION OF
SIEMENS LOW SPEED SHRUNK-ON DISK LP ROTORS
H. R. fhansale
DETAIL 10 DISK DESIGN 8 DISK DESIGN 6 DISK DESIGN
Year Introduced 1960'S 1987 1995
No. of Rotors 57 29 4.+ 6 on order
Manufactured
No.o/Disks 570 232 24 + 36 on order
No. of Disk Hours 2x 10
7
6x 10
6
4.5 X 10
4
Blade Layout Disks 1 & 2: Drum Disks 1 & 2 Disks 1 &2
Disks Free combined giving 1 combined and Disks
Standing drum and 3 free 3 & 4 combined,
standing giving one with drum
. type, and two with
free standing
Material 2 - 3.5% NiCrMoV 3.5% Ni, . 3.5% Ni,
1.50/0 Cr-Mo-V 1.5%
.Keyways 3 or 5 Keyways on 3 or 5 on Only disk 1 has
downstream downstream
keyways
Net Stress Surface 80-100% of yield < 500.!o of yield < 50% of yield
Comprehensive Not common Yes Yes
Residual Stress by
Heat Treatment
Shot peening, None Yes Yes
.Honina. and Rot/ina
.Comprehensive None Only caulking area Yes
Residual Stresses
in Blade attachment
areas
InspeetJon interval > 50,000 hours >100.000 hours > 100, 000 hours
Der NRC Criteria
Disk cracking One disk with sec, "None No Data
experience and two with non-
specified indications
Figure 1
Ten Disk Rotor DesIgn

----
ll!HI
,,,'...
EIght DIsk Rotor DesIgn
Ic:aMI..... ... &12 Dlel'
Wf- _
SIx DlakltCltcr Dellgn
PIwwIow-:&aDl..)
------_..-
Figure 2
Slrass '!tensity Fae1or. K
13
H. R. Jhansale
Figure 3: Stress Corrosion Crack
Initiation Tests on LP Rotor Steels
Figure 4: Stress Corrosion Crack Initiation
of LP Rotor Steels, 0.2% Yield Strength <
1000 MPa
High Purity
RSCCI>1.1
Steel Surface with Inclusions.
Mechanical Scratches.
Favourable Crevices
Stress Corrosion Crack
Initiation
T Strength
o .
0,9
1.2
0,5
1400
MPli
1200 1000 800
Pure Water
SmoottI and Natchecl $pl!eimens
.09 to 1.0 Test Stress to
Yield Stnngth Ratio
10$'+--"T--+--""'-+---.-
100
Hours
10",-----.--------.-----
c
.2
Gi
1'i
f
o
E
i=

Yield Strength (Rpo.2)
Figure 5: Yield Strength Dependence of sec Crack Growth on LP Turbine Rotor Steels
s-CoftoD&Ioft Cndc
of fIoto< Sceela III
HIgh Purttr w_
o nSF (10.(;) }

eu.s.. Nuc:tMr Plcnt lIftoulIa

,
..0
.-
-
.
-
14
Figure 6: Measures to Produce Residual Compressive Stresses
H. R. Jhansale
O!l;:k Heat Treatment
Water
Supply
Special ....chinlng Procedure fo4"
Biade Grooves to IIlnfm!ze
TensUe SUrface Stresses
Compressive Stf'e&s
Zone kom
Heat Treatment
A
ROLUNG AND HONING OF KEYWAY AFTER SHRINKING
Detail A
Disk 11
DetailS
Figure 7: LP Rotor Half, Isothermal Lines, c
15
16
GE Dovetail Stress Corrosion Cracking Experience and Repair
Presented By
Carl DeCesare
Dennis Lessard
Jack Nolan
17
18
By
C. DeCesare
S. Koenders
D. Lessard
J. Nolan
Schenectady, New York
19
20
By
C. DeCesare
S. Koenders
D. Lessard
J. Nolan
Schenectady, New York
Introduction
Stress corrosion cracking (SeC) in steam turbine components has been a concern in
the industry for over 50 years as evidenced by ASME involvement in creating sodium
hydroxide equilibrium curves in the 1930ss for the typical steam conditions of the times.
Since then, a substantial amount of laboratory and theoretical work has been done in
an effort to prevent and characterize the behavior of sec in turbine components. But,
perhaps more importantly, a wealth of knowledge has been gained by actual
experience in all manufacturers fleets. This paper presents information on finding and
repairing dovetail cracks. the most recent sec problem to emerge as a critical issue in
the bid to prolong the useful life of steam turbines. Pertinent to this discussion is some
background which defines the potential units and locations for dovetail see to
develop, the NDE technique for locating and sizing possible cracks and the options to
repair the rotor if cracking is found.
Background
see occurs when the "right" match of conditions exist between stress, material and
environment. Despite the concern in the early days of steam turbine manufacture, it
wasn't until the 1960's that dovetail cracking was first evidenced. As turbine sizes were
increased to larger and larger ratings during the 1950's, 60's and 70's, manufacturers
were faced with the need to employ stronger and stronger material in order to withstand
the higher stresses that became unavoidable. At the same time, boiler pressures were
increasing and the once-through boiler and the nuclear reactors were introduced as
steam supply equipment. In addition, techniques for scavenging oxygen, removing or
preventing deposits and regulating pH all introduced chemical impurity levels beyond
those previously experienced. The combined change in material, stress and
environment led to a few more instan.ces of dovetail cracking in the 1970's, several
more in the 1980's and the frequency is increasing in the 1990's.
The sec phenomenon that occurs around the Wilson line in steam turbines that have
experienced corrosive contamination is well documented. Corrosive products can enter
a turbine through many avenues during the operation of the unit such as: improper
control of the phosphate treatment process. condenser tube leaks, introduction of
untreated water such as throughattemperating sprays, cleaning processes used to
remove deposits and use of sodium sulfite as an oxygen scavenger. While the
21
Carl DeCesare, Dennis Lessard and Jack Nolan
potential for corrosive contamination exists on all units, once-through boilers present a
greater susceptibility to carryover.
While the extensive work done investigating see has shown that it is very often
associated with "corrosive" contamination by chemicals such as sulfides, hydroxides
and chlorides, it can also occur in pure water. The history of the 1980's has
demonstrated that the highly oxygenated steam found in n u ~ l r steam cycles creates
an environment conducive to sec in the low pressure sections.
In-Service Inspections
Due to the potential for a rotor burst to occur from cracks that might develop at the rotor
bores after operation at elevated temperatures and cycling conditions, GE established
a program to perform nondestructive examination and evaluation of rotors in the
1950's. The test scope has been modified through the years to address other areas of
potential concern such as dovetail cracking from creep rupture. high cycle fatigue and
sec.
Dovetail Ultrasonic History
Tangential entry wheel blade attachments (wheel dovetails) in GE steam turbine
rotors have been inspected with ultrasonic equipment since the mid-1960's.
Initial inspections were limited to those rotors that had unusual operating
excursions or evidenced steam chemistry problems such as caustic ingestion
during a boiler tube cleaning process. The ultrasonic inspection of wheel
dovetails became one of the standard in-service inspections recommended for
some specific stages of high pressure and intermediate pressure rotors in 1968.
This recC?mmendation was driven by the potential for the high temperature
stages to experience creep rupture cracking. The ultrasonic testing of the last
two stages of tangential-entry dovetails in low pressure rotors was added to the
in-service inspection list in the early 1970's. The inspection standard was further
expanded in the mid-1970's to include wheel dovetails susceptible to sec in
both fossil and nuclear turbine rotors. Additional wheel dovetail stages were
added to the inspection list in the 1980's and again in 1993. Today,. all large
steam turbine rotors with tangential-entry dovetails have some stages
recommended for ultrasonic inspection.
The basic approach for ultrasonically inspecting tangential-entry dovetails
developed in the mid-1960's has not changed. In essence, the ultrasonic search
unit is aimed radially toward the outside of the wheel with a beam angle that
directs the ultrasound at the inner bottom radius of the hook to be inspected
from the opposite side of the wheel (Figure #1 scan plan). The operator
positions the ultrasonic search unit in a given radial position and adjusts the
ultrasonic instrument to display the expected reflectors on the screen at the
required amplitude. The operator then moves the search unit circumferentially
around the wheel looking for reflectors to show at the expected crack site. The
signals that meet the reporting criteria are recorded for engineering analysis and
disposition. Despite the fact that the basic technique has not changed,
continuous improvement in fixtures, ultrasonic instruments, transducers and
calibration techniques have been made over the years to address the desire to
detect smaller and smaller cracks. The most recent equipment applied to
ultrasonic inspections of wheel dovetails is computerized digital data acquisition.
22
The system collects ultrasonic data as the wheel is rotated and stores the
information in a file. At the completion of the scanning process, the inspector
displays the stored data on the computer screen. The display shows the
circumferential position, time of flight and amplitude (color coded) for all
ultrasonic scans in the selected segment of the wheel. The ultrasonic
information is also displayed in an lIAn scan form in the segment displayed
(Figure #2). This technique allows the inspector to see both ,the overall view of
what has been recorded during the test as well as precise details by positioning
the cursor to areas of concern in the display and observing the "A" scan. The
computerized digital data acquisition (Auto Dovetail System) has been in use
since 1993.
Detection and Sizing
The correlation of wheel dovetail ultrasonic inspection results to magnetic
particle inspection results are rarely one to one, even today with all the
refinements in ultrasonic equipment. The results of one such correlation study
are shown in Figure #3. To gather the data, the ultrasonic test was performed
and then the buckets were removed and the depths and lengths of the cracks
estimated during removal by local grinding. The depths ranged from less than
0.02" to 0.50
1t
The aspect ratio of length to depth for these cracks showed no
consistency and varied from 1: 1 to > 500: 1. This wide variation in aspect ratio is
one of the contributors to the uncertainties in detection and size estimations in
this application of ultrasonic testing. Despite the substantial variation that occurs
in stress corrosion cracks. the capability to correctly diagnose the situation is
increasing as the database of testing experience increases. Two of the major
features of the ultrasonic inspection data that are presently considered most
reliable in discriminating a potential crack reflection from other types of
reflectors are shadowing effect (LOE) and circumferential extent (estimated
length). The type
l
location, number of ultrasonic indications, degree of bucket
lifting, fleet history and the units' operating parameters are some of the other
factors considered in any recommendation to investigate further by removing
buckets.
Future Developments
Presently there is in development a technique that will supply an additional
ultrasonic feature to improve the reliability in discrimination of crack type
reflectors from others. This ultrasonic technique will provide an estimated depth
of crack-like reflectors. Although the data collected with this technique is
Iimited
l
the correlation of crack depths estimated ultrasonically to the depths
determined during the removal of cracks by grinding is encouraging. An
example of one such study is shown in Figure 4.
23
Transdncer
Position
4.7
Inches
Wheel Thickness
Figure 1- Ultrasonic Inspection S:8D Plan
15
14
13
12
11
18
9
e
7
6
5
4
3
2
1
8
x FSH

)(-POS :4.186
Y-POS :3.557
III
#.
r
1/.
;
+3.162
+4.2.1.6
..PRIMT PALETTE VELOCTY OFFSET
+6 324.----
Figure 2 - Inspection Presentation
24
12 .-
10
8
!11 MP Lengti1l
I_Auto UT length:
4
2

. .... '
Figure 3 - Dovetail Crack Length Estimates
T
--
I
I
i.
I
0.2 f------t--
0.05
1
015
1
1 1
I- I I
0.1 _ .. ;. .
:;
c(
0-- 1.
4
.__.
o 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Grinding Depth (inch)
Figure 4 - Dovetail Crack Depth Estimates
25
Repair Options for sec of Wheel Dovetails
Even when a mechanism such as sec causes severe damage to a highly stressed area of a
rotor, such as the dovetail, it does not necessitate replacement. GE has developed a number of
repair alternatives which can often restore or, in many cases, improve the integrity of the rotor.
When coupled with positive steps to improve steam chemistry, a repair is often the most cost
effective option and can provide a long term solution to the problem. Four repair alternatives
which provide a range of cost, repair cycle time and implementation lead time are:
enhanced dovetail machining and shot peening
long shank buckets and shot peening
a class IV weld repair and shot peening
a class VII weld repair and shot peening
Enhanced Dovetail Machining and Shot Peening
When dovetail cracking is found, the quickest and most cost effective repair is to re-machine the
dovetail profile to the most optimum design possible. The following considerations influence the
potential use of this repair technique:
crack location and depth, and
rotor material and condition
In many cases, if indications are detected early enough, machining is the only operation
necessary for continued long term operation of the unit. This is particularly true if GE is able to
enhance the dovetail. In cases where the risk of cracking is high, machining may also be used
as a preventative technique in order to perform the work in a planned fashion and reduce the
possibility of developing a crack that grows too deep to machine out. On the other hand, if
indications are found to already exist and to be deep, a more extensive repair may be
necessary.
Long Shank Bucket Replacement and Shot Peening
When dovetail cracking is found to be very deep, it may not be feasible to simply re-machine the
existing dovetail. In this situation, it may be desirable to install long shank buckets to avoid the
loss of the row of buckets, weld repair or rotor replacement.
Long shanking involves machining a new dovetail at a smaller diameter, in an uncracked portion
of the rotor forging. Buckets are specially designed with weight reducing slots or holes to
reduce the additional centrifugal forces placed on the rotor as a result of long shanking. Other
features such as enhanced dovetail profiles and shot peening can be used in conjunction with
long shanking to provide the optimum repair.
The principal disadvantage of using the long shank design is the cycle time required to
manufadure special buckets. Additionally, the geometry of the original wheel and dovetail
design may not allow room to perform a long shank repair. If a long shank repair is not feasible,
a weld repair would be the next option available.
26
Rotor Dovetai I Weld Repairs and Shot Peening
GE has had many decades of successful experience weld r p i ~ i n g thousands of applications
on several tUrbine components and has pioneered in the understanding of complex
mechanisms such as creep, low and high cycle fatigue, SCC, and relaxation. GE is also
responsible for the development of a number of new welding technologies. In 1957, GE
published one of the first practical methods of performing long term weld repairs on Cr-Mo-V
steel, the steel that most high and intermediate pressure steam turbine rotors are made of.
Nonetheless, because of the high operating stresses in turbine rotors, the severe consequences
if a weld failure were to occur, the relatively low incidence of severe cracking problems on GE
rotors, and the availability of other repair options, GE did not elect to develop specific rotor
welding techniques until the 1980's.
In the earty 1980's, GE developed and performed a rotor weld repair for AEP's Mitchell unit. The .
repair was performed at GE's Schenectady facility. GE and AEP co-authored a paper on this
repair and presented it at an EPRI conference in 1985. The unit continues to operate
successfully.
In 1989, GE opened its first rotor welding "Center of Excellence" in Bangor, Maine, a facility
dedicated to performing high quality, economical repairs of small turbine rotors. In 1993, GE
opened a second "Center of Excellence" at its Chicago Service Center designed to perform high
quality and economical repairs of large turbine rotors. The Chicago Service Center has the
capability to perform repairs on smaller rotors as well. Both facilities now have experience
performing all types of rotor weld repairs.
In 1996, GE began development of an innovative state-of-the-art on-site rotor welding process.
The process is derived from the Fine LineTM Welding technology used in the GE Nuclear Energy
business. Repair of a steeple on a gas turbine wheel of another manufacturer's turbine has
been successfully completed using this process.
Class IV and Class VII Repairs
The Hartford Steam Boiler Co. has deveJoped a classification system for the methods commonly
used to repair rotors which is now generally recognized in industry. The most common
classifications are as follows:
Class I:
Class II:
Class III:
Class IV:
Class V:
Class VI:
Class VII:
Weld buildup of gland, seal, or bearing journal areas.
Weld repair of blades.
Weld restoration of individual rotor steeples.
Replacement of integral wheel rims with weld buildup
Deep groove weld repair of circumferential cracking.
Sectional replacement of rotor with welded on stub shaft.
Replacement of integral wheel rims with welded-on forged ring
While GE performs all of these classes of repairs, the class IV and class VII repairs are
particularty useful in repairing dovetails.
27
One recent example of a Class IV repair done at GE's facility in Chicago was on a Ni-Cr-Mo-V
rotor dovetail with SCC. A proprietary shielded metal-arc welding and heat treating process was
used to bUild-up a new dovetail entirely with weld metal. The heat treating was done with the
rotor turning while horizontal to avoid potential loss of control and damage to the rotor. During
the machining process, an enhanced dovetail was machined using the eNG. The Class IV
process can often be done within a normal outage cycle without extensive advance planning.
Another recent example is a Class VII repair performed on a 'gas turbine wheel. A forged ring is
used as a replacement dovetail instead of a weld buildup. The weld is exceptionally narrow and
is confined entirely to the low stressed area of the wheel, allowing the higher stressed steeple to
be entirely new forging material. Furthermore, the properties of the weld itself are exceptional,
with defect levels no greater than acceptable forging indications. In the case of this wheel, the
NiCrMoV dovetail was replaced with a forging of the same material. However, repairs are being
done with dissimilar materials optimized to achieve maximum repair life. The Class VII repair
requires pre-planning due to the lead time for the new forging.
Shot peening
Since the inception of see in turbine components. GE has performed tens of thousands of
hours of laboratory testing simulating environmental, material, and stress conditions in an effort
to improve our understanding of and ability to prevent sec. Included in this work has been
various methods of changing the surface condition of the components, such as coatings,
hardening and shot peening. This testing has shown that shot peening can significantly
increase the resistance to SCC. For instance, GE has over 20 years of successful experience
with sh.ot peened dovetail pins as replacements for pins that started cracking via sec in the
early 1970's. Based on this success and the research performed by GE and others, with any of
the above repairs, GE shot peens the newly machined dovetail surface.
Conclusion
Dovetail sec has emerged as a leading concern in the bid to prolong the useful life of low
pressure rotors. While the replacement of some of these rotors will be required due to other
issues that impact the potential life or will be desired in order to increase the thermal
performance or output, there are cost effective alternatives to replacement. Key to dealing with
this potential is tightened water chemistry guidelines and operating practices, periodic NDE and
early action for high risk rotors or those with known problems. When a rotor dovetail is found to
be cracked, a thorough evaluation of the intended plant life and economics associated with any
modernization that can be obtained with a new rotor may lead to a desire for short or long tem
repair. GE has a range of options available to allow a repair that is responsive to each owner's
unique situation. For several decades, GE has been investing in the advancement of NDE
technology, preventative techniques such as coatings and repair technologies and equipment
such as portable welding. Programs are ongoing in all these areas as we search for solutions
that minimize risk and are increasingly more cost and time effective for equipment owners.
28
Qualification of Welding Rotors with 12Cr
Stainless Steel to Improve sec Resistance
Presented By
DenmsAmos
Westinghouse Electric Corporation
29
30
Qualification of Welding Rotors with 12Cr Stainless
Steel to Improve sec Resistance
D. R. Amos
E. A. Lay
Westinghouse Turbine Generator Plant
PO Box 7002
Charlotte, NC 28241-7002
s. A Bachman
Westinghouse Power Generation Business Unit
4400 Alafaya Trail
Orlando, FI 32826-2399
31
32
Qualification of Welding Rotors with 12Cr Stainless Steel to Improve
SCC Resistance
D. R. Amos
E. A. Lay
Westinghouse Turbine Generator
PO Box 7002
Charlotte. NC 28241
S. A Bachman
Westinghouse Power Generation Business Unit
4400 Alafaya Trail
Orlando. FI 32826-2399
Stress corrosion cracking (SeC) of low pressure steam turbine rotor blade attachment areas has become a
severe industry wide problem within the last several years. In an effort to assist turbine operators with this
problem, Westinghouse developed a methodology to improve sce and other generic corrosion
resistance[l. 2. 3]. The methodology involved combining two very successful technologies. The key to
success was to capitalize on the corrosion resisting properties of our 12% chromium rotating blades, which
to date have demonstrated a long operating history without problems related to SCC. The other technology
used as a basis ofthe program was Gas Tungsten Arc Welding (GTAW). the Westinghouse rotor repair
process which has now been used on over 300 LP rows to replace the blade attachments by (solid) weld
buildup. Most ofthese rotors came from severe sec environments but all have since operated successfully,
many dating back to the early 1980s. Thus, the new technology consists of a 12% chromium, all weld metal
deposit, or buildup, for blade attachments. or other areas requiring greatly improved SCC and other generic
corrosion resistance. The program has been successfully completed. The repaired discs met all the
requirements and in most cases, such as the mechanical properties, have far exceeded the requirements.
The process has been implemented on 42 rows on four nuclear LP rotors. In addition to upgrading
operating units, several rows on new nuclear LP rotors have been offered. The paper will discuss in detail
all aspects of development. testing, mechanical properties, heat treatment, nondestructive evaluation, fuJI
scale prototype, and implementation.
33
Dennis Amos
Background
Repair of rotor steeples and blade attachment areas by weld buildup has become routine. Since our first
commercial repair in 1979, we have weld repaired over 300 LP rows on rotors or discs. Since 1988, when
we completed our first control stage repair, we have welded 70 rows on 34 HP or HPJIP rotors, some welds
operating at 1050F. In the early 90's, the rotor welding program was successfully extended to combustion
turbine discs of the 140 ksi class and rotors of both LP and HPIIP turbines. Repairs Has
have been made on all the rotor alloy compositions used by domestic' turbine and generator manufacturers
and covered by ASTM A470 for rotors manufactured since 1948.
Not only have aJl types of components been repaired, but also there have been a variety of different types of
repairs. Welds on bores ofrotors, bores of discs, coupling holes, control stages, journals, locking slots, and
many others have successfully been accomplished. However, the majority of experience gained has been
steeples or blade attachment areas. Steeples repairs have included side entry configurations such as straight,
skewed, curved, as well as individual steeples from one to eight. By far, the most commQn repair has been
the entire row, i.e. a 360
0
buildup. A large data base exists including base metal properties, welding
consumables, and different welding parameter schedules or heat input criteria.
For more than 20 years, our company has been aggressively evaluating all phases of processes, equipment
and many other refinements for rotor welding. Key to all these developments has been the area of filler
metals. One of our newest developments, welding of 12Cr stainless steel to the rotor periphery to improve
stress corrosion cracking resistance, is an innovative and cost effective solution to an industry wide
problem.
Design Qualification
Use ofa 12Cr steeple in areas that had been traditionally low alloy, NiCrMoV forged material, required
several important investigations. First, the yield strength needed was cakulated baSed on the steeple stress
level for the minimum level required. The maximum level was kept as low as possible to improve sec
resistance.
Because the I2Cr material was the same weight density as the NiCrMoV forging that it replaced, there was
no change in the weight of the rotor. Also, because the areas of the rim have no effect as to stiffuess ofthe
rotor in comparison to the original rotor, there was no change in lateral critical speeds or torsional
frequencies of the rotor or ofthe rotor train.
A very important aspect of 12Cr that required careful and thorough analysis was its lower coefficient of
thennal expansion as compared with NiCrMoV forging. The significance of this differential occurs when
cooling the rotor from PWHT where thennally induced residual stresses result. This occurs because the
weld shrinks at a slower rate than the rotor material adjacent to the weld, resuiting in residual stresses which
are compressive tangentially in the weld and radial tension across the heat affected zone. In addition, minor
residual levels of weld stress remain because stress relief never results in zero weld stress. The thermally
induced stresses are opposite in sign to the residual weld stresses.
A transition zone of varying Cr levels was used between the base metal and the weld to provide a gradual
transition ofthermal expansion coefficients and hardness. This lowered the thermally induced residual
stresses across the heat affected zone and minimized the metallurgical stress riser.
Nominal operation stresses were calculated and combined with residual stresses at operating temperature.
This solution included the different coefficients of expansion, the differing elastic moduli, the temperature
change between room temperature and design temperature, and centrifugal loads.
34
Dennis Amos
Development and Testing
Our design and manufacturing philosophy in developing new welding technologies has proven to be highly
effective and reliable as none of our welds have ever experienced problems in the field. Foremost in our
philosophy is that the weld repair must preserve the operational integrity and reliability of the component.
To that end, there have always been 3 important considerations in any decision regarding development of a
process, filler metal. and welding procedure:
] ) Superior weld properties
2) Acceptable heat affected zone properties
3) Consistent weld soundness, free of rejectable indications
Process Selection
Both submerged arc and gas tungsten arc processes are used for rotating components. Each
process has its advantages and disadvantages. Submerged arc data already existed for 12% Cr"and
it was unacceptable for critical components. Even without sufficient data, experience has proven
that the very coarse dendritic structure of submerged arc results in reduced toughness. Therefore,
submerged arc was eliminated until improvements can be made.
Filler Metal.Development
All commercial wires of ] 1Cr or more were investigated, including many non-standard wires not
presently classified. However, none were adequate. In general most were too high in strength and
too low in toughness. Some had the proper strength and toughness but did not have the chemical
makeup to optimally resist stress corrosion. Thus a plan to fonnulate a specific chemistry was
devised. The highly successful blade chemistry was studied and the new filler metal specification
closely followed. After adjustments were made to the specification, the exact target properties
were achieved. Seven heats have now been tested and qualified for use. .
Procedural Development
It was understood very early that the differences in thermal expansion coefficients of the 12Cr
weld and rotor forging were substantial enough to possibly result in very high stresses upon
cooling down from the postweid heat temperature. A decision was made to produce a transition
weld with a different wire. As the thermal expansion coefficient is primarily governed by
chromium, wires with lower chromium were evaluated, resulting in an intennediate value of Cr
filler wire. The transition consisted of depositing several layers of a lower Cr wire which were
precisely controlled to achieve an exact bead thickness and a predetermined dilution resulting in a
graduated or incI:easing chromium percent. Thus. the chromium content increased. layer by layer,
gradually from about 1.8% in the base metal to about 12% in the weld metal.
Mechanical Testing
The tensile and impact test results of test plate welding are given in Table 1 as averaged data. Note that in
all cases 12Cr all weld metal properties equal or exceed those of wrought 12Cr steel. Tensile and impact
test results of the prototype disc are shown in Table 2, and in addition to all weld metal, a comparison is
made to transverse weld data. These tests were performed at room temperature. The importance of
transverse tensile testing is to determine where fracture will occur and what is the strength level. Transverse
testing includes the base metal, HAZ, the transition zone oflower Cr deposits, and finally, the full 12Cr
weld metal. There were no failures in the HAZ or in the transitional zone. The slight variation in yield
strength between the all weld and transverse weld is caused by directionality and is normal.
High cycle fatigue tests were performed on weld and base metal. The tests were designed to simulate steam
turbine environments, i.e. low oxygen and high oxygen steam. Both smooth and notched bar specimens
35
Dennis Amos
were tested to assess what, if any, were the effects ofnotches in high cycle fatigue. Such understanding is
critical when a design incorporates many small radii. Le. steeples Dr blade attachments, subject to bending
stresses. Previous experience with the submerged arc process resulted in the all weld metal being notch
sensitive. However, the present GTAW data confirms that there is no notch sensitivity. For all specimens
and conditions, 12Cr is equal or better than both wrought 12Cr and NiCrMoV rotor material.
Results of coefficient of thermal expansion testing is shown in Figure 1. Samples were machined from base
metal, an all 12Cr weld deposit and two areas of the transition zone. One area of the transition zone was
adjacent to the base metal, and the other was adjacent to the 12Cr weld metal. The tests were conducted
using the thennomechanical analysis technique. While the COE at operating temperature is not a concern,
there is a large differential at the maximum temperature ofPWHT. The differences were critical and taken
into account in the design analysis.
Microstructure examinations were made from the HAZ, through the transition zone, and well into the 12Cr
weld metal. All tests were performed after PWHT. The grain structure in the welds was fully refined, and
very little as-cast structure was found. The fine grain tempered martensitic structure resulted in excellent
.strength and toughness and even exceeds that of wrought 12Cr steeI. This very high qual ity weld structure
is the product of the GTAW process and carefully selected welding parameters.
To properly assess stress corrosion crack susceptibility, it was decided to use several types ofU-bend
specimens. After 12 months of exposure, predictably the NiCrMoV rotor material test specimens
experienced the most distress, especially for those exposed to aggressive environmental conditions. No
cracking has been reported in any ofthe 12Cr weld metal specimens. ln all testing environments the
response of 12Cr all weld metal and the 12Cr blade material specimens are very similar, with the 12Cr weld
being marginally better than w r o u ~ 12Cr.
Postweld Heat Treatment
Defining a PWHT cycle that resulted in the correct amount of weld tempering without overtempering the
base metal required extensive stress analysis and process development. Welded test plates were stress
relieved at different time-temperature combinations to determine the optimum stress reliefparameters.
The maximum cooling rate was established to minimize thermal stresses and the minimum cooling rate was
limited to prevent embrittlement. Impact (CYN) tests, resulting in acceptable toughness, confinned that the
minimum cooling rate did not embrittle the weld. A full scale prototype disc, which will be explained in
more detail later, was postweld heat treated with the optimum parameters. Results of residual stress
measurement after PWHT indicated acceptable stresses which verified that the hold time was adequate to
reduce stress and that the coolil'!g rate did not induce further stresses.
Postweld heat treatment ofthe 88381 FI rotors was the largest ever performed in Charlone. PWHT of these
large rotors required defining and executing a heater setup and cycle that was different than the method
conventionally used. A transient finite element analysis of the PWHT cycle determined allowable thermal
gradients in the rotor body to keep thermal stresses in an acceptable range. The setup was accurately
modeled including heater locations, boundary conditions, and temperature loading. Temperatures were
monitored at critical locations to assure the gradients were kept within the limits specified: FEA predicted
gradients and temperatures agreed well with actual measured data.
Nondestructive Examination
The subject forgings were purchased using ultrasonic inspection (UT) to a criterion of no indications 50%
or greater ofa 1/16" diameter FBH (Flat Bottom Hole). This is also the requirement in various ASTM
36
Dennis Amos
specifications. However, test plates, prototypes, and operators are subjected to 50% of a 1/32" FBH to
more closely scrutinize the process and welding operators.
Additionally, magnetic particle examinations (MT) are used before and after postweld heat treatment to
verify surface or near-surface integrity. No linear indications are acceptable.
All test plates. the prototype disc, operators, and production welds met the above criteria.
Full Scale Prototype
One of the final and the most important phases in the development and qualification of 12Cr rotor welding
is the welding and testing of a full-size prototype.
The prototype was welded on a rotor disc similar in size to Row 9 (one of the largest rows in diameter) of
the first 2 production rotors by using the same procedures and wires as developed from test plate welding.
The 360
0
prototype weld was welded to the comparable size (width and steeple height) as to the weld for
Row 10 (the largest weld in size to be welded for the repair) ofthe production rotors. The weld was
finished at 2.00" in height and 9.62" in width. The rotor wire consumption was 1699 lbs, which included
both types of alloys. The weld volume was large to fairly evaluate the residual stress measurement. One
arc ofthe disc was side entry machined to more accurately detennine residual stresses at the steeples.
The prototype was postweld heat treated using the selected parameters from the testing and development
phase. Tests made on the prototype were tensile, impact, microstructure, hardness, residual stress
measurements and dimensional measurements.
Prototype Disk Material
A service exposed LP NiCrMoV rotor disc was used for the prototype. The disc was tempered
prior to weld prep machining to match the strength of the production rotors. The disc material
properties and chemistry were tested, the results are given in Tables 3 and 4. The prototype disc
had equivalent mechanical properties a's the BWR LP rotors.
Nondestructive Test
The prototype was magnetic particle and ultrasonically inspected. Four indications randomly
located in the weld were detected to a 50% of a 1/32" diameter flat bottom hole reporting level, but
no indications were detected at the level (50% of a 1/16" diameter flat bottom hole).
The indications were later sectioned out from the prototype and examined in lab. These
indications were determined to be gas porosities.
Dimensional Measurements
The disc run-outs were charted before welding, after welding but before PWHT, and after PWHT
and final machining at 45
0
intervals. The most critical dimension was the bore diameter but inlet
and outlet faces as well as the OD were measured. No measurable distortion occurred,
Residual Stress Measurement
A significant number of residual stress measurements were made on the weld, HAZ and base metal
after postweld heat treatment. The exact locations were carefully selected to cover the entire weld
such as the 12Cr and the transition zone of several layers between the 12Cr and the HAZ. Most
measurements were made using the hole drilling method, ASTM 837-94, although some X-ray
diffraction measurement was used on very small radii of finished machined steeples.
Measurements were made at the surface and nominal depths of .001, .005, and .011 inches.
Measurements at these depths gave a clearer understanding ofthe machining residual stress
distribution. In summary the data show small levels, less than 8 ksi, of tensile residual stress in
37
Dennis Amos
the base metal and HAZ. The residual stresses measured in the 12Cr and transitional zone were in
a compressive or in the very low tensile range.
Implementation
Two large nuclear FI LP rotors (20 rows per rotor) and two separate discs have been repaired with 12Cr
weld. with approximately 42,000 pounds of filler wire applied. An example of one large rotor is described
in Table 5. On smaller jobs, a single torch is used in a conventional manner. However, in the case of20
rows on one rotor, it was necessary to apply innovative tooling and multiple torches. An example of this is
Figure 2 which depicts a rotor with 8 simultaneous welding arcs. The rotor is shown again in Figure 3
being finished machined. .
There are several other methods to reduce cycle time, such as increasing wire feed speeds or lowering
preheat temperatures. However, such methods may jeopardize weld quality. Close observation was
to keep the diameters fairly equal so that the welding (surface) speed was within specification.
Conclusions
12Cr weld has slightly higher yield strength and much better toughness than v,rrought l2Cr steel.
The tensile and impact strength of the weldment meet the design requirements.
In low oxygen stearn environment, the fatigue strength of the 12Cr weldment is higher than that of
v,rrought 12Cr blade material, and meets the design requirements. For equivalent tensile strength
between 12Cr and NiCrMoV rotor steel, the 12Cr weld fatigue strength is comparable to NiCrMoV
steel.
In high oxygen stearn environment. I2Cr weld has higher fatigue strength than NiCrMoV rotor
steel.
The 12% Cr all-weld metal and the 12% Cr blade test specimens have shown a similar response in
terms of limited corrosive attack, and lack of cracking, when exposed to a variety of chloride rich
environments. No difference in sec resistance was noted between the 12Cr weld and l2Cr v,Tought
specimens.
The prototype welding and testing demonstrated that the selected posrnreld heat treatment can
effectivel\' eliminate hicll welding stresses. After PWHT. the residual stress of the welded disc was
within -
Hardness transverses after PWHT taken across the heat-affected zone and transition weld showed
no sign of disc and weld materials softening or over hardening. The hardness profile showed a gradual
transition between the rotor and weld.
The tensile and impact strengths of the prototype are comparable to the strength of the weld test
plates.
38
Dennis Amos
I U.S. Patent 5,024,582 by Bellows, et aI., dated 1991, "Steam Turbine Rotor Having Graded
Weldments. "
2 Patent application filed, "High Chromium Content Welding Material to Improve Resistance of
Corrosion Method and Device."
3 U.S. Patent 5,591,363 by Amos, et aI., dated 1997, "Optimized Welding Technique for NiMoV
Rotors for High Temperature Applications."
39
Dennis Amos
Table 1
12Cr Weld Test Plate Properties
40
Table 2
12Cr Weld Prototype Disc Properties
Table 3
Tensile Properties of Rotor or Discs Forgings
Dennis Amos
Table 4
Chemistry Results of the As-Received LP Rotors
and Chemistry Requirements for ASTM A-470, Class 7
Table 5
Pounds of 12Cr Used on a Large Nuclear BB381 FI Rotor
41
Dennis Amos
Coefficients of Thermal Expansion
9
8.5
'?
8
lo<
""" 7.5
c
]
7
6.5
6
100 300 500 700 900 1100 1300
Temperature of
Base Metal _.. 12 CrWcld
Laboratory Stress Corrosion Cracking Experience in
Steam Turbine Disc Steels
Presented By
Dr. S. R. Holdsworth
GECAlsthom
43
44
LASORATORY STRESS CORROSION CRACKING EXPERIENCE IN
STEAM TURBINE DISC STEELS
S R Holdsworth:
t
M Nougaret
B W Roberts:
D V Thornton:
GEC ALSTHOM Large Steam Turbines
t Willans Works, Newbold Road, Rugby CV21 2NH, UK
t
141 Rue Rateau, 931 26 La Courneuve, France
45
46
LABORATORY STRESS CORROSION CRACKING EXPERIENCE IN
STEAM TURBINE DISC STEELS
S RHoldsworth
t
M Nougaret
t
B W Roberts
t
D V Thornton:
GEC ALSTHOM Large Steam Turbines
: Willans Works, Newbold Road, Rugby CV21 2NH, UK
t
141 Rue Rateau, 931 26 La Coumeuve, France
Abstract
The risk of stress corrosion cracking (SCC) in turbine components is generally assessed on the
basis of prior service experience and the results of laboratory tests. Care is required in the use
of experimental data since unreliable predictions of service behaviour are possible if the see
properties used in assessment are not collected under conditions representative of those
encountered during operation.
For this reason, extensive laboratory testing has been performed by GEe ALSTHOM in pure
low oxygen condensing steam to investigate the influence of alloy chemistry, strength level
and surface condition on sec properties, and to characterise the mechanisms of crack
initiation and growth in 3-3Y2NiCrMoV turbine disc steels. Long tenn tests have also been
conducted to characterise the properties of weldments.
In addition, the testing programme has considered the effect on SCC properties of modifying
the pure low oxygen steam environment by (i) increasing oxygen concentration, (ii) two
shifting, (iii) treating with ammonialhydrazine, (iv) increasing conductivity levels through
sodium sulphate doping, and (v) molybdenum disulphide contamination. The results of these
studies are also reviewed.
Introduction
Since the Hinckley Point disc failure in 1969 (1], significant effort has been devoted to the
experimental study of stress corrosion cracking (SeC) in turbine materials [2-8]. Initially,
stress corrosion crack initiation (SCC!) and crack growth (SCCG) properties were
investigated in relatively aggressive environments such as hot caustic solutions (eg. the CEGB
studies in [2]). Subsequently, most of the laboratory tests conducted by GEe ALSTHOM
were in flowing pure low oxygen condensing steam and water based environments [2-8].
The risk of sec in turbine components is generally assessed on the basis of prior service
experience and the results of laboratory tests. Care is required in the use of experimental
information since unreliable predictions of service behaviour are possible if the SCC property
data is not collected under environmental conditions responsiple for the same crack initiation
and gro'Wth mechanisms as those encountered in disc steels during operation. Tests performed
in aggressive aqueous media may characterise sec properties in a short time scale (eg. [2,9]),
47
S.R. Holdsworth
but the information has to be treated with caution, in particular when the mechanism of
cracking is different to that observed in service. Unlike the results from investigations of this
nature, detailed inspections of many long term tests (up to 100,OOOh) performed in flowing
condensing steam at 95e confirm that the appearance of corrosion and sce arising from
these laboratory conditions are closely representative of features observed in the wet stages of
large steam turbines [1-8]. A sample of sec properties determined in laboratory steam
environments are reviewed in the following paper.
Materials
The review is restricted to the SCC properties generated for 3-3~ N i r M o V LP turbine disc
and rotor steels with RpO.2 strength levels inthe range 695-1035MPa. The properties of older
3%CrMo steels are covered elsewhere [2-4]. Data has been collected for fourteen
3-3 Y2%NiCrMoV steels manufactured by a variety of melting practices. These include
conventional purity forgings produced from silicon deoxidised, vacuum carbon deoxidised
and electroslag refined ingots and three superclean rotors.
Pure Low Oxygen Condensing Steam
Environment
Most tests have been performed in pure low oxygen condensing steam at 95C. This is
generated from feedwater which is de-ionised twice by passing through nuclear grade mixed
bed exchange resin columns both prior to and following oxygen stripping using a sulphite
resin bed column [5]. The conductivity and oxygen concentration of the feedwater and bulk
condensate at 20
0
e are consistently maintained at less than O.1f..lS/cm and 10ppb respectively
(these being typical of that expected and achieved in well run power stations). The feedwater
is converted to steam using a once-through boiler arrangement and passed directly into the
environmental chamber. Feedwater flow rates are adjusted to ensure that there is always a
small positive pressure of steam in the chamber.
This environment was devised to 'model' the condensing steam conditions existing in the
later stages of an LP steam turbine. The steam pressure and flow rates are clearly different to
actual conditions, but an important similarity is the deposition of a thin condensate film on
specimen surfaces in which local solute concentrations are significantly raised and gas
concentrations reduced due to partition thermodynamics.
Crack Growth Properties
Crack growth properties have been detennined using standard pre-cracked 25mm thick wedge
open loaded (WOL) testpieces [2]. Initially, tests were performed in single specimen tests.
Testpieces were loaded to the required initial stress intensity factor (K) and exposed to the
environment for times up to 50,OOOh. Following test, the WOL specimens were broken open
by fatigue to clearly delineate any sec growth prior to physical measurement. The measured
crack extension was simply divided by the exposure time to detennine the rate of SCCG.
Crack growth rates are plotted as a function of the initial K (eg. Fig. 1). Above a threshold
level referred to as K
1SCC
, crack growth rates are relatively independent of K until the critical
stress intensity factor, K
c
, and instability is approached.
48
S.R. Holdsworth
Stress corrosion is a stochastic process and this is reflected by the scatter in SCC property
data. Nevertheless, it was recognised that the width of the SCCG rate scatterband was at least
partly dependent on the exposure duration of the test. Crack extension is typically preceded
by an incubation period, after which growth rates tend to be relatively constant (Fig. 2) due to
the K characteristics of the WOL geometry. Consequently, in short term tests, propagation
rates determined in the way described above could he unduly non-conservative (the same
effect is responsible for non-conservative estimates of K1sed. A multi-specimen testing
practice was therefore subsequently adopted, such that a number of specimens were mounted
for the same material/K conditions and systematically discontinued for fractographic
inspection at various life fractions. Plateau crack growth rates were subsequently determined
from the resulting crack extension time plots (eg. Fig. 2).
Within the scatter exhibited by the SCCG properties for 3-3Y:!%NiCrMoV steels (Fig. 1), the
higher strength materials exhibit the highest growth rates and the lowest values of K
1see
. For
a given strength level, there is not a consistent influence of melting practice on SCCG
properties.
Crack Initiation Properties
The first sec initiation tests were performed on bent beam specimens pre-loaded in self-
contained rigid fixtures to maintain a constant outer fibre stress of known magnitude until the
onset of cracking [2-5]. Specimens were inspected for evidence of crack initiation every
1,OOOh using a low power binocular microscope. More recently, bent beams have been
replaced by a tensile version of the self loaded testpiece, these providing more accurate
control of the applied stress [5-8]. A further refinement has been the adoption of a multi-
specimen approach similar to that used to determine crack growth properties. This provided
the opportunity for destructive metallographic inspection at various life fractions, leading to a
greater understanding of the process of stress corrosion crack initiation in 3-3 Y:!%NiCrMoV
steels [5-7].
The results of this metallographic inspection strategy are summarised in a series of
Rpo.rSCCI crack development diagrams [5-8]. For a given applied stress to proof strength
ratio, these demonstrate the improvement in SCCI endurance properties with decreasing
strength level. Fig. 3 shows an example of such a diagram, for O"app,,=RPO.2' indicating
endurance data covering 'the formation of engineering size stress corrosion cracksI in pure
steam.
In the case of medium strength steels, with RpO.2 less than -850MPa, SCC microcracks
generally nucleate from pits formed at the sites of surface breaking MnS bearing inclusions
[5,6]. Depending on strength level, these may initiate after only 1,000 to 3,OOOh. The
metallographic evidence indicates that a relatively small proportion of these microcracks
develop to become engineering size cracks. The local electro-chemical conditions may
encourage side face corrosion and blunting at the expense of crack extension. or the local
environment may simply become diluted by the bulk environment before K exceeds K
lsec
Those microcracks which do propagate take at least another 10,OOOh to 30,OOOh to develop
into macrocracks, even at applied stress levels of yield point magnitude. In this strength
1 Cracks are referred to as engineering size cracks (or macrocracks) when their surface length exceeds -I mm.
Such cracks are typically >250 deep. Cracks with a depth less than -2501lffi are referred to as microcracks.
49
S.R. Holdsworth
regime, the probability of SCCI appears to be greater in steels produced by melting routes
leading to higher proportions ofMnS bearing inclusions [6J.
At higher strength levels, microcracks increasingly initiate from other metallurgical
discontinuities (eg. surface breaking grain boundaries). The greater number of potential
initiation sites and the inherently lower resistance to crack propagation displayed by the
higher strength steels are largely responsible for their shorter and notably less scattered
microcrack initiation and crack development times. The probability of sccr in higher
strength steels appears to be relatively independent of melting practice.
Surface Finish
The majority of SCCI tests have been performed on specimens prepared with a gently ground
surface finish. However, testpieces with a range of surface conditions have been tested to
investigate the effect of surface finish and surface residual stress on the crack initiation
properties of a medium strength steel. The results of these tests are
summarised in Fig. 4. There is a tendency for rougher surfaces to lead to shorter initiation
times, but within the overall databand established with the standard specimen. Residual
surface machining stresses do not appear to have a consistent influence on SCCI times for
these steels, and this is probably due to the role played by pits in the initiation process and
their depth relative to the depth of the residual surface machining stresses (typically <100 and
-70f..lm respectively).
Weldments
In addition to the work on parent materials, the SCCI properties of weldments have been
investigated. Coarse grain regions of heat affected zone (CGHAZ), even in post weld heat
treated (PWHT) welded joints, can exhibit higher hardness and therefore be potentially more
prone to SCC than adjacent parent and weld metal microstructures. However, while the
density and types of inclusion in the CGHAZ are identical to those of the parent steel, the
number of surface breaking non-metallic particles associated with this region can be relatively
low (since the surface area of CGHAZ exposed to the environment is low). Moreover, the
maximum stress which can be imposed on the CGHAZ of PWHT weldments is limited by the
properties of the lower strength parent and weld materials, and is consequently at a lower
fraction of RpO.2 than in those adjacent microstructures
2
. The probability of stress corrosion
microcrack nucleation and growth in CGHAZ microstructures is therefore
not as high as might first be anticipated. This has been demonstrated for a conventional
3%NiCrMoV and a superclean steel [8].
SCCI tests have been performed on simulated CGHAZ and cross weld specimens. The
simulated CGHAZ microstructures were developed to represent the transformation product,
grain size and hardness of those characterised in actual PWHT weldments fabricated from the
two steels. The SeCI resistance of both the conventional and superclean homogeneous
CGHAZ conditions was less than the equivalent properties displayed by the respective parent
steels. In contrast, this higher susceptibility to SeCI was not evident from the results of cross
weld tests performed on specimens taken from the characterised weldments. In these tests,
2 In practice, the applied stress is significantly less than the proof strength of the parent or weld materials due to
the location of the weldments, being only..; 100MPa.
50
S.R. Holdsworth
actual PWHT CGHAZ regions were no more prone to crack initiation than the adjacent parent
and weld metal structures [8].
Influence of Environment
Oxygen Concentration
In response to the findings of an EPRI survey of US service experience in the early 1980s
[10], a series of tests were mounted to investigate the effect of oxygen concentration in pure
steam. At that time, the evidence indicated that plants with feedwater oxygen levels normally
below 5ppb and never above 15ppb did not develop SCC, whereas the most widespread and
severest cracking was observed in plant with the highest oxygen concentrations, ie. BWRs
with 20 to 30ppm 02-
A series of tests was therefore perfonned on one steel in pure condensing steam at 95C
generated from feedwater with oxygen concentrations of (i) <1Oppb, (ii) 100ppb and
(iii) 20ppm. Conditions (ii) and (iii) were achieved by passing deoxygenated feedwater at a
controlled rate through silicon rubber tubing surrounded by oxygen, prior to entry into the
boiler. The required concentrations were achieved by controlling the rate of oxygen diffusion
through the tube wall.
The results of this study indicated that there is not a significant effect of feedwater oxygen
concentration in pure condensing steam on sec growth rates (Fig. 5). The crack growth
mechanism exhibited by the 3'li%NiCrMoV steel exposed to both the low and high oxygen
containing pure steam environments is consistently intergranular.
Increasing oxygen concentration from 7ppb to 20ppm leads to a change in corrosion
behaviour from one of discrete pit formation and limited broad front oxidation to one
increasingly influenced by broad front corrosion but still involving pitting attack and
associated blistering. However, this change in feedwater oxygen concentration does not
appear to have a deleterious effect on seCI resistance (Fig. 6).
Two Shifting
Activity during the 1970s and early 19805 concentrated on constant environment testing to
cover the base load turbine operating mode typical of that era. Base load turbines were run for
long periods of time and only shut down in the case of malfunction or for inspection!
overhaul. The operating environment therefore remained essentially constant for a large
proportion of life and turbine components only experienced aerated condensing steam
conditions once every several thousand hours. With the increase in two-shift operation in the
UK, tests were mounted to examine the implications of this change in practice on the risk of
SCC. In two-shift operation, sets are typically run during a 16h peak demand period and
. switched off for the remaining 8h of the day. During off-load periods, air is admitted into the
turbine establishing aerated condensing steam conditions for several hours on a regular daily
basis. Pitting is known to develop at a greater rate in warm moist air at -70C than in low
oxygen steam at higher temperatures, and to act as a precursor to sce.
In order to represent this mode of operation, pure low oxygen condensing steam was fed into
the chamber for a period of 16h. A timing device then (i) switched off the tube boiler heater,
(ii) switched on thermostatically controlled strip heaters attached to the side of the test
51
S.R. Holdsworth
and (iii) activated a solenoid valve which ventilated air to the test vessel. For the
next 6h, de-ionised deoxygenated water trickled into the aerated test vessel maintained at a
temperature of 70
D
e. At the end of the 24h cycle, the timing device (i) switched on the tube
boiler, (ii) switched off the strip heaters and (iii) closed the air inlet solenoid valve.
In the two-shift tests the stress corrosion cracking mechanism involved a significant
transgranular component, consistent with published observations from tests in aerated
condensate [1 I]. Nevertheless, the see propagation rates measured in the two-shift steam
environment were within the databand established for continuous pure steam (Fig. 5).
Pit nucleation and growth kinetics were enhanced in the two-shift environment, although this
did not appear to affect the SeCl resistance of the steel (Fig. 6).
AmmonialHydrazine Treatment
Ammonia is added to boiler feedwater to control pH within the range 9.0-9.6, and to thereby
minimise the rate of corrosion of steel components predominantly in the boiler but also in the
steam path [10). Hydrazine is added to scavenge oxygen and act as a corrosion inhibitor. The
effectiveness of the ammonialhydrazine all-volatile treatment (AVT) was investigated in both
. continuous and two-shift steam tests.
For these laboratory tests, feedwater was produced in the standard way, but with 03ppm
ammonia and 20ppb giving a pH of -9'. Crack growth rates measured during the
course of continuous ammonia/hydrazine treated steam tests are similar to those observed in
untreated tests (Fig. 5). However, the SCCG rates measured in treated two-shift tests are
significantly below the lower bound plateau growth data line for NiCrMoV steels (Fig. 5).
Pitting susceptibility was markedly reduced by treating steam with ammonialhydrazine, and
the treatment has a positive effect on the crack initiation resistance of 3-3:1z%NiCrMoV steels
(Fig. 6).
Increased Conductivity Levels
Most leading turbine manufacturers recommend that the cation conductivity of power station
steam is maintained at -0. IJ.lS/cm with the result that the incidence of pitting in service is not
a major problem. Nevertheless, higher conductivity levels may occur due to fault conditions
and limited transients. Such conditions were investigated by laboratory simulation (but on a
continuous basis) by metering a O.8ppm sodium sulphate solution into the steam line as a fine
droplet spray, immediately after the boiler. The doping rate was controlled to give a bulk
condensate conductivity of (with an estimated specimen surface condensate film
conductivity of -4f..lS/cm). A version of this environment buffered to pH9 by the addition of
ammonia was also generated.
Measured values of SCC growth rate in the high conductivity environment fall within the
established databand for 3-3lh%NiCrMoV steels exposed to a pure steam environment. There
is not a significant effect of increasing the cation conductivity to 0.5JlS/cm on long crack
propagation rates in (Fig. 5).
52
S.R. Holdsworth
On the other hand, the SCCI resistance of 3Y2%NiCrMoV in low oxygen condensing steam is
significantly reduced by increasing cation conductivity, irrespective of the level of pH (Fig.
6). Macrocracking is observed after -lO,OOOh exposure in' steam with A::=
compared with>16,OOOh for this and other: steels with the same strength level exposed to a
pure steam environment.
Molybdenum Disulphide Contamination
Molybdenum disulphide was widely used as a lubricant for turbine disc-on-shaft constructions
until its deleterious effect on SCCI properties was recognised during the 1970s. In laboratory
SCCG tests, the effect of MoS
2
contamination has been investigated by wedging impregnated
fabric in the roots of the machined notch pre-crack starters. Using this approach, long crack
growth rates were apparently not significantly affected (Fig. 5), possibly because the
contaminant could not be physically located at the pre-crack tip.
In the SCCI tests, MoS
2
impregnated bandages were tied to the specimen gauge lengths, and
in these circumstances the contaminant had a markedly deleterious influence on crack
initiation properties by promoting early and deep pitting (Fig. 6).
Concluding Remarks
The stress corrosion crack initiation and growth properties of a number of 3-3 V
steam turbine disc steels have been characterised by the results from an extensive programme
of long term tests (up to lOO,OOOh) performed in a pure low oxygen con4ensing steam
environment representative of service.
In this environment, the influence of variations in material, surface and steam conditions on
the sccr properties of low to medium strength steels depends to a large extent on their effect
on pitting characteristics and, in particular, the characteristics of those pits formed at the sites
of surface breaking MnS bearing inclusions. For a given strength level, the probability of
seCT times approaching the established lower bound increases with the proportion of MnS
bearing inclusions in the steel and with surface roughness. Increasing cation conductivity and
contamination with MoS
2
reduce the resistance to SeCI below this lower bound. The
probability of SCCI increases with increasing strength level.
For a given strength level, seeG properties are relatively insensitive to melting practice and a
variety of steam based environments. Growth rates tend to increase and K[scc decrease with
increasing strength level.
References
D Kalderon, Proc. Inst. Mech. Engrs., 1972, 8, 341.
2 R P Harrison, D de G Jones & D V Proc. Conf. Mechanics and Mechanisms of
Crack Growth, Churchill College, Cambridge, April 1973, Paper 7.
3 D V Thornton, P B Mould & E C Patrick, Proc. Conf. Grain Boundaries, Jersey. April
1976.
4 B W Roberts & P Greenfield, Corrosion NACE, 1979,35, p402.
53
S.R. Holdsworth
5 S R Holdsworth & G Burnell, Proe. COST Conf. High Temperature Materials for Power
Engineering, Liege, September 1990, Ed: D Coutsouradis et al.
6 S R Holdsworth, G Burnell & C Smith, Proe. EPRI Conf. Superclean Rotor Steels,
Sapporo, 1989, 1.1, 3.
7 S R Holdsworth & D V Thornton, Proe. Jaffee Memorial Symposium on Clean Steels
Teelmology, ASMrrMS Materials Week, Chicago, November 1992, Ed: R Viswanathan,
]97.
8 S R Holdsworth, M. Nougaret & B Vittemcnt, Proc. EPRI Workshop Clean/Superclean
Steels, London, March 1995.
9 R J Lindinger & R M Curran, Proe. ASMEIlEEE Power Generation Conf., 8t Louis,
October 1981.
10 F F Lyle & H C Burghard, EPRI Report NP 2429-LD, June 1982.
11 A McMinn, F F Lyle & G R Leverant, Corrosion NACE, 1985,41, 9, 493.
54
S.R. Holdsworth
1..Q6 ,..--------------------------,
3-3.SCXNiCttv1JVsteels, pure steam,
<E 1.07
E
E-
1E-08
:::
o
c
u
u
\/:; IFr09
no grcMthdata
o
@o
o
lowstrength - open points
mediumstrength - greypoints
high strength - black points
100 90 70 20 30
lE-lO ---:0;;.... --'- ----'
o 10
Fig. 1 The effect of strength on SCCGproperties for 3-3Y2%NiCrMoV steels in
pure low oxygen condensing steam at 95C
5.0 ,..---------:-----------------------,
4.0
3.5mOM>Vsteels,
Pure steam,
o lowstrength
omediurnstrength
4O,(XX) 35,<m 30,(0) 25,(0) 2O,<XXl 15,00:> 10,(00 5.(0)
0.0 ______'______'_____'___---'
o
1.0
E
E
Z
Q 3.0
V)
E-
><
20
u
EJ<KSURE1UvlE, h
Fig. 2 Stress corrosion crack extension with time in two 3V2%NiCrMoV rotor
steels exposed to pure low oxygen condensing steam at 95C
55
S.R. Holdsworth
1200 ,----------------------------,
3-3.5o/oNiCrrv1oVsteels, (japp -
Pure low02
.. ,
,
,
,
,
. ,
. .. '
. ......,
"-
,
.....
'
. ..
0
a 0
0
macrocrack fonnation - solid points
no macrocracking - open points
800
1000
jOO,OOO
10.000
600 '-__---'-__"'------L-----'''---'...........--O-.......... ---L-_----''---'----'-----'----'--''-.........
1.000
TIME,h
Fig.3 The effect of strength on secr properties for steels in
pure low oxygen condensing steam at 95C
100.000 .- --,
o o
3-3.5OfcNiCrMlVsteels, condensing stearn, 95C, cr.w-1%.2
(crack initiation - solid points; 00 rmcrocracking - op;:n points)
..s:::
uS
:E
1= 10,000
o
U
if)
10000boun:iJor pure lowU2 steam

1.000 '- --'

Fig.4 Influence of surface condition' on SeCI times for a medium strength
3!/2%NiCrMoV steel in condensing steam at 95e
56
S.R. Holdsworth
1&06
..:::
.......
E
1..07
::t
1E-08
f-
:s
0
et:
'0

U
U
!JJ
lE-lO
datahandfor pure lowO

2
steam
3-3.5o/oNiCrMoV steels, condensing steam, 95C

...t::
\l,)
\l,)
lrl 0\
Cc
+-' t::
t::
::r:: ::c:
.S
t::
I:: s:: ..c .-
......
\l,) \l,)
__ N
0.. 0. ee
\l,)
.-
Cc
00
00 00 ::
=l-.
S'
S ;;>-.
>,
:.a
Q)"'O
s
><:
x >c: 00>'
::>-' u u
0
0 0
r/)

....... .......
I
r.IJ r.IJ
+-'
.t:J
.D
8
0
:i. :::t
::::::
0.
o .-
o'2
0
0.
0.
0.
8
V) V)
u
0.
0
0.
0 0 0 N
0
0
0 0.8 . 6 r.IJ .......
........ N
58
t'"'"" 8
11 II
0
........
ro -< -<
Fig. 5 Influence of environment composition on plateau SCCG rates for a
mediwn strength steel in condensing steam at 95C
too,em ,......--------------------------.,
-=

o o o o
o
lowzrbotm:JJor pwe lowU2 steam

3-3.5o/oNiCr:MoVsteels, condensing steam, 9SOC, (japp RJJo2

crack initiation - solid points; no macrocracking - open points
1,000
...t::
Q)
Q)
lrl 0'.
t::
....... ::::::
t::
::r:: ::c:
0
t::

:::::: 'N
..c 'N

v 0. 0.
\l,)
00
(I)
ro
::::::
00
00
::
I:: l-o

E E
;;>-.
>.
>-.
:E
(1)""0
s
>c:
x
x 00>'
a::>- u u
0
0
0
00 >,...t::

....... ....... ro
..c
iZl r./J
.......
.t:J
.t:J
S
0
::::1. ::t
c
0..
0'-
Cf.l._
0
0.
0.. 0.
E
0
t::
V) V)
u
0.
0 0.
0
0 0
('I
0
o.E
e
-
0
0
II
rn
N
o.E
a
II
0
Sro ro <: <:

Fig. 6 Influence of environment composition on SeCI times for a medium
strength 3Yz%NiCrMoV steels in condensing steam at 95C
57
S.R. Holdsworth
Figure 2 View showing simultaneous welding of 8 rows, of 20, on a BWR, BB381 rotor
58
S.R. Holdsworth
Figure 3 View showing final polishing after PWHT and prior to side entry machining
on a BWR, BB381 rotor
59
60
Management of Rotor Disk Rim Stress Corrosion.Cracking
Presented By
Tony Lam
Stress Technology, Inc.
61
62
Management of Rotor Disk Rim Stress Corrosion Cracking
Ton)' C-T Lam
Stress TecbnologJ Inc.
Rochester, New York 14623
Robert P. Dewey
Stress TechnologJ Inc.
Thomas H. McCloskc)'
Palo Alto, California 94304
Abstract
This paper discusses a process developed for selecting bem'een the long-term and short-term strategies
currently offered to the industry, and the limitations associated with each strategy when dealing with the
distinctive features of see in the disk attachments of LP turbine rotors. Presented within this approach is the
detailed relationship between chemistry, materials and the types ofanalysis from which repair or replacement
strategies may be assessed. Further development ofthis approach is discussed to transform it into a practical
tool and guideline for plant managers to rapidly identify an effective run-repair-replacement strategy best suited
to their specific unit.
1. Current Challenge Faced by the Industry _ _ _ .
2. Mechanisms Distinctive to Disk Attachment SCC _ _. " 2
3. Five Stage Involved with SCC in the LP Rotor Disk Rim _ _. 4
4. Limitations of Long Tenn Solutions Currently Offered to the Industry __ .. _ _ _. 5
5. The Challenge to Assess Short Term Strategies _ __ . _ _ _ 7
6: Recommendations _ _ _. . . . .. 11
7. References _ _ _. . . . . . . . .. 12
63
64
1. CURRENT CHALLENGE FACING THE INDUSTRY
A survey of the U.S. nuclear industry has recently confinned that stress corrosion cracking (SeC) in low
pressure steam turbine rotors is emerging as one of the most serious threats to the reliable operation of large
turbine generator units. The EPRI PSE survey documents a massive industry movement presently underway
to replace the original shrunk-on disk type rotors with monoblock and welded drum-type rotors. Of the 109
active units, 27 reported that they have completed or are planning LP section replacements, precipitated by the
development ofSCC in the rotor wheel bore, in keyways and/or in the disk attachments. Using a conservative
cost of$7.5 million dollars per rotor, it is estimated that over $200,000,000 in replacement pars alone has been
spent to date coping with the rotor SCC problem. [IJ
Cracking in the disk-rim attachment remains as a distinctive SCCtype ofproblem - As early as 1986 it was
predicted that the cumulative fraction ofsteam turbine rotors with cracked disks in U. S. nuclear power stations
would reach over 90 percent within their envisaged lifetime of 100,000 operating hoursyJ Despite these early
warnings, reliance on metallurgy to identify more corrosion resistant materials has clearly proven to be
inadequate in preventing stress corrosion cracking of the keyway and disk-rim attachment regions of these
original shrunk-on rotors. Wh{le the newest generation of replacement monoblock and welded drum-type
designs have conveniently eliminated the keyway as an issue, cases of sec in the disk attachments are still
being reported. [3] Principally affecting those stages which operate in the Wilson line region of the turbine,
sec in the disk rim has been reported on all three of the most common blade attachments; straddle mount
dovetail, pinned finger, and fir-tree axial entry types. [4] [5]
Time is a principal constraint which limits the effectiveness ofcurrent analytical tools to deal effectively
with disk-rim sec - In connection with this massive retrofit effort now underway, the more immediate
challenge is to select and schedule effective short-term strategies which manage the see in the disk-rim
attachments once cracks have reached a detectable length. Without sufficient time to prepare the sophisticated .
finite element, fracture mechanics and probabilistic models needed to calibrate the risks and assess the damage
tolerance, managers in charge are often forced to make snap judgements and assume unknown risks in
attempting to avoid the substantial penalties associated with the unscheduled downtime required by many
repair strategies.
Can another massive replacement effort be avoided? - Unlike the keyway problem, elimination of the disk
rim attachments is not a viable solution. If no measures are undertaken to identify effective preventive
measures or treatments for sec in the disk-rim, and should this new generation ofreplacement rotor materials
prove to be just as susceptible to see as their predecessors, the industry could find itself facing another
massive rotor replacement scenario in 7-10 years. In the wake of deregulation current predictions anticipate
an average of 15% of base load capacity units to be displaced by the IPPls alone.[6] Faced with capital
expenditures of $20-$30million dollars per unit to once again replace the LP rotors, the decommissioning of
the Trojan plant could be foreshadowing of the fate in store for these units if adequate measures are not
undertaken now to address the mechanisms which are distinctive to disk rim stress corrosion cracking.
65
Tony Lam
2. MECHANISMS DISTINCTIVE TO DISK RIM STRESS CORROSION CRACKING
Presented at the most elementary level as shown in figure 1, intergranular'Stress corrosion cracking can be
made to occur from pits when alloy steels are subj,ect to tensile stresses in a corrosive environment.
Similar to a keyway, the disk-blade root
interface geometry affords an opportunity for
condensates within the steam to p r e i p i t t e ~
collect and concentrate in regions ofhigh stress.
Similar to a keyway, rim cracking re-introduces
the risk ofa catastrophic failure-forced outage
shutdown.
Figure 1: Contributors to sec Damage
However, the failure process due to sec becomes more complex where features ofthe disk rim involve
additional mechanisms which have been demonstrated to influence crack initiation and crack propagation:
Disk rim attachments involve the contact of dissimilar blade and rotor materials which have the
capability of creating an electric potential. Results from tests perfonned by Karato et.aJ. [7) indicate
that this electric potential can accelerate the surface oxidation and corrosion process, and so promote's
the formation of sec. (See Figure 2)
Unlike a keyway, the disk rim attachment experiences a vibratory stress superimposed from the
rotating blades. Results from further tests published by Karato et.a!' [7J reveal how a superimposed
dynamic stress can further reduce sec initiation life. (See Figure 3) Dynamic stress is also the
critical factor which detennines the remaining safe operating life of the rotor.
Unlike a keyway, the disk attachment is more directly exposed to the steam flow. Results published
by lnagaki et.a11 {SJ suggest how oxygenation treatments performed to protect turbine boilers may also
promote the initiation of sec in the turbine attachments. (See Figure 4)
Unlike a keyway, stresses in the rim attachments are typically higher and often localized yielding in
the,notch radii occurs. Results published by Nesbitt et.a!.19] demonstrate how repeated load cycling can
reduce a materials original resistance to SCC. (See Figure 5)
These examples of published research into sec of rotor steels suggest that features of the disk rim involve
several distinctive mechanisms each of which requires some consideration before any of the solutions now
available in the growing arsenal ofcompetitive strategies can be acknowledged as being truly effective. What
is less clear to decision makers is how to best to discriminate from this information as to what is relevant in
tenns ofevaluating the effectiveness ofcompeting' strategies, and howtest results can be related in a consistent
manner to best judge which options are still viable. To date there has been no standard methodology or
guideline which fonnally organizes the management of disk rim sec as a comprehensive procedure.
66
Tony Lam
IGSCC TGSCC
o
o
()
General Oxidalion
Corrosion
'2l
--@-- HOJH2 Equll.ibrlum Uno at 1300C
-@-- Oz IH20 EquWtltlurn line at 130C
_-_._-. equilibrium line 81 100"C } [EPRI ]
---- Equlllbilum line at 200C Fe-H,O system Repoll
O.s.r-:-,
...." ..
.0.5
w-
:I:
en
'" >
Figure 2: Effect of Applied Electric Potential and PH on
Crack Initiation [7]
Results of these experiments by Karato, ltoh and Kondo
suggest that crack nucleation and crevice development occurs
only under specific electrochemical conditions. The potential
charge created by dissimilar metals in a wet steam
environment may therefore playa significant role for sec in
disk rim attachrncnts.[7]
pH al T
Slress ampUlUde Q. (MPaJ
100
Figure 3: Effect of Superimposed Vibrational Stress on
Crack Initiation [7]
Further results of experiments conducted by Karato et.al.
also reflect how a vibrational stress amplitude superimposed
on a steady stress can drastically decrease the crack initiation
period from that without a vibrational stress.
Applied StN!!;S Ratio
1.0 y .S.= Cracked!: liD Cracking
J ,7 Y.5. : Cracled n: 110 C..-acl;"9
Figure 4: Relationship Between Oxygen Content and SCC
Initiation on 2-3NiCrMoV Steels [8]
-
W,ooo -
30.000_--":"-------.,
....

10.000 -
)NiCrMc>V 5tee I
Test results by Inagak,i, Miyazaki and Kashiway also show
howthe oxygen content ofthe surrounding environment can
influence the see initiation time for 3% NiCrMoV rotor
steels. This suggests that the benefits of oxygenated water
treatment for the boiler should be weighed against the
potential damage caused to the disk rim attachments.
u
u
o ... p-pra----'
(Deaerat ton l
i
Defect ' Defect

I
... Cyclic Load en 120-

Figure 5: Effect of Stress Hold Time on the Reduction of
Q,
'0 0 > Static Load 00-

Crack Initiation Cycle 19}
"$.
l>
..
100-
-
<X
..
00-
..
ae.
Test results produced by Nesbitt, Beech and Howard show
iii
l
eo
-
how specimens at cyclic load levels of 90%-100% of their
ii
Co
yield stress failed at 10
3
_10
4
cycles on load. Under static
<:
6Q
<>eo
load conditions only one defect free specimen failed. Pitting
-
....
was generally more severe in the cyclic load specimens than
40
10 1lF 10
3
10' 10'
the static load specimens.
67
Tony Lam
3. FIVE STAGES OF sec INVOLVED IN AN LPROTOR DISK RIM
To best evaluate the effects of see on disk attachment for the convenience of selecting effective repair-
replacement strategies, the failure process itselfcan be separated into five distinctive stages as shown in figure
6. The five stages are summarized as (1) incubation, (2)"initiation, (3) stable crack growth,(4) accelerated
crack growth, andfinally (5) fracture.
NewRotor
.lncubatiC)ll
a='O
Degredation

8>10 mils
sec
Slable:Grcwtb
daldt =constlnt .8 =100";500mils
FRACTURE

Figure 6
Five Stages of Stress Corrosion Cracking In 3 Disk Rim
Incubation and initiation relate directly to the operating environment. They involve chemical mechanisms
which promote the slow process of degradation, causing the rim attachments of a new rotor to eventually
become pitted over time. As tiny pits form, sometimes within the crevices between grain boundaries, they
create a stress singularity at the bottom ofthe pit. Depending upon its location, this stress concentration may
or may not be sufficient to result in the formation ofa micro-crack. The governing mechanism for both these
steps is prIncipally steam chemistry which controls the process until the crack reaches about 10 mils.
At this point the dominating mechanism changes to stress corrosion; crack propagation under a combination
of steady stress and chemical attack. This third stage is the most recognized and best understood by utility
engineers in terms ofprevious experience drawn from treating rotor keyway sec problems. During this stage
crack growth rate is stable and can be predicted using da/dt curves obtained from material tests. The principal
factor influence crack growth is material ductility (yield strength) and steam temperature. Stress has a marginal
impact on sec growth. This portion of the failure process can take anywhere between 5-50 years.
Because the disk rim must contain features which allow for the attachment of blade rows, a fourth and less
predictable stage involves the eventual take-over of the growth process by low cycle fatigue due to start-stop
operation ofthe unit, and high cycle fatigue from vibratory stresses imposed on the attachments by the blades.
The dilemma posed to the decision maker is what to do when cracks are first discovered in the disk rim. In
accordance to thetime line, the operator must best detennine howclose the stable grovvth process is to entering
the accelerated growth stage. For disk rims, this transition win occur when the dynamic stress (a function of
excitation, damping and blade frequencies) exceeds the threshold intensity factor (defined by material
tests) at the crack tip. Accelerated growth generally takes over when the cracks reach between 100-500 mils
depending upon the combination of cited variables.
Ifleft untreated the cracks will eventually result in fracture, with the potential consequences of a forced outage
due to blade failure when redundant features such as multiple fingers or hooks are overloaded by the
redistribution of stress.
68
Tony Lam
4. LIMITATIONS OF LONG TERM STRATEGIES CURRENTLY OFFERED
How does this time line related to assess and choose from the long term and short term strategies currently
offered to deal with sec in LP turbine disk rims? Figure 7 presents to most commonly offered solutions to
show at which step of the failure process they are most likely to have an impact.

cs#O
' .l..
1-. ---.
... - . .
..Degredat.ion
oc'",n,itiati(
. '1l>10mlls
.HCForLCF
'.

_. - .-l.
.. - ,. --::
... -.-
. ... - ..
FRACl1JRE

Figure 7
Measures Available to Prevent see and Extend Rotor Service Life
Long-term solutions currently offered to the utility industry include:
(1) Purchase of monoblock or welded drum-type rotor replacements.
(2) Applications of coatings and shot peening to resist or inhibit the initiation of pits and cracks.
(3) Rebuilding of selected attachments with a high chrome (12%), corrosion resistant material.
(4) Replacement with rotors comprised of "clean" steels
Capital investment in a long term strategy, particularly rotor replacement is typically predicated on the
guarantee that one or more of the following has been applied to compensate for a deficiency of the original
design:
.. Use of rotor forgings with higher nickel and chromium contents, and lower yield strengths.
.. Application of shot peening in sec sensitive regions .
.. Use of attachments which have beenTedesigned to reduce attachment fillet stresses
It is important to recall from the failure sequence that the effectiveness of any of these strategies is best
measured by how they resist the incubation and initiation of cracks. This is achieved by controlling the rate
in which surface deterioration occurs. In terms of improved metallurgy, nickel content in rotors forged after
the mid-1980's has been purposely increased from 1% to 3.5%. Test results published by Inagaki et.al. shown
in Figure 8 however, indicate the increase in nickel appears to cause only a limited benefit on crack growth
rate beyond a certain tensile strength (about 70% ofyield).18]
69
Tony Lam
Apcl;f"d S'tt'e'u RbtlC
Ct.1 "I'.S :tnc:k.ed: 0 hOC....,1Cl11f'1g
I.e '.s... :C..
1.2 '.'S.
to PS- (b.ne" CO"t.&1 .. 5telnt
l..-prrdurr :
20'_[ I
.
Tl:ft'S1;f!$lrtngtl'lof
"Clean" steels focus on the reduction/elimination oftrace

elements from the rotor forging, Chrome content is
unchanged. To date no comparison has been made to
assess how well any of these improved materials deal
with the aforementioned issues associated with stress,
pitting, electric potential, or oxygenation water
treatments, K
rscc
, daldN etc.
I
Ter.,lle cf
10.000
o
Figure 8: Relationship Between Nickel Content
and SCC Initiation Time (8J
Rebuilding of selected attachments with high chrome
(12%) weld material represents a most recent repair
solution. Ifthis remedial action proves an effective one-
time fix, it could also represent a cost effective retrofit
solution for the newer generation of replacement rotors.
200 240
[MN'm-
3/2
]
100
stress intensity, K
1
10
12

o
stress intensity, KI [kSi"/fr;' ]
o 50 TOO 150 200
TO-. l--T--.....L,-..,..--,..;..-.,......,--r'.....,...-,--+--,
steam t ...bine ,otor st.els.SeC.... H20
.3%Ni sleel
10-. 2Cr - lNi steel
.. "clean steels", speciaiy alloyed sleels
E
Figure 9: Effects of Stress Intensity on SCC
Growth Rates of Different Rotor Steels
The long term effectiveness of any strategy which reiies
principally on reducing stress in the attachments is
anticipating benefits for resisting or retarding the initiation
of stress corrosion cracks. Material test results published by
Magdowski and Speidel (Figure 9) confino that unless the
stress intensity is kept below 10-20 ksi,(in, the effect of
lowering stresses will be of marginal benefit. For most disk
rim attachment designs, such levels are impossible to achieve
once the initial pits are formed. 1111
Coating and shot peening are offered to the industry as
lower cost preventive measures which can be effective if implemented at stage 1. In regards to extending the
incubation period, it is not apparent whether the initial advantage ofa more resistant compressive surface layer
caused by shot-peening is outweighed once a pit penetrates to the subsurface tensile layer, e.g the extended
incubation phase is compromised by an acceleration ofthe crack initiation phase. The use ofsacrificial coating
materials is expected to represent an attractive option for
reducing sec. Currently however, no OEM offers or
promotes the use of coatings as a method for providing
further see resistance. Despite this, based on original EPRI
sponsored research utilities such as Southern California
Edison have regularly coated LP turbine blades for over a
decade [Ill]. In subsequent inspections it has been noted that
further see in the disk attachments appears to have also
been inhibited from treatments to the blades. This
consequence is due to the reversal ofthe electric potential of
the disk with respect to the coating materials ofthese blades.
70
Furthennore, reducing fillet or notch stresses
presumes that stress-corrosion cracks will
initiate and grow within regions of maximum
stress concentration. As shown in figure 10
a recent experience with sec in the root
attachments of several fossil units revealed
how cracks had initiated in the disk finger
ledges, and not in the pin holes where stresses
are 2.5 times higher. There are aJready early
indications that the long term mechanical
reliability of monoblock and welded-drum
type rotors may be compromised if too much
reliance is placed on reducing stresses and
yield strength, while other possible
mechanisms are discounted or ignored.
Tony Lam
Figure 10: Distribution of stress in disk rim finger
attachment which experienced sec
5. THE CHALLENGE TO ASSESS SHORT TERM OPTIONS
Short-Term Strategies available to the industry typically involve one or more of the following actions:
(1) Operating the rotor for a limited period of time until a major LP turbine overhaul is scheduled.
(2) Attempting to remove the cracks without significantly increasing the stresses in the notch radii.
(3) Installing pressure plates, thereby sacrificing performance for the sake of reliability.
(4) Installing spacers over the notched portion of the rim to reduce centrifugal stresses at an initiation site
(5) Installing drop-notchllong-shank blade replacements on damaged disks
The immediate challenge facing most plants is to determine just how far into the third stage of crack growth
the process has advanced. It is at this point where analysis is required and is cost effective ifapplied properly.
A remaining life assessment (crack initiation to accelerated crack propagation) of the disk rim requires
supportingdata ofstresses, frequencies, and material properties. The viabilityofrepair or replacement options
are detennined using a combination of both detenninistic and probabilistic analyses ofthe disk attachments.
However, while any of the listed strategies might extend the short-tenn operating life of the rotor, each has a
significant economic penalty associated with it. This penalty is worsened if the affected rotor cannot be
returned to service until a time when the maintenance can be scheduled.
As shown in Figure 11 a comprehensive engineering methodology can be added to the original failure time
line which shows how to integrate the distinctive mechanisms found within the three principal areas of
chemistry, metallurgy and analysis, in order to assess the viability of strategies which are currently offered to
the industry. The organization ofthis approach can also help to guide decision makers through the confusing
procedure of what key parameters are required and how best to use them to assess short-term decisions
. associated with disk-rim stress corrosion cracking. Analytical models of the disk rim provide a focus for the
input of rotor-specific details such as material properties, NDE test results, rim attachment geometry and
dimensions, blade-attachment tolerance fits, blade-disk frequencies ~ unit load cycling history.
71
Tony Lam
Figure 11: Organization of Analysis For Supporting Disk Rim see Decision Management
I

1
1

I
1.400
I.JOD
0.20D
1.100
0.050
I.OJD
O.OUJ
0.005
Allowable
Flaw Size
Figure 12: Example of contour plot which maps
allowable sizes of surface defects based on calculated
stresses for an assumed number of cycles
Fracture mechanics is used to identify limits where the
critical crack size is approached, and accelerated
Featured .in the approach is the integration of the three types of analyses most commonly employed; stress
analysis, fracture mechanics andprobabilistic analysis. Stress analysis is needed to provide essential details
of disk attachment steady and dynamic stresses, and for some attachment types to assess the, influence of
tolerance fit. The blades themselves must be included to account for the three dimensional nature of the
attachment stresses and to reflect their impact in subsequent fracture mechanics or probabilistic studies. FE
results such as those shown in figure 12 also provide the plant with a map for pre- and post NDE inspection
, where allowable flaw sizes based on growth due to low cycle fatigue growth or see are identified for an
estimated period or cycles of operation. Similar "maps" can be provided to prescribe limits for fillet
enlargement, material removal ("skim cutting") and other remedial maintenance activities for specific stage
and root attachment types. Finally, the blade-disk
model is used to ensure that any alterations to the
original attachment do not significantly affect the
original blade-disk natural frequencies. If a resonant or
near resonant condition is created, the fai lure process
will move into the accelerated growth mode. To fairly
compare the merits of competitive replacements an/or
the sec behavior for different attachment geometries,
attachment stresses should be derived using a consistent
analysis program and procedure. The current version of
EPRl's BLADE-ST program is ideally suited to
perfonn this role by eliminating the subjectivity of FE
mesh density, and selection of boundary conditions.
72
Tony Lam
gro"'th is predicted. The operator would be advised to effect a removal for cause decision, i.e. either remove
the blades and install pressure plates, or remove the entire rotor for off-site rim repair. Pursuant to this
requirement, the FRANC program has been used to extend the original BLADE-ST stress analysis by
modeling the crack gro'Wth process within the disk attachment as shown in figure 13. Key features of the
FRANC program include a capability to handle multiple crack locations within the multiple fillets of typical
root attachments, account for possible interaction of existing cracks as they propagate, account for the
influence ofa three-dimensional non-unifonn stress field that exists in most attachment designs. and iteratively
adjust the cracked attachment model to represent the change in geometry caused by the enlarging crack, and
recompute the stress field accordingly_ [4][5J[12[1 3 Unlike conventional FM models, FRANCtakes into account
the complicated stress field and load redistribution which occurs when a crack propagates in a disk rim.
Prediction of remaining life is accomplished in three steps, using results such as those illustrated in figure 14.
Step I uses the FRANC model to calculate the root stress intensity factor as a function of different crack
depths. Curve fitting is then perfonned for each of the calculated values as the second step. From this
curve, the remaining life is plotted as a function of initial crack depth and number of cycles to failure to
complete the process.
Figure 13:
Example showing details offracture mechanics model used to
evaluate fatigue damage to disk-rim attachments
"
;.-- --_.... .... _-
DISK REMAINING LIFE UNDER lCF
.oJ . ,..
iDala .
Curve Fit -
.. ..... ...o..l2 ,., U ox -t,Jf .
CRACK DEPTH (In)
FINGEFI ROOT STRESS INTENSITY FACTO'=!
Figure 14:
Example of stress intensity and remaining life
curves derived using FRANC Model of the disk-
rim attachments
73
Tony Lam
Mont.-Carlo Failure Slmubltlon
.- Failure
20
(E 1
17
15
12
Improbable
<10
4
Probability oflbnsbold Del K
19
14
Figure 15: Probabilitty density histogram for LiK threshold values
and results of probabilistic analysis
o 1.11 3.7'! UJ 7.S ,,,, 'l
DtI K (lCSll..... I1.2)
tto.;...:tijl LowRisk
Acceptable
Probable Occasional Remote
10.
3
1010"" 10
4
to lO'S ufO
frequent
>10.3
_HighRisk
Medium Risk
Critical
ICat IIJ
M_rginal
(Cat III)
Negtibte
fCatfV)
Typically plant operators find themselves trying to decide what the probability is that the rotor is likely to reach
the fourth stage of the failure process. Because ofthe variety ofuncertainties which are imposed on a disk rim
cracking problem, aprobabilistic analysis becomes a necessary step. In addition to the original history profile
of hours and start-stop cycles the operators are required to provide initial estimates of crack sizes at one or
more critically stressed locations on the designated diskirim attachment. Such details can be based either on
unverified projections of crack growth,
from a hypothetical assessment of
possible crack lengths based on limited
inspections ofthe or from extensive
measurements made by NDE
examination.
As previously discussed the influence of
dynamic variabilities such as forcing,
frequency, root fixity, damping etc.
should not be ignored. Figure 15
illustrates the results of a probabilistic
approach used by Lam and Puri
(13
)
which introduced these variables as
random factors into the BLADE and
FRANC models to develop a series of
statistical data bases to account for each
variable from which histograms and
probability density functions of stress
intensity were derived. In tum these are associated with various degrees of attachment cracking. Similar
histograms and PDF's can be prepared for the assembled material properties. The probability of failure is
computed as the intersection ofthe two independent conditions which satisfy the criteria for an in-service
failure, e.g. high stress (steady or
dynamic) and lower material
strength. Using this model,
failure estimates and risk factors
can be optimized by reflecting
the sensitivity of the projected
probabilities offailure to changes
in attachment stresses, assumed
crack sizes, and material
properties etc. To facilitate the
interpretation of probabilistic
results, a hazard matrix for safety
can be used to mandate actions.
An example of such standards
used by the American military
suppliers is shown in Figure 16.
For a steam turbine critical
component a remote frequency of
10-
5
to 10-6 is recommended.
Figure 16: Hazard matrix derived from MIL-Standard 882C(1)
74
Tony Lam
6. RECOMMENDATIONS
The approach presented in this paper has evolved to address the cited trends occurring within the industry, by
performing a critical review of the mechanisms distinctive to crack initiation and growth in the disk rim, and
relating them to both the short-term and long-term strategies which are currently being promoted. The
following objectives are identified which could further assist the present capabilities of the approach for
widespread future management decision making:
1. Material Treatments: To support organizations which have already invested in new LP rotor and section
replacements, further research should focus on the systematic assembly, comparison and evaluation of
advanced materials and treatments which may control or influence the earliest stages ofcrack initiation ofSCC
in the disk-rim attachments ofmonoblock and welded-drum rotors. Effects ofload-cycling, pitting, electrical
potential, and oxygenation which are sec contributors, and distil'lctive features ofthe disk-rim attachment
design should be evaluated for the most commonly used 3.5CrMoV rotor materials. Coated, shot peened and
high chromium weld or clean steel materials should also be qualified to s s s ~ and compare the genuine
benefits ofthese measures.as potential preventive and repair solutions. The results from all these comparisons
should also be tabulated in a consistent format and supplemented with relevant portions of original sec
material properties data bases which have already been developed.
2. Rotor Material Properties: While there is an abundance of da/dt curves to characterize crack propagation
due to SCC, and S-N curves to address crack initiation and propagation from LCF and RCF, better infonnation
is needed to establish the thresholds at which cracks can be expected to initiate from SCC and RCF. Also
more statistical data on material properties is needed to obtain better probabilistic distributions. Whilemany
original rotor materials were selected for their resistance to stress corrosion and corrosion assisted fatigue what
is not well documented is the behavior of these materials after initial cracks or pits are formed. The same
question remains as to how the aforementioned coated components, or the new high-chrome and clean steels
offered today will behave under similar circumstances. Because test data points are often obtained using
methods to accelerate the process, whether these accurately portray long-term effects is questionable. Existing
data needs to be further expanded to include comparisons of pitted versus unpitted specimens, as well as
specimens which have been subjected to overload cycles (preloaded). Other factors such as the influence of
chemical environment, oxygen content and electrical potential also need to be more thoroughly examined to
identify and/or confirm the influence of these contributing mechanisms in the initiation of sec.
3. Disk stresses and blade natural frequencies: When SCC cracks are first detected or confirmed, there
is rarely sufficient time for a plant to initiate the development of sophisticated damage tolerance or
probabilistic models ofthe type described which are essential in guiding a plant's run-repair decision making.
Without these models, the ability to reasonablycalculate the risks associated with a "continue to run" decision
is significantly compromised. The principal criteria offast-resjJonse could be achieved by pre-modeling the
blade-disk attachment regions for the wet stages ofthe most common families ofLP turbines that comprise
the majority ofUS fleets. Stress and frequency results then could be tabulated and organized as data bases to
be used in conjunction with the relevant material properties. Simple.interfaces could further allow operators
to input their individual rotor operating histories and features of the steam chemistry to make see life
assessments machine specific. Following the management approach, the procedures for undertaking damage
tolerance and probabilistic failure analysis of disk attachments could also be formalized to process this data
and provide recommended outage intervals for scheduling and selecting run-repair-replacement options.
75
Tony Lam
4. Dedicated Probabilistic Models - Because ofthe prohibitive cost of removing large numbers of blade rows
to inspect the attachments for cracks there is a need to develop industry guidelines for handling the
uncertainties related to known crack depths and distributions is handled by means ofa probabilistic, rather than
an empirical approach. Current probabilistic analyses of root attachment cracks may be offering expediency
at the cost ofaccuracy if the complex multiple loading andtolerance effects present in the disk rim attachment
problem are ignored or oversimplified. The advantages offered with probabilistic modeling of disk
attachments also may be compromised ifthey fail to account/or the three-dimensional nature ofmost disk-rim
stress fields, e.g. by substituting weighting functions which are applied to stresses computed with a 2D FE
disk attachment model along a designated crack path.
7. References
L Dewey. R. "Turbine Upgrade Guidelines" EPRI TR-106230, Project 3186-46, Plant Support Engineering
Nuclear Power Group, January 1997
2. Jaffee. R. Metallurgical Transactions, Vo117A, 1986
3. Ting, K. "Review o/thefraclure problems at the blade roots and steeples o/the low-pressure steam
turbines in Taiwan Power nuclear plants." 1996
4. Hong,C.,Larn, T., Wan,E. "Remaining Life Analysis a/CrackedDisk Attachment in a Wet Steam Environment",
ASME P\VR-VoI18. 1992
5. Lam, C-T, Omprakash,V., Gruwell, D., McCloskey,T. "Life Extension Strategies ofCracked Disk
Attachment for LowPressure Steam Turbines. " ASME PWR-Vol 26, 1994
6. Parkes,J. "Proactive Operations and Maintenance Plan 'I AES 92111815-2-1, EPRl, March 1994
7. Karato,H., Itoh,H., Kondo, Y. "Studies on SCCfor Steam Turbine Rotor and Disk!! Proceedings, Fossil Steam
Turbine Disk Cracking Workshop, EPR! GS7250, April 1991
8. Inagaki,S., Miyazaki,M., Kashiwaya,H., "Effects ofMaterial and Environmental Factors an Stress Corrosion
Cracking ofNiCrMoVRotor Steels", Proceedings, Disk Cracking Workshop, EPRl GS7250, April 1991
9. Nesbitt,J., Beech,S. and Howard,N. "Slress Corrosion ofNiCrMoV LP Disc and Shaft Steels Under Cyclic
Loading", Proceedings, Fossil Steam Turbine Disk Cracking Workshop, EPRI GS7250, April 1991
10. Ortolano, R. I'Guide For the Use ofCorrosion-Resistant Coatings on Steam Turbine I ~ d e s EPRI CS-5481,
Project 1408-01 Final Report, EPRI, November, 1987
11. Magdowski, R., Speidel "Clean Steels For Steam Turbine Rotors - Their Stress Corrosion Cracking
Resistance" Metallurgical Transactions A, Volume 19A, June 1988
12. Lam,T.,Dewey,R.,McCloskey,T., Wawrzynek,P. "Assessment o/Remaining Life ofa Control Stage
Blade with Solid Particle ErosiC?n Damage", 93-GT-24I, ASME 1993
13. Lewis'!., Roberts,B., Pun.A., Sherlock,T. "Cracking ofHydrogen Cooling Holes in Older Generator
Rotors", PWR-VoI26. ASME 1994
14. Lam. C-T, Puri,A. "RiskAssessment ofCumberland Unit 2 L-O Blades" ASME IJPGC Houston, ]996
76
Management of Rotor Disk Rim
Tony C. T. Lam
Robert P. Dewey
Stress Technology Incorporated
and
Thomas H. McCloskey
March 19, 1997
Current Challenges
Disk Rim Cracking is a distinctive see problem
which involves many mechanisms in the Wilson Line
Stage.
For a short term fix, time is a principal constraint
limiting the effectiveness of gathering technical data
and performing critical analysis.
Long term strategies need to be qualified now to
protect the new generation of LP rotors.
Tony Lam
77
Tony Lam
78
Mechanisms Unique to
Disk Rim sec
Dissimilar metals operating in a wet environment.
Vibratory stress imposed on the blade.
Direct exposure to the steam flow.
High stress concentrations subject to load cycling.
Five Phases of Stress Corrosion
Cracking in an LP Turbine Disk Rim
Current Measures Used to Prevent
see and Extend LP Rotor
Service Life
see Damage Root Cause Analysis
Tony Lam
79
Tony Lam
Organization of Analysis for Supporting Disk Rim sec
Decision Management
O'il . , . . /1"
-
E I I /; I
i
"I
_'____.J

J.(II"l" - 17 PfIlI\.C"O. - 1.......
EfTecl of Tempernturc: on & d"/dl
3.5 NiCrMoV Rotor Material sec Properties
80
Tony Lam
PDF of Measured Root Attachment
Gap Clearances
'" \
\
\
"'I
PDF of Measured Surface Damage
NDE & Blade Condition Assessment
Steady Stresses
I ~ ~
~
Effect of Root Tolerance
0.0 0.5 1.0
G1pin 0.0
41.2 ksJ. 48.] ksi. 48.1 i'si
'Iq> !b:lk. 0.5
49.9 Irs:. 69.0 ksi 68.7. Ic..si
6ail.) 1.0
48.-; i'.s:.. 7j.0 ksi 83.8 k$J.
81
Tony Lam
IRJIl.
-"-JlIlOll.SllI
i I
(-A I
'
_:1....,.,.,.
--..a.'.:lZ

:112:114 .,.
U:l ., ....
.'"

....
-....
I
:::::::; I -loll') .

(:!II.'
Using Analysis to Map Allowable Flaw (Defect)
Sizes for a Weld Build-up 3.5 Ni-Cr-Mo-V Disk
Attachment
Allowable

0.300
0.250
0.200
0.150
0.100
0.050
0.025
0.000
82
Tony Lam
Advanced Fracture Mechanics to Identify
Critical Crack Size
Pin
Disk
Finger
I
I
Finite Element Based Fracture Mechanics
Model for 3D Stress Field and Load
Redistribution
83
Tony Lam'
--=;:.=
I I I I

__
f---i--+ - --to --
DUC I IF"t ttm)tR LeT
.. -t-='...-+--.-
rr--r-t-..
'_\+- I I f
,.,,1: ,I t I I
,.,K = 12453"+ 463 3"'-13173
I I i
.---,-
; t
FINGER ROOT SlRES5
"r INTENSITY FACTOR
... j i
".1: ,I I I
Crack Propagation Model
Showing Remaining Life as a Function of Initial Crack Depth
and Number of Cycles to Failure

.L,.;o-__--_Cun''l"I:I(

...
I : :i ., _:
Mode 3 =348 Hz Mode 2 =256 Hz
Dynamic Variables
Forcing, Natural Frequencies, Damping, Dynamic Stress
Mode 1 =155 Hz
84
Tony Lam
..
11...!ftl".- .. t..oJII,,"w1ftW
Statistical Data on Blade Damping
Statistical Data of Threshold Values
:i.iS 563 7,5 '-).38 11-15 1:l.lJ 15
Dd K (KSI :n'll2)
Probability Of Threshold Del K
1/(1
...
Threshold Del K
50
10.--,,------------,-----,
85
Tony Lam
Results of Probabilistic Analysis
(100,000 Virtual Blade Samples)
/
/
86
Haz;ard Matrix to Interpret Probabilistic Results
MIL-STANDARD -882C
FREQUE<NT PROBABLE OCCASIONAL REMOTE IMPROBABLE

SEVERITY >10" 10"" to 10" 10" to 10"" to" to 10" " 10
I (Al IBI (Bl rOl eE)
...'
CATASTROPHIC
.i ..
':
a 12
lCATl1
CRlTICAL 3
>,:
'6
.
1S
{CAT II) ...
.
.
...
MARGINAL 7 9 11 14 17
ICATml
NEGUGIBLE 13 16 18 19 20
(CAT IV)
Recommendations
Need systematic assemblyI comparison and evaluation of
advanced materials and material treatments.
Need more information to establish the thresholds at which cracks
can be expected from see and HCF.
Need better statistical data for establishing probabilistic
distributions.
Need pre-modeling for databases of disk stresses and blade
frequencies.
Need guidelines for handling uncertainties related to known crack
depths and distributions probabilistically.
Tony Lam
87
88
Remaining Life and Risk Assessment of Low Pressure (LP) Rotor Dovetails
Due to Stress Corrosion Cracking (SeC)
Presented By
Avinash Sarlashkar
89
90
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
REMAINING LIFE AND RISK ASSESSMENT OF LOW PRESSURE (LP)
ROTOR DOVETAILS DUE TO STRESS CORROSION CRACKING (SCC)
Avinash V. Sarlashkar
Tony C.T. Lam
1800 Brighton-Henrietta Town Line Road
Timothy J. Szumski
Arizona Public Service Company
Palo Verde Nuclear Generating Station
P.O. Box 52034, Mail Station 6301
Phoenix, Arizona 85072-2034
91
92
REMAINING LIFE AND RISK ASSESSMENT OF LOW PRESSURE (LP)
ROTOR DOVETAILS DUE TO STRESS CORROSION CRACKING (sec)
Avinash V. Sarlashkar
Tony C.T. Lam
1800 Brighton-Henrietta To\Vll Line Road
Timothy J. Szumski
Arizona Public Service Company
Palo Verde Nuclear Generating Station
P.O. Box 52034, Mail Station 6301
Phoenix, Arizona 85072-2034
Abstract
A technical approach is presented to provide electric utilities with easy-to-use tool for managing
sec damage on LP rotor dovetails. This approach generates a database of fracture mechanics
data for the specific dovetail geometry before the start of a scheduled outage. This tool uses
information from the database to present the user with specific fracture mechanics data
instantaneously. A specific example of such a use by an electric utility is presented. It is
believed that this approach shows promise and can be further improved to address such issues as
probabilistic modeling and plasticity.
Background
Arizona Public Service (APS) is experiencing main turbine low pressure (LP) rotor dovetail
cracking at their three Palo Verde units. This cracking was originally discovered after only
50,000 hours of turbine operation. The original equipment manufacturer (OEM) has identified
the failure mechanism as stress corrosion cracking (SeC). In response to the cracking, .APS
developed a list of repair options that would maximize unit output and availability while
minimizing repair costs. The short tenn repair options are:
LP rotor replacement,
Long shanking LP rotor wheels,
Reducing dovetail stresses using
93
Avinash Sarlashkar
Titanium buckets or
Titanium blocks (de-vaned buckets) and
Pressure plating.
The first and last options were rejected due to the high cost of rotor replacement and the lost
revenue and moderate cost of pressure plating. The most cost effective approach was to reduce
dovetail stresses using titanium buckets or blocks (windowing). this option did not
eliminate the possibility of pressure plating should very deep cracks be present. The long shank
option eliminated the concern with pressure but the cost was moderately high. In order
to reach a balance between repair costs and the risk of unit output reduction, APS decided to
implement a combination of the long shank option and the titanium replacement option. The
strategy would be to use titanium buckets for shallow cracks, titanium blocks for deeper cracks
and implement the long shank option only ifthe cracks were too deep for the previous two
options.
Implementing the titanium replacement option involved determining the maximum crack depth
at which titanium buckets could be installed without exceeding rotor wheel material properties.
This involved perfonning a fmite element analysis of the rotor wheel dovetaiL The challenge in
perfonning the analysis was to develop a tool that could be used to quickly analyze and predict
dovetail crack depths and locations identified in the field without the time delay associated with
modeling and running the computer code. Stress Technology Inc. (STI) met this challenge by
developing a spread sheet that APS could use at the site. The spread sheet is a compilation of
computer runs for different crack depths that envelope all expected situations. By specifying
crack depths and locations, the spread sheet would specify whether titanium buckets could be
installed. If not, .titanimn blocks would be used. The disadvantage of installing blocks instead of
buckets is the turbine power loss associated with the missing bucket vanes and the potential
diaphragm damage associated with the cyclic steam loading of the block "window" as it passes
the stationary diaphragms. That's why the spread sheet results were so valuable.
In swnmary, the benefit of using the spread sheet is that the assessment of the cracking can be
performed on-site without running any complicated programs. In addition, the spread sheet \\'ill
analyze the actual crack depths and locations identified in the field. This approach saves critical
outage time and reduces the potential conservatism associated with computer results that don't
exactly match the field crack conditions.
The Technical Approach
The technical approach followed in this investigation has several different aspects. The
following are the main issues that were addressed. This paper describes item 2 in detail while
items 1, 3 and 4 are briefly discussed for the sake of completeness.
94
Avinash Sarlashkar
1. 3-D stress analysis for the "notch group."
2. 2-D fracture mechanics for various crack-size configurations.
3. Change in natural frequency and mode shapes for the notch group.
4. Simplified analysis of crack propagation due to high cycle fatigue CHeF).
Note that all the analyses perfonned are for the "notch group" and assume that the middle three
buckets in the group are made of titaniwn ""hile the end buckets are made of 12Cr steel. Also
note that in all the analyses, the middle bucket for the notch group is supported by the
neighboring buckets through the use of platform pins.
3-D Stress Analysis for the Notch Group
The objective of this analysis was to estimate the effect of crack excavations on the peak steady
stresses for the disk hooks. Figure 1 shows the finite element model ofthe disk sector for the
specific case of excavations. It is important to note that this geometry corresponds to the
configuration at "time =0" (i.e. the end ofthe outage before turbine start up). It is this geometry
that should be used for any fracture mechanics calculations used to detennine dovetail stress
levels at the next inspection. The results of thj.s specific case indicate that the resulting peak
steady stresses for the specified excavation depths are similar to those for the original geometry.
This is due to the use of titanium buckets which are approximately 50% lighter than the steel
buckets. This is an important piece of information since titanium buckets were installed after the
cracks were ground out. Therefore, the expected time to crack initiation for the new
configuration should be similar to the original design.
Fracture Mechanics Analysis Using a 2-D Axisymmetric Model
A major portion of the investigation was devoted to the development of a database containing the
relevant fracture mechanics results for different crack configurations at each of the three dovetail
hook locations. Figures 2A and 2B show the finite element mesh for the 2-D fracture mechanics
model. Note that the model used in the 2-D analysis represents the bucket next to the notch
bucket. This is implied through the additional load applied to the 2-D model to account for the.
centrifugal load of the notch bucket which is supported by the neighboring buckets through the
platform pins. The fracture mechanics database was generated using the crack-tip element in the
general purpose finite element program ANSYSTM. The following are some of the relevant
details:
1. The 2-D fracture mechanics model includes both the disk and the bucket root. This is
essential from the view point of addressing the load redistribution between different dovetail
hooks resulting from the change in flexibility of the hooks.
95
Avinash Sarlashkar
2. The 2-D model assumed that the crack lengths are symmetric with respect to the plane of
synunetry for the disk dovetail.
3. 2-D contact elements were used in the analyses to allow for any separation
ofthe root/dovetail hooks.
4. For each of the hooks, it was asswned that the crack lengths at the end of 6-years would be in
the range of 0.325" - 0.550" due to SCC. This expectation was based on the field data at the
Palo Verde station as well as the material test data for 3.5NiCrMoV. A total of8 cases (:;:;
2x2x2 for 2 crack lengths at each of the three hooks) were analyzed, where crack lengths of
0.325" and 0.550" were assumed for each of the hooks.
5. The results ofthe 8 cases were used to generate the fracture mechanics database used for
interpolation ofarbitrary crack length configurations. Note that the 0.325" crack length case
corresponds to no crack at the beginning of the next 6-year inspection interval, and the
0.550" case corresponds to a crack length of 0.225" at the beginning ofthe next 6-year
period.
Dynamic Characteristics of the Notch Group
Calculations were performed to estimate the effect of excavations on the natural frequencies for
the notch group. The objeCtive of this analysis was to compare the natural frequencies ofthe
original all-steel notch group with those of the titanium notch group in the presence of multiple
dovetail hook excavations. This information was used in estimating the magnitude of resonant
stresses and the likelihood of HCF driven failure for the notch group. It was assumed that the
transfer function between the uncracked steady stress and the corresponding LiKStatic for a
specified crack length is applicable for the case of dynamic stress as well. The estimated value
of &<'Dynamic(Resonant) based on this assumption was compared to the material property, M<.thresnold ,
to assess the likelihood of crack propagation. The assumed value of M<.threshold was 3 ksi"in.
In summary, it was found that there is a high risk of HCF driven failure for the worst case of
0.550" long dovetail cracks. For this case, the estimated value of LiKDynamic(Resonant) was twice as
large as the material property M(Threshold. It was estimated that for those cases where the buckets
are not operating at or near resonance or for those cases where the cracking was not as severe as
the worst case scenario mentioned above, the risk of dovetail HCF was low.
Estimation of Fracture Mechanics
Traditionally, the risk assessment for turbomachinery components has relied mostly on stress
based criteria, such as average shear stress for the hook ligament, rather than fracture mechanics
based criteria. However, in this investigation, fracture mechanics based criteria were usedI. In
I It should be recognized that for crack lengths that are a significant portion ofthe depth. of the original hook
ligament, additional calculations to address gross plasticity may be required.
96
Avinash Sarlashkar
particular, the two material properties of fracture toughness K
IC
and the threshold value
were used. The value ofK
rc
was estimated using the value of Charpy impact energy
(CVN) in the Rolfe-Novak relation.
(
J
' ( J
K
IC
- CVN
- =5 ---0.05
cry cry
(1)
For the values ofCVN=45 ft-Ib and cry =115 ksi, the estimated fracture toughness is 150 ksi"in.
As mentioned previously, it was asswned that a reasonable value of LlKthreshold is 3 ksi\l'in.
Prediction of Crack Growth and
The LP rotor inspection interval for the Palo Verde wilts is 6 years. It is important to be able to
predict the rotor condition and assess the risk of failure at the end of this 6 year inspection
interval. During the maintenance outage, all SCC damage is removed. New SCC cracks are
expected to develop and grow before the next inspection. The crack growth is determined by, A)
the degree of current damage, B) the incubation time, and C) the crack groVlth rate. Because
sce incubation time and the range of growth rates are well documented for steam
environments, a deterministic model was developed-to predict the fracture properties for a given
operating period before the next inspection. The user is required to input the current crack sizes,
the expected incubation time and the crack growth rate. The model will then predict the final
crack configurations and the stress intensity factors in all the three hooks. In addition, safety
factors based on a given value of material toughness will also be calculated.
When the stress intensity factor .6K at the crack tip exceeds the threshold K
rscc
, the crack will
grow due to sec. The crack size for a given period can be predicted as:
a =Rate x (Inspection interval - Incubation time)
(2)
For 3.5NiCrMoV disc material, the K
1SCC
and the growth rate under SCC can be estimated from
the published data shown in Fig. 3 and Fig. 4 [Reference 2]. The sec incubation time in the L-3
stage envirorunent is estimated to be 2 years. For the L-3 stage disc, the da/dt ranges from 50
mils to 90 mils per year. Palo Verde records show an average growth rate of 82 mils per year.
\\Then the future crack configuration is determined, the stress intensity factor can be calculated
and the risk of failure can be assessed. However, the time which can be allowed for perfonning
the stress and fracture mechanics calculation is very limited during the outage. The real
challenge is to provide plant management with technical information in a timely manner so that
repair decisions can be made. This problem was addressed by developing a technique to
97
Avinash Sarlashkar
calculate the fracture properties of any given crack configuration. The Lagrangian interpolation
is employed in the model:
., ., .,
K
T
= N i i k K i ~
;=1 )=1 k=l
a T _aT a,\.-1 _aM QJ B _as
N 2 I _ ~ __
121 == T T aM _ a Af as _ aB
a
2
-a, 2 I 2 1
T T M .'1.1 B B
a a
J
Q
2
- a a
2
- a
N211 = T T M M B B
a
2
- a
J
a - a a - a
2 1 2 I
where:
(3)
KT, K
M
, K
B
Interpolated stress intensities at top, middle and bottom hook each for the
measured crack lengths of aT, aM, as respectively.
KTijk Stress intensity factor calculated at the top hook for crack configuration 'ijk'.
Indices i,j and k each can have values of 1 or 2. For i=j==k=l, the crack length is
0.325" at each of the hooks, for i=j=k=2 the crack length is 0.550" at each of the
hooks.
N
ijk
The shape functions. For example, the shape function N112 correspond to a case
of crack size of 0.325" (i.e. i=l) in the top hook and the middle hook (i.e. j=l) and
a crack size of 0.550" (i.e. k=2) in the bottom hook.
aT, aM and as The measured crack lengths at top, middle and bottom hook notches.
M b T M B M B
K ijk' K ijk The vector K ijk IS replaced by K ijk and K ijk to get the values of K and K
respectively.
The database of stress intensity factors for each configuration was generated by using the fmite
element model described previously. A minimum oftwo crack sizes for each hook are needed
for Lagrangian interpolation. Assuming a symmetric root attachment with three hooks, a total
of 8 ( 2x2x2) cases of crack configurations were included in the matrix. This minimum model
allows a linear interpolation to be perfonned for any combination of cracks which may be found
98
Avinash Sarlashkar
in the field during the next outage. A more advanced model with non-symmetric cracks and
quadratic interpolation can be developed which requires a 3x3x3x3x3x3 matrix or 729 finite
element calculations. The 2x2x2 database used in the Palo Verde. evaluation is shown in Table 1
as a Microsoft Excel spread sheet. The r s u ~ s can be obtained instantaneously after the data is
input.
Table I: The Database
Palo Verde Unit #1 see L-3 Stage Investigation, Stress Intensity Factor Predictions
Crack Size Stress Intensity Factor
(in.) (Ksi-in**.5)
Top Middle Bottom Top Middle Bottom
Hook Hook Hook Hook Hook Hook
0.325 0.325 0.325 58 43 35
Database 0.325 0.325 0.550 66 54 42
from 0.325 0.550 0.325 74 43 47
Fracture 0.325 0.550 0.550 92 59 60
Mechanics 0.550 0.325 0.325 58 67 47
Calculations 0.550 0.325 0.550 70 87 59
0.550 0.550 0.325 90 80 76
0.550 0.550 0.550 133 129 115
Hypothetjcal Case I
daldt Inspection
years
Incubation
Years
Tonghness KIC
Input 0.082 6 2 150 150 150
Growth Parameters
Results of
Estimated
Crack Size & DK
at the next Inspection
Input current top middle bottom crack
sizes below (in inche.s)
0.35 0.05 0.5
Projected Crack Sizes
(in inches)
Top Middle Bottom Hook
Hook Hook
0.678 0.378 0.828
Safety Factor =>
Stress Intensity Factor
(Ksi-in**.5)
Top Middle Bottom
Hook Hook Hook
122 164 117
1.23 0.91 1.29
99
Avinash Sarlashkar
Hypothetical Case II
daldt Inspection
years
Incubation
Years
Toughness KIC
Input 0.082 5 2 150 150 150
Growth Parameters
Results of
Input current top. middle. bottom crack
sizes below (in inches)
0.35 0.05 0.5
Projected Crack Sizes
(in inches)
Stress Intensity Factor
(Ksi-in**.5)
Estimated
Crack Size & DK
at the next Inspection
Top Middle Bottom Hook Top Middle Bottom
Hook' Hook Hook Hook Hook
0.596 0.296 0.746 69 103 62
Safety Factor => 2.18 1.46
.... ~
..t.."t_
In the hypothetical Case I sho\\on in Table 1, the three crack sizes were 0.350", 0.050" and 0.500"
before these cracks were ground out. After two years of incubation time, new cracks will initiate
and grow to effective lengths of 0.678", 0.378" and 0.828" respectively. The effective length is
defined as the depth of excavation plus the new crack length. The estimated stress intensity
factors are 122 ksi in1/2 , 164 ksi in1/2, and 117 ksi in1/2 respectively. The corresponding factors
of safety are 1.23,0.91 and 1.29 respectively. Therefore, the middle hook will become the most
critical location from the view point of crack stability. Therefore, the inspection interval of 6
years is unacceptable. For the hypothetical Case II, all the data are the same as the Case I except
the inspection interval. The inspection interval is set at 5 years instead of 6 years, i.e., the
expected cracks will be on an average 82 mils shorter compared to Case 1. The results clearly
show that the calculated stress intensities are significantly lower with the result that each hook is
"safe". It is quite clear that this tool can be used very easily to perfonn "what if'. analyses to
generate an acceptance envelope for considering different crack size configurations .
Additional Analysis of Hypothetical Case I. The results for the hypothetical Case I show
that although the middle hook has a much smaller crack than the other two hooks, it has the
highest value of stress intensity. This is because with the smallest crack size, the middle hook is
stiffest and shares more load than the other two hooks. As a result, the stress is highest in the
middle hook notch and so is the stress intensity factor. "When the stress intensity factor reaches
the material toughness, the crack in the middle hook will become unstable and therefore will
grow. This will make the middle hook more flexible. As the load re-distributes to the top and
bottom hooks, the stress intensity factor at the middle hook will reduce and the crack in the
middle hook may retard. For the LP dovetail investigated, the criteria for titanium buckets verses
titanium blocks is based on whether a hook will fail in 6 years. To be conservative, any possible
crack retardation after the stress intensity reaches the toughness ofthe material is disregarded.
100
Avinash Sarlashkar
Conclusions and Recommendations
A detenninistic model was developed to predict the crack configuration and the fracture
mechanics properties for the L-3 stage dovetail hooks. The fracture mechanics model is
finite element based and has the re-meshing capability. The stress intensity factors were
calculated to generate a 2x2x2 matrix for a set of crack configurations. Lagrangian
interpolation was employed for quick evaluation of the future crack configuration.
During the Unit 1 outage, technical infonnation was generated in a timely manner and
provided to plant management. Using the results of this analysis and other data, APS was
able to install titanium buckets on the generator end L-3 notch group.
A more sophisticated model which can provide more accurate information is being
considered. This model will be used in future outages for all three Palo Verde units.
Based on the work in this project, a probabilistic model can be developed which will be able
to provide more realistic risk assessment based on statistical data for incubation period as
well as rates of SCC crack propagation.
Inclusion ofplasticity in the analysis may be required to analyze "deep" cracks.
References
1. Sarlashkar Avinash V., Tony C. T. Lam, "Damage Tolerance Assessment for sec Cracking
on Palo Verde Unit #1," Teclurical Report PB 1137, Stress Technology Incorporated,
Rochester, NY 14623.
2. Speidel M. 0., J. Denk., B. Scarlin: "Stress Corrosion Cracking and Corrosion Fatigue of
Steam-Turbine Rotor and Blade Materials,:' Commission of European Communities, 1991,
ISBN 92-826-2017-4.
AS:jah
d: \avilepri3cc.doc
101
Avinash Sarlashkar
/

Figure 1. 3D Finite Element Model of the Disk Dovetail with 0.3" Excavation
I

7\" :\ /,
- - __.I--,
Details of the Finite Element Mesh around the Crack
Tip
L._
Figure 2B. \ jl
\\,Q/ ,j
\. ;
'-----"---....,.
2D Axisymmetric Finite Element MOdel
for the Blade Root/Dovetail Assembly
Figure 2A.
102
Avinash Sarlashkar
1200
sec, condensing steam, 3.5NiCrMoVl26NiCrMoV145
1100
ro
a.. 1000
6
900
0)
C
~ 800
..,
(/)
"0 700
<I>
~
600
o 10'000 20'000
time [hours]
30'000 40'000
Figure 3. Dependence of Different Stages of sec on the Material
Yield Strength. [Ref. 2]
stress intensity, K
I
[kSh'iil]
0 50 100 150 200
.----.
.---..
10-
5
r-
eD
~
steam turbine rator steels, sec,in H:O
CD
>0-
s:
3%Nisteel
.........
.J:.
10-
6
.2Cr-1Ni steel
10
3
(,)
.E
~ I
A "clean steels",speciaDy aloyed steels
~ I
4i
10-'
10
2
0)
....
Cl:l
-ttl
.J:.
...
-
10-
8 .J:.
3
10
- 0
3
s-
O
Dl
...
~
0)
(J 10-
9
~
ttl (,)
...
to
0
...
c:
(,)
0
10
10
10-'
C
'(;j
0
0
'0
...
0 ...
-.-
0
...
10-"
...
0
10-
2 0
Ul
(,)
en rJ)
~
rJ)
CD
-
10'2 r-
f,/'J
- 0 100 200 240
(/)
stress intensity, K
r
[MN'm-
312
]
Figure 4. Effect of Temperature and stress Intensity on the Growth Rates
of Stress Corrosion Cracks in Steam Turbine Rotor Steels. [Ref. 2]
103
104
Stress Corrosion Crack Initiation
in Steam Turbine Steels
Presented By
Fred F. Lyle, Jr.
SOl;lthwest Research Institute
105
106
STRESS CORROSION CRACK INITIATION IN
STEAM TURBINE STEELS
Fred F. Lyle, Jr.
Department of Materials and Mechanics
Southwest Research Institute
P. O. Drawer 28510
San Antonio, Texas 78228-0510
Abstract
This paper describes a research program that was conducted to identify factors which promote development of stress
corrosion cracks in steels used in discs of low-pressure (LP) steam turbines in electric power generating plants. Typical
3.5NiCrMoV disc steels made in the United States and in Japan and a special high-purity disc steel were exposed in three
simulated power plant environments. Factors found to influence stress corrosion crack initiation were: oxygen
concentration of the environment; yield strength of the disc steel; tensile stress level; and tight crevices and localized
corrosion within crevices. Factors found to have no significant effect were: disc steel composition; inclusions; and
pitting. The stress corrosion cracking threshold stress intensity. KIscc, in the environments studied was 7.2 MPa"m.
Introduction
In 1969. a catastrophic rupture of a forged. shrunk-on, a 3CrMo steel disc in a low-pressure (LP) steam turbine occurred
during an overspeed test at the Hinkley Point A nuclear power plant in England. (l ,2)- The cause of the failure was found
to be propagation to critical size of a stress corrosion crack in a keyway. No specific causative chemical species were
identified, but NaOH was suspected because it had been used for pH control of feedwater for an extended period. After the
Hinkley Point failure, inspections of other LP turbines in the United Kingdom (U.K.) revealed stress corrosion cracks in
keyways of discs in 50 other turbines. Subsequently. similar stress corrosion cracks have been found in keyways. on
bore surfaces. and on blade-attachment slot surfaces of discs in more than 100 LP steam turbines in service in the United
Kingdom, Australia, South Africa. Taiwan, Japan. Canada, and the United States (U.S.).
The Hinkley Point failure and widespread cracking that was "found in other nubines after the failure resulted in a number
of laboratory research programs to identify factors responsible for stress corrosion cracking (SeC) in LP turbine disc
materials. The subsequent discovery of cracking in LP turbine discs in the United States and other countries led to
additional research programs. Results of many of these studies were summarized in an earlier paper.(3)
Analyses of experience and of previous laboratory studies conducted in several countries showed that SCC of LP turbine
disc steels can occur in high-purity watery as well as in a number of more corrosive environments (e.g., concentrated
hydroxides). It has been established that stress corrosion crack growth is dependent upon: the presence of liquid water;
disc yield strength; disc temperature; tensile stress level; and the aggressiveness of the environment, including oxygen
concentration, concentrations of ionic species. and pH. Also, a limited amount of data indicates that tight crevices (on
the order of 0.04 nun wide or less) between mating surfaces significantly increases the susceptibility of discs to sec. (4)
The availability of crack growth rate data and the methods of linear elastic fracture mechanics have enabled accurate
prediction of times required for cracks to grow to the critical size necessary for disc rupture.(5-r)
However, very little crack initiation data have been reported. This paper describes a research program that was conducted:
to identify factors which control stress corrosion crack initiation in LP turbine disc steels; and to provide data needed to
predict crack initiation times. The effects of mechanical. environmental, and metallurgical factors on stress corrosion
crack initiation in modem 3.5NiCrMoV turbine disc steels were evaluated in a comprehensive experimental program.
Experimental Program
Test Parameters
Experience and data from earlier studies indicated that liquid water is required for stress corrosion crack initiation. Six
additional factors identified as potentially important to crack initiation were: chemical composition of the water, in
particular the oxygen concentration; steel composition and phases present, in particular the Mri and S concentrations and
the numbers of MnS inclusions; tensile stress level; tight crevices between mating components; and temperature. All of
these factors except temperature were evaluated in the experimental program described in this paper. All the tests were
conducted at 130C, the temperature of the 'Wilson line" where water fIrst condenses in some turbine designs.
(*) Superscript numerals in parentheses refer to references.
107
Fred Lyle
Three aqueous environments were used. A ''high-purity'' water environment was representative of secondary-side water in
a presurized water reactor (PWR) plant during normal operation; the water contained less tban 5 ppb of oxygen and had a
conductivity of less than 0.3 pS. An ''inteJDlediate-oxygen" environment represented typical short-teJDl upset conditions
in a PWR. secondary-side water system; the water contained 40 ppb oxygen and 20 ppb NaCl. And. a
llcontamjnated-water" environment represented a severe secondary-side upset condition; the water was saturated with air,
and contained 6-8 ppm of oxygen and 20 ppb of NaCl.
Three steels were selected to provide a wide range of Mn and S concentrations to evaluate the effects of MnS inclusions
on stress corrosion crack initiation. One steel was a typical American-made 3.5NiCrMoV steel. The second was a
typical Japanese-made 3.5NiCrMoV steel. The third was from a special high-purity heat of 3.5NiCrMoV steel. The
compositions of the steels are given in Table I. The typical American and Japanese steels had similar compositions.
except that the sulfur content of the American steel was about 6.5 times that of the typical Japanese steel. The special
high-purity steel was produced with very low levels of Mn and S to prevent formation of MnS inclusions. Metallog-
raphy showed significant concentrations of MnS in the American steel. The typical Japanese steel contained silicate
inclusions, but no MnS inclusions were found. The high-purity steel was essentially free of non-metallic inclusions.
The typical American-made and 1apanese-made steels were heat treated to produce yield strengths representative of the
upper and lower ends of the yield strength range in LP nubine discs. i.e., approximately 725 and 930 MPa. Specimens
of both steels with both yield strengths were tested. The high-strength and low-strength American steels were designated
as "Steel AH" and "Steel AL," respectively. Similarly, the heat-treated high-strength and low-strength Japanese steels
were designated "Steel JH'1 and "Steel JL." The high-purity steel was designated "Steel JP;II it was tested in the
as-produced condition. Mechanical properties of the five test steels are given in Table n.
British test results indicated that stress corrosion cracking did not occur at stresses below .50% of steel yield strengths and
when crevice widths were greater than about 0.05 mm(4). Consequently, specimens of the five steels in this program
were tested at tensile stresses equal to 70 and 90% of the yield strengths and in excess of the steel yield strengths. and
both creviced and uncreviced s p ~ n s were exposed. A crevice width of 0.04 m.m was employed.
Exposure periods ranged in length from a minimum of 3 months to a maximum of 24 months. The times used were
determined by the severity of corrosion and degree of sec experienced in each of the three environments. The number of
replications of test conditions was dictated by space available in the autoclaves used and by a limited amount of Steel JP.
Duplicate specimens were exposed for each test condition in all but a few cases.
Procedures
Specimens were exposed in two high-pressure, recirculating autoclave test loops. Tests in the contaminated-water and
intermediate-oxygen environments were conducted in a loop containing three 3.8-liter titanium autoclaves. Tests in the
high-purity water environment were conducted in a loop containing two 3.8-liter Type 316L stainless steel autoclaves.
Three-paint-loaded bent-beam specimens were used for specimens stressed to 70 and 90% of steel yield strengths; V-bend
specimens were used for specimens stressed above their yield strengths. Creviced and uncreviced conditions were achieved
by placing two specimens in contact with each other in the same loading frame. Typical bent-beam specimens are shown
in Figure 1. The tension-loaded surface of the uncreviced (ourer) specimens were exposed to the bulk solution, while the
tension-loaded surfaces of the creviced (inner) specimens were exposed to the environment within the crevices between the
mating surfaces. Slots 3 mm wide and 0.04 mm deep were machined across the centers of the compression-loaded
surfaces of uncreviced specimens to ensure crevice geometries similar to those used in British studies.(4)
Exposed specimens were examined visually and with optical and scanning electron microscopes to establish the extent of
corrosion damage and cracking, before corrosion products were removed. Analyses of corrosion products were performed
in situ or on deposits removed from specimens. Specimens were cleaned electrolytically to reveal surfaces that had been
covered with corrosion products. Metallographic cross sections through selected cracks and other features were prepared
. and examined to characterize cracks, to determine relationships between crack initiation and features such as pits and
intergranular corrosion, and to define the deptbs of features from which cracks initiated.
1.0 MPa - 6.895 ksi
108
Fred Lyle
Application of Linear Elastic Fracture Mechanics
The driving force for crack initiation and growth is the plane strain stress intensity factor. KI. which is dependent upon
tensile stress, flaw depth, and flaw geometry. The expression for KI for a semi-elliptic crack growing inward from a
surface under axially-applied loading is:
KI - ([(1.21)(1t)(a)2(a)]/Q)1/2 (1)
where: o.is the tensile stress; a is the flaw depth; and Q is a flaw shape parameter. The value of Q is dependent upon the
ratio of tensile stress to steel yield strength (o/Oys) and the ratio of the flaw depth to flaw length, (a/2c), where 2c is the
flaw length on the surface. Equation (1) shows that KI is inversely proportional to Q. This means that for two flaws of
the same depth but different lengths, KI will be larger for the longer flaw because the value of Q decreases as (a/2c)
approaches zero.
Cracks initiate when KI exceeds KIscc, the threshold stress intensity factor for SCC. The value of KIscc for steels in
water at LP turbine temperatures has been reported to be relatively constant for a variety of different steels. regardless of
steel composition or yield strength, over a temperature range of 100 to 288
Cl
C. and to range between about 10 to 30
MPa..Jm .(8,9)* Over the majority of the period of crack growth, the crack growth rate is independent of KI. Other than
the presence of tensile stresses and a liquid phase, the crack growth rate of typical LP turbine disc steels has been shown
to depend upon only two factors _. disc temperature and disc yield strength .- in accordance with the following
0)
In (daldt) - - 4,968 - 7t 302JT + 0.0278 Oys (2)
where: (daldt) is the crack growth rate, in incheslhour; T is the disc temperature
t
in degrees Rankine (OF + 460); and Oys
is the disc yield strength in ksi.
Metallographic examinations indicated that in the vast majority of cases stress corrosion cracks initiated from corrosion
trenches along grain boundaries; in some cases, cracks initiated along boundaries between phases within individual grains;
and, in a very few instances, cracks initiated from pits. KIscc was calculated from Equation (1) using: the depth of a
flaw at which the crack began to grow; the tensile stress at the location of the flaw; and appropriate values of Q for the
applicable (a/2c) and (aloys) ratios. In the case of a crack growing from a corrosion trench, the length of the trench on
the surface at the time the crack initiated could not be measured, and the value of Qcould not be detemrined .accurately.
Extreme values of (a/2c) were used to bound the values of Krscc. The (a/2c) values used were 0.5 (semi-circular crack)
and 0.0 was infinitely long in comparison to a).
It should be noted that Equation (1) gives the stress intensity factor for a semi-elliptic crack growing inward from a
surface under axially-applied loading. In calculating stress intensity factors from specimens exposed in the test program,
it was assumed that Equation (1) also was applicable to three,..point-loaded bent beams. This assumption is justified
because the depths of corrosion trenches and pits from which cracks initiated were so small that there was essentially no
difference in stress levels at specimen surfaces and at the bottoms of the flaws.
Results and Discussion
All exterior surfaces of specimens from the contaminated-water environment were covered with red-brown hematite
(Fe203). Black and gray magnetite (Fe304) was found within the crevices between mating tension-loaded and
compression-loaded surfaces. Within crevices, specimens were uncorroded except in localized etched zones along
specimen edges. The typical appearances of mating bent-beam. specimen surfaces are shown in Figure 2. Exterior
surfaces of specimens exposed to the intennediate-oxygen and high-purity water environments were covered with thin
layers of magentitet but were essentially uncorroded. However, specimens exposed to both of these environments
displayed localized corrosion along the edges of mating surfaces within crevices; the corrosion was essentially identical to
localized corrosion found at the similar locations on specimens exposed in the contaminated-water environment.
Intergranular stress corrosion cracks initiated and grew in all five test steels, including high-purity Steel JP. In all three
environments, cracks initiated from COrTosion trenches that formed within the etched zones on mating surfaces within
crevices. The vast majority of cracks were initiated from corrosion trenches that formed along grain boundaries or along
boundaries betWeen phases within grains. Typical grain boundary trenches are shown in Figure 3. Figure 4 is a
1.0 MPa..Jm - 0.91 ksi..Jin.
109
Fred Lyle
high-magnification micrograph through a typical crack. Data for specimens exposed to the contaminated-water,
intermediate-oxygen. and high-purity water environments are summarized in Tables ill through V. respectively.
Time to Initilltion
Experimentally determined times to initiation for each of the test steels at various stress levels are shown in Tables III
through V. However, these data do not allow estimation of initiation time more closely than the length of the test period
in which cracking was flfSt observed. For example, in the high-purity water environment (Table V), severe cracking
occurred in specimens of high-strength Steels JH and AH that were exposed for 12 months at stress levels equal to 90%
of the steel yield strengths. but no cracks were found in specimens stressed to the same level that were exposed for 6
months. From these data. it is not possible to tell when within the second six months of exposure the cracks were
initiated. In the absence of further information, conservatism requires use of six months for the initiation time in a
lifetime prediction.
Times to initiation can be more closely estimated using: the crack depth (ainiV corresponding to KIscc; the deepest crack
found for a condition (amax); and the crack growth rate from Equation (2). The difference between the two crack depths
divided by the growth rate yields the time required for a crack to grow from initiation to largest depth measured. The time
to crack initiation then can be calculated by subtracting the time required for crack growth from the exposure time.
Times to crack initiation calculated in this manner for the five steels in the three test environments are tabulated in Table
VI. Initiation times for low-strength Steels AL, n... and JP are indicated as >17,520 hours (2 years), since cracks did not
initiate in specimens of these materials in two years at tensile stress levels below the steel yield strengths. Fracture
mechanics analysis is not valid for stress levels in excess of the steel yield strength.
The high-purity water environment is of greatest interest because it most closely simulates nonnal PWR operating
conditions. Estimates of times to crack initiation for use in remaining lifetime calculations for high-purity water
environments are as follows:
H a:S 0.7 ays; use 17,520 hours. This concb.lsion is based on the absence of stress corrosion
cracks in all steels at tensile sttess levels equal to 70% of steel yield strengths in the two-year test.
If 0.7 ays :$; a:$; 0.9 y ~ and ays S 780 MPa; use 17,520 hours. This conclusion is based
on the absence of cracJting in the low-strength steels at tensile stress levels equal to 90% of the
steel yield strengths.
If ~ 0.7 Oys and Oys > 780 MPa; use 8,585 hours. This conclusion is based on the observation
that no cracks were initiated in 6-month test specimens of the high-strength steels that were stressed
to 90% of the steel yield strengths, but severe cracking occurred in specimens tested for 12 months
under the same conditions. The initiation time used is the shorter time calculated for the two
high-strength steels (Table VI).
Controlling Parameters
Analyses of test results showed that five factors had significant effects on stress corrosion crack initiation in the test
steels; these were oxygen concentration of the liquid phase; steel yield strength; tensile stress level; the presence of tight
crevices and localized corrosion within the crevices; and the KIscc values of the materials.
The test steels were significantly less susceptible to crack initiation in the high-purity water environment. The major
differences between the three environments were the oxygen concentrations and the resulting electrode potentials of the
test steels. Typical oxygen concentrations in the contaminated-water environments were between 6 and 8 ppm. and
electrode potentials ranged between 0 and -100 mV vs a AglAgCl (O.Ol M KCl) reference electrode. The typical oxygen
concentration in the intermediate-oxygen environments was 40 ppb, and electrode potentials varied between -200 and -500
mV. And. typical oxygen concentrations in the high-purity water environments were less than 5 ppb. with electrode
potentials ranging from -600 to -800 mY. These results clearly indicate that oxygen concentration (or the corresponding
electrode potential) was a primary factor that controlled initiation of cracks.
Crack initiation was very strongly dependent upon the steel yield strengths. In all three environments, cracks were longer
and deeper and were present in greater numbers in specimens of the two high-strength steels. AH and JH. Further. severe
cracking of bent-beam specimens in the high-purity water was limited to the two high-strength steels.
110
Fred Lyle
Crack initiation also was strongly dependent upon tensile stress level. In all cases in which V-bend specimens (that were
stressed above the steel yield strengths) were susceptible to cracking, either no craclcs were formed in the bent-beam
specimens (which had tensile sttess levels of 70 or 90% of the steel yield strengths), or cracks in V-bend specimens were
greater in number and deeper than cracks in bent-beam specimens. Similarly, in the contaminated-water and intermediate-
oxygen environments, cracks were deeper in bent-beam specimens loaded to 90% of steel yield strengths than in
specimens loaded to 70% of the yield strengths. In high-purity water. cracking of the low-strength steels occurred only in
U-bend specimens. while cracking in high-strength steel bent-beam specimens was limited to specimens that were
stressed to 90% of the steel yield strengths.
In all three environments, cracks were' found only along the edges of tension-loaded surfaces within crevices. and the
cracks initiated under corrosion products deposited within crevices on mating surfaces. Corrosion products shielded the
metal surfaces beneath them and minimized the amount of oxygen reaching the shielded -surfaces. The result was reduced
oxygen concentration beneath the corrosion deposits within crevices (relative to the oxygen concentrations in the bulk
solution that was present at surfaces outside the crevices). These differential aeration cells between creviced and
uncreviced portions of specimens caused localized corrosion (trenching) under the deposits along grain boundaries and
boundaries between phases within individual grains.
Metallographic measurements and calculations performed on specimens from all environments in which cracks were
found indicated that KIscc for the five test steels ranged between about 5.6 and 11.5 MPa-Jm. and that the mean value
was 7.2 MPa..Jm. These results are consistent with findings of other investigators.(8,9)
Stress corrosion cracking occurred in specimens of all five test steels in the contaminated-water and intermediate-oxygen
environments. and in four of the five steels in the high-purity water environment. Differences between crack sizes and
numbers of cracks in steels of different compositions but the same general strength level were not significant, with the
exception that specimens made from the low-strength. typical-Japanese Steel JL did not crack in the high-purity water
environment. Steel JL also generally was more resistant to cracking in the other environments than were the two other
low-strength materials.
The fact that cracking occurred in specimens of all five steels indicates that steel composition and the metallurgical
phases present had little effect on stress corrosion crack initiation. In particular. the test results indicate that crack
initiation did not depend upon Mn and S compositions or the number and type of inclusions. High-purity Steel JP was
produced with very low levels of Mn and S that prevented formation of MnS inclusions, and inclusions in the typical
Japanese-made Steels IH and 1L largely consisted of AI-Mg-Ca silicates. It is evident that the number and type of
inclusions were not primary factors that determined susceptibility to crack initiation.
As has been noted. a few instances of cracks growing from pits were observed. However. while those instances show
that cracks can be initiated at pits, hundreds of cracks initiated from corrosion-induced trenches along grain boundaries or
boundaries between phases. These observations indicate that initiation from corrosion trenches was strongly favored over
initiation from pits. This probably is because the depth-la-length (al2c) ratios for pits generally were larger than for
trenches. and. consequently, the flaw depth necessary to initiate a crack was larger for a pit than for a trench.
The absence of a primary relationship between pitting and stress corrosion crack initiation in this study is not in
agreement with results of several other studies in which stress corrosion cracks were found to initiate from pits.(11-14)
However. the results of the present study are consistent with experience in LP turbine discs in u.s. nuclear power plants.
In U.S. plants, pits and cracks frequently were found together within crevices at bores. keyways. and blade-attachment
slots, but there was no relationship between the presence of pits and initiation of stress corrosion cracks.(15-16) The
current results also are consistent with results reported from a Japanese study and from a recent Australian study, both of
which found no relationship between pitting and stress corrosion crack initiation.(17.18)
There are significant differences between experimental procedures used in some of the studies that found crack initiation
dependent upon pits(4,9.l1,14) and procedures used in the present sOldy and in the Japanese and Australian studies'<17,18)
In several cases, different test temperatures were employed, and temperature might have some effect on whether cracks
initiate from pits or at grain boundaries. Also, slow-strain-rate, testing was employed by some investigators. Severe
strains. as occur in this test, can cause inclusions to fracture or separate from the surrounding metal. which can result in
formation of crevices at metal-inclusion interfaces, localized corrosion within the crevices, and initiation of cracks under
conditions where cracking may not have occurred at less severe strains and stresses more representative of in-service
loading. However, the Australian study (18) that found no relationship between pitting and cracking used slowstrain rate
testing, so this test technique. alone, was not solely responsible for the observed differences in crack initiation.
111
Fred Lyle
In most of the studies in which cracks were found to initiate from pits. uncreviced test specimens were used.. The absence
of crevices in those studies did not accurately simulate the creviced conditions present in bores. keyways. and
blade-attachment slots where SCC has been found in U.S. LP steam turbines. and the lack of crevices would be expected
to influence test results. However, crack initiation from pits on freely exposed (uncreviced) specimen surfaces in a
number of the studies implies that there are other important experimental differences, since cracking was not found on
freely exposed surfaces in the present study, in the Japanese study. or in the Australian study. and, to the author's
knowledge. sec on freely exposed surfaces of LP steam turbine discs has occurred only in one plant in the United
States. and pitting was not involved in that incident.<7.15) The difference may be the result of periodic exposure of
specimens to air. In several of the studies that showed a dependence of crack initiation upon pitting. it was reponed that
test specimens were removed periodically from the environments and were examined in air to track growth of pits and
cracks. (9,11,14) In the current study and the Japanese and Australian studies{l7,18) which found no relationship
between pitting and crack initiation. specimens were removed from the test environments only at the ends of exposure
periods.
Exposure of wet steels to air is known to oxidize ferrous ions (Fe++) to femc ions (Fe+++).(19) and several investiga-
tors have reponed that exposure of wet specimens to air results in accelerated corrosion rates. The higher corrosion rates
encountered under such conditions are thought to result from reduction of the (Fe+++). formed upon exposure in air, to
(Fe++) when the specimens are returned to a deaerated. or low-oxygen environment. (20-23) A similar effect has been
reported in a Canadian study of microbiolgically induced corrosion (MIC)'in underground pipelines. It was found that
where MIC was occurring under anaerobic conditions in the presence of sulfate reducing bacteria. ferrous sulfides were
fanned. Introduction of air into the environment converted the ferrous sulfides to ferric oxides. and the corrosion rate in
the anaerobic enviromnentbefore and after exposure to air increased by an order of magnitude from 0.1 to 1.1 mmlyd
24
)
The implication that exposure of wetted steel components to air can accelerate corrosion rates is of practical interest to
LP turbine operators. The presence of pitting and SCC on freely exposed surfaces of wetted turbine steel specimens
exposed to air implies that power plant operators should consider taking steps to minimjze exposure of wet turbine discs
to air during shutdowns or other operations in which air may enter, the turbines. Procedures to minimize air inleakage
also should help to minimize localized corrosion and subsequent cracking within bores. keyways, and blade attachment
slots. since the results of the current study show that the number and size of cracks initiated within crevices is strongly
dependent upon the difference between the oxygen content of water present on freely exposed surfaces and within crevices.
Conclusions
The results of the present study have established that stress corrosion crack initiation in typical and high-purity
3.5NiCrMoV LP steam turbine disc steels exposed to aqueous environments depends upon:
oxygen concentration of the liquid phase (or the electrode potential corresponding to
the oxygen concentration);
steel yield strength;
tensile stress level;
the presence of tight crevices and localized corrosion within crevices; and
a combination of tensile stress and flaws of size and shape sufficient to produce a KI value
greater than KIscc. i.e. greater than about 7.2 MPa..jm.
In general. of LP steam turbine discs to crack initiation increased with increasing oxygen concentration and
steel yield strength. Crack initiation was much more likely to occur in the' presence of high tensile stresses on creviced
surfaces, as are present within disc keyways. bores, and blade attachment slots. Crack initiation was been found to be
independent of disc steel composition, the presence of MnS or other inclusions, and pitting. although it was found that
cracks could be initiated at pits formed on steel surfaces at inclusions if the value of KI at a pit exceeds KIscc.
These results have practical implications for turbine operators. The results indicate that water system chemistry should
be carefully controlled to maintain oxygen concentrations at low levels, e.g., < 5 ppb. Comparison of results of this and
other programs. in which cracks initiated from pits on wetted. freely exposed surfaces, implies that operators should
minimize exposure of wet LP turbines to air during shutdowns or other operations in which air may enter the turbines.
For new or replacement rotors, the lowest yield strength steels permitted by other design requirements should be used.
Similarly, tensile stresses should be maintained at the lowest possible levels (in particular, stress concentrations that
produce tensile stresses greater than disc yield strengths should be avoided), and crevices should be minimized or
eliminated where possible (e.g., by use of monoblock. welded, and keyplate rotor designs).
112
Fred Lyle
The current program also confirmed that the stress cracking threshold, KIscc, for disc steels, is on the order of
10 MPav'm, as had been reported previously, and it has provided estimates of crack initiation times that can be used to
increase accuracy of lifetime predicitions in typicallP turbine discs.
References
L D. Kalderon, Proceedings ofthe Institution ofMechanical Engineers, Vol. 186, No. 31, p. 341, 1972.
2. J. L. Gray, Proceedings o/the Institution o/Mechanical Engineers, Vol. 186, No. 32, p. 379, 1972.
.3. F. F. Lyle, Jr., "Stress Corrosion Cracking in Low-Pressure Steam Turbines -- An Overview," Paper No. 219,
Proceedings CORROSION/94, NACE International. Houston, TXt 1994.
4. B. W. Roberts and P. Greenfield, Corrosion, Vol. 35, p. 402, 1979.
5. N. S. Cheruvu and B. B.Setb., Proceedings: Fossil Steam Turbine Disc Cracking Workshop. Electric Power
Research Institute, Palo Alto, CA, 1991.
6. J. W. Cardinal and F. F. Lyle, Jr., Proceedings: Steam Turbine and Generator Life NDE, Assessment, and
Maintenance Workshop, Electric Power Research Instiwte, Palo Alto, CA. 1991.
7. F. F. Lyle, Jr. and H. C. Burghard, Jr., Steam Turbine Disc Cracking Experience, Final Report, EPRI
NP-2429-LD. Electric Power Research Institute, Palo Alto, CA, 1982.
8. R. M. Magdowski. Stress Corrosion Cracking ofLowAlley Steel in Water, Diss ETII No. 9432, Dissertation,
Swiss FederallDstitUte of Tecbnology, ZUrich, Switzerland, 1987.
9. M. O. Speidel and R. M. Magdowski, Proceedings: The International Conference on Advances in Materials
Technology for Fossil Power Plant.s. Electric Power Research Institute, Palo Alto, CA, 1991.
10. W. G. Clark, B. B. Seth, and D. M Shaffer, Joint ASME'IEEE Power Generation Conference Paper No.
81-JPDC-Pwr-31, ASME, 1981.
11. H. Karate, H. Itch, and Y. Kondo, Proceedings: Fossil Steam Turbine Disc Cracking Workshop, Electric Power
Research Institute. Palo Alto, CA, April 1991.
12. M. O. Speidel, J. Denk, and B. Scarlin. Stress Corrosion Cracking and Corrosion Fatigue of St.eam-Turbine
Rotor and Blade Materials, Catalog No. CD-NA-13185-ElN-C, Commission of the European Communities,
Luxembourg. 1991.
13. K. Klemetti and H. Hinninen. Proceedings of the Second International Symposium on Environmental
Degradation ofMateriaIs in Nuclear Power Systems--Water Reactors. American Nuclear Society, La Grange Park,
rr.... p. 70., 1986.
14. S. R. Holdsworth and D. V. Thornton, ofthe RobenI. Jaffee Memorial Symposium on Clean
Materials Technology, ASM International. Metals Park. Ohio, November 1992.
15. F. F. Lyle, Jr., and H. C. Burghard, Jr., Steam Turbine Disc Crac1dng Experience, EPRI Report NP-2429,
Research Project 1398-5, Electric Power Research Institute. Palo Alto, CA, June 1982.
16. Personal Communication with B. B. Seth, Westinghouse Electric Corporation, October 10,1990.
17. S. Inagaki, M. Miyazaki, H. Kashiwaya, and K. Nakadi, Proceedings: Fossil Steam Turbine Disc Cracking
Workshop, Electric Power Research Institute. Palo Alto. CA, 1991.
18. A. Oehlert and A. Atrens, Corrosion Science, 1996, Vol. 38, pp. 1159-1169.
19. A. J. Gould and U. R. Evans. Journal ofthe Iron and Steel Institute. Vol. ISS, 1947, pp. 195-200.
20. A. Nishikata, S. Kumagai, and T. Tsuru, Corrosion Engineering, Vol. 43, 1944, pp. 109-120.
21. A. Nisbikata, S. Kumagai. and T. Tsuru, Corrosion Engineering, Vol. 43, 1944, pp.225-234.
22. T. Tsuru, A. Nishikata, and J. Wang, Materials Science andEngineering. Vol. A198, 1995, pp. 161-168.
23. M. Stratmann and H. Stteekel, Corrosion Vol. 30,1990, pp. 697-714.
24. T. R. Jack. A Practical Manual on Microbiologically Influenced Corrosion, Volume II. J. G. Stoecker. editor,
NACE International, Houston, TX, in press, 1997.
113
Fred Lyle
Table I Chemical Compositions of Test Steels
Steel Elemental Composition (Weight Percent)
&w Si Mn .e .s .Cr
y
Al .s.n &. S2
AHandAL 0.27 0.02 0.27 0.007 0.013 3.50 1.81 0.55 0.09 <0.01 <0.01 <0.01 <0.01
JHandIL 0.28 0.06 0.28 0.007 0.002 3.32 1.57 0.35 0.10 0.004 0.001 0.006 0.0013
JP 0.24 0.02 0.04 0.0032 0.0014 3.72 1.73 0.40 0.12 NR <0.00 <0.003 0.0018
Table IT Mechanical Properties of Test Steels
SlW
Mechanical Propc:rrics
Yield Tensile Red
Steel Strength, Strengdl Area. Elong.
MPa MPa (flO) (%)
AH 934 1,041 62.0 19.0
AL 780 895 66.8 25.5
JH 941 1,038 66.2 19.0
JL 741 851 74.8 23.5
JP 738 853 72.0 23.8
Table m' Summary of Stress Corrosion <::racking Test Results in the ContaminatcdWater Enviroruncnt
3 Months 6 Months
Steel &, Crevice Maximum
#C/lfI"a)
Max. Crack
tlCl#fta)
Max. Crack
Yield Strength Condition Stress Depth (mm) Depth (mm)
C O.7YS 1/1 0.013 2/2 0.2S4
U 0.7YS 0/1
-
0f2 -
Steel AH, e O.9YS 1/1 0.381 212 0.406
934MPa U 0.9YS 011 - 0/2
-
e >YS 111 0.406 212 2.845
U >YS 1/1 0.762 212 1.945
C 0.7YS 0/1 - 011 -
U 0.7YS 0/1
-
011
-
Steel AL. e 0.9YS a/I
- 0/1 -
780MPa U O.9YS 0/1
-
011
-
C >YS 1/1 0.051 110 0.787
U >YS 0/1
- 011 -
e 0.7YS 1/1 0.076 0(2
-
U 0.7YS 011 - 012
-
SteelJH. e 0.9YS a/I - 11'2 0.508
941 MPa U 0.9YS 0/1
- 0/2
-
c >YS 1/1 0.038 212 1.930
U >YS 1/1 0.737 .
212 2.381
C 0.7YS 0/1
-
0/'2 -
U 0.7YS 0/1 - 0/2
-
Steeln.., C O.9YS 0/1
-
212 0.028
741 MFa U O.9YS 0/1 - 0J2
-
C >YS 0/1
-
212 0.076
U >YS 0/1
- 012 -
C 0.7YS 1/1 0.223
U 0.7YS 0/1
-
Steel JP, C 0.9YS (Steel JP was not tested in the III 0.279
738MPa U 0.9YS 3-month exposure) Oil
-
C >YS 1/1 0.229
U >YS 011
-
(8) IC/#T-No. of Specimens CrackcdlNo. of Specimens Tested
114
Table IV Summary of Stress Corrosion Cracking Test Results-Intermediate-Oxygen Environment
Fred Lyle
3 Months
6 Months
12 Months
Steel & Crevice Maximum
Max. Crack
Max. Crack
.
Max. Crack
.
#CI#T Depth (mm) #CI#T Depth (nun)
Yield Strength Condition StreSS #CI#T Depth (mm)
C 0.7 YS III 0.468 III 0.160 1/1 0.500
U 0.7 YS a/I -
0/1 - all -
SleeIAH. C 0.9YS 1/1 0.460 1/1 1.600 III 2.220
934 MPa U 0.9 YS 0/1 - 0/1 -
all -
e > YS 1/1 0.340 1/1 0.841 1/1 3.175
U > YS 011
0/1
0/1 -
C 0.7 YS III 0.060 1/1 0.130 III 0.190
U 0.7 YS 0/1 -
011 -
011 -
Steel AL. e 0.9YS 1/1 0.070 1/1
0.220 1/1 0.220
780 MPa U O.9YS 0/1 - 0/1 - 0/1 -
e >YS III 0.060 1/1 0.190 1/1 1.550
U >YS 0/1 - 0/1 -
0/1 -
C 0.7YS 0/1 -
1/1 0.160 1/1 0310
U 0.7YS 0/1 -
011 - 0/1
-
Steel1H, e 0.9YS 1/1 0.127 1/1
0.250 1/1 2.290
941 MPa U 0.9YS 0/1 -
Oil -
0/1 -
C >YS 1/1 0.219 III 0.510 1/1 2.030
U >YS 0/1
0/1 -
0/1 -
C 0.7YS 1/1 0.020 011 -
1/1 0.117
U 0.7 YS 011 - 0/1
-
0/1 -
Steel JL, C 0.9YS 1/1 0.020 0/1 - 1/1 0.260
741 MPa U O.9YS 0/1 -
011
-
0/1 -
C >YS 0/1 -
1/1 0.160 III 0.130
U >YS 0/1
011 - 011 -
C O.7YS 1/1 0.130 111 0.140 1/1 0.290
U 0.7YS 0/1 -
011
-
011 -
Stee11P, C 0.9YS 1/1 0.103 1/1 0.210 1/1 0.440
738 MPa U 0.9YS 0/1 -
011 -
0/1 -
C >YS 1/1 0.127 111 0.143 1/1 1.130
U >YS 0/1 - 010 - all -
* #CI#T-No. of Specimens Cra.ckedlNo. of Specimens Tested
Table V SUIIUDIrY of Stress Corrosioa Cilcking Test R.esults-High-Pmity Water Environment
3 Months 6MDDIhs 12Montbs 24Montbs
St=l& O'cvice Max.
~ >
Mu.Crack Max.Qact Max. O'ack Max.Cntck
Yield Strength Condition Stress Depth (imn,)
::J#1t&}
Depth(mm)
tK:JWta)
Depth (mm) ~ Depth(mm)
C 0.7YS 011
-
012
-
012 - 012
-
U 0.7YS 011
-
012 - 012
-
012 -
SteelAH. C O.9YS 011 - 012 -
112 0.089 1/2 2.130
934MPa U O.9YS 0/1 - QI2
-
0/2 - W2
-
C >YS 011
-
0/2 - 212 2.064 '2/l 3.175
U >YS 0/1 - 0/2 - 1/2. 1.588 2/l 3.175
C O.7YS 0/1
- 011
-
011
-
011
-
U 0.7YS 011
- 011
-
011
-
011 -
Steel AI.. C 0.9YS 011
-
011 -
0/1 - 011
-
780MPa U 0.9YS 011
-
011
-
0/1
- 011 -
C >YS 011
-
011
-
VI 0.124 1/1 2.160
U >YS on - 011 -
0/1
-
111 2.340
C O.1YS 011 - 012 -
on. - 012
-
U 0.7YS 011
- 012
-
012 - 012 -
SICClJH. C 0.9YS 011 - 012 -
III 0.061 '2/l 0.182
941 MPa U 0.9YS 011
-
012 - 012 - Oil -
C >YS 011
-
012 - 2/2 2.064 112 2.222
U >YS 011 - 012
-
012
-
ZI2 2.381
C O.7YS 0/1
-
012
-
0/'2 - 012 -
U 0.7YS 0/1
-
012 - 012
- 0f2 -
SteelJL., C 0.9YS 011 - 0/2
-
012 - on.
-
741 MPa U 0.9YS 011 - 012 - 012 - 012
-
e >YS 011
-
0/2 - 0/2
-
OIl -
U >YS 0/1 - 012
-
0/2 - OIl
-
C O.7YS
NT") ~
011 - 011
-
0/1 -
U 0.7YS NT NT 011 - 011
-
011
-
SteelJP. C O.9YS NT NT 011
-
011 - 011 -
738MPa U O.9YS NT NT OIl -
0/1
-
011 -
C >YS NT NT 0/1
-
011 - 1/1 0.429
U >YS NT NT 011
-
0/1 - 011 -
<a> l#CIfl'-No. ofSpecimeDs CrackedlNo. of Specimens Tested
(b) NT-DOt lCS1Cd.
115
Fred Lyle
116
Table VI Time to Initiate Qacks in Creviccd Regions from Frac:turc Mechanics Analyses
Tune To Initiate Cracks (Hours)
YIdd Strength High-Purity Water Intermed.i.&re-()xygen Contaminared-Warer
Steel (MPa) Environment Environment Environment
AH 934 8.585 425 1,289
AL 780 > 17.520 1.924 1.989
m 941 8,625 1.866 2.106
JL 741 > 17.520 2,190 3,979
JP 738 > 17.520 1.683 2.650
Figure 1. Bent-beam specimens from intermediate-oxygen environment.
l.2SX
JH66
. , .
I
7 i _ -'
4 : -<1J \ .'1 "',:". ..
.' ": ... ,;,- , I I \01\ .... ..
JH65
Figure 2. Mating surfaces of cleaned bent-beam specimens from
intermediate-oxygen environment. 1.9X
Figure 3. Intergranular cracking along grain boundaries in etched zone on
bent-beam surface. environment. l00X
Figure 4. Large crack in bent-beam specimen.
Contaminated-water environment.
200X
Fred Lyle
117
118
Stress Corrosion Cracking of Steam Turbine Rotors
Presented By
Darryl Rosario
119
120
Stress Corrosion Cracking of
Steam Turbine Rotors
Prepared for:
EPRI Steam Turbine Stress Corrosion Cracking Conference
March 19-20, 1997, Baltimore, MD
Prepared by:
D. A. Rosario, C. H. Wells and G. J. Licina
3315 Almaden Expressway, Suite 24
San Jose, CA 9S118-1557
R. Viswanathan
3412 Hillview Ave.
Palo Alto, CA 94303
121
122
Stress Corrosion Cracking of Steam Turbine Rotors
D. A. Rosario, C. H. Wells and G. J. Licina
Structural Integrity Associates
3315 Almaden Expressway, Suite 24
San Jose, Ca 95118
R. VlSWanathan
3412 Hillview Avenue,
Palo Alto, Ca 94303
ABSTRACT
In the wake 0 f the catastrophic failure of an LP turbine disk at
the Hinldey Point Nuclear Station in 1969, considerable R&D
work has been devoted at EPRI and elsewhere to the problem of
stress corrosion cracking (SCC) in steam twbine roton. The
principal factors affecting the susceptibility of rotors to SCC have
been identified to be disk yield strength, applied stress level and
surface filmIcrevice chemistry; microstructure and cleanliness of
the steel were found to have relatively little effect. Advances in
steel making and forging technology during the last twenty years
have provided manufacturers with additional design and material
options to mitigate the stress corrosion problem. Increases in the
forging size capabilities of steel companies and welded
construction of rotors now permit designing with integral and
partial integral rotors which utilize lower yield strength (hence
more sec resistant) materials as well as eliminating the sec
problem in bores and keyways. However, a recent survey of
utilities by EPRI has shown that sec in the blade attachment
region ofLP rotors has become a problem of increasing concern.
This problem has led to the development of repair and
refurbishment methods for rim attachments, especiaUy weld
buildup of rims wilh corrosion resistant alloys. Life prediction 0 f
rotors under SCC conditions currently involves estimating crack
growth time from assumed defcds to critical size. The factors that
govern the location and time of crack initiation are inadequately
understood.
INTRODUCTION
Stress corrosion cracking is an ubiquitous problem in power
genenation equipment subject to flowing or condensing wet surface
films. The problem is particularly insidious at the highly stressed
mating surfaces of steam turbine rotors, wherein crevice corrosion
conditions, crack initiation and the early growth of cracks may go
undetected for many years. The consequences of undetected
cracking have been repeatedly demonstrated over almost three
decades, commencing with the Hinkley Point disk burst in 1969
and the discovery of bore cracks in sister turbines. The financial
and safety risks of bore cracking, especially in nuclear plants,
prompted a massive response from the power industry world-wide
to cope with the complex issues of rotor design, material
properties, steam chemistry, and nondestructive inspection that
govern the economics of turbine maintenance.
Among major milestones was the EPRI research project
RP1398, which addressed the root cause of stress corrosion
cracking and produced a survey of the extent of disk cracking
experienced by nuclear and fossil utilities up until 1981. Key
findings were the dependence of crack growth rate on disk yield
strength, temperature, applied stress level and steam chemistry,
with secondary influence of steel composition, and the
concentration of cracking at the farthest upstream stage of the
turbine at which water droplets fonned, commonly referred to as
the Wllson Line. Combined with inspection procedures, this
information allows estimates of the time for crack growth to
critical size. Recently Lyle [1] reviewed this early research and
combined with it the results of internatiorial projects on the
characterization of the susceptibility of rotor steels of various
composition and strength levels to environmental factors,
including his own results with simulated crevices. This recent
review, combined with other observations, underscores some
significant gaps in our understanding of crack initiation and the
ability to predict the time and location of the onset of cracking.
Recent months have seen the introduction of refurbishment and
repair methods to reduce or eliminate the risk of stress colTOsion
cracking as well as improvements in steel making and processing
for more resistant forgings. At the same time the industry's efforts
to reduce the risk of bore and keyway cracking through integrally
forged and welded rotor construction have shifted concern
somewhat to the susceptibility of rim attachments. In particular it
was considered important to determine whether this cra<;king
123
Darryl Rosario
mode should be expected to become more prevalent with aging of
the fleet and what steps utilities could take to mitigate the problem.
These issues led EPRI to commission a survey of US utility
experience and of the efficacy of repair methods, and the results
of this survey are summarized in this paper. In addition the paper
addresses the adequacy of life prediction and nondestructive
examination methods to provide the basis of maintenance
decisions, as well as an overview of key research results
generated during recent years.
see and Ufe Prediction Methods
seeIniliation: Cheruvu and Seth (2) have investigated factors
that control initiation and have found that oxygen and caustic were
the most aggressive however, rotor materials exposed at
less than 200
0
P or which had a yield strength less than 110 ksi
exhibited no significant cracking. Lyle [3] has recently publiahcd
the rcsulls of a test program that addressed the influence of water
chemistry and material composition and hardness on relative rates
of see over a 24 month test duration. These results confinned
the dominant role of yield strength and applied stress in
establishing temperature was a constant 130"C
(266F) in these tests. Estimates were made of time to cracle
initiation in typical disks on the basis of the results and a
generalized SCC crack growth rate correlation. Specimens were
tested at stress levels equal to 70% and 90% of yield strength and
at stresses in excess of yield strength. The initiation times were
on the order of one to two years as shown in Table 1 (3) for
typical American and Japanese made high and low strength steels
in various environments. The short initiation times in the
laboratory tests do not appear consistent with field experience,
which indicates that the first detection of sec in certain disk rims
is generally observed after many yean of operation. Lyle also
concluded that crevices and crevice chemistry playa strong role
in crack initiation, but that steel microstructure and chemistry
effects were relatively weak. Oxygen was reported to be the
dominant impurity. Only a few instances 0 f cracks growing from
pits were observed on test specimens; initiation from corrosion
trenches was strongly favored over initiation from pits.
Rau et. al. (4) have proposed a quantitative model for SCC
initiation time. using a nonnalized initiation time. tN' given by:
tN = td ( rR (1)
where t
d
is the actual or calculated initiation time from field .
experience, (alayJ is the ratio of surface tensile stress to yield
stress. and R is the sce crack growth rate (computed using data
from EPRI RP2408-1 as descnbcd below) to account for the
depc:ndenc:eofinitiation time on disk temperature, yield strength,
manganese content and tempering temperature. The effect of
surface stress was accounted for using a third power dependence
based on laboratory test data. [5.6]. The nonnalizcd initiation
times were then fit statistically to yield an initiation probability. P,
given by:
(2)
124
where 0:, Pand to are curve fit constants. Figure 1 is an example
of initiation probability data compiled by Rau for disk keyway
cracks.
Endo et al. [7) recently published results of their laboratory
study on environmentally assisted cracking (SCC and corrosion
tid:igue) in 3.5NiCrMoV steels with yield strengths of 700 to 1050
Mpa (102 to 152 ksi). Testing was perfonncd at stress levels
equal to the disk yield strength. The three damage mechanisms
investigated were stress corrosion cracking. dynamic stress
corrosion cracking and corrosion fatigue. The stress corrosion
cracking process was reported to consist of three stages: (i) pit
growth, (ti) crack fonnalion and (iii) macroscopic crack
propagation, as shown in Figure 2. Cracks were observed to
initiate at corrosion pits. (Note that Lyle [3] has shown that pitting
is not as a precursor to SCC). A correlation was seen by
Endo between the number of pits and the inclusion content or
impurities. Higher material yield strength and higher
temperatures were found to reduce crack initiation time. Crack
initiation from a microscopic pit was predicted for a critical pit
size 2C
cr
, which occurs when the applied stress factor
exc:ccds a critical value (KISCC>P for static stress or where
corrosion fatigue is operative. Increasing numbers of start-stop
cycles and longer hold times were found to reduce crack: initiation
time, a phenomenon referred to as dynamic stress corrosion
cracking. The dependence of stress corrosion crack growth rate
on yield strength and temperature per Clark et. a1. [8] was verified
by Endo.
A summary of Siemens experience and R&D studies on LP
rotor sec is provided by W. David et. aI. (9). Only one of 430
disks in Siemc::nsIKWU nuclear turbines had SCC; none occurred
in the rim attachment region. Extensive SCC initiation testing was
performed on low-alloy LP turbine rotor and disk materials for
various applied strcss-to-yield strength (RSCCJ) ratios and
environments (high purity water up to 212P, condensing steam
at 203
G
P and 30% NaOH solution at 212-392F). Results are
summarized in Figure 3. As shown, an improvement in the
operating environment pennits high stress levels up to and above
the yield strength level of the material. Even in a severely
corrosive environment, sec initiation does not occur for stress
levels below balfthe material yield strength (0.5 Rsca). Siemens
has concluded that the absence of stress corrosion cracking in the
1970 vintage rotor! in nuclear power plants is probably due to the
high-purity water -condition in those plants, which pennits
operating sIress levels up to 10% above yield strength (1.1 Rsca)
without initiating cracking. In contradiction with Endo et al. [7J
but consistent with Lyle [3] and Holdsworth et a1. (10), the
Siemens data indicated that in high-purity water sec initiation is
not affected by the cleanliness of i.e.. non-metallic
inclusions do not serve as initiation sites for corrosion pits.
Part of the problem of predicting the initiation life of a disk rim
is the uncertain history of attachment crevice chemistry and
existence of a liquid fIlm. EPRI is conducting a program that
includes monitoring of droplet nucleation and growth in the steam
path, simulating the enthalpy drop in a bypass loop, and
measuring the chemistry of a crevice placed in the flow loop [11].
Condensation can take place on the upstream rim face if the
surface temperature is sufficiently low and the total pressure is
sufficiently high. In addition water may condense on adjacent
surfaces, such as the web of a disk or ID stator shroud and flow
past the rim if no seal is present. To the writers' knowledge, no
mapping of surface temperature or liquid films has been reported.
Early studies [12] concluded that cracking occurred at the blade
stage just: downstream of the point of thermodynamic equilibrium
moisture formation, or Wtlson Line, since that stage would
cxpcricnce the highest fIlm temperature. In fact, it appeared that
upstream blade rows were primarily susceptible to cracking in the
1980 survey. It follows that downstream rows, assuming the
pmscnce of moisture, should also eventu&lly crack with increasing
exposure time. The picture is complic:atcdby the possibility of
excessively wet operation at some point in a unit's history, as was
attributed in some units to moisture separator/rehcater
malfunction.
Once the solute concentration in a crevice has increased
because of off-design operation or other factors, the vapor
pressure is lowered, and moisture may remain stable at
temperatures above the Wllson line (11).
Another factor complicating the prediction of sec initiation is
the uncertainty in local surface tensile stress, which requi.ra
knowledge of residual machining stress and the machining
tolc:rances and variation of the blade-kHlisk fitup. Finite element
models exist for most of the blade attachments susceptible to sec
[13], although few are in the public domain. A quantitative data
base should incorporate the nominal attachment stress, with
statistical variability in fit-up.
SCC Crack growth rate: The most widely accepted model for
crack growth rate was given by Clark ct aI. ofWcstinghouse in
1981 [8], expressed by the fonowing equation:
In R = -4.968 - (7302l1) + O.02780ys (4)
wIlCI'C is the crack growth rate in inchlhr, T is the temperature in
Rankine (i.e., "F+460), and aya is the room temperature disk: .
yield strength in bi.
This equation has been accepted by the U.S. NRC for
remaining life predictions of turbine disks with cracks [14]. Since
that time, additional crack growth rate data has been generated
under EPRI RP2408-1 in 1985 [15]. Rec:alibration of the two-
variable Westinghouse model yielded the following equation:
In R = -8.8 - (4040/1) + 0.02310
ys
(5)
The standard deviation of residuals was 1.20 for the above
equation, which is based on 228 data points, versus 0.587 for the
Westinghouse model. which was based on 70 data points:. The
study also identified two more signifIcant variables: Manganese
content and tempering temperature for disks which were
specifically tempered below the ASTM A471 standards (lOSO"F)
to obtain high strength. The three-variable model with the effect
of manganese is given below:
In R = -4.74 - (9270/n + O.03370ys ... 4.53 Mn (6)
where Mn is the weight percent of manganese. The standard
Darryl Rosario
devation for this model is 1.17. For disks tempered above
lOS0F, the effect of tempering temperature was included as
fonows:
In R " 7.04 - (927011) .. O.0337oys .. 4.53 Mn - 0.0127 T.,.(7)
where TT is the tempering temperature in degrees Kelvin. The
standard deviation for this model is 1.20.
Data on fossil disk crack growth rate was presented by both
U.S. and foreign researchers at the EPRI Fossil Steam Turbine
Disk Cracking Workshop. in October 1990 (16]. In summary. the
roles oftempera1Ure, stress, environment, material condition. cte:.
on cnlCk growth rate have been extensively investigated; however
the wide scatter in the data suggests that further studies and
modeling would be required to better characterize the data.
Life prediction codes: Two crack propagation-based life
prediction codes have been developed and pUblicized for LP blade
attachments, Lifetime Analysis of STeeples (LAST) by Southwest
Research Institute [17] and Disk Crack Evaluator (DICE)
developed under EPRI RP1929-14 as part of the EPRl guidelines
for predicting the life of steam turbine disks exhibiting stress
corrosion cracking [18].
TIle LAST code was developed for two Westinghouse design L-
5 and L..{) steeples and one General Electric L-1 steeple. The code
assumes an existing crack, and propagation is simulated by SCC
and corrosion fatigue. Remaining life is the time required to grow
the existing crack to critical size. Typically, very little life
remains once the crack reaches the fatigue threshold depth.
Propric:tary corrosion fatigue threshold and crack growth data are
used by the code. Two crack geometries, an edge crack and a
semi-elliptic surface crack, can be evaluated. The SCC growth
rate is simulated using the Westinghouse equation (Equation 4).
Sample remaining life results are shown in Figure 4 as a function
of initial crack depth and alternating stress level.
'The DICE code performs stress intensity versus crack size and
critical crack size calculations for an edge crack and a semi-
elliptic surface crack. The user specifies a stress distribution and
fracture toughness value. These calculations are part of the
ovenill1989 EPRI guidelines [18] for predicting the life of steam
turbine disks with sec. These guidelines offer two levels of
analyses. Level 1 or Level 2. depending upon the amount of
infonnation available. see growth rates are simulated using
Equations 5, 6 and 7 listed above, which were: developed as part
ofEPRI RP2408-1 [15].
In summary, there does not appear to be a single commercially
available code that combines the expcrience-based crack initiation
probability, incorporating distributions of stress, temperature,
presence of wet surface films (flowing/stagnant, refluxing.
condensing, etc.) and material properties, with a probabilistic
crack growth rate analysis.
U.S. Industry Experience
Typical locations and orientations of cracks documented in the
original EPRl industry survey through July 1980 {12) are
illustrated in Figure: S. This survey included one decommissioned
plant and 33 ofn operating U.S. nuclear planlS at that time; 29
PWRs with steam generators made by Westinghouse, Combustion
125
Darryl Rosario
Engineering (CE) and Babcock & Wilcox (B&W), and.5 BWRB
made by General Electric (GE). Power ratings ranged from 100
to 1100 MWwith operating times from 6 months to 20 yean. The
33 plan1s had a total of79 LP turbines; 76 ofthcse had disks made
of qucnched-and-lemperm 3.5NiCrMoV steels, two tuIbines had
disks made ofnorma.l.ized-and-tempeted 2..5NiCrMoV steels, and
one had disks made of lCr-laMo steel. Cracking was found only
in Westinghouse rotors in 20 003 plants; 19 oftbese plants were
PWRs and one was a BWR. Cxacking was found in 70 disks from
36 LP rotors; 3S disks had keyway cracks, 5 disks had bore
surfacc cracks, 7 disks had face crack, and 30 disks had rim
attaclnnc:nt cracks. Crack depths ranged from 2.5 to 7S mm (0.1
to 3.0 in.). Cracks were predominantly intergranular and
branched, and usually were 61led with iron oxides; it was
generally agrcc:d that lhe operative mechanism was stress
conosion cracking. Survey data indicated a correlation between
disk cracking and the following variables: (l) Presence of a liquid
environment - all cracking occurred at or dOwnstream of the
saturation tine where wet steam is usually prcsc:nt. (n) Oxygen
level- the most widC3pn:ad and most severe cracking occurred in
a pJant with the highest oxygen level of20 to 30 ppm; plants with
oxygen levels nonnally below 5 ppb and never above 15 ppb did
not develop cracking. The source of oxygen in most of the plants
was air inleakage through the condenser which typically occurs
during plant starWps. (iii) Crack growth rate was higher for disks
with higher yield strengths, and also increased with disk opcntting
temperature.
Thc 1980 EPRI survey included only 34 U.S. jossil-fueled
plants with 45 LP"tmbincs - 18 made by Westinghousc, 19 by GE
and 8 by ABis Chalmers. Disk cracking was reported in 49 disks
in 31 turbines of22 plants - 31 disb had web face cracks (29 of"
these were Westinghouse), 16 disks had rim-attachment cracks (14
of these were GE), one disk had a bore crack (Allis-Cbalmen) and
one disk had a keyway crack (Allis-ehalmers). Cracking
occurred in 22 of 24 rotors used in plants with once-through
boilers and all-volatile water treatment compared with cracking in
only 9 of 21 rotors in pIamwith drum boilers and phosphate water
treaJmcnt. Survey data for fossil-fueled plants were not sufficient
to establish relationships belwecn disk cracking and system
variables; the only apparent trend was for boiler type and water
treatment as summarized.
Since the 1980 survey, inspection of GE turbines used in
nuclear plants in the U.S. and in other countries has revealed disk
keyway crack indications in approximately 60 disks [1]. In the
U.S. at least four nuclear units with GE shrunk-on disk LP designs
have had their LP rotors replaced with the new GE monoblock
design to eliminate the risk associated with catastrophic failure of
disks with borclkcyway cracks. With regard to CoslIil units. it was
feh that n:hcat machines of up to SOO MW output were immune to
SCC. This belief changed in 1987 with the of SCC in
turbinc disks in fossil reheat units. and was highlighted in
February 1989 with the burst of an IP/LP rotor disk from 200
MW reheat unit in the U.K. Subsequent to this incident and in
response to utility concerns. EPRI organized a workshop in
October 1990 to gather and consolidate pertinent industry
experience regarding the problem of sec in fossil turbine disks
[16]. GE reported that 1049 disks in 8S of the older fossil
126
turbines of GE design had been inspected. Of these, 36 disks
'WCJ'C reported to have ultrasonic indications in the keyways; 12 of
these disks were removed and analyzed. and SCC was confmned
in 6 of these It is GE's assessment that keyway crack-like
indications may be found in any of the LP stages, though the tint
and last stages appear to be considerably less susceptible.
Westinghouse also provided a summary of their disk: cracking
experience at the 1990 EPRI workshop. Of a total of 220 fossil
disks inspected, no cracking of any type was reported in these
disks. For nuclear units. Westinghouse reported a total of 1475
disks inspected of which 119 (or 8.1 %) were cracked; 105 of
these disks has keyway cracks and 27 dish had bore cracks. No
instances of cracking in rim-attachments or on disk faces was
reported.
Over the last five years or so, industry concern has shifted
toward addressing cracking in rim-attachments of LP roton.
which has increased dramatically over this period. The following
discussion focuses on this problem and the findings of a recent
EPRI survey which wu conducted to address this problem.
LPDisk Rim Blade Attachment Cracking
Cracking in rirn-attac:hments of low-pressure steam twbine rotor
disks has increased dramatically since the original EPRI survey in
1980[121. The 1980 survey documented rimattachment cracking
in only seven nuclear units and a few fossil units. In response to
mcart industry concerns with regard to the increasing number of
rotors that have either been repaired or retired because of this
problem, EPRI sponsored an industry-wide survey of rim-
attachment cracking in both nuclear and fossil units. The
objectives ofthis survey (RP900S-01[19J) were to collect data on
LP rotor rim cracking to document the location. extent and
mechanism of cracking, and repair methods used and to
investigate possible relationships between cracking experience and
various design and operating parameters. Disk cracking
in U.S. nuclear utilities and several fossil utilities was
determined from a review of literature data and a questionnaire
survey which wu mailed to"these utilities. Infonnation on rotor
repair methods was obtained from a separate survey of repair
vendors and OEMs. Data was gathered through December,
1995. Key findings ofthc survey arc summarized below.
U.S. Nuclear Unils: The survey included all 109 currently
operating nuclear units in the United States; LP rotor rim
attaehmentcracking was reported in 41 (38%) ofthesc units. The
cracking mechanism reported was predominantly stress corrosion
cracking, with a few instances of corrosion-fatigue and one
incident of high-eycle fatigue in the L-Q stage, attributed to high
back pressure operalion. Based on the survey data, cracking does
not appear to be related to steam generator (Figure 6) or turbine
manufilcturer (Figure 7) or power rating (Figure 8). It should be
noted that the survey database included primarily GE and
Westinghouse turbines; data on ABB and GEC turbines was too
limited to be included in the overall comparison. Rim cracking
was reported in nuclear units with operating times ranging from .
33,000 to 140,000 hours (Figure 9). In GE turbines cracking was
most severe in the L-2 and L-3 rows. which have circumferential
entry, straddle-mount blade attachments (Figure 10); temperatures
at these rows are approximarcly in the 220
0
P to 260
0
P range. No
cracking was reported in the L-o and L-1 rows of GE turbines,
which have a finger and pin attachment design. In Westinghouse
turbines, cracking was primarily in the L-o through L4 rows,
which have axial-entry blade attachments (Figure 11);
temperatures at these rows are approximately in the 120
D
F to
260F range. Most of the reported cracking propagates across the
steeples with a few cracks propagating ladially downward from
the base of the steeple. Higher incidences of clacking were
reported in units with longer last stage blades. nata was
insufficient to establish a reJationship between the incidc:ncc: of rim
cracking and the following variables: operating time, number of
startups, type of water treatment, oxygen level!. condenser
cooling wa1er and condenser leakage rate. Limited rim adachmcnt
crack growth data (Figure 12) suggests that the Westinghouse
equation [8] to estitnatc crack growth rates is an adequate
for use in life prediction. Compared with the
resuhs of the 1980 survey where rim-attachment cracking was
reported only in certain rows of Westinghouse turbines (Figure
13), the 1995 data shows (a) a significant shift in the number of
cracks by row number to dOW118trcam rows (Figure 11), and (b)
a significant rate of clacking in General Electric rotors, not
reported up to 1980 (Figure 10).
U.s. Fossil Unils: The survey included 757 fossil units from 33
utilities. The incidence of cracking was ten times higher in
supercritical units with once-through boilers than in subcritical
units (Figure 14). The cracking mechanism reported was
predominantly sec, with a few instances of corrosion fatigue and
reportedly excessive attachment stress attributed to an overhung
blade design. Higher incidences of cracking were reported in
units with Combustion Engineering and Foster Wheeler boilers
(FtgUre 15) and in units with Westinghouse turbines (Figure 16).
Data on unit operating time was insufficient; however the number
of uni1s with rim-auachment cracks by year (Figure 17)
indicates that the largest: &action of units with rim cracks went into
service between 1966 and 1975 (i.e., approximately 100,000 to
200,000 operating hours). Compared with nuclear units, the
appamrtly longer time required for initiation of cracking in fossil
units is related to steam conditions (wetness and temperature)
during operation, which are typicaUy drier for fossil units. For
GE tUIbincs no rim cracking was reported in sub-eritical units; in
supcr-critical units, cncking was reported to be most severe in the
L-l and L-2 rows (Figure 18); temperatures in these rows are
approximately in the 160 to 200"F range. Cracking in GE
designs was also reported at various locations (pin holes, base
mdius, ledges) in the multi-finger pinned type of attachment used
in the L-G and L-1 rows of four units (of 39 units in the survey)
with 33.5-inc::h last stage blades. For Westinghouse turbines rim
cracking reported was most severe in the L-o and L-l rows
(Figure 19); temperatures in these rows are approximately in the
100" to I60"F range. Most of the reported cracking propagated
across the steeples, with no radial cracking reported at the base of
the steeples.
Inspection Meliuxis: A review of disk' blade attachment
irispection methods in the U.S. focused on the GE straddle-mount
and Westinghouse axial-entry attachment designs for which most
of the rim-attachment cracking has been reported. For GE
stroddk-molUll designs, unless blades are removed to expose the
Darryl Rosano
rim surfaces. inspection is generally perfonncd using ultrasonic
methods. The ultrasound is introduced from the side of the disk
immediately below the blade such that the sound is directed
radially upward into the rim toward the blade hooks on the
opposite side of the rim attachment. The inspection is conducted
in a pulse--ecbo mode using shear waves C5-waves) at the
appropriate refracted angles as shown in Figure 20. The beam is
directed toward the base of the different hooks by moving the
transducer radially along the face of the disk, by changing the
refracted angle of the transmitted beam, or a combination of the
two. Depending on the exact geometry of the attachment it is not
always possible to direct a beam toward all the hooks, particularly
the outennost hook. The presence of a crack or pit typically
resulrs" in disruptions to the uniformity of the hook reflections and
is discriminated by comparing time and domain signals.
This inspection can be implemented either manually or automated.
For Westinghouse axial-elllTy designs, inspedions at the
exposed disk faces can be accomplished using surface inspection
methods including magnetic particle and eddy current. Eddy
current provides certain advantages in tenns of its inherent
capacity for automation. which provides for comprehensive scan
covemge and digital acquisition and storage of data. For interior
flaws, ie. those originating along a hook away from the exposed
end W:es, ultrasonic puise-echo and pitch-(:8,tch inspections must
be employed as shown schematically in Figure 21 [20]. For
portions ofthe attachment where the transducer can be positioned
directly in line radially (or near radially) with the area of interest,
pulse-echo inspection can be perfonncd. For the central portion
of the attachment above the disk web, where the transducers
cannot be positioned radially beneath the area of interest,
catch inspections must be perfonned with the transmitter on one
side of the disk and the receiver on the opposite side of the disk.
This, of course, requires relatively precise coordination of the
transducer positions and beam directions from two opposite sides
ofthe disk. Regarding delectability, Southwest Research Institute
[21] has reported good reliability of detection down to 1 mm crack
depth in the absence of significant pitting or machining scratches,
using fixed focus lens contact transducers at compound angles.
EPRI is cum:ntly conducting a three-phase program at the NDE
CeDfCr to characterize and improve inspection capabilities for disk
rim inspection. Phase I activities comprised a study of inspection
practices currcnl1y being implemented, development of a
depth focused phased array transducer assembly to assess
potential improvements and advantages of such a system [22] and
compilation of an inspection guide to document the Phase 1project
findings [23]. Phase II, which is nearing completion, compares
the detection and sizing capabilities of conventional, broad-bcam
and focused-bcam inspection approaches for the GE straddle-
mount design [24}. Phase III activities will address the more
difficult axial-c:ntry configuration.
Repairs: U.S. nuclear utility operators reported the use of the
foUowing types of mechanical repairs for rim-attachment cracking:
(I) flaw excavation, (ii) shot peening, (ill) blade removal, (iv) use
of steel/titanium notch blocks, (v) pressure plates and ('\Ii) long
shank buckets. Though weld repair of cracked rims has been
successfully implemented for fossil rotors. it has not yet been
reported for U.S. nuclear rotors; however, weld repairs are now
127
Darryl Rosario
being initiated for some foreign nuclear rotors. In addition to the
mechanical :repairs listed for nuclear rotors, U.S. fossil utility
operators reported that the repair of choice was a 360 weld
repair of the rim; 21 of 49 fossil units with rim cracks were weld
repaired. No recurrence of rim cracking in weld repaired rotprs
was reported, with 20,000 to 40,000 successful operating hours.
Remedies
10eremedy for borc-<:onnected cracking is usually replacement
ofthe disk: or rotor with a daign that eliminates the keyway or the
bore itself. Axial keyways have been eliminated by substituting
inter-locking tabs to prevent slippage. Replacement rotor designs
that feature integrally forged or welded disks afford the
opportunity to reduce strength level in order to provide more
resistance to stress corrosion cracking in the rim attachmcnb (25).
Welded construction allows the usc of smaller forgings than
required for integral forgings, although excellent control of
microstructure has been reportedly achieved with both methods of
construction (25). Weldability of 3.SNiCrMoV has also been
excellent. The stress conosion resistance of weldments of this
material has been reported by Holdsworth et a1. [10] to be
equivalent to parent materiaL.
Replacement of disks or rotors with forgings of the same
strength level and without redesign of the attachments leaves the
rim area susceptible to stress corrosion cracking, -and cracking of
integral or partial integral rotors has reportedly occurred. In such
eases, recourse must be taken to weld repair.
Weld Repairs
Utilities and OEMs have demonstrated that damaged low
pressure (LP) turbine rims may be repaired and refurbished.
Over the pest ten yean or so, weld repair of LP turbine rotors has
become a routine method for returning a rotor to a condition as
good as new or. possibly, even better than new. This trend is
illustrated in Figure 22 (26]. Weld repairs offer a distinct
advantage since the composition and mechanical properties of the
wcld-deposited replacement material can also be optimized so that
the repair has greater resistance to SCC. The development of
rotor weld repair technology has required extensive qualification
ofprocesses, weld :6Ilcr mdals, and testing to demonstrate that the
weld metal and the heat affected zone have adequate
properties. High purity, vacuum melted filler wire and carefully
controlled welding procedures and post weld heat treatments are
specified.
Carefully controDed weld build-ups have been used succcssfuLly
for a variety ofturbine applications [27,28,29,30,31] as illustrated
in Figure 23. When defects are found in one or two blade
attachments. local repairs may be adequate. Local repairs require
extreme care during pre-heating, welding, and post weld heat
treatment to avoid distortion and residual stresses in adjacent
areas. More often, the damage is more widespread and a full
360
0
repair is performed. Full 360
0
repairs offer distinct
advlU'llages with respect to the level of residual stress that is likely
to be encountered. Most of the vendors surveyed indicated that
all or nearly aU of their repairs Were full 360 repairs.
The majority ofwcld repairs to LP rotors have used filler metal
with a composition very similar to that of the rotor. Slight
128
modifications to the composition are made to improve weldability
and to achieve improved resistance to sec [30,32,33.34]. Weld
filler metal compositions similar to that of Type 403 or Type 410
stainless steel with additional requirements on many of the
alloying and trace elements have been selected for some of the
weld repairs in order to produce good weJdability and deposits
with the desired combination of mechanical properties and sec
resistance [7,27,33.35]. The SCC resistance of martcnsitic
stainless steels is a strong function of the strength level. Lower
strength material (e.g., tensile strength less than 100 kli; hardness
less than R
c
32) exhibits a high resistance to see in all of the
solutions of interest (i.e., hydroxides, sulfates, chlorides, etc.)
[36-(0).
Both the VfAW process, using a temper bead technique, in
which the heat of welding serves to temper beads applied earlier
in the sequence. and the SAWprocess have been used for repairs.
Both processes are readily amenable to automation, a key
consideration for applications where large amounts of weld metal
must be deposited and the quality requirements on the deposit are
critical.
The major advantage of the GrAW process is the flexibility and
capability for control that are afforded by separate controls on arc
parameters and filler metal addition. The major disadvantage is
the limitation on deposition rates. Use of multiple wire and hot
wire techniques can improve deposition rates. The GTAW
process, described in several patent disclosures for rotor repair
[41-45], offm some advantages in toughness of as-deposited weld
metal for In general, the fracture toughness of SAW
deposits is less than that of GfAW or GMAW deposits.
Nonetheless, hundreds of rows of turbine rotors have been
successfully repaired using SAW.
Shot Peening
While some OEMs have reportedly adopted shot peening for
additional protection against stress corrosion, at least one other
OEM avoids peening both in initial manufacturing and in repair.
However, the process is increasingly employed in the field by
utilities in conjunction with the removal of cracks by grinding,
generally as a temporary measure until a more pennanent repair
or replacement can be implemented. No perfonnance data was
available for the prevention of the recurrence of stress corrosion
cracking by shot peening in low aUoy steels, albeit the benefit has
been demonstrated for other materials.
Effect of Steel Composition
The 3% erMo disks of the original Hinkley Point design
combined the adverse effects of high strength, high operating
stress and low fracture toughness. It has long been recognized
that nickel additions incn::ase fracture toughness and lower FATI,
but it was thought that increased nickel concentration relative to
chromium would reduce the resistance to stress corrosion
cracking. For this reason 2% CrNiMo steel was selected by
certain OEMs in preference to 2%-3.5% NiCrMoV used
elsewhere. It is now recognized that in addition to improved
brittle fracture resistance, increased nickel level provides greater
resistance to sec, although the effect of yield strength clearly
dominates both the initiation and propagation stages of cracking.
The superior through-thickness hardenability and toughness of
3 %NiCrMoV have been cited as the basis for selecting this
material for welded rotor construction.
The introduction of vacuum degassing and vacuum carbon
deoxidizing into steel making practice by 1975 allowed the
reduction of sulfur levels to the vicinity of 10ppm and the
achievement of total concentrations of tramp clements ofless than
20ppm. The reduction of sulfur in turn allowed steel makers to
lower manganese content to around 200ppm without incurring the
risk afbet shortness [15]. The results were a dramatic reduction
in the size and density of nonmetallic inclusions and the virtual
elimination oftcmpct' cmbrittlement associated with segregation of
tramp clements on prior austenite grain boundaries.
Thanks to a major effort by EPRl and the dedication of Dr.
R. I. Iaffee. the steel industry was able to produce superclean
NiCrMoV for low-pressure turbine rotors in sizes up to 9 fl. in
diamctcrand ingot weights up to 570 tons [46]. One rotor
manufacturer. Iapan Steel Works. has reported delivery of 28
fossil low pressure monoblock rotors [46]. Several potential
advantages have been claimed for superclean rotor forgings. The
most dramatic improvements have been achieved in elevated
temperature properties. bOth in the resistance to temper
ernbrittlement and the increase of creep properties from 350
0
C to
soocC. the latter attributed to the elimination of sulfur-containing
intergranular inclusions. These improvements allow additional
t1exibility in design that result in more efficient and cost effective
turbine designs, such as an increase in the cross-over steam
temperature for a fossil plant and combined cycle steam turbines
with HP and LP sections on the same rotor [47]. The cross-over
temperature has heretofore been limited by the concern for
embrittlement of conventional NiCrMoV rotors. By the same
token, the elimination of embrittlement allows increased stress
levels for the same critical crack size affording the option of
increased last stage blade length.
From the standpoint of stress corrosion cracking the benefits of
superclean low pressure rotors art less dramatic, and the amount
of benefit more problematic. Early stress corrosion .tests of
superclean 3.5 NiCrMoV demonstrated clear superiority with
respect to crack initiation. albeit not much difference in crack
growth rates. For example. Scarlin and Dcnk [48] compared
conventional and superclean 3.5 NiCrMoV at a yield strength
level of 860 MPa in 30% NaOH at lOO"C and demonstrated
incmlSCd initiation life at stresses from 700MPa up to yield. Such
increased life would be expected in view of the reduced size and
density of inclusions in the superclean steel, which are known to
govern the fonnation of pits that may be the site of microcracking.
However. testing of superclean 3.5 NiCrMoV in pure steam at
95
c
C by Holdsworth et aI, [10] failed to establish any superiority
over conventional steels at the same stress level and at stresses
close: to yield exhibited significantly inferior initiation resistance
to that of conventional steels of comparable strength level, i.c.
875 Mpa. Testing was reported to be in progress on superclean
steel wilh a yield strength of74O MPa in pure steam in the attempt
to resolve this apparent discrepancy [10]. Holdsworth et al.
suggested that the discrepancy might be the result of a noted
susceptibility to creep at 95C when loaded to yield point level
stresses, which necessitated repeated loading to counterbalance the
Darryl Rosario
relaxation of stress during the test.
Despite substantial differences in pit depth between superclean
and conventional stee13, it appeared that extensive cracking had
occurred in the superclean steel at times prior to microcrack
initiation in the conventional steel. Holdsworth ct al. observed
that initiation of stress corrosion cracks at pits obtained in low to
medium strength steel with yield stress less than 850 MFa, but at
higher strength levels (above 900 MFa), cracks could initiate at
grain boundaries by a film rupture process. Denk (49) reported
observations of pitting and crack initiation in conventional and
superclean 3.5 NiCrMoV steels at a yield strength level of 700-
7S0 MPa. Pitting occurred readily in stagnant. aerated water,
with much larger density in the conventional steel. However. in
dcaerated water the superclean steel did not ~ t while pits
developed in the conventional steel. Stresses near the yield level
were required to fonn active pits. In 30% NaOH. intergranular
cracks occurred at the same stress level in both steels; initiation
was attributed to a strain controlled, fUm rupture mechanism.
Denk commented that in addition to environmental conditions. the
fihn rupture mechanism is strongly influenced by loading history
effccts.
Lyle [3] investigated the crack initiation lifetime of superclean
and conventional3.5NiCrMoV steels at a yield strength of 740
Mpa (108 lcsi) in 130C (266F) water of three different purity
levels under simulated crevice conditions. The times to initiate
cracking in the superclean steel were approximately 70 percent of
those for the conventional steel. and the cracks were generally
dcc:pc:r. Lyle observed that cracks initiated preferentially in grain
boundary trenches rather than at pits, which he attributed to the
existence of an oxygen depleted zone under the crevice corrosion
fJIm. He commented that this lack of association of cracks with
pitting was consistent with examinations of cracked disks.
The conclusion that can be drawn from the published data on
stress corrosion cracking of superclean steels is that the relative
lifetime in crack initiation is inadequately understood at present,
in particular the role of transient, low temperature creep under
crevice corrosion conditions.
SUMMARY
Comparison of the results of the current industry survey with
those of the 1981 survey indicates that the downstream stages of
nuclear low pressure turbines eventually crack despite lower
surface temperature and presumably cleaner steam. This
observation appears consistent with the dominant effects of
material strength level, oxygen inlea.kage and crevice conditions
on susceptibility to stress corrosion and the effcct oftc:mperature
on the rates of crack initiation and growth. It appears that the
incidence of rim aIIacluncnt cracking is increasing for both nuclear
and fossil turbines.
The current industry approach to rotor integrity under wet f1lm
conditions assumes that cracks have initiated unless direct surface
inspection indicates their absence. When ultrasonic examination
is employed, the presence of cracks with depth dimension equal
to the detectability limit is assumed. Remaining life estimates for
boreJkeyway or rim-attachment cracking are either performed
detenninistically using best-estimates for key variables or
probabilisticaUy to account for the scatter in these variables.
129
Darryl Rosario
Typical estimates of remaining life range from less than 1 year to
10 years or more depending upon the specific disk being
evalWded. Local grindout followed by shot peening is considered
by some to be a suitable temporary repair for a rim attachment.
'There does not appear to be either test data or analysis to support
an extended initiation lifetime for a peened surface. Accordingly
a more permanent repair is planned. or the affected disk is
replaced, or an integral rotor may be substituted to eliminate bore
cracking (and rim cracking, if the yield strength and applied strcu
levels are suitably low). The chemistry of the forging docs not
appear to influence signiflCantly the initiation or propagation
resistance. provided the strength level constant.
In order to extend the interval between turbine outage. it would
be desirable to be able to predict the time, put or future, at which
stress corrosion cracks initiate and, in the proccsa, to take
advantage of improvements in steam chemistry control. Such
prediction would entail knowledge of the corrosion conditions in
crevices ofthe bore or rim attachments, in addition to strc:ngth of
the disk. stn:ss level and temperature. Monitoring of the presence
of wet surface films would aid in assessing the long-term
accumulation of stress corrosion under variable operating
conditions; monitoring film chemistry would obviously enhance
tha assessment, and monitoring of crevice chemistry or potential,
ifit were feasible, would be even better. The current international
effort to monitor droplet nucleation, growth and chemistry in a
bypass circuit of the main steam flow is intended to provide this
local chemistry. To make full use of this infonnation, it would
appear that our understanding of crack initiation in crevices should
be revisited. Several issues require resolution, beginning with the
basic mechanism ofinitiation, which recent research now suggests
may involve strain rate controlled. film rupture rather than pitting.
If this is indeed the case, the accelerating effect of periodic
unloading, the influence of surface cold work on film formation,
the benefit of surface compressive stress, the allowable stress
level relative to the yield strength, and the low temperature
transient creep of the material may come into play.
REFERENCES
1. F. F. Lyle, Jr., "Stress Corrosion Cracking in Low-Pressure
Steam Turbines - An Overview," Paper No. 219, Corrosion
94, 'The Annual Conference and Corrosion Show Sponsored
by NACE International.
2. Sastry Chcruvu and B. B. Seth, "Key Variables Affecting the
Susceptibility of Shrunk-on Disks to Stress Corrosion
Cracking," PWR-Vol. 21, The Steam Turbine Generator
Today; Materials, Flow Path Design, Repair and
Refurbishment, ASME 1993.
3. F. F. Lyle, Jr., Stress Corrosion Crack Initiation in Low-
Pressure Steam Turbines Disk: Steels,lJ Sixth International
Symposium on Environmental Degradation of Materials in
Nuclear Power Systems - Water 'Reactors, The Minerals,
Metals and Materials Society, 1993.
130
4. C. A. Rau, et aI., "Probabilistic Assessment of Crack
Initiation and Growth in Shrunk-Cn Disks, in Proceeding:r
Fossil Steam Tyrbine Djsk Cracking Worksbo.p, EPRl GS-
7250, April 1991.
5. Caligiuri, R. D., et a.i, "Microkinetics of Stress Corrosion
Cracking in Steam Turbine Disk Alloys," EPRI Project
RP1929-8, Final Report. April 1983.
6. Eiselstein, L. E. and Caligiuri, R. D., Stress Corrosion
Cracking of A471 Turbine Disk Steels," EPRl Project
RPI398-12, rmal Report, June 1987.
7. T. Endo et aI., "Material Aspect for the Prevention of
Environmentally-Assisted Cracking in Low Pressure Steam
Turbine: PWR-Vol. 21, The Steam Turbine Generator
Today; Malerials. Flow Path Design, Repair and
Refurbishment, ASME 1993.
8. W. G. Clark, Jr.; B. B. Seth,; and D. H. Shaffer,
"Procedures for Estimating the Probability of Steam Turbine
Disk Rtq:ture from Stress Conosion Cracking," ASME Paper
at ASMEnEEE Power Generation Conference, October
1981.
9. W. David, d. aI., "Disk-Type LP Turbine Rotor Experience,"
The Steam Turbine (hneralor Today: Materials. Flow Palh
De.s;igTl. Repair and Refurbishment, PWR-Vol. 21, ASME,
New York, 1993,83-92.
10. S. R. Holdsworth et aL. "Stress Corrosion in 3.5NiCrMoV
Welded Joints," Clean Steel: Superclean Steel, Conference
Proceedings, 6-7 March 1995, London UK, Book 635
Published by the Institute of Materials, 1996. .
11. O. Jonas and N. Rieger, "Turbine Steam, Chemistry. and
Corrosion: EPRl Research Project 9002-01, Final Report
TR-I03738, August 1994.
12. EPRI Report NP-2429. "Steam Turbine Disk Cracking
Experience: Volumes 1 through 7, Research Project 1398-5,
June 1982.
13. R. Plummer et aI, "BLADE Analysis Capabilities and
Experiences." In Proceedings of the Steam and Combustion
Turbine Blading Conference and Workshop 1992, EPRI TR-
102061, April 1993.
14. Nuclear Regulatory Commission (NRC) Docket Nos. So.
3380L and 5o.3390L, circa 1979-80.
15. EPRI Report NP-4056. "Stress Corrosion Cracking in Steam
Turbine Disks: Analysis of Field and Laboratory Data," May
1985.
16. EPRl GS-71S0. "Proceedings: Fossil Steam Turbine Disk
Cracking Workshop," April 1991.
17. J. W. Cardinal and F. F. Lyle, Jr., Southwest Rcsc:arch
Institute, -Remaining Life Assessment Methodology for LP
Disk Rim Cracking" Proceedings: EPRI Turbine and
Generator NDE, Life Assessment, Maintenance and
Monitoring & Diagnostics Workshop, July 16-19 1991,
Charlotte, North Carolina.
18. EPRJ Report NP-6444, -Guidelines for Predicting the Life of
Steam Twbine Disks Exhibiting Stress Corrosion CrackUlg:
Volumes 1 and 2, July 1989.
19. D. A. Rosario et al, "LP Rotor Rim-Attachment Cracking-
Survey of Utility Experience," EPRI Research Project 9005-
01, Final Report, October 1996.
20. D. L. Goodlin etal., Southwest Research InstitUlc, J. E. Jilek
et al., WISconsin Public Service Corp., "Automated
Ultrasonic Inspection of Steam-Turbine Disk-Rim Blade
Attachments, Case Study: Kewaunee Nuclear Power Plant"
Proceedings: Fourth EPRI Turbine and Generator NDE, Life
Assessment. Maintenance and Monitoring & Diagnostics
Workshop, August 1-41995, Milwaukee. WISCOnsin.
21. Personal communication, J. Crane, Southwest Research
Institute, February 1996.
22. L. Nottingham, K. Solomon, and H. Presson. "Phased Array
Ultrasonic Approach to Turbine Blade Auachment
Inspec:tion." grlb, 12
th
International Cooferc;nce
00 Ngndestructiye Evaluation in the Nuclear and Pressure
Vessel Industries. Metals Park, OH: ASM International.
1994.
23. Inspection of Turbine Pisk Blade Attachment Guide -
Volyme 1 Baclcgtound aust Inspection Princjples. Palo Alto,
CA: Electric Power Research Institute, August 1995. TR-
104026.
24. D. Lessard, C. Rose, and P. Sabourin. "Comparison of
Ulbasonic Techniques for the Purpose of Inspecting Straddle-
Mounted Dovetail Blade Attachments." Proceedings' Fourth
EPBI Turbine/Generator WQrksbo,p. Palo Alto, CA:
Electric Power Research Institute, August 1995. TR-I05613.
25. R. Viswanathan, "Application of Clean StecVSupen:lean
Steel Technology in the Electric Power Industry - Overview
,of EPRI Research and Products," Clean Steel: Superclean
Steel, Conference Proceedings, 6-7 March 1995, London
UK, Book 635 Published by the Institute of Materials, 1996.
26. F. D. Mansfield and R. Munson. "Ten Years of Welded
Repair on Steam Turbine Rotors, An Insurer's Perspective.-
Proceedings: Welding and Repair Technology for Fossil
,Power Plants, 1994 EPRI International Conference, March
23-25. 1994, Williamsburg. Virginia.
Darryl Rosano
27: K. Fuentes, K. Oberle, "Laboratory Study of a Steam
Turbine Rotor Weld Repair", ASME Paper 87-JPGC-Pwr-15,
The American Society of Mechanical Engineers, New York.
NY, 1987.
28. "Large Steam Turbine Repair Survey", EPRI NP-7385,
Project RPl929-15, 1991.
29. C. E. Chaney, "Weld Repair of Gcothennal Turbine Roton" ,
GeOlheT71UJl Resources Council - Transaclions, Vol. 13.
1989.
30. A. TIptOn, "Steam Turbine Rotor Weld Repair Development
Based Upon F"1lnc:ss for Purp>sc Philosophy", De.sign. Repair
andRefurbishRJe1l1ofSzeam Turbines, PWR-Vo113, ASME.
New York, 1991,249-253.
31. "Welding & Repair Technology for Fossil Power Plants".
EPRl Conference, March 23-25. 1994.
32. G.W. Galanes, T.D. Spry, J.E. Indacoches, "Filler Metal
Evaluation for Repairing CrMoV Steel HP Rotors", Design,
Repair. and Refurbi.shnu!nt ofSteam Turbines. ASME, 1991,
241-248.
33. T. J. Stiebler, R. M. Nugent, R. P. Wilson, "Welding for
Life" Advances in Steam Turbine for the Power
Ge1lUatio" Industry, PWR-Vol. 26, ASME, 1994, 143-158.
34. D. R. Amos et aI., Westinghouse, S. McQueen, Houston
Lighting and Power Co., M. English, Public Service
Company of Oklahoma, "Repair Welding Experiences for
Low Pressure Rotor Dovetail and Finger Dovetail Blade
Attachments Proceedings: Fourth EPRI Turbine and
Generator NDE, Life Assessment. Maintenance and
Monitoring & Diagnostics Workshop, August 1-4 1995.
Milwaukee,
35. E. Kramer. et aL, "Advanced Reliable Low Pressure Steam
Turbine Retrofits", Advances in Steam Turbine Technology
for the Power industry, PWR-VoI. 26, ASME.
New York. 1994, 89-98.
36. "Valve Stem Failures", United States Nuclear Regulatory
Commission IE Information Notice No. 85-59. July 17,
1985.
37. Failure Analysis and Prevention, Metals Handbook. Ninth
Edition, Volume 11, ASM International.
131
Darryl Rosario
38. CJ. Czajkowski, of Failed Type 410 Stainless
Steel Shafts from Main Steam Isolation Valves at Two
Operating Plants", CORROSION/S7, Paper No. 97, NACE,
Houston, TX, 19S7.
39. H. Tanczyn, "Properties of 12% Cr Alloys Modified with
Small Cb Additions", Advances in rhe Technolcgy of
Slainless Steels and Related Alloys, ASTM STP 369, 1963.
40. R.W. Staehle, ct. aI., "Corrosion and Corrosion Cracking of
Materials for' Water-Cooled Reactors", FCC-'n04, EPRI
Final Summary Report (1-1-75 through 12-31-77), EPRI
Project RP311-1.
41. US Patent No. 4,958,431, "More Creep Resistant Turbine
Rotor and Procedures for Repair Welding of Low Alloy
Ferrous Turbine Components" .
42. US Patent No. 4,897,519, "More Creep Resistant Turbine
Rotor and for Repair Welding of Low Alloy
Ferrous Turbine Components".
43. US Patent No. 4,893,388, "Method of Modifying Turbine
Rotor Disks" .
44. US Patent No. 4,940,390, "Turbine System Having More
Failure Resistant R.ot:ors and Repair Welding of Low Alloy
Ferrous Turbine Components by Controlled Weld Buildup" .
45. US Patent No. 4,903,888, "TUrbine System Having More
Failure Resistant Rotors and Repair Welding of Low Alloy
Ferrous Turbine Components by Controlled Weld Buildup".
46. Y. Ikeda et a1., "Production and Properties of Superclean
Monoblock LP Turbine Rotor Forgings," Clean Steel:
Superclean Sleel, Conference Proceedings, 6-7 March 1995,
London UK, Book 635 Published by the Institute of
Materials, 1996.
47. M. Yamada et aI., A New Superclean High Prcssure--Low
Pressure Rotor Steel for Advanced Combined Cycle Power
Plants, .. Clean Sleel: Superclean Steel, Conference
Proceedings. March 1995, London UK, Book 635
Published by the Institute 1996.
48. R. B. Scarlin and J. Denk, Stress. Corrosion Craclcing
Behavior of a Clean 3.5% NiCrMoV Steel," Report GS-
6612. RPI403-32, Electric Power Research Institute, Palo
Alto, CA, July 1990.
49. J. Denk, A New Superclean High Pressure-Low Pressure
Rotor Steel for Advanced Combined Cycle Power Plants,
Clean SteeL' Superclean Steel, Conference Proceedings, 6-7
March 1995, London UK, Book 635 Published by the
Institute of Materials, 1996.
Table 1
Time to Initiate Corrosion Cracks in Crevieed Regions by Lyle [3)
(All tests at 130C or 266F)
Time to Initiate Cracks (Hours)
Yield Strength High-Purity Water
Intermediate- Contaminated
Steel 1
Oxygen Water
MPa (lail Environment 1
Environment J Environment .-
AH 934 (13S) 8.585 425 1,289
AL 780 (113) > 17.520 1,924 1,989
JH 941 (136) 8,625 1,866 2.106
JL 741 (l08) > 17,520 2.190 3,979
IP 738 (107) > 17.520 1.683 2,650
Notes:
132
1. High-strength and 3.5NiCrMoV American steels a.re designated AH and AL. respectively, while
Japanese steels are designated JH and JL. respectively. The Japanese high-purity steel is designated lP_
2. Representative of secondary-side wa.ter in a PWR during nonnal operation with less "than 5 ppb of oxygen.
3. Representative oftypic.al short-tenn upset conditions with 40 ppb oxygen and 20 ppb NaCl.
4. Representative of a. severe secondary water upset condition with 6-8 ppm oxygen and 20 ppb of NaCI.
Darryl Rosario
S132Bl Steeple. Type: Crack
Hook "umber 2
. :CCCC
Figure 1. SCC Initiation Probability versus Initiation Times
Normalized to account for Stress and sec Growth Rate (Rau et.
al.[4])
.6
I:
....
.
l:
.2
'0
.1
.a
.06
.a
"
a:
.0
..
.02
il ,
a
,
.01
...
".
.006
"
5 10
_ulUa" UN
50
o
o
o
I
.1
Figure 4. Sample Remaining Life Results of SWRJ's LAST
Computer Code [l7]
PIt
growth
-..r--
olSe RUuBu.oE
ATTACl-IMtNT .loRE).

CRACK ORIENTATlONS

..2. -Cireumf,rential

LCC..1Tl0NS
CD RIm -- AI:cIj-Clraanfetelllial
Onento""" I TrIM' I )
CD l<im --A:lial-Roaial
Onent..,i... IT1pall l
<D
On_Ii01\
o Wee -Allial-Itodial
Ori.,t"'_
CD Ke,-, -- Asial-Ra<liCI
enatita1iati
o Bore Surloce--AxiaI-Rad
ia1
Orientation
I
!
i i
I
- \:
pn i
Figure 2. Stress Corrosion Crack Initiation and Growth Model
per Endo et. al. [7]
s_ c:rac_lni1iatiDn,
Rscc o-....'Rpa,z
0-. LacaI Tn..... opet2li"9 $_.......
Figure 5. Illustration of Typical Locations and Orientations of
Cracks in U.S. LP Rotor Disks (12]
Figure 3. Stress Corrosion Crack Initiation Results bv Siemens
[9] .
133
Darryl Rosario
---1
--
-----.....,1'
OCRACUD
CUNCRACKEO
0-
201

601
101
10
10
lo
1001 1201
200
10
10
10
to
800
800
1000
12011

Unit _ (MW)
30
--
=1 L:_
10-
3S.
Figure 6. Rim Attachment Cracking in Nuclear Units by Reactor
Type and Manufacturer
Figure 8. Rim Attachment Cracking in Nuclear Units by Power
Rating (MW)
I.
--I
16
1Jl:IlACKl:tl
IJUNCAAI'Kl:D
14
..
12
::l
10
,;
::l
Z
lI-
t
'0
z
Rolor Opeqtin; Houra (Th""aanda)
Figure 7. Rim Attachment Cracking in Nuclear Units by
Turbine Manufacturer
Figure 9. Rim AtIachmem Cracking in Nuclear Units by Total
Operating Time
134
Darryl Rosario
ti-
t
12a
"";
1 ...
'1.
111

U
'1
1
I
I
. C_s.-o,OalII(.GIleI
!

-
.
.
I
i
I
.
-
.
- - - -
.
. .
-
-I-
- -
-
!
I
f I
I
!
i
i
A
1
+
Ar
I i
I
I
I
I I
I +
I
i
i
I
i I
I
i
I
I
!
!
I I
i
!
! !
I
i
I
I
i
!
I I
: !
lfo07
Ui:1j4
i 'f-G.
i
1E-otS
Figure 12. Apparent Crack Growth Rates Versus Disk Yield
Strength from 1995 Survey Compared with Mean Westinghouse
Data and a Best Fit to the 1980 Survey Data
, 80'"

- .-------J
--- - -. --j
---l
---,. . --'-1
-_._J
.j
I
20
Figure 10. Rim Attachment Cracking in Nuclear Units with GE
Turbines by Row Number
toO
eo..
L.J
L-" L-Ei L-7 L--8

ElV'"'
------...........-
L7
Figure 13. Rim Attachment Cracking in Nuclear Units with
Westinghouse Turbines by Row Number, Per the 1980 Survey
Figure 11. Rim Attachment Cracking in Nuclear Units with
Westinghouse Turbines by Row Number
Figure 14. Number of Fossil Units Surveyed by Steam Type
(Supercritical or Subcritical) and Fraction with Rim Attachment
Cracking
135
Darryl Rosario
Figure 15. Rim Attachment Cracking in Fossii Supercritical
Units by Boiler Manufacturer
Figure 16. Rim Attachment Cracking in Fossil Units by Turbine
Manufacmrer

:: lOCI
:.
i
..
!
...
Figure 17. Rim Attachment Cracking in Fossil Unies by In-
Service Year
136
. Figure 18. Rim Attachment Cracking in Fossil Units with
GE Turbines by Row Number
Aow........... m.ml,.uISeto-
Figure 19. Rim Attachment Cracking in Fossil Units with
Westinghouse Turbines by Row Number
BOT
70
~
60
"iii
e-
50
Q)
a:
'0 40
~
30
E
::::l
Z
20
10
0
Betore 1983
1983
Darryl Rosario
1988 1990 1992
1989 1991 1993
Figure 20. Ultrasonic examination of GE straddle-rnountdesign
rim-attachment
~
I (
I \
I \
Type 1 S ~ l e C:adt lnsped:iDn
Cow:rage Zones. PuJse.iIo ~ ZICIIWS
'10M depic:t2d as ovals and pitdKateh .::0ver-
age zcmaU c:iJda
Figure 21. Ultrasonic examination of Westinghouse axial-entry
design rim-attachment
Figure 22. Number of Weld Repairs To Utility Stearn Turbine
Generators by Year [26]
._........... ~ ..
...... r.-.......,
~
Figure 23. Illustration of the Types of Weld Repairs That Have
Been Performed [261
137
138
Probabilistic Assessment of Failure by Stress Corrosion Cracking in
Shrunk-On Disks on Low Pressure Turbines
Presented By
Darryl Rosario
139
140
---------------------------------
Probabilistic Assessment of Failure by
Stress Corrosion Cracking in Shrunk-On
Disks of Low Pressure Turbines
Prepared for:
EPRI Steam Turbine Stress Corrosion Cracking Conference
March 19-20, 1997, Baltimore, MD
Prepared by:
Darryl A. Rosario
Blaine W. Roberts
Tony Khalid
Michael F. Steakley
Tennessee Valley Authority
_ _ _ _ ~ _ .._. _..---_._._--_.. _-----_.----- _.-.__._-----------------------
141
142
PROBABlusnc ASSESSMENT OF FAILURE BY STRESS CORROSION CRACKING IN
SHRUNK-ON DISKS OF LOW PRESSURE TURBINES
Darryl A. Rosario
Structural Integrity Associates
San Jose, CA
Blaine W. Roberts, Tony Khalid and Michael F. Steakley
Tennessee Authority
Chattanooga, TN
ABSTRACT
Scvend largeLP twbinc:s in thi:: TVA system utilize shronk-on disks
which an:: keyed to the shaft. The 7th, 8th, and 9th stage disks
experience wetness in operation which renders them potentially
subject to stress Corrosion cracking (SCC) in the keyway ofthe shrink
fit area. To minimize sec concerns in the disk keyway, TVA has
embarked on a phased approach to refurbish the LP rotors with a "fat"
shaft and a '"tab" design replacing the rectlingular keyways. Non-
destructive examinations have been performed on rotors being
refurbished andthose continuing in service to assure that the schedule
for refurbishment does not place operating units with the original
design at undue risk.
Many variahIes are involved in the quantification ofthe risk: of disk
failure from sec: incubation time, stress intensity distribution as a
function ofcracksize, material sec rate as a function ofstrength and
environment, fatigue crack growth rate, loading history (cspc:cially
overspeed events), and fracture toughness. For a given set of
assumptions and parameters, a deterministic model may be used to
assess the risk of failure. Unfortunately, if mostly conservative
assumptions arc made for the random variables, it can be concluded
that small initial flaws will grow to critical sizes in two years of
additional operation.
As an alternative to the deterministic a probabilistic
Monte Carlo simulation was performed using ten critical variables.
Variable distributions were examined by binning data values within
threc standard deviations above and below the mean. One million
iterations wen: performed for each disk.
The probabilistic evaluation concluded that initial crack depths of
0.45, 2.35, and 2.95 inches were tolerable on the 7th, 8th, and 9th
stage disks, respectively, for two years of addition operaticn with a
very lowcumulative probability offailurc of 0.0001. Since no cracks
approaching these sizes had ever been found by NDE and confinncd
when unstacking the disks, it was concluded that continued operation
of a rotor with replacement scheduled in two years posed a minimal
risk: and there \VBS no need to accelerate the refurbishment schedule.
INTRODUCTION
The TVA power system, comprised of coal-fIred. nuclear.
pumped storage, combustion turbine, and hydro generating units.
has a total system capacity of over 32,000 Megawatts (MW). The
coal-rued generating capacity is over 17.600 MW and is
comprised of 59 units located at cleven sta.tions in Tennessee,
Alabama. and Kentucky. The average age of the coal-fired units
is over forty years.
There are 2SS active fossil fueled steam tuIbine and generator
rotor forgings. Additionally, there are 63 other rotors including
spares and rotors recommended to be retired by the OEMs. This
fleet represents a diverse cross section of the equipment supplied
by Westinghouse, General Electric. Parsons. and Asea Brown-
Boveri.
In low press (LP) turbines, TVA has several GE machines
which utilize the shronk-on wheel design commonly employedby
GE since the mid-1940s [1}. At the Paradise Fossil Plant, Units
1 and 2 (650 MWeach) and Unit 3 (1150 MW) have this design
feature. In all three units, two LP double flow (LPOF) turbines
operating at 1800 RPM are used. This paper addresses the
shrunk-on disks for the Paradise Units 1 and 2 LPOF turbines.
Each flow in these turbines consists of nine stages. The major
is with stress corrosion cracking (SCC) in stages 7-9 (L-
2, L-1, and L-o respectively) which run wet either intennittently
or continuously during operation.
Stress corrosion cracking of shronk-on disks in LP turbines has
received considerable attention over the last 25 years because of
the large number of cracks which have been detected during
inspections and the OCCUrrence of an occasional rotor burst [1-3].
Since recognition of this iSsue, TVA has non-dcstnJctively
examined (NOE) the affected rotors during planned outages.
Additionally, a program to upgrade the GE LP fleet to minimize
the threat of sec with an improved fat shaft and tab design was
initiated in 1993.
During the March 1995 Paradise Unit 1 planned outage, the
LPB rotor which was removed for upgrading was inspected using
the Ultrasonics Data Recording and Processing System (UDRPS).
Inspection of the L2 disk, which is most critical because of a high
fracture appearance transition temperature (FATT), revealed sevcral
boreIkcyway indications, ranging from about 0.1 to 0.67 inch 'in depth.
The immediate question was whether to return the LPA rotor to
service without performing a UDRPS examination since LPA Totor
removal had not been plaJUled. Depending on the assumptions, critical
flaw sizes on the:: order of 0.3 to 0.7 inch in depth had been estimated
for the critical L-2 disk. Based on the LPB findings, it was decided to
remove the LPA rotor and perfonn a UDRPS examination. The
inspection for LPA showed smaller indications (less than 0.070 inch
in depth for the critical L-2 stage). A lower bound detenninistic
analysis demonstrated that the LPA rotor could be returned to service
with operating restrictions (prewarming and restrictions on overspc:ed)
for the two year interval before its planned replacement by an
upgraded rotor.
143
Darryl Rosario
Resulting hoop stRss distributions at speeds of 0, 900,1800,
1980 and 2160 rpm are shown in Figure 3 for the 7th stage disk;
stress distributions for the 8th and 9th stages were similar with
bore stresses of Sl and 45 ksi at 1980 rpm, respectively. These
stresses do not account for the stress concentrating effect of the
keyway which was addressed separately as described below.
where R is the keyway depth which was reported to be 0.625 inch.
Fmite element slrcss distributions modified to incorporate the
keyway stress concentrating effect are shown in Figure 4 for the
7th stage disk:. Because average yield strength values for the disk
material were above 90 lcsi. any yielding or plasticity due to the
keyway stress concentration was assessed to be highly localized.
Stress Coocentrating Effect or Keyway: A schematic illustrating
stress concentration factors (KJ of 2.2 for the existing axial
keyway design and 1.2 for the proposed tab design is shown in
Figure 4. Stress concentrating effects in the GE axial keyway
design have been studied in detail under prior EPRl research
projects [5]. The gradient or die-away in the peak stress
concenlration factor (KJ of 2.2 with increasing radial distance (x)
from the crown of the keyway has been given by the equation:
1+(1<.-1) + O.7S(..E-)4] (2t
R+x R+x
shrunk-on disk! were perfonned to determine the distribution of
stress in these disks due to the combined effect of shrink fit ar
centrifugal mechanical loads. The contribution of thenn...
transient effects to stresses in the disk during startup was assessed
to be insignifiC8Dt relative to the mechanical loads, because of low
tc:mpera1UrC3 and ramp rates for these disks which are located ncar
the exhaust end of the LP rotor.
Two-.dimensional (20) axisymmetric fmite element (FE)
models of the subject disks were constructed and analyzed to
detemUne nea.r-blre disk stresses due to shrink tit and centrifugal
effects using the ANSYS computer program [4]. The FE model
ofthc 7th srage shrunk-<>n disk and rotor shaft is shown in Figure
2. This FE model, which is typical for the 8th and 9th stage disks,
includes one-half of the disk with a symmetry boundary condition
imposed at the midplane of the disk, as shown in the figure.
Centrifugal forces generated by the blades were applied as traction
loads at the periphery of the disk and shrink fit betWeen the disk
and rotor shaft: was simulated using 2D gap elements. The amount
of radial intefen:nce between the disk and rotor shaft was not
provided by the manufacturer but was calculated based on a
complete loss of shrink fit for this design at 120% of rated speed
or 2160 rpm. Two stress analyses were performed; one at 2160
rpm to calculate centrifugal stresses and determine shrink
fit, and, the other at at 0 rpm to calculated stresses due to shrink
fit only Stn:sses due to both shrink fit and centrifugal effects
at any iJltcnnediate speed (RPM) were calculated using the
relationship:
(1)
[ (
RPM\2J [(RPM-,1]
u.. 0 1 - 2i6O) + 0,... '2i'6OJ
Since a similar outage scenario (replace one rotor, continue to
operate the other) was scheduled for the Unit 2 Fall 1995 outage. it
was decided that a more comprehensive probabilistic assessment was
warranted to properly quantify the risks associated with continued
operation of a potentially flawed rotor. Such an analysis offered the
potential for predispositioning indications found during the Unit 2
outage. This paper focusses on the approach utilized for the
probabilistic analysis and the associated results.
The ultrasonic results for both LPA and LPB rotors are
shown in Table 1. All stages had ultrasonic indications. In the
initial treatment of the data for LPB, keyway cracks as deep as
0.52 in. depth and isolated bore cracks as deep as 0.67 were
noted. In subsequent refinement of the LPB data. the maximum
keyway depth was revised to 0.25 in. For LPA. the maximum
keyway crack depth observed was 0.325 in. while the maximum
bore craclcwas estimated as 0.080 in deep. The critical 7th Stage
for LPA had no keyway or bore craw exceeding 0.070 in. depth.
INSPECTION
The ultrasonic inspection of LP turbine shrunk-on disk bores and

keyways is a geometrically complex but an ultrasonically simple
task. Ultrasonically, the marerials are quiet, acoustically non-
dispersive forgings. The keyway face, when present, produCQ an
unmistakable geometric reflector which servQ as a landmark for
all observations. A clear demonstration of adequate sensitivity is
the "backwall
Tl
roll produced by the disk In bore surface
texture.
The geometric complexity imposed by the design of the
individual disks and by the clearances between them create a
number of problems. The major problems to be overcome are:
Design and fabrication of multiple degree of freedom
scanner that permits manipulation of the transducers
over as large a fraction of disk surfaces as possible.
Proper selection of transducer wedge and/or skew
angles.
The scanner/manipulator is mounted on a scaffold platfonn
and the turbine is rotated past the transducers. Thus,
circumferential bands of observation are made for 360 degrees
around the Pulse-echo techniques are U3ed under the hub
and trough areas; pitch-eateh techniques are used in the midplane
of the web; e.g., the plane of the disk. Pulse-echo techniques may
also be used on selected web faces to explore the central part of
the shrunk-on area. 1be latter observations are not sensitive when
made from web surfaces that are inclined at less than 10 degrees
with respect to the midplane of the disk. Figure 1 illustrates the
approach used for the ultrasonic examinalion .
The skew angles used and the angles of incidence employed
are calculated from simple measurements of disk geometrY.
The data are digitized and fully recorded using the UDRPS
system. 'These data fonn a pennancnt record of the inspection for
future reference. The data are reviewed and analyzed for the
pn:sc:nce of indications and to establish the dimensions and nature
of indU:ations.
STRESS ANALYSIS
Finite element stress analyses of the 7th, 8th and 9th stage
144
Darryl Rosario
CRITICAL CRACK SIZE DETERMINATION
A key factor in estimating remaining life 0 f the 5ubjca disks
is determination of critical crack size or maximum tolerable size
before unstable fracture occurs. Linear elastic fracture mechanics
principles were employed because the effects of yielding near the
keyway crown were predicted to be highly localized. Stress
intensity factors (KJ were romputed as a function of crack size (a)
for various crack: aspect ratios, which then permit determination
of critK:al crack size (a
er
) for a given fracture toughness
CorrelaJion [8] given by the equations:
" 5 (CVN)
E
and,
T:IIIft = 215 - 1.50yo
(41
(5)
Factor Solutions: The solution typically used
by the manufacturer is given by the equation:
Fracrure Toughnus: Disk-specific fracture toughness (Klc) values
were estimated from I'OOlll4.emperature charpy impact (CVN) data
and small-sample test data provided by TVA. A statistical
summary of CVN data along with other relevant mechanical
properties is provided in Table 2. Worthwhile noting is the fact
ttwlow CVN and high FAIT values for the 7th stage disk. which
is constn.Jcted of lCr-O.SMo steel. make it much more susceptible
to brittle failure than the 8th and 9th stage disks which are of
1.75Ni-1.2SCr-O.5Mo-O.1V steel.
Disk 7: During startup disk temperatures are in the 7S"F to
lS00P range; because FAIT values are in the -78F to 140F
range. the transition-temperature relationship was assessed to be
most appropriate. Three methods were used to estimate from
CVN. First. using the Barsom-Roife Two-Stage CVN-Kftl"K
k
where, 0 ill bore stress, Sl8-inch is the keyway radial depth,
and, a is crack depth measured from the keyway surface. This
solution, which equates to the formulation for an edge-.;racle in a
uniform stress field, is overly conservative because it does not
represent typical keyway cracks which are semi--elliptic in proflle.
does not include the effect of the decreasing st1'css gradient and is
far more compliant than the actual geometry which is a radial-
axial crack emanating from the inside surface of a thick hollow
cylinder.
To obtain a more accurate K
l
solution. the model for a semi-
elliptic crack at the 10 of a hollow cylinder, shown in Figure 6,
was investigated using 51' s in-house fracture mechanics code pc-
CRACx: [6]. Several crack aspect ratios {radial depth, a. to
su.rface length, 1. ratios) were ronsidcred for this model: (i) Planar
(a/]=O), (ii) 10:1 (aI1=O.I), (iii).5: 1 (all=O.2), (iv) 2: 1 (aII=O.5);
with the planar crack representing a worst-case scenario and the
semi-.;irc;:ular cracle (2: 1) representing the most favorable crack .
geometry. Finite-length crack models are believed to be more
representative of actual cracks observed in shrunk-on disks than
the conservative planar assumption. For example, most of the
cracks documented for the failed Yankee Rowe LP turbine disks
were approximately semi-.;ircular in shape [7].
"K! versus a" results for the various crack aspect ratios arc
shown in Figure 7 for 7th stage disk at 10% overspeed (1980
rpm). The manufacturer's solution (Equation 3) is also shown in
this figure to illustrate the over-eonservatism relative to the 51
models. Results for the 8th and 9th stage disks are similar to
those shown for the 7th stage disk.
A lower bound K
1c
value of 181 ksi/inch was estimated for
the minimum reported CVN value of 70 ft-lbs and an average
yield strength of 101 ksi (see Table 2).
(6)
= 15.873 {CVN)3Ill
Disk 7 K1<; estimates for lower bound and average CVN
values of23.5 and 72.6 ft-Ibs were 101 and 128 ksi./inch.
Finally, small punch fracture toughness testing was pefonned
on four samples, two from thegenerator--end disk and two from
the turbine-end disk [10]. The lowest estimated value was
54.7 ksi./inch for the generator-end disk with an uncertainty of
+1- 25% based on published data [11]; Le., a possible lower
bound toughness of 41.0 ksi./inch. The estimates of 92.8 and
96.5 ksiv'inch for the turbine-end disk were more consistent with
the K
1c
estimates from CVN data described above.
Disks 8 and 9: These are Ni-Cr-Mo-V disks, with FAIT
values in the -135F to OF range (Table 2). Therefore for
normal operation these dislcs are in the upper shelf region. The
following upper-shelf relationship was used [8]:
(K
1C
)2 = 5 (CVN - ,0.05)
Oy Oy
a lower bound K
Ic
value of 108 ksiv'inch was predicted for the
lowest measured CVN value of:!3.5 ft-Ibs, and an average yield
strength of 89 ksi. The estimate for an average CVN of 72.6
ft-Ibs was 153 lcsiv'inch.
Next, in the above procedure. equation 4 was replaced by the
Corten-Sai/crs relationship [9] for K;d:
Critical Crack Sizes
With the Kt versus a results and fracture toughness estimates
discussed above, deterministic estimates of critical crack size are
as follows:
Disk 7: Critical crack sizes are summarized in Table 3 at
75F (room temperature) and at-100F and 150F corresponding
to start up without and with prewarming at 900 rpm. Room
temperature K
1c
values of 100 ksiv'inch and 55 ksi./inch
rom:sponding to minimum CVN and small-punch measurements
were included in the calculations. Minimum critical crack sizes
predicted without and with prewanning were 0.14 and 0.4S inch,
respectively.
Disks 8 and 9: . Ktc values at nannal operating conditions are
in the upper shelf regime, and are therefore independent of
operating temperature. Critical crack sizes were estimated at
speeds of 900. 1800 and 1980 rpm. Results .3.re summarized in
Critical crack sizes predicted were larger than 1.7 inches
(3) K.. = 2aJ5J8" .. a
145
Darryl Rosario
for Disk 8 and larger than 3.5 inches for Disk 9, for the estimated
lower bound toughness of 180 ksiv'inch.
The methodology and results for assessing initial and critical
crack siza have been discussed in detail. The issue of crack .
growth rate is addressed below.
Stress Corrosion Cracking (SCC)
To date, the most widely accepted SCC crack growth
equation uxd in the analysis of cracking in bores and keyways of
LP turbines from Westinghouse research [12], is given by:
REMAINING UFE ESTIMAnON
Remaining life of the LP rotor disks with regard
to keyway cracking is controlled by the initial or detected crack
size (ai)' the critical crack size (a... ) and the crack growth rate
(da/dl), given by the following relationship:
Deterministic Remaining Life Estimates
Using equations 8 and 9 described above, deterministic
c:stimaI.es of remaining life (t
rem
) were computed assuming wOTSt-
case and mean properties as a function of detected initial crack
size (llj). Detenninistic results for the 7th stage disk without
prcwarrning, shown in Figure 9, illustrate that it would be
extremely difficult to justify continued operation for 2 years
wilhout prewanning because critical crack sizes are small. With
prewanning, however, continued operation for 2 more years could
be justified for initial crack depths up to 0.36 inch (10:1 flaw).
Also apparent is the wide scatter in deterministic predictions for
worst-case versus mean SCC growth rates.
Results for Disks g and 9 using worst-case properties indicate
that large initial crack sizes upto 2.6 and 3.4 inches, respectively,
can be tolerated for two additional years of operation. Therefore,
the probabilistic approach described below will focus on the 7th
stage disk which is most life limiting.
of lOS ksi, corresponding to the Powcrgen data, was used in
equation 9. The comparison. illustrates lhat the US nuclear dis:
data bounds both the Powergen fossil disk NiCrMoV and CrMo
data, further substantiating the applicability of equation 9 to the
Paradise NiCrMoV and CrMa disks.
Probabilistic Remaining Life Evaluation
The large scatter in detenninisticaUy predicted remaining
lives for wom-case and mean data discussed above, suggests that
variability in modelled data cannot be adequately characterized
detenninistically. The use of worst-case assumptions stacks
conservatisms on conservatisms, resulting in an overly
conservative estimate of remaining life which does not represent
a realistic outcome. It is unlikely that all of the worst case
conditions would occur simultaneously, and therefore a
probabilistic analysis that considers position in the scatterband can
provide a mote realistic assessment of the suitability of a given
component for continued service.
The first step in a probabilistic evaluation requires
identification of appropriate random variables and determination
of a statistical distribution associated with each variable. Then
probabilistic results can be generated using a Monte Carlo
simulation technique which involves successive detenninistic
remaining life calculations using randomly selected values of
inputs. The probabilistic approach is illustrated schematically in
Figure 10.
Random Variables: The and distribution of random
variables for the 7th stage disk is summarized in Table 6 based on
data evaluation and discussions between TVA and SI personnel.
A brief description of selected variables is given below.
Vgrigb/( #1 Sims or KJ-tmZL
1be FEM analyses were assumed to provide accurate results
with a maximum error of 5% due to errors in either geometric
infonnation or applied loads. A nonnal distribu[ion was assumed.
Vqriqblc #2' IQ! Overspna' PQUibiliry
The possibility of a 10% overspeed event (1980 rpm) was
assessed to be no more [han %of the total number of startups.
VqWI, 1t3. FraclUu Toughnns at RQQm Tmwcraruu
A fairly large variability in fracture toughness was noted for
the 7th stage disk. A normal distribution was assumed with a
mean of7S ksiv'inch based on the CYN data and a lower bound (3
(8)
(9)
a -lit
t = -"'-
.... . daldt
In(daldt) =C, - (7302JT) .. O.02780
y
Crack Growth Rate
The most significant crack growth mechanism in low
pressure rotor disk keyway craclcing has been identif1ed
to be stlcss conosion cracking (SCC). Crack growth due to low-
cycle fatigue startups and shutdowns is typically insignificant
relative to SCC, especially in the case of base-loaded units like
Paradise for which low-eycle fatigue crack growth was estimated
to be less than 0.001 inch/year assuming 10 starts per year. This
value is two orders of magnitude lower than typical SCC growth
rates, discussed below, and is therefore insignificant.
where, daJdt is the growth rate in inches/hour, C
1
is a
constant having a mean value of 4.968 and an estimated standard
deviation of 0.587, T is the operating temperature of the disk in
oR, and, 0y is the yield strength in Scsi.
This equation has been accepted by the u.s. Nuclear
Regulatory Commission (NRC) for use in analysis of cracking in
bores and keyways ofLP turbines in U.S. nuclear plants. While
this equation is applicable to 3.5NiCrMoV steels, research studieS
[13J have shown thalcraclc growth rates are relatively independent
of steel composition so long as the microstructures are alike. It is
therefore reasonable to use this relationship for the 7th stage disks
which are constructed of Cr-Mo steel. A summary of SCC crack
growth rates using equation 9 for the 7th, 8th, and 9th stage disks
is provided in Table S. Results for both mean and upper bound
(three standard deviations above the mean) data are included,
illustrating the wide scatter in the data.
The above data was developed to address SCC in nuclear
turbines. To asSC$S the applicability oftrus data to fossil turbines.
SCC data on fossil NiCrMoV and CrMoY disks presented by
Powergen, U.K. [14] at the 1990 EPRI fossil disk cracking
workshop was compared with the above data as shown in
Figure 8. To make the comparison, an average disk yield strength
146
standard deviations below the mean) of 41 ksiv'inch based on a -
25% error in the lowest small-punch estimate of 54.7 ksiv'inch.
Variables #4 5' Disk Temperq,"" at Keywgy wirhoyrlwith
Prmqmting
As discussed earlier, the fracture toughness for the 7th stage
disk, which is in the transition regime and then:fore dependent on
the disk temperatun:, can be significantly improved by
prewarming. Assuming the disk is not prewanncd. the local
keyway temperatun: was assumed to have a normal distribution
with a mean of 100"F and a standard deviation of 5F. The mean
disk tempetature achievable with prewanning was estimated to be
150"F. Prewarming was pnxiicted to have no significant effect on
toughness for the NiCrMoV Disks 8 and 9. which wen: reported
to have FAIT values below OF.
Vgriqble #6 Crack Models
To reduce the conservatism associated with a planar crack,
probabilities of occurrence were assigned to crack aspect ratios
as foUows: 5% for a planar crack, 60% for a 10:1 elliptic crack,
25% for a 5:1 crack: and 10% for a scmi-.circu1ar crack (2: 1).
This distribution was considered to conservatively represent the
non-planar configuration of documented disk: cracks.
Yqriahl( #7' T1!itiq[ Ccd Size
Information provided by TVA indicated a 15% variability in
detected crack: size. A rangc of mean initial crack sizes was
investigated for each disk so that thc probability of failure was
centered around a value of lxlO-4.
Vqrjqblt '8' Stress Corrosion Crack Growth CoefficienT C
1
Per reference [12], C
1
was assumed to be distributed
norrnally wilh a mean of-4.968 and a standard dcviation of 0.587.
Vqriable 19' Stress COUQfion Crack Grnwrh Keywqy
Temperature
1be keyway tcmpera1Ure for see growth was assumed to be
distributed nonnally with a mean equal to the average of the disk:
tip$tIeam and downstn:am temperatures and 3 standard deviations
corresponding to the upstream/downstream temperatures.
Yqriqbk '10' Sluss Corro:rion Crack GrQwlh - Yield Srrrngrh
Disk yield strength values for use in the SCC growth rate
equation 9, were assumed to be distributed nonnally with mean
and standard deviation values based on available material property
data (Table 2).
Risk-Tolerance Considerations
The goal of the probabilistic evaluation was to provide TVA
with an estimate of risk associated with continued operation
expressed in the fonn of a probability of failure versus time.
Acceptable levels of risk must be established by the utility. One
means of establishing this risk level is from a review of historical
failures. 1be "Bush curvc" representing rotor reliability based on
historic turbine rotor failures compiled by Dr. S. Bush [151 is
given by the following equation from Bush's paper:
H(T) = 0.119 TO.57 (101
where, H(T) represents the cumulative hazard or probability of
failure in percent (%) versus operating time, T. in years. Using
this equation, the historical failure probability fOT the Paradise
rotors which have operated for approximately ::!.5 years is 0.75 %
or 0.0075. TVA conservatively elected to limit the failure
probability to lxlO-4 representing a safety factor of 75 compared
Darryl Rosario
with historical failures.
Probabilistic Simulation Results
Using the Monte Carlo technique, one million iterations
(lifetime calculations) representing two orders of magnitude above
the desired probability level of Ixl0'" wen: performed for each
disk. Reproducibility and consistency of results was tested using
various random number generators and random number seeds.
Variable distributions were examined by binning data values
within three standard deviations above and below the mean.
7th stage disk: Results are plotted in Figure 11. Failures in
excess of 100 in I million iterations violate the failure criterion of
lxl0-4. Simulations were performed for two separate cases
referred to as "Undamped" and "Clamped". "Unclamped"
simulations represented no restriction on selection of variables
outside of three standard deviations abovelbelow the mean.
"Clamped" simulations limited the lower-bound fracture toughness
to three standard deviations below the mean. The clamped"
simulations also assumed that planar cracks were unrealistic,
increasing the probability of occurrence for the 10: I crack from
60% to 65%. The objective behind the "clamped" analysis was to
provide a more realistic set of results consistent with expected
toughness values and crack prof1les. The results
illustrate the significant benefit of prcwarming. Without
prewarming, the acceptable failure criterion of lxl0-4 is not
achievable. With prewanning at 900 rpm to 150"F, the
"clamped" results predict that an initial crack size of 0.45 inch is
tolerable for two additional years of operation.
8lh and 9th Stage Disks: Tolerable initial crack sizes of2.35
and 2.95 inches for the 8th and 9th stage disks, respectively, wen:
predicted to satisfy the lxlO-4 failure criterion for two additional
years of operation.
CONCLUSIONS
The seventh stage (L-2) disk which was constructed of ler-
0.5Mo stccl and had a high FAIT in relation to operating
temperature, was identified as the most crucial in the analysis.
Based initially on a detenninistic analysis, with reasonable
assumptions that were consistent with crack depths estimated from
a UDRPS inspection, the LPA rotor for Unit 1 containing
indications of small depth was returned to service for two
additional years with the following operational restrictions:
Prewann the L-2 stage to 130"F prior to attaining
900RPM.
Restrict rotational speed to 1800 RPM until the L-2
stage temperature is 150 F or higher.
Limit the number of overspeed events and assure that
they do not exceed 110 percent.
Befon: any intentional overspecd, assure that the rotor
has soaked to achieve the estimated 17S"F temperature
for the L-2 stage.
These same operational restrictions were imposed on Unit 2.
In the period from March 1995 to May 1995, a probabilistic
analysis was perfonned to: (a) provide guidance on rotors to
inspect for Unit 2 in the October 1995 outage, and (b) provide
guidance on the upgrade schedule to an improved fat shaft with a
tab design. Based on the previous Unit 1 inspection results and
the quantitative risk estimates from the probabilistic analysis. TVA
deferred UDRPS inspection oi the one remaining rotor to be
147
Darryl Rosario
upgraded and continued to plan for replacement as scheduled in
1997. This action resulted in about a $"...50,000 cost avoi.d.aooe.
For the materials used in the TVA Paradise Units 1 and 2,
the probabilistic analysis confinncd that the recommendations of
the OEM are overly conservative relative to the risks of rotor
bunt from SCC. The probabilistic approach avoided the use of
worst<ase assumptions by considering the statistical distributions
associated with key variables, thereby providing a more realistic
assessment of the risk associated with continued operation.
REFERENCES
1. "A Review ofGE Fossil Stf:am Turbines Roton with Shrunk
On Wheels," A. Mortensen, and J. J. Pcpe. R. T. Bievenue,
GE Report GER-3539A, GE Power Systems, Schenectady,
NY. 1987.
2. "Advanced Reliable Low Pressure Steam Turbine Retrofits,"
E. Kramer. N. LanncIors, and Brandon Scar1in, Adyances ig
StC3m Turbine TechnQlogy for the Power Generation
lnWmo:. PWR-Vol. 26. pp. 89-98. ASME, 1994.
3. "Investigation into the Consequences of the Failure of a
Tutbine--Gencrator at Hinkley Point A Power Station, " I. L.
Gray. Proceedings Institute of Mechanical Engineers, pp.
379-390, 1972.
4. ANSYS Engineering Analysis System. Vol. I and II, Rev
4.4A, 1989, Swanson Analysis Systems, Inc., Houston,
Pennsylvania.
5. H. G. Pennick and C. H. Wells, "Stress and Fracture
Analysis of Shnmk-Qn Steam Turbine Disks" EPRI Report
NP-3340, Research Project 1398-5, January 1984.
148
6. Structural Integrity Associates, wpc_CRACK
M
, Version 2.1.
1991.
7. EPRI Report, NP-2738, MMetallurgical Evaluation of a Faileo
LP Turbine Disc", December 1982.
8. J. M. Banom and S. T. Rolfe. "Fracture and Fatigue Control
in Stnlctures
M
, Second Edition. Prenticc.-Hall Inc., 1987.
9. W. L. Server and W. Oldfield, wNuclear Pressure Vessel
Steel Data Base", EPRI933, Electric Power Research
Institute, Palo Alto, Ca, December 1978.
10. Letter from J. R. Foulds to M. Steakley, "Report on Small
Punch Testing for Fracture Toughness <KrJ. Paradise Unit
1A LP Turbine Stage 7 Disks (SF20273)", March 17, 1995.
11. J. R. Foulds et aI, "Fracture Toughness by Small Punch
Testing." I. Testing & Evaluation, Vol. 23, No.1, Ianuary
1995. p.3.
12. W. G. Clark, B. B. Seth. and D. M. Shaffer, "Procedures for
Estimating the Probability of Steam Turbine Disc Rupture
from Stress Corrosion Cracking", presented at Joint
ASMElIEEE Power Generation Conference, October 1981.
13. M. O. Speidel and R. M. Magdowski, MClean Steels for
Steam Turbine Rotors - Their Stress Corrosion Cncking
Resistance. M Proceedings: The International Conference on
Advances in Materia.ls Technology for Fossil Power Plants,
Sponsored by EPRI and ASM. Ohio, September 1987.
14. D. Annstrong. "Assessment of Probability of Survival of
Built Up LP Turbine Rotors with Discs Containing Semi-
circular Keyways", Proceedings of the Fossil Steam Turbine
Disc Cracking Workshop, EPRI GS-7250, April 1991.
15. S. H. Bush, "A Reassessment of Turbinc.-Generator Failure
Probability," Nuclear Safety, Vol. 19, No.6. November-
December. 1978, pp. 681-698.
Table 1
Results from UDRPS Examination of Paradise Unit 1 LPA and LPB Rotors.
Darryl Rosario
Keyway Maximum Depth, 10. Hore Maximum
Stage ' Depth, In. Comments
Original Estimate FIWlI Estimate
Rotor LPB Remoud (or UPlJ'ade to Improved Design
7TE 0.32 <0.10 0.174 Isolated Mostly small random bore indications
8TE <0.050 No Est. Essentially no bore indications
9TE <0.20 <0.050 No Est. Small random bore indications
7GE 0.52 0.25 0.67 Isolated Mostly small random bore indications
8GE No No No Est. A single bore 'indication wall found
9GE No No 0.180rig. Mostly small random bore
Rotor LPA - Inspected Based 00 LPB Results; Returned to Service
7TE NA <0.070 <0.070 A single bore indication
8TE ,NA <0.050 <0.050 14 small bore indications
9TE NA <0.050 <0.050 33 small bore indications
7GE NA No <0.050 22 small bore indications
8GE NA 0.050 0.050 4 small bore indications
9GE NA 0.325 0.080 27 small bore indications
Notes: TE - Turbine End GE - Generator End NA - Not Applicable
Table 1
Statistical Summary of Disk Mechanical Properties
FATT CVN 0.2% Yield Ultimate
"F ft-Ibs ksi ksi
1Cr-Q.5Mo Steel (Disk 7)
Average 18.0 72.6 89.3 114.3
Minimum -78.0 23.5 78.5 104.8
Maximum 140.0 113.0 110.0 130.5
Std.deviation 61.6 27.4- 6.: 5.7
1.75Ni-1.2SCr..o.5Mo-o.lV Steel (Disks 8 and 9)
Averaee -58.1 97.5 100.5 118.5
Minimum -135.0 70.0 86.1 107.0
Maximum 0.0 128.0 112.5 128.0
Std.deviation 41.96 13.0 6.3 4.9
149
Darryl Rosario
Table 3
Disk 7 Critical Crack Sizes
i50
Critical crack size (inch) for a min. CVN value of23.5 ft-Ibs (or K of 100 ksi JiiiCii )
TOF
75 100 150
K,,, ksiv'inch 100 115 145
RPM 900 1800 1980 900 1800 1980 900 1800 1980
Crack: Model
Planar 1.13 0.5:5 0.45 >1.20 0.88 0.75 > 1.20 > 1.20 >1.20
10:1 >1.20 0.90 0.78 >1.20 >1.20 1.13 >1.20 >1.20 > 1.20
5:1 >1.20 1.20 1.03 >1.20 > 1.20 >1.20 >1.20 >1.20 >1.20
2:1 >1.20 >1.20 >"1.20 >1.20 > 1.20 >1.20 >1.20 > 1.20 >1.20
Critical crack size (inch) for a min. small punch K,.. value of 55 ksiJinch
T OF
15 100 150
Kr.. ksiJinch 55 10 100
RPM 900 1800 1980 900 1800 1980 900 1800 1980
Crack Model
Planar 0.14 0.084 0.075 0.30 0.1:5 0.14 1.13 0.55 0.45
10:1 0.24 0.13 0.12 0.58 0.28 0.23 >1.20 0.90 0.78
5:1 0.34 0.17 0.14 0.80 0.40 0.32 >1.20 1.20 1.03
2:1 1.20 0.63 0.52 >1.20 > 1.20 LOS > 1.20 >1.20 >1.20
Table 4
Disks 8 and 9: Critical Crack Sizes
Critical Crack Size (inch) for a min. CVN of70 tUbs
DISKS DISK 9
K
1c
ksi,fiiiCh 180 180
RPM 900 1800 1980 900 1800 1980
Crack Model
Planar 4.50 3.05 1.73 5.20 3.85 3.53
10:1 4.50 3.06 2.78 5.20 3.85 3.53
5:1 7.50 4.66 4.18 >8.00 5.94 5.45
2:1 >8.00 >8.00 >8.00 >8.00 >8.00 >8.00
Table 5
Suaunary of Stresa COtTO&ion Craek Growth Rates'
Statis.
SCC(4)
Value
C (I) TOFa)
Yield (bd')
daJdt
'-_I ...,
DISK 7 U.bound -3.207 240.0 110.0 0.203
Mean -4.968 207.5 89.3 0.012
DISK 8 U.bound -3.207 170.0 112.5 0.068
Mean -4.968 160.0 100.5 0.007
DISK 9 U.bound -3.207 150.0 112.5 0.047
Mean -4.968 140.0 100.5 0.005
Darryl Rosario
Notes:
I.
2.
3.
4.
Upper bound values for C
t
inelude 3 standard deviations above the mean.
Upper bound keyway temperature, T, eorresponds to Ihe disk upSU'eam temperature. mean keyway
temperature eorrespoDds to the average of the disk: upstream and downstream temperatures.
Upper bound yield corresponds to the maximum measured value in Table 1.
SCC growth rates arecaleulated per equation 9.
Table 6
Disk 7: Summary of Random Variables
v.... 1J V,.,,;,.hl..
U .."n Vol"..
Stress ADalvsis:
1 Stress ot" 1<, Error - 3 5%
2 10% OYerspeed possibility -
-
2%
Fracture Mecbaaics:
3 Toughness @ RT. ksiJinch 75.0 3 34.0
Kevway temperature (oF) for critical crack size calc.:
4 No soak: 100 3 15
5 Soak at 900 rpm 150 3 15
6 Crack Models (probabilities of occurrence):
Planar (infinite lenlZ1h) - - 5%
10:1 -
-
60%
5:1
- - 25%
2: 1 (length:depth)
- - 10%
7 Initial Crack Size (in.) ran2e 0.070 to 0.75 3 15%
Stress Corrosion Crack Growth:
8 Cl -4.968 I 0.587
9 T (oF); keywav temp. (240 +175)12 I (240-175)16
10 Yield (l:si) 95 I 5
151
Darryl Rosario
Stress Concentration Factor (n Wheel Bore
Axial Key Way
Design
..
'I
-1-- ')'J
_1-- Kt=_._ ..
Tab Design
Figure 1. Transducer Placement and Beam Paths for the LP Disk
Ultrasonic Inspections
, Figure 4. Schemalic illustrating Stress Concentration Factors (KJ
for the Existing Axial Keyway Design and Proposed Tab Design

7!h .... .....*:tC...,
Figure 5. Combined Centrifugal and Shrink-Fit Hoop Stress
Distributions in the 7th Stage Disk at 0.900, 1800, 1980 and 2160
rpm with a Keyway of 1.2
i
i
\\
,
1\\\

!

:
i '\\... '"
I
:
.
I
,'"

I

I
"
r-
-l-
i
I
: I I I
"

'" o
.
:
..
i
/
$loriM.lil
lntn(N"'C"
!
i Lv
Un! t, 1.2 LP Rotar "'-2 (7th 5t!Ql;') Stv"vrok:-QI"I Olslooi
2. 'fwo.DimensinaJ Finite Element Model of the LP Rotor
7th Stage (L-2) Shrunk-on Disk
........ ..,..1&%

'-- I
I
I
I
-
I
-
I
......
I
i
I
i
I I
!
,.
o
I _aJIPII ,_.... _1_"'l1li
MM_ i
Figure 3. Combined Centrifugal and Shrink-Fit Hoop Stress
Distnbutions in the 7th Stage: Disk at 0,900, 1800. 1980 and :160
rpm
Figure 6. Schematic of the Crack Model used for K, Solutions -
Semi-Elliptic Crack at the 10 of a Hollow Cylinder with Crack
Depth, a, and Surface Length, I
152
Darryl Rosario
Probabilistic Analysis
UMa".J\".P' r_ .... o..
K ..... ...
f
t '111 -",........c::::....::p.--'""':::....__-=_+="-=-I-.....-===-
11
i
i
.l..-__---'i--__ -l. ....L... _L_____'
O.
... .. -1
Figure 7. Stress Intensity Factor (K() Versus Crack Size (a)
Results for Disk 7 at 1980 rpm (10% overspeed).
Figure iO. Schematic Illustrating ProbabilisLlc Remaining Life
Approach
e.....-el ................ ICCC..
................. 11..

"...... ........
o. 01
-_:'----1
......,.
..
_c.....,... ; ..
I I I
I I
I J
I I
I I
I J
/ /
I / II
l
V
-----
.-/
.
,
Ii
i
_IH .l..-_---...;__-J..__--L__.J-__ __.....L.. __
0.-
E
o.a- o.a_ .. - I
_'.GMNCfIIho_ _".EIe"MICfIM _".GNlItC:........
::;:".O"'ttIte.... -._.P'...... l:na.1IU

!....

Figure 8. Comparison of SCC Crack Growth u.s. Data
(Equation 13) and Powergen. U.K. Data
Figuze 11. Probabilistic Results for Disk 7: Predicted Number of
Failures in 1 Million Iterations for Two Years of Operation, (a)
without prewanning, and, (b) with prewanning to 150F

"-....... l,......... WWlIW.......,...
r .
I
....
'-
1
'"
!
""-... i
......

-;" I

r ..... .. '+-:.:-.--i
I' ,i
--i
O
... - . _
......1_. e 1
1---''''-
_____ ..1. ....
.- ...
---.-.-.. ........ -.
.-..
. - ..."'-._1......
--''''''-1
-- ......-............
Figure 9. Disk 7 Deterministic Remaining Life Estimates without
and with Prewarming
153
154
Major Influences on the Growth Rates of Stress Corrosion Cracks
in Steam Turbine Rotor and Blade Materials
Presented By
Swiss Federal Institute of Technology
155
156
MAJOR INFLUENCES ON THE GROWTH RATES OF
STRESS CORROSION CRACKS
IN STEAMTURBINE ROTORAND BLADE MATERIALS
Markus O. Speidel and Ruth Magdowski
Institute of Metallurgy
Swiss Federal Institute of Technology, ETH
CH 8092 Zurich Switzerland
157
158
IN STEAMTURBINE ROTOR AND BLADE MATERIALS
Abstract
The effects of alloy composition, heat treatment, and yield strength as well as
stress intensity, environmental composition, and temperature have been
systematically studied for steam turbine rotor and blade materials. The objective
of the study was both, to provide useful data for remaining life time calculations
and to provide a data base for comparison of experimental results for different
groups of steam turbine steels. The methodology was based on fracture mechanics
and crack growth studies in high temperature water. It is concluded that both,
strength of the material and temperature of the testing environment are the
major parameters affecting stress corrosion crack growth rates. Of far less
importance are stress intensity (within the plateau region), alloy composition,
and environment composition within the usual range for steam turbines.
Introduction
Intergranular stress corrosion cracking can occur in both, steam turbine rotor
steels and steam turbine blade steels as well as in their intended replacement
materials. For a quantitative evaluation of the resistance to stress corrosion
cracking, extensive data sets are necessary concerning initiation of stress corrosion
cracks as well as stress corrosion crack growth rates. The present paper deals only
with the latter. Crack growth rate studies of the various steels in high
temperature water are necessary not only to evaluate the relative resistance of
various materials but also for quantitative evaluations of the remaining life time
of cracked components. In addition to publications todayl-n, further quantitative
comparisons of stearn turbine rotor and steam turbine blade materials are
presented to shed some light on the different behavior of such diverse materials
as the standard rotor material (e. g. 3.5NiCrMoV), 12% chromium steels (e. g.
X20CrMoV121) and various precipitation hardening steels (e. g. 15-5 Mo PH, 13-
8 Mo PH) intended for steam turbine blades.
159
Markus O. Speidel
Experimental
Fracture mechanics type double cantilever beam specimens of the dimensions
10x60x120 mm were cut from commercial steels of the above mentioned types.
Specimens were then fatigue precracked, wedge loaded to a predetermined stress
intensity and exposed to water of an initial conductivity of less than 0.3 JlS/ em in
static autoclaves. Typical testing times were 1000 hours. After that, the specimens
received the final fatigue crack and were then broken open. Crack growth rates
were evaluated by dividing the length of the intergranular stress corrosion crack
visible on the fracture surface by 'the total testing time. The resulting crack growth
rates were plotted primarily as a function of the applied stress intensity.
Experimental Results
Steam turbine rotor steels
The way stress corrosion crack growth rates depend on stress intensity is
illustrated in Figure 1. At extremely low stress intensities, stress corrosion crack
growth rates, if any, are too slow to be measured. As the stress intensity exceeds 10
to 20 MPa-vm, the observed crack growth rates qUickly rise to a "plateau" value
which is not further increased even as the stress intensities are raised to extreme-
ly high values. This means that over a very broad range of stress intensities, there
is no effect of stress intensity on the stress corrosion crack growth rate. When in
the later part of this paper we plot crack growth rates versus other parameters
such as temperature or yield strength, we use always data taken from the plateau
range.
The influence of yield strength on the stress corrosion crack growth rate of steam
turbine rotor steel 3.5NiCrMoV is illustrated in Figure 2 for two different
temperatures, 100 and '160C. It is obvious that two different yield strength ranges
may be recognized: one below 1100 I\1Pa where the effect of yield strength on the
stress corrosion crack growth rate is moderate, and another, above 1100 MPa,
where increasing yield strengths result in extremely strong increases of the stress
corrosion crack growth rates.
The influence of temperature on the growth rates of intergranular stress
corrosion cracks in the clean 3.5NiCrMoV steam turbine rotor steel is shown in
Figure 3. Obviously, there is a complicated joint influence of yield strength and
temperature and, moreover, not all metallurgical effects have been investigated
in sufficient detail. Three essential experimental observations may be noted,
however. First, when heat treated to a high strength, the steel exhibits a
maximum stress corrosion crack growth rate near SOC. Second, the strong effect
of yield strength is observable both, above and below SOC, down to DOC and up to
100C. Third, it is possible, but not well established, that for the same steel in the
lower yield strength condition, a single activation energy of about 47 kJ /Mole
controls the temperature dependence of the stress corrosion crack growth rate.
The data shown in Figure 3 appear particularly helpful when only the lower
straight line is considered as turbine rotors with yield strength exceeding
160
Markus O. Speidel
1100:MFa are no longer manufactured. The line in Figure 3 for steels with yield
strengths below 800 MPa is reproduced in Figure 4 and compared there with the
upper bound of actual stress corrosion service experience in steam turbine rotors
(dashed line)1, 6-10. It is quite remarkable how well the upper limit of service
experience and the upper limit of our laboratory test results coincide all the way
from 60 up to 170 e. Thus, laboratory data indeed provide a conservative
estimate of the crack growth rates to be expected in steam turbines.
One of the goals of the present paper is to demonstrate the major metallurgical
influences on stress corrosion cracking of steam turbine materials. In steam
turbine rotor steels, the major variable is the yield strength as influenced by the
tempering temperature. This is shown in Figures 5 and 6 and will be used below
to compare similar influences in steam turbine blade materials. Obviously, as the
tempering temperature is increased, the yield strength comes down and so does
the stress corrosion crack growth rate in high purity 3.5NiCrMoV rotor steels,
Figures 5 and 6. Whether this is primarily a consequence of tempering
temperature or of the concomitant decrease of the yield strength is difficult to say,
and this is one reason for the further evaluation of similar effects with blade
materials below.
Blade materials
The effect of yield strength on the stress corrosion crack growth rate in steam
turbine blade materials is illustrated in Figure 7. Note that 12% Cr steel
X20CrMoV121 has a stress corrosion crack growth rate below the yield strength of
760 MPa which is so slow that it cannot be measured in the laboratory with a
standard 1000 hour test. However, already a yield strength of 800 11Pa and
certainly all Yield strengths above that result in clearly measurable crack growth
rates. A similar behavior is. apparent for the two precipitation hardening stainless
steels, Figure 7. Note, however, that in these steels at high strength levels a
considerable scatter in crack growth rate is apparent in Figure 7. This, however, is
not due to uncertainties in measuring stress corrosion crack growth .rates, as is
obvious from Figure 8 which shows that in one particular heat in one particular
heat treatment crack growth rates are measured quite reproducibly. Thus, the
scatter in Figure 7 may relate rather to different microstructures introduced by
different heat treatments. Further metallurgical influences, in particular
precipitation hardening, on the growth rates of stress corrosion cracks in two
stainless steels are illustrated in Figures 9 and 10. ObViously, peak hardness
corresponds to peak crack growth rates, and averaging is beneficial.
Further Comparisons and Discussion
The upper limits of all the crack growth rate data of all types of turbine steels
considered in this paper are plotted in Figure 11 versus the annealing
temperature. Obviously, with all these steels, the design engineers is free to
choose an appropriate compromise between the tolerable stress corrosion crack
growth rate and the yield strength as influenced by the annealing temperature. A
comparison of all crack growth rates, strictly versus yield strength, is illustrated
in Figure 12. Below 1100 MPa yield strength, the 3.5NiCrMoV rotor steels have
161
Markus O. Speidel
significantly higher crack growth rates compared to all the stainless steels at any
given yield strength. Above 1100 MPa yield strength, most steels have the same,
exceedingly high stress corrosion crack growth rate, and this is probably due to the
effect of hydrogen. In any case, design engineers wishing to avoid stress corrosion
problems are obviously well advised to design with the lowest yield strength
material of any group.
References
1. R. Magdowski and M. O. Speidel, "Clean steam turbine rotor steels - their stress corrosion
resistance", Metall.Trans. vol. 19A, 1988, pp 1583-1596
2. Ruth Magdowski, "Stress corrosion cracking of low alloy steel in water", Ph.D. Thesis No.
8432, ETH Zurich, 1987 '
3. M. O. Speidel, J. Denk, B. Scarlin, "Stress corrosion cracking and corrosion fatigue of steam-
turbine rotor and blade materials", Commission of the European Communities, Luxembourg.
1991, COST Report EUR 13186 "physical sciences"
4. S. Inagaki, M. Miyazaki, H. Kashiwaya, K. Nakadi, "Effects on material and
environmental factors (Xl. sec of NiCrMoV rotor steels", Proc. of Fossil Steam Turbine Disc
Cracking Workshop, October 1990, Duke Power Company, EPRI, GS-7250
5. W. David, K. Schleithoff, F. Schmitz, and F. Ewald, "Stress corrosion cracking behavior of
turbine rotor disk materials", Proc. of Fourth Conference of COST 501 and 505, Liege, Belgium,
September 1990, pp 577-588
6. J.Y. Liu, E.E. Lai, c.c. Su, H.C. Lai, and D.H.R. Lin, "Life prediction for nuclear LP rotor SCC
cracks", Proc. of Fifth Int. Symp. on Environmental Degradation of Materials in Nuclear
Power Systems - Water Reactors, 1993, San Diego, pp 596-603
7. M. O. Speidel and J. E. Bertilsson, "Stress corrosion cracking of steam turbine rotors, in
"Corrosion in Power Generating EqUipment", Markus O. Speidel and Andrejs Atrens, eds.,
PlenumPress, New York, 1984, pp 331
8. W. Engelke, H. Jestrich, K. Schleithoff, and H. Termiihlen, "Design, Operating and
Inspection Considerations to Control Stress Corrrosion of ,LP Turbine Disks", Proc. American
Power Conference, Vol. 45 (1983), pp 196-206
9. }.M. Hodge and I.L. Mogford, "Ul( Experiences of Stress Corrosion Cracking in Steam Turbine
Discs", }?roc. of the Inst. of Mechanical Engineers, Vol. 193 (1979) No 11, pp 93-109
10. EPRl Report NP-2429-LD, Vol. 1-7, "Steam Turbine Disc Cracking Experience", Research
Project 1398-5, Final Report, 1982
11. R. Magdowski and M. O. Speidel, "Environmental Assessment and Life Time Prediction of
Low Alloy Steam Turbine Rotor Steels", Proc. of Seventh Int. Syrnp. on Environmental
Degradation of Materials in Nuclear Power Systems - Water Reactors, 1995, Breckenridge,
pp 921-928
162
Markus O. Speidel
-4
10
steam turbine rolor steels, sec in hot water
o 0 clean rotor steel
J
-5
10
0 commercial rotor steel

...
-6
10
<l
ai
E 10
-7
.c
0.1
3:

0 .d
0,
10
-8

(,)

III
i3
-9
--
c:: 10
0
16QOC -I- -. I 'ii)
I- , e
;" .
(;
10
10
-.
(,)

III
C1l
<n
-11
10
yield strength, Rpo.2, [MPa]
600 800 1000 1200 1400 1600 400

sec, steam turbine rotor steels
deionized water. 160C, aerated and
3-3.5%Ni steel deaerated
.2Cr-1Ni steel
A "clean steels ", specially alloyed steels
stress intensity, K
r
, [kSi-v'iil j
a 50 100 150 200
10-6 I--r--..,.--L-r---r--,.....I...,...--r--,.J.-,...--r---!---,
10
13

o 100 200 240
stress intensity, K
11
[MN'm-
312
]
Figure 1. Effect of stress intensity on the
growth rates of intergranular stress
corrosion cracks in steam turbine rotors.
Figure 2. Effect of yield strength on the
corrosion cracks in steam turbine rotor steels
at two different temperatures.
temperature, T, [0C]
temperature, T, [oC]
Figure 3. Growth rates of intergranular
stress corrosion cracks in 3.5NiCrMoVrotor
steel: The effect of temperature and yield
strength.
reciprocal. temperature, 1fT, [11K]. 10
3
300 200 100 0
-6
1ii'
10
i
intersran"lar ,trcss corrosion cracking
of "team turbine rotor steels in service,

-7
in steam arid in WOlter
10
- ;rI H20. RoO.2 < aOOMPo
<I
-- $crviee experlcnce

steam, RoO.' C 763MPa
-8
deaerated steam Rm =830M?a

10
deionized H20, Ap < BOOM?a

.c
0
-9
0,
10
(,)
<tl
ti
-10
c:: 10
0
"iii
E
a
-11
u
10
IIJ
IIJ
<n
-12
10
1.6 2.0 2.4 2.8 3.2 3.6 4.0
Figure 4. Comparison of intergranular stress
corrosion crack growth rate data from
Figure 3 (straight line), from service
experience (dashed line, Ref, 1,6-10) and
from tests in different laboratories (Ref. 4-
6). Note the good agreement, particularly
with data from Figure 3 and the upper
bound of service experience between 60
0
and
170C, Ref.11.
40 20 0 -20 10080 300 200
reciprocal temperature, 1fT, [11K] .10
3
-s
10
stress corrosion crack growth in 3.5NiCrMoV steel
in high temperature water

-6

oS
10
.-' .
-7
' '\ <l
10
11)-

Ci:i
...
Ai t
-8
( .."
"i
10
0 / ..
...
r
f
.'., t
C)
-9 c.>
10
c.>
s::
0
'iii -10
E
10
0
P\
c.>
",
-11 yield strength, MPa
",
10
135<l
I. .1211
1ij
1190
- .
-12 <800
.,
10
1.6 2.0 2.4 2.8 3.2 3.6 4.0
163
Markus O. Speidel
1400
I
high purity steam turbine
1300

rotor steel
1200

ro
a.
1100
N
IIa
ci
1000
Q.
a::

..t:
0,
900
c
~
800
iii
"0
0
Q)
700
0>,
600
9200 (1f2h)lair/8500 (112h)/oillT (2h)/air

0 920
o
/air/850o/oiI1620o/air (as received)
500
o 300 400 500 600 700 800
tempering temperature, T, [0C]
-6
10
I I I I I I
sec of HP steam turbine
en rotor steel, H20, 100C,
E
-7
deaerated
-
10
fo-
~
ro
-8
<J
- 10 l-
<D
ro
...
..t:
-9
~
10 fo-
-
0
...
OJ
~
(,) -10
-
ro 10 I-
...
T
(.)
c:
0
"c;; -11
-
0
10 -
...
0
high purity steam turbine rotor steel (.)
en
-12.
920(1I2h)/air/B50
o
(1/2h)/oillT {2h)Jair
en
10
- 0 as received 920
o
Jair/850o/oil/620o/air
-
~
(;)
-13
r I f I I I
10
0 300 400 500 600 700 800
tempering temperature, T, [Oe]
Figure 5. High purity stearn turbine rotor
steel 3.5NiCrMoV: Effect of tempering
temperature on yield strength.
Figure 6. High purity steam turbine rotor
temperature on stress corrosion crack growth
rate.
100 120 80 60 40 20
.1"'=.---.
~ r ?-9-
I
I stress corrosion cracking 01 blading materials
H
2
0, 160
D
C, deaerated
.155Mo PH R
POo2
'" 1025MPa
o 1S5Mo PH R
pO
2
'" 740MPa
13-8Mo PH R p o _ ~ '" 910MPa
1
16
13
'--_--JI.-_--'-__-'-__-'-__"""-_---J
o
en
stress corrosion cracking of blading materials
H
2
0, 160C. "deaerated
,
E
10-
6
/0
0 X20CrMoV12 1
~
10. 15 - 5 Mo PH
Cii
-
138 Mo PH
<1
10-
7
af
e
8
.s::
10-
8
~
~
0
0,
I
10-
9
I
(,)
I
co
,-
...
(,)
l:
.10.10.10.
I
0
1<)"'0
0
Ui
10..
0
' .10. .,.
...
Il!' 1i/
0
(,)
-1 ,
/ ..;''''
en 10 -ffl(t-
en
~
(ii
161 2. '--_--1__-'-__-'-__..........__"'-_---'
400 600 800 1000 1200 1400 1600
yield strength, R
pO
2
, [MPa] stress intensity, K
1
, [MN.m-
3
/
2
]
Figure 7. Both, 12 percent chromium steels
(X2OCrMoV121) and precipitation
hardening steels exhibit increasing stress
corrosion crack growth rates with increasing
yield strength, particularly at high yield
strength levels.
Figure 8. Stress corrosion crack growth rates
in precipitation hardening stainless steels
intended for stearn turbine blade
applications.
164
Markus O. Speidel
600 500 400 300
SCC of precipitation hardening steel 13 - BMc PH
H20, 160"C. deacr"ted
10.
,2

20
'iii"
0-
1600
70 N
75
.2
SS ti
:l
"C
SO
see of precipitation hanlening stainless steel 15-S Mo PH
H
2
0, 160C, aerated
:J H
2
0, 160C, deaeralod
.. H
2
0, 9SoC, de"erated
'iii"
0. 1400
annealing temperature, T, rOC] annealing temperature, T, [DC]

Figure 9. Precipitation hardening affects
both, mechanical properties and stress
corrosion crack growth rates. Note that
averaging reduces. the maximum stress
corrosion crack growth rate.
Figure 10. Comparison of the effect of
averaging for two types of precipitation
hardening stainless steels.
o
stress corrosion cracking

steam turbine steels,
H20, 160
c
C, deaerated
rOlor steel 3.SNiCrMoV
o X20CrMoV12 1
A 15 - 5 Mo PH
13 - 8 Mo PH
-10
10
-11
10
-12
10
-9
10
-8
10
-6
10
-7
10
-s
10
400 600 800 1000 1200 1400 1600 800
see in 160
o
e, deaeraled.
upper limits of scalterbands
300 400 500 600 700
-10
10
1612" '--_----1__.-1.__--'-__-'-__--'-_---'
200
-11
10
annealing temperature, TA' [DC] yield strength, RpO.2 [MPa]
Figure 11. Comparison of the maximum
stress corrosion crack growth rates of all
materials discussed here: The effect of
annealing temperature.
yield strength.
165
166
IN STEAM TURBINE ROTOR AND BLADE MATERIALS
Abstract
The effects of alloy composition, heat treatment, and yield strength as well as
stress intensity, environmental composition, and temperature have been
systematically studied for steam turbine rotor and blade materials. The objective
of the study was both, to provide useful data for remaining life time calculations
and to provide a data base for comparison of experimental results for different
groups of steam turbine steels. The methodology was based on fracture mechanics
and crack growth studies in high temperature water. It is concluded that both,
strength of the material and temperature of the testing environment are the
major parameters affecting stress corrosion crack growth rates. Of far less
importance are stress intensity (within the plateau region), alloy composition,
and environment composition within the usual range for steam turbines.
Introduction
Intergranular stress corrosion cracking can occur in both, steam turbine rotor
steels and steam turbine blade steels as well as in their intended replacement
materials. For a quantitative evaluation of the resistance to stress corrosion
cracking, extensive data sets are necessary concerning initiation of stress corrosion
cracks as well as stress corrosion crack growth rates. The present paper deals only
with the latter. Crack growth rate studies of the various steels in high
temperature water are necessary not only to evaluate the relative resistance of
various materials but also for quantitative evaluations of the remaining life time
of cracked components. In addition to publications today1.11 / further quantitative
comparisons of steam turbine rotor and steam turbine blade materials are
presented to shed some light on the different behavior of such diverse materials
as the standard rotor material (e. g. 3.5NiCrMoV), 12% chromium steels (e. g.
X20CrMoV121) and various precipitation hardening steels (e. g. 15-5 Mo PH, 13-
8 Mo PH) intended for steam turbine blades.
167
Markus o. Speidel
Experimental
Fracture mechanics type double cantilever beam specimens of the dimensions
10x60x120 mm were cut from commercial steels of the above mentioned types.
Specimens were then fatigue precracked, wedge loaded to a predetermined stress
intensity and exposed to water of an initial conductivity of less than 0.3 IlS/ em in
static autoclaves. Typical testing times were 1000 hours. After that, the specimens
received the final fatigue crack and were then broken open. Crack growth rates
were evaluated by dividing the length of the intergranular stress corrosion crack
visible on the fracture surface by the total testing time. The resulting crack growth
rates were plotted primarily as a function of the applied stress intensity.
Experimental Results
Steam turbine rotor steels
The way stress corrosion crack growth rates depend on stress intensity is
illustrated in Figure 1. At extremely low stress intensities, stress corrosion crack
growth rates, if any, are too slow to be measured. As the stress intensity exceeds 10
to 20 MPa-vm, the observed crack growth rates quickly rise to a "plateau" value
which is not further increased even as the stress intensities are raised to extreme-
ly high values. This means that over a very broad range of stress intensities, there
is no effect of stress intensity on the stress corrosion crack growth rate. When in
the later part of this paper we plot crack growth rates versus other parameters
such as temperature or yield strength, we use always data taken from the plateau
range.
The influence of yield strength on the stress corrosion crack growth rate of steam
turbine rotor steel 3.5NiCrMoV is illustrated in Figure 2 for two different
temperatures, 100
0
and 160C. It is obvious that two different Yield strength ranges
may be recognized: one below 1100 MPa where the effect of yield strength on the
stress corrosion crack growth rate is moderate, and another, above 1100 MPa,
where increasing yield strengths result in extremely strong increases of the stress
corrosion crack growth rates.
The influence of temperature on the growth rates of intergranular stress
corrosion cracks in the clean 3.5NiCrMoV steam turbine rotor steel is shown in
Figure 3. Obviously, there is a complicated joint influence of yield strength and
temperature and, moreover, not all metallurgical effects have been investigated
in sufficient detail. Three essential experimental observations may be noted,
however. First, when heat treated to a high strength, the steel exhibits a
maximum stress corrosion crack growth rate near 50C. Second, the strong effect
of yield strength is observable both, above and below 50C, down to OC and up to
100C. Third, it is possible, but not well established, that for the same steel in the
lower yield strength condition, a single activation energy of about 47 k}/Mole
controls the temperature dependence of the stress corrosion crack growth rate.
The data shown in Figure 3 appear particularly helpful when only the lower
straight line is considered as turbine rotors with yield strength exceeding
168
Markus O. Speidel
1100 MPa are no longer manufactured. The line in Figure 3 for steels with yield
strengths below 800 MPa is reproduced in Figure 4 and compared there with the
upper bound of actual stress corrosion service experience in steam turbine rotors
(dashed line)l, &-10. It is quite remarkable how well the upper limit of service
experience and the upper limit of our laboratory test results coincide all the way
from 60 up to 170C. Thus, laboratory data indeed provide a conservative
estimate of the crack growth rates to be expected in steam turbines.
One of the goals of the present paper is to demonstrate the major metallurgical
influences on stress corrosion cracking of steam turbine materials. In steam
turbine rotor steels, the major variable is the yield strength as influenced by the
tempering temperature. This is shown in Figures 5 and 6 and will be used below
to compare similar influences in steam turbine blade materials. Obviously, as the
tempering temperature is increased, the Yield strength comes down and so does
the stress corrosion crack growth rate in high purity 3.5NiCrMoV rotor steels,
Figures 5 and 6. Whether this is primarily a consequence of tempering
temperature or of the concomitant decrease of the yield strength is difficult to say,
and this is one reason for the further evaluation of similar effects with blade
materials below.
Blade materials
The effect of yield strength on the stress corrosion crack growth rate in steam
turbine blade materials is illustrated in Figure 7. Note that 12% Cr steel
X20CrMoV121 has a stress corrosion crack growth rate below the yield strength of
760 MPa which is so slow that it cannot be measured in the laboratory with a
standard 1000 hour test. However, already a yield strength of 800 MFa and
certainly all yield strengths above thatresult in clearly measurable crack growth
rates. A similar behavior is apparent for the two precipitation hardening stainless
steels, Figure 7. Note, however, that in these steels at high strength levels a
considerable scatter in crack growth rate is apparent in Figure 7. This, however, is
not due to uncertainties in measuring stress corrosion crack growth rates, as is
obvious from Figure 8 which shows that in one particular heat in one particular
heat treatment crack growth rates are measured quite reproducibly. Thus, the
scatter in Figure 7 may relate rather to different microstructures introduced by
different heat treatments. Further metallurgical influences, in particular
precipitation hardening, on the growth rates of stress corrosion cracks in two
stainless steels are illustrated in Figures 9 and 10. Obviously, peak hardness
corresponds to peak crack growth rates, and averaging is beneficial.
Further Comparisons and Discussion
The upper limits of all the crack growth rate data of all types of turbine steels
considered in this paper are plotted in Figure 11 versus the annealing
temperature. Obviously, with all these steels, the design engineers is free to
choose an appropriate compromise between the tolerable stress corrosion crack
growth rate and the yield strength as influenced by the annealing temperature. A
comparison of all crack growth rates, strictly versus yield strength, is illustrated
in Figure 12. Below 1100 MPa yield strength, the 3.5NiCrMoV rotor steels have
169
Markus O. Speidel
significantly higher crack growth rates compared to all the stainless steels at any
given yield strength. Above 1100 MFa yield strength, most steels have the same,
exceedingly high stress corrosion crack growth rate, and this is probably due to the
effect of hydrogen. In any case, design engineers wishing to avoid stress corrosion
problems are obviously well advised to design with the lowest yield strength
rnaterial of any group.
References
1. R. Magdowski and M. O. Speidel, "Clean steam turbine rotor steels - their stress corrosion
resistance", Metall.Trans. vol. 19A, 1988, pp 1583-1596
2. Ruth Magdowski, "Stress corrosion cracking of low alloy steel in water", Ph.D. Thesis No.
8432, ETH Zurich, 1987
3. M. O. Speidel, J. Denk, B. Scarlin, "Stress corrosion cracking and corrosion fatigue of steam-
turbine rotor and blade materials", Commission of the European Communities, Luxembourg,
1991, COST Report EUR 13186 "physical sciences"
4. S. Inagaki, M. Miyazaki,' H. Kashiwaya, K. Nakadi, "Effects on material and
environmental factors on sec of NiCrMoV rotor steels", Proc. of Fossil Steam Turbine Disc
Cracking Workshop, October 1990, Duke Power Company, EPRI, GS-7250
5. W. David, K. Schleithoff, F. Schmitz, and F. Ewald, "Stress corrosion cracking behavior of
turbine rotor disk materials", Proc. of Fourth Conference of COST 501 and 505, Liege, Belgium,
September 1990, pp 577-588
6. J.Y. Liu, E.E. Lai.. C.C. Su, H.C. Lai, and D.H.R. Lin.. "Life prediction for nuclear LP rotor SCC
cracks", Proc. of Fifth Int. Symp. on Environmental Degradation of Materials in Nuclear
Power Systems - Water Reactors, 1993, San Diego, pp 596-603
7. M. O. Speidel and J. E. Bertilsson.. "Stress corrosion cracking of stearn turbine rotors, in
"Corrosion in Power Generating Equipment", Markus O. Speidel and Andrejs Atrens, eds.,
Plenum Press, New York.. 1984, pp 331
8. W. Engelke, H. Jestrich, K. Schleithoff, and H. Tenniihlen, "Design, Operating and
Inspection Considerations to Control Stress Corrrosion of LP Turbine Disks", Proc. American
Power Conference.. Vol. 45 (1983), pp 196-206
9. I.M. Hodge and I.L. Mogford, "UK Experiences of Stress Corrosion Cracking in Steam Turbine
Discs", Proc. of the Inst. of Mechanical Engineers, Vol. 193 (1979) No 11, pp 93-109
10. EPRI Report NP-2429-LD, Vol. 1-7, "Stearn Turbine Disc Cracking Experience", Research
Project .1398-5, Final Report, 1982
11. R. Magdowski and M. O. Speidel.. "Environmental Assessment and Life Time Prediction of
Low Alloy Steam Turbine Rotor Steels", Proc. of Seventh Int. Symp. on Environmental
Degradation of Materials in Nuclear Power Systems - Water Reactors, 1995, Breckenridge,
pp 921-928
170
Markus O. Speidel
yield strength, Rpo.2, [MPa)
-
-
-
-
-
-
-
1000 1200 1400 1600 400
-4
10
10
1Z
L...__.!- ' __.!- '_--l'__.J' 'l-_-'
600 800
I I I I I
steam lurbine rotor steels, see in hot water
(It 0 clean rolor steel
10-
5
I- 0 commercial rotor steel i
0
10-
6
I-

2- .,
-7 _ -0
10 - 0./

.d

,.4
10-
9
I- .,...
,'
-10
10 I-
-11 00
10 I-
stress intensity, K
r
, [kSiYin']
0 50 100 150 200
10-
6
10:'>
sec, steam turbine rotor steels
E
deionized water. 160C, aerated and
10-
7
3-3.5%Ni steel
deaerated
-
10
2

.2Cr-1Ni steel
... "clean sleels ",specially alloyed steels
(1) 10-
8
10
.s::
:!:
10-
9
0
II .A.. -
-
OJ
r:....

10-
10

()
10-'
ro
U
I
t:
; 0 10-
11
10-
2
'05
0
...
0
10-
12
()
10-
3
U'l
(fl
+'
10
13
(JJ
0 100 200 240
stress intensity, K
r
, [MN-m-
3/2
]
Figure 1. Effect of stress intensity on the
corrosion cracks in steam turbine rotors.
Figure 2. Effect of yield strength on the
corrosion cracks in steam turbine rotor steels
at two different temperatures.
temperature, T, [DC]
temperature, T, [DC]
300 200 10080 40 20 0 -20
300 200 100 o
3.6 4.0 3.2 2.8 2.4 2.0
I
inlergranular stress corrosion cracking
of steam turbine rOlor steels in service,
in 51eam and ;n water
- in H20, RoQ.2 < 800MPa
-- st'!fvice experience
steam, Ape.2 =763MPa
... deaeraled steam Am =830MPa
deionized H20, Rp < 800MPa
reciprocal temperature, 1rr, [11K] 10
3
1.6
Figure 4. Comparison of intergranular stress
corrosion crack growth rate data from
Figure 3 (straight line), from service
experience (dashed line, Ref. 1,6-10) and
from tests in different laboratories (Ref. 4-
6). Note the good agreement, particularly
with data from Figure 3 and the upper
bound of service experience between 60 and
170C, Ref.11.
-6
10
.s
4.0 3.6 3.2 2.8 2.4 2.0
yield strength, MPa

1350
Ji. 1211
1190
<800
stress corrosion crack growth in 3.5NiCrMoV steel
in high temperature water
./.
.
. .", ..
.' "-
'. + l':) "-,

+/+ '+
, t
reciprocal temperature, 1rr, [11K] .10
3
1.6
-12
10
-5
10
Figure 3. Growth rates of intergranular
stress corrosion cracks in 3.5NiCrMoV rotor
steel: The effect of temperature and yield
strength.
171
Markus O. Speidel
1400
high purity steam turbine
1300

rotor steel
1200
m

a.
1100
N
..
c:i
1000
a-
D:
.
-
.
900
c:
11>
~
800
in
'0 0
a> 700
';;'
600
9200 (1/2h)/air/8500 (1/2h)/oilrr (2h)/air

0 920
0
/air/850
%
il/620o/air (as received)
500
a 300 400 500 600 700 800
tempering temperature, T, [Oel
-6
10
I I I I I I
see of HP steam turbine

!!!.
rotor steel, H
2
0, 100C,
Eo
-7
deaerated
- 10 ~
~
Ctl
-8
<J
~ -
11>-
10
(;j
...
~
-9
~
10 ....
-
0
~
OJ
~
t) -10
-
co 10
~
t
r
c:
0
'(jj -11
- -
E
10
0
high purity steam turbine rotor steel (.)
U'l
-12
920
0
(1I2h)/air/8500 (1I2h)/oilfT (2h)/air
U'l
10
- 0 as received 920
o
/air/850
%
i1l6200/air
-
~
(;)
-1:;
I I t I I r
10 -y
o 300 400 500 600 700 800
tempering temperature, T, [Oel
Figure s. High purity steam turbine rotor
temperature on yield strength.
Figure 6. High purity steam turbine rotor
steel 35NiCrMoV: Effect of tempering
temperature on stress corrosion crack growth
rate,
stress corrosion cracking of blading materials
H
2
0, 160C, deaerated
....L.....-_..
~ c 9-9-
U)
stress corrosion cracking 01 blading materials
H
2
0, 160C, deaerated
,
E
10-
6
/e
0 X20CrMoV12 1
~
... 15 w 5 Mo PH
Cti
13 - 8 Mo PH
<:l
10-
7
11>-
~
8
~
10-
6
~
~
0
...
,
OJ
10-
9
I
(.)
I
m
,- (.)
c:
AAA
/
0
10
10 0
-
f/) A
E
'. A .,.
...
~ ti/ 0
/ .. .... 0
-11
f/) 10 -9r.i?t-
f/)
e
't;
Ui'
E 10-
7
all:'
<JI<J
11> 10
a
~
..:::::
~
o
OJ
~
t)
~
t)
c:
o
-en
~
o
t)
CIl
CIl
~
en
15-5Mo PH
o 15-5Mo PH
13-8Mo PH
R
pO
.
2
= 1025MPa
R
po
.
2
= 740MPa
R
pO
2
= 910MPa
20 40 60 80 100 120
stress intensity, K" [MNm
o3/2
J
1013 l-_---l__-l..__--L__-L__..L....__
o 600 800 1000 1200 1400 1600
yield strength, R
pO
2
, [MPaJ
Figure 7. Both, 12 percent chromium steels
(X20CrMoV121) and precipitation
hardening steels exhibit increasing stress
corrosion crack growth rates with increasing
yield strength, particularly at high yield
strength levels.
Figure 8. Stress corrosion crack growth rates
in precipitation hardening stainless steels
intended for steam turbine blade
applications.
172
Markus O. Speidel
~
ll.
see 01 precipitation hardening steel 13 aMo PH
~ 1600 H20, 160"C, deaerated
N
~
0 00 without intermediate anne:)1
a:.Q.
1400
with inlermediate anneal
~
.
-- $tcel15 - 5 Mo PH
. ~
~
rn
~
!::
~
1200
~ - -
- ~
N
(;i
~ ----
cO
"0
.,/
.,/
" ~
CJ "ai
1000
.//----- RpO.2 Q; ';;'
~
I'll
,/
.!:
!::
7ii
.2
~ '0
1
:J
"0 10'9
~
3
ca
<:I
~ 10"0
"
..t=
~ 10.
,1
0
0>
::t:-
o
10"2 Q
t 20 300 400 500 600
60
70
75
55
50
65
600 500 400 300
sec of precipitation hardening stainless sleel 155 Mo PH
H
2
0, 1GO"e, aeraled
o H
2
0, 160"C, deaerated
... H
2
0, 95"C, deaerated
ca
ll. 1400
~
~ 10" I-
o
0,
i'5 1<)'2 '---"r---l.. -.l- --J.. ...J...--.J
~ 20
u
annealing temperature, T, (0C] annealing temperature, T, [0C]
Figure 9. Precipitation hardening affects
both, mechanical properties and stress
corrosion crack growth rates. Note that
averaging reduces the maximum stress
corrosion crack growth rate.
Figure 10. Comparison of the effect of
averaging for two types of precipitation
hardening stainless steels.
SCC in H
2
0, 160C. deaerated.
upper limits of scalterbands
400 300 400 500 600 700 800
5
10
~
.s -6
10
~
co
<J
-7
aJ-
10
'
.:
-8
'i
10
e
C'l
x
-9
U
co
10
ti
l::
0
-10
Vi
10
e
0
u
In
-11
In
10
~
Cil
-12
10
-
I I I I
1
stress corrosion cracking
steam turbine steels,
0
,...H20, 150C. deaerated
-
rotor steel 3.5NiCrMoV
o X20CrMoV12 1

.... 15 - 5 Mo PH
~ . . 13 - 8 Mo PH
-
. ~
/;
'-
;}"
-
'-
..... 1.... -
........ ~
.......-..i(. /./...
0 ............
- -
I' .....
/ ~ ; ~ .
- I-
rrnf
I 1 I 1 I
600 800 1000 1200 1400 1600
annealing temperature, TAl [OC] yield strength, RpO,2 [MPa]
annealing temperature.
yield strength.
173
174
Unit-Specific Assessment of Low-Pressure Steam Turbine
Rotor Stress Corrosion Cracking
Presented By
Dr. Sanjeev Srivastav
Failure Analysis Associates, Inc.
175
176
Proceedings, EPRI Steam Turbine Stress
Corrosion Cracking Conference
March 1997
Unit-Specific Assessment of Low-Pressure Steam Turbine Rotor
S. Srivastav
J. R. Foulds
S. J. Nunes
149 Commonwealth Dr.
Menlo Park, California 94025
M. F. Steakley
Tennessee-Valley Authority
2N 55A Blue Ridge Place
Chattanooga, TN 37402
177
178
Unit-Specific Assessment of Low-Pressure Steam Turbine Rotor
S. Srivastav
J. R. Foulds
S. J. Nunes
149 Commonwealth Dr.
Menlo Park, California 94025
M. F. Steakley
2N 55A Blue Ridge Place
Chattanooga, TN 37402
Abstract
Assessment of stress corrosion cracking in low pressure steam turbine rotors, whether
at rims/blade attachments or at disk keyways, requires, among other things,
knowledge of two sets of variables, (a) the steam path temperature and pressure
conditions, and (b) the rotor metal temperatures. These variables govern the wetness at
a specific location in the turbine and are sensitive to the operating conditions.
Prediction of wetness helps identify turbine locations/sub-components at risk of
corrosion damage and crack growth. Further, the effect of operating cycles on the risk
of wetness and corrosion damage can be used to modify cycles to reduce risk.
A study on a TVA LP rotor exemplifies the unit-specific aspects of the corrosion
damage issue. Variability in wetness is quantified as a function of the condenser
pressure/temperature conditions, and the input steam for partial and full load
operation. A steam-path analysis is used to obtain steam conditions at intermediate
stages, while a SAFER-PC finite element model of the rotor is used for computing the
. corresponding temperature state of the rotor. The study illustrates that LP turbine rotor
stress corrosion cracking susceptibility and tolerance evaluation need to be unit-specific
with respect to operating and local steam conditions.
179
Sanjeev Srivastav
Background
Over approximately the past three decades, there has been a considerable accumulation
of experience and research helping further the understanding of stress corrosion
cracking in stearn turbines (e.g., [1-4]). Following the Hinkley Point turbine disk failure
in 1969 [5], the electric power industry has wrestled with the shrunk-on disk bore and
keyway stress corrosion cracking (SeC) problem, and more recently with the disk rim
or blade attachment cracking problem. While the factors affecting the susceptibility of a
particular unit and unit location to sec are now reasonably well-documented,
predictions of risk associated with the phenomenon remain uncertain. Indeed, major
rotor design changes and costly retrofit options develoPed for eliminating the
bore/keyway sec problem reflect this uncertainty. Reducing the uncertainty in sec
risk prediction is becoming increasingly important where design changes or retrofit
options are limited or non-existent as in the case of the disk rim or blade attachment
problem.
Of the many variables influencing sec susceptibility such as material strength,
operating stress, environment, and exposure temperature, a clearly definable and
quantifiable variable relates to the fact that in order to have SCC, the component must
be wet. Very simply, where metal temperatures exceed the local steam saturation
temperatures, component wetness and therefore, SCC, is highly unlikely. While local,
near-surface steam conditions may not be precisely known and uncertainties typically
exist in estimation of component surface temperatures, the estimation of a wetness
condition provides the simplest, first approximation to prediction of the relative risk of
see within a given unit and from unit-to-unit. For example, conservatively derived
predictions of component dryness can help exclude any further concern for see. While
the wetness consideration is an obvious one, there appears to be little attention paid to
determining wetness conditions, particularly as a function of unit operating cycles. This
paper illustrates the application of a steam path analysis and rotor thermal analysis to
development of a picture of relative SCC exposure in case of a fossil plant low-pressure
stearn turbine. The relative wetness picture helps prioritize locations meriting
examination and evaluation for sec. Both, the disk bore/keyway and the rim regions
can be assessed via the same steam path/thermal analysis.
The Tennessee Valley Authority (TVA) has identified its Paradise Station, Units 1 and 2
low-pressure (LP) steam turbines as being potentially at risk of sec at the shrunk-on
disk keyways and at the disk rim blade attachment regions of the last three stages. The
potential disk keyway sec problem is being addressed by a hardware modification to a
Ufat shaft" and tab key configuration initiated in 1993. The modification does not,
however, mitigate the potential for disk rim see which remains a concern. The 1800
RPM units are dual flow, General Electric-manufactured, 650 MW each, consisting of
nine stages, with a service date of circa 1972. Figure 1 is a sectional view of the LP rotor
model of a single flow path reproduced from the EPRl SAFER-PC [6] analysis
180
Sanjeev Srivastav
conducted for rotor thermal conditions. The inlet steam is nominally at 150 psia and
720F. TVA's prior evaluation of the condition of these LP units included a risk
assessment of disk failure at the keyway by sec [7]. The assessment considered the
indications of cracks by ultrasonic inspection, a stress analysis for local stress
conditions, and fracture toughness estimation/measurements by genericCharpy V-
Notch (CVN) data and disk-specific sample small punch testing. An expectedly
conservative deterministic analysis indicated that critical crack depths for the keyways
of the disks of the last two stages (8 and 9) were very large (2.6 and 3.4 in., respectively),
indicating that near-term failure is highly unlikely. In contrast, conservative
predictions of disk keyway cracking in case of the 7th stage disk showed a potential for
near-term failure under normal operation. To preclude the possibility of a 7th stage
disk failure, a pre-warm requirement was imposed on the turbine for the near-term
period (two years) over which the hardware modification was being implemented.
The following study examines the wetness aspect of sec susceptibility for the latter
stages (including the 7th stage of concern) via a steam path analysis and a rotor thermal
analysis for steam saturation and metal temperature determination. The results
illustrate the importance of including component wetness considerations in the sec
susceptibility evaluation.
Paradise LPDual Flow R.otor
stage 1 2 3 4 5 6 1 B 9 Exh.
Figure 1. Model of a single flow path of the dual flow LP steam turbine rotor studied.
181
Sanjeev Srivastav
Models and Procedures
Steam Path Analysis
The steam path analysis for the subject turbine has been carried out using the "Steam
Turbine Performance Evaluation" software package from Encotech Inc. [8] This
package calculates the steam pressure, enthalpy, temperature, and steam quality,
among various other quantities related to the thermodynamic state, at each stage of the
turbine. These quantities are calculated as a function of the stage geometry.. steam input
conditions, and steam extraction data.
To study the performance of the rotor.. three different power levels, 58 MW, 77 MW,
and 85 MW, were chosen to represent a low, intermediate, and full power level.
Similarly, three different values of condenser pressure (vacuum), 1 inch of Hg, 1.5 inch
of Hg, and 2.5 inch of Hg, were chosen to represent the variation expected in the
exhaust pressure. The changes in exhaust pressure can be associated with, for example,
changes in atmospheric conditions. Thus a total of 9 cases have been analyzed to
consider all possible combinations of the chosen power and exhaust pressure levels.
Heat balance diagrams, generated for the operation of these rotors under various
conditions, have been used for specifying the steam conditions at the turbine inlet and
exhaust. Any operation of the turbine which affects these nominal steam conditions
would require a separate steam path analysis. Table 1 is an example of the tabular
output obtained from the software. Using the predicted steam conditions, the steam
saturation temperatures have been estimated using a steam table lookup. Note that the
software analyzes the performance of the HPlIP and LP rotors together. However, all
thermodynamic calculations referenced in this paper pertain to only one side of the
dual flow LP rotor. Hence the 85 MW level refers to the power contribution of half the
LP rotor, corresponding to a nominal unit full load of 650 MW.
Table 1. Steam conditions for 85 MW power level and 1 inch Hg exhaust pressure
Location Static Pressure (PSIA) Static Enthalpy (BTUILBM)
Root Pitch Tip Root Pitch Tip
Stage 1 147.828 147.828 147.828 1386.00 1386.00 1386.00
Stage 2 86.603 86.569 86.536 1327.91 1327.91 1327.91
Stage 3 48.098 48.063 48.031 1271.19 1271.19 1271.19
Stage 4 31.673 31.661 31.650 1235.44 1235.44 1235.44
Stage 5 19.356 19.343 19.332 1197.02 1197.02 1197.02
Stage 6 11.408 11.397 11.387 1159.96 1159.96 1159.96
Stage 7 6.073 6.059 6.047 1120.44 1120.44 1120.44
Stage 8 3.081 3.069 3.060 1085.06 1085.06 1085.06
Stage 9 1.402 1.391 1.383 1046.80 1046.80 1046.80
Exhaust 0.455 0.434 0.423 996.54 996.54 996.54
182
Sanjeev Srivastav
Thermal Analysis
The thermal analysis for estimating rotor temperatures is performed using the EPRI
SAFER-PC computer program [6]. Figure 1 shows the axisymmetric finite element
model of the rotor used in SAFER-PC. The slightly curved geometry of the disk faces
on this rotor has been simplified by flat faces. Inlet and outlet steam conditions needed
for this thermal analysis have been obtained from the results of the stearn path analysis
described above. Conditions for the various stages, and at the interstage seals are
computed within SAFER-PC based on user-specified geometric data. Steady-state
temperatures have been used for estimating the wetness at various stages. Figure 2
shows an example of the rotor temperature plots generated using SAFER-PC. Although
SAFER-PC can also be used for estimating stress 'levels and remaining life, that portion
of the program has not been exercised for this analysis.
It should be noted that the accuracy of hath the steam path analysis and the thermal
analysis can he improved by the imposition of additional boundary conditions, such as
interstage seal steam conditions.
Contour Viewer I!l!llIil E.i
Steady state temperature (deg. F)
84.0 127.0170,0 213.0256.0299.0342.0385.0 428.0471.0514.0557.0600.0643.0 686.0
Figure 2. Example of a steady-state rotor thermal condition developed from an EPRI
SAFER-PC analysis for the 85 MW (full load), 1 in. Hg condenser pressure
operating condition.
183
Sanjeev Srivastav
Results
Of the nine operating conditions analyzed, results from four of these are detailed here
in order to illustrate the main observations. Figures 3 through 6 show steam saturation
temperatures in comparison with steady-state metal temperatures at the latter turbine
stages for the 58 MW, 1 in. Hg (Fig. 3), 58 MW, 2.5 in. Hg (Fig. 4), 85 MW, 1 in. Hg (Fig.
5), and 85 MW, 2.5 in. Hg (Fig. 6) conditions. The verticals represent interstage regions
and the central portion between the verticals represent the disk rim locations. The
thermal profiles permit prediction of wetness potential or susceptibility via comparison
of the steam saturation temperature with the metal temperature at locations of interest
along the turbine. Wetness susceptibility increases as the difference between the metal
temperature and the steam saturation temperature decreases. The disk rim or blade
attachment region is represented by the data points between the verticals or interstage
lines. The steam saturation conditions are shown both for normal and choked flow, the
latter being applicable only to the keyway wetness determination. The metal
temperature data are shown for the keyway region at the near-interstage location
(points on the verticals) and for the rim region (points midway between verticals). An
approximate smooth curve could be drawn through all of the metal temperature points
suggesting little or no difference between near-bore and rim temperatures.
It can be"seen from Figs. 3 through 6 that, in general, the last (L-O) or 9th stage shows a
susceptibility to wetness at the keyway and at the disk rim. This is independent of the
operating conditions (load level and condenser pressure). In contrast, the L-2 or 7th
stage shows little or no susceptibility to wetness at the keyway and rim. In fact, it may
be inferred from the profiles that the disk rim of this stage is immune to wetness and
therefore, see. The keyway at this stage also shows a large, albeit smaller (than for the
rim) difference between the steam saturation temperature and the metal temperature,
with the smallest difference being for the downstream face of the disk under full-load
(85 MW) conditions and 1 in. Hg condenser pressure (40F per Fig. 5). This worst-case
condition for the stage also assumes choked flow. It can be generally inferred that the
L-2 stage disk is expected to remain dry under most operating conditions. This inferred
weak susceptibility to wetness at this stage was not previously known, nor could it
have been determined without the steam path and thermal analysis conducted here.
Further, the wet steam conditions at this L-2 stage (steam quality <1) would, by itself
and without a metal temperature estimation, suggest a misplaced concern for sec.
The L-l or 8th stage disk rim shows a weak susceptibility to wetness at the rim (40F
difference between metal and steam saturation temperature at full load and 1 in. Hg;
Figure 5) with an obvious susceptibility to wetness at the keyway under choked flow
conditions. Under the re-built fat shaft modification, only the rim is of potential
concern. Under the fat shaft retrofit, it is inferred from the profiles, that the L-O or last
stage needs to be carefully studied for see at the disk rim, other stages being of far less
concern.
184
Sanjeev Srivastav
300 ,...-----,-----,------.,---------r-----,
250
200
Temperature
(F) 150
100
50
......
- .. - -
............... i
';--.--: -. - -
.... ~
MetalT
Steam S;tluratioll T
Steam Saturation T
(choked)
O------'- ....L...- .L...- -l.-----'
6 7 8 9 Exhaust
STAGE NO.
Figure 3. Disk metal temperature and steam saturation temperature profiles for the
latter stages of the LP turbine under conditions of 58 MW and 1 in. Hg
condenser pressure; verticals represent interstage regions and the midpOint
betw'een verticals represents the ~ s k rim region.
300 ,...---.......----,....-----,..-------,---,
250
200
Temperature
(F) 150
100
50
I i
..-1-. -t'
-....-.
.... -- ..... --
! ._- -.-.
Metal T
Steam Saturation T
Steam Saturation T
(choked)
I
- -.
I
i
i
O----------'--- .L...- -l.-----'
6 7 8 9 Exhaust
STAGE NO.
latter stages of the LP turbine under conditions of 58 MW and 2.5 in. Hg
condenser pressure; verticals represent interstage regions and the midpoint
betw'een verticals represents the disk rim region.
185
Sanjeev Srivastav
300
250
200
Temperature
a.
...
(F)
150
...: .- -
.........
100
50
0
MetalT
Steam Saturation T
Steam Saturation T
(choked)
6 7 8 9 Exhaust
STAGE NO.
latter stages of the LP turbine under conditions of 85 MW and 1 in. Hg
between verticals represents the disk r i ~ region.
300 ,----__----,---------------,
250
200
Temperature
(F) 150
100
50
.......
MetalT
Steam Saturation T
Steam Saturation T
(choked)
.... -:-:: --
..... -...L....- .--
!, .- ......
\ ~ ...l..._ ___'_ l....-__J
6 7 8 9 Exhaust
STAGE NO.
latter stages of the LP turbine under conditions of 85 MW and 2.5 in. Hg
between verticals represents the disk rim region.
186
Sanjeev Srivastav
Effect of Power Level and Condenser Pressure
Power Level. For a given operating condenser pressure, power levels do not influence
disk metal temperatures. Further, at condenser pressure levels that are typical (-1.5 in.
Hg), there is virtually no effect of power level on wetness susceptibility. Comparison of
the profiles of Figs. 3 (58 MW, 1 in. Hg) and 5 (85 MW, 1 in.Hg) illustrates the point
(little or no change in the difference between metal and steam saturation temperature).
Under high condenser pressure conditions (2.5 in. Hg), there appears to be a slight
increase in wetness susceptibility of the L-1 or 8th stage disk, and what appears to be an
expansion of the wetness susceptibility zone in a direction upstream from the exhaust.
The effect is slight, but shows a trend that can be of interest where operating power
levels change simultaneously with high condenser pressures being operative.
The exercise, in general, shows a weak or negligible effect of operating power level for
these units and for the manner in which they are typically operated. Naturally, changes
in steam throttling practice and other controls may change this situation. As with any
turbine, steam path and rotor thermal analyses are recommended for development of
thermal profiles and a wetness susceptibility picture.
Condenser Pressure It would intuitively appear that increases in condenser pressure
would increase the steam saturation temperature at the latter stages and produce an
increase in disk wetness susceptibility. For the turbine and operating conditions
analyzed here it was found that that, while stearn saturation temperatures do increase
at the latter stages with increasing condenser pressure, latter stage disk metal
temperatures also increase, resulting in little or no change in wetness susceptibility. A
comparison of Fig. 3 with 4, and of Fig. 5 with 6 illustrates the effect. Again, these
conclusions could only be drawn from a unit-specific steam path and rotor thermal
analysis, highlighting the importance of the exercise.
Conclusions
The steam path and rotor thermal analysis exercise conducted for TVA's Paradise Units
1 and 2 LP turbines shows the importance of including considerations of disk wetness
when evaluating the potential for SCC.
Specifically, the units, when analyzed for reported operating cycles/conditions, showed
a very weak, if at all, susceptibility to disk keyway wetness at the 7th or L-2 stage. This
contrasts with a concern for sec that could be generated by simply examining the
quality of steam, which is indicated to be significantly wet at this stage.
The same analyses provide useful insight ~ t the possibility of disk rim/blade
attachment SCC on the basis of wetness susceptibility. For these units and their typical
187
Sanjeev Srivastav
operation, it appears that only the 9th or last, L-O stage can be concluded to be
definitively at risk of wetness and sec, with the 8th or L-1 stage showing a relatively
weak rim wetness susceptibility.
An intuitive inference that could be made on the basis of steam and operating
conditions was found to be at odds with the results of the analyses. For example,
increased condenser pressures increased the latter stage steam saturation temperatures,
but did not change wetness susceptibility due to accompanying increases in metal
temperature.
The unit-specific analyses described herein highlight the importance of examining LP
turbines for disk wetness on a u:r:it-specific basis.that includes consideration of specific
operating conditions and rotor design.
References
1. F.P. Lyle, Jr. and H. C. Burghard, Jr., "Steam Turbine Disk Cracking Experience,"
EPRI NP-2429, Final report on project RP1398-5, Electric Power Research Institute,
Palo Alto, CA, 1982.
2. "Guidelines for Predicting the Life of Steam Turbine Disks Exhibiting Stress
Corrosion Cracking," Vols. 1 and 2, Electric Power Research Institute, Palo Alto,
CA,1989.
3. "Proceedings: Fossil Steam Turbine Disc Cracking Workshop," EPR! GS-7250"
Electric Power Research Institute, Palo Alto, CA, April 1991.
4. D. A. Rosario et al., "LP Rotor Rim-Attachment Cracking - Survey of Utility
Experience," Final Report on EPR! Project RP9005-01, Electric Power Research
Institute, Palo Alto, CA, October 1996.
5. D. Kalderon, "Steam Turbine Failure at Hinkley Point'A'," Proc., Inst. Mech.
Engineers, London, UK, v. 186, 1972, p. 341.
6. SAFER-PC, Ver. 0.5, Steam Turbine Rotor Life Assessment - Cyclic Life Evaluation
of Rotors, EPRI Project RP2481-06. Developed by Failure Analysis Associates, Inc.
7. D. A. Rosario, B. W. Roberts and M. F. Steakley, "Risk Assessment of Failure by
Stress Corrosion Cracking in Shrunk-On Disks of Low Pressure Turbines," Proc.,
1996 International Joint Power Generation Conference, ASME, New York, NY, 1996.
8. Steam Turbine Performance Evaluation, Ver. 5.1, Encotech Inc., Schenectady, NY.
188
Presented By
Jay Ely
Northeast Nuclear Energy Company
189
190
J. F. Ely
Northeast Utilities
Rope Ferry Rd.
Waterford, Ct 06325
191
192
Abstract
An inspection performed on the shrunk-on wheels of the Millstone Unit 1 steam turbine rotor in May, 1991
revealed unusual disk face cracks. The unit had approximately 160,000 hours of operation and previous;
1984 and 1987 examinations had found indications only at the keyway or emanating from the wheel bore.
The 6TA wheel had three indications, at 91,337, and 355, emanating from the hub face. To determine
the axial extent of the indications, the hub lip was removed. After verifying the indications ultrasonically,
the front 2" of the disk was removed to allow destructive examination of the flaws.
The flaws were intergranular stress corrosion cracks, nearly semi-circular in shape with a maximum size of
1.4" long by 0.85" deep. This was the first time GE had discovered flaws on the hub face. The inspection
technique was not optimized for this orientation or location. Flaw growth was conservatively higher than
previously seen, and its effect on the probabilistic results was evaluated.
Introduction
In May 1991, a routine wheel bore ultrasonic inspection was performed by GE on the low pressure 'A'
(LPA) steam turbine shrunk-on wheels. In addition to indications at the axial keyway, three indications of
large amplitude were detected along the upstream hub face ofthe 6th turbine end wheel (6TA or L-2) at
locations well removed from the axial keyway. Further ultrasonic and fiberoptic examinations confirmed
the presence of cracks on the upstream hub face. Following removal of a ring containing the cracks
(figure I) a metallographic examination confirmed that the cracks were branched and intergranular, typical
of stress corrosion cracking. The axial depths ranged from 0.43" to 0.85", and were semi-circular. While
large, the flaws were well below the wheel's critical crack size.
Inspections of the remaining LPA wheels, including a new ultrasonic technique for the 4th, 5th, and 6th
wheel faces, did not find any additional flaws.
The face cracks were unique in that similar cracks had not been previously seen by GE any nuclear
turbines. Initiation of stress corrosion cracks at the wheel face had been considered extremely unlikely due
to the low operating stresses, approximately 50% of the materials yield strength. Also unique was the
relatively large depth of the cracks, which when combined with the lack of earlier indications during
inspection of the 6TA wheel, implied a higher than expected crack growth rate.
As a result of the cracks at Millstone #1, and the implications to the nuclear turbine safety assessment, GE
perfonned an extensive investigation, with the assistance of Northeast Utilities. The objectives were:
1. Determine the root cause of the hub face cracking.
2. Assess the implications of the face cracking on the inspection intervals for the GE
nuclear turbines.
Wheel Bore Inspection Procedure
The standard wheel bore ultrasonic inspection procedure, applied up to May 1991, was designed to detect
cracks originating from the wheel bore or keyway region where the probability of stress corrosion cracking
was considered highest. Due to the testing geometry's used, the test had a reduced sensitivity to flaws
along the wheel face region. Following the Millstone inspection, an additional series of ultrasonic and
193
Jay Ely
fiberoptic inspections were incorporated to provide a much higher sensitivity to wheel face cracks. Over
16 rotors had been inspected through 1991 with the new procedures, and no additional face cracks have
been found.
Investigation of the Root Cause for Face Cracking
The investigation into the root cause of the face cracking included:
I. Review of Millstone and GE Nuclear plant steam chemistry records, and evaluated
deposit samples to check for corrosive contaminants.
2. Review of Millstone plant operating records and NRC Greybook data to check for
unusual start-ups and shutdowns, or other unusual load cycling of the unit.
3. Determine if a mechanism exists for creating a more corrosive environment along the
wheel face, for example, by concentrating dissolved oxygen levels.
4. Determine if there was anything unique about the 6TA wheel material that would
increase its susceptibility to stress corrosion cracking. .
The results of the investigations were:
1. No evidence of poor steam chemistry or exposure to corrosive contaminants was found.
2. The number of start-ups and shutdowns and the load cycling duty of Millstone I is
typical of nuclear units.
3. The tight gap (less than 5 mils) between the 5TA and the 6TA wheel faces may have
played a role in the face cracking. Calculations by GE indicate that oxygen can
concentrate in stearn condensate formed along the wheel faces when the gap between
wheels is tight. While stress corrosion crack initiation is more pronounced in creviced
regions, many wheels have less than 5 mil gaps and no additional face cracks have been
found.
4. All mechanical, metallurgical, and manufacturing parameters of the 6TA wheel are
typical of similar vintage NiCrMoV wheels.
Therefore, although a number of influencing factors were identified, no specific cause could be found for
this occurance of cracking.
Assess Implications to Wheel Inspection Intervals
Based on the result of the face cracking assessment and available inservice inspection data, GE treated the
face cracking as a random occurance. This reflects the view that SCC can only be predicted
probabilistically, not deterministically. Higher stress levels and more corrosive environments produce
more cracks, but there is sufficient scatter in the data for occasional anomalies.
While the Millstone 1 face wheel SCC may be a unique occurrence, an appraisal of its impact on turbine
missile probabilities was made. The reason for evaluating this incident is that current wheel inspection
intervals were based on results obtained prior to introducing the increased sensitivity of wheel face
inspections.. Also, the evaluations used to set those inspection intervals did not factor in the possibility of
wheel face cracks.
GE assessed wheel face cracking as a random event; modifying the methodology, the effect on the
probabilistic analysis, the fracture mechanics parameters, the probability of detection versus size, the crack
growth rate distribution, and the probability of crack initiation. The important effects were the addition of
the probability of face crack initiation, and the maximum estimated crack growth rate. The maximum
crack growth rate was two standard deviations above the keyway data, but it was conservatively decided to
include all the wheel face estimates within a revised limit.
194
Jay Ely
With ]7,000 wheel-years of operation, averaging between two and three inspections per wheel, only one
wheel has exhibited face cracking. Monte Carlo simulation was used to estimate the probability of
initiation for the GE nuclear turbine wheel population. The estimated mean time for face crack initiation
from this analysis is about two orders of magnitude longer than the mean time for keyway cracking.
Conclusions
A method was developed by GE to account for the impact of wheel face see on turbine wheel missile
probabilities. The method applies to GE nuclear LP turbines with shrunk on wheels, and all involved
rotors have been re-evaluated using this methodology. The resulting increases in missile probability were
very small, less than two percent of original values. When this was considered in terms of the wheel
inspection intervals, it was deemed insignificant
A more sensitive wheel face inspection procedure was implemented following the Millstone examination,
and is being used in conjunction with the wheel bore ultrasonic inspections. GE has decided that baring
new wheel face cracks, GE will evaluate wheel missile probabilities per the standard GE procedure, which
includes keyway cracks, but not face cracks. If wheel face cracks are detected during future inspections,
new calculations, including the face crack effect, will be required, and recommendations for wheel
inspection intervals to maintain the unit within NRC limits will be impacted.
195
tT1
/'
Figure 4
7 - 5GAWheel- Keyway Indication; Web = 0.25"; Non-Measurable
8 - 6TA Wheel - Hub Outside Diameter (Discharge); Maximum Indication
Length = 0.38" at 260
0
; Crack Depth == 13/32 It
17 1/8" Long Indications Revealed on Hub Outside
Diameter Continuing into Lifting lug Groove
9 - 6TA Wheel - Dovetail Cracking - Indications Still Present After
Grinding to a Depth of 3/8"; Maximum Depth = 1/2" in Top
Dovetail Hook When Removed for Laboratory Investigation;
Remaining Hooks Removed by Machining
10 - 5TA Wheel - Admission Face; Severe Water Cutting at 10
0
; Pitted
Out Area; 5/16" Deep
Northeast Nuclear argy Company
Millstone Point #1
Turbine #: 170x346 Serial #: FD2152
LPA Rotor

F-- -- I 91 tr' )
1 - 8TA Wheel- Hub Outside Diameter (Admission); 16 Indications;
1/32" - 3/16" Long Between 180
0
- 210
0
2 - 6TA Wheel - Web Keyway Indications; 0.44" Long Measured on Newly
Machined Surface; q.30" long Measured Ultrasonically
3 - 6TA Wheel - Hub Face Indications at 91
0
(0.82" x 0.17" x 0.42"); 33]0
(0.5" x 0.75"); 355"(0.85" x 1.4"); 1.6" of the Hub
Removed to Investigate Cracks
4 - 4TA Wheel- Keyway Indications; Hub =0.72"; Web =0.43"
5 - 4TA Wheel- Admission Face; Severe Water Cutting at 175"; 1/2"
Wide Pitted Out Area; Approximately 1/8" Deep
6 - 4GA Wheel - Admission Face; Severe Water Cutting at 3; 1" Long x
1/2" Wide Pitted Out Area; Approximately 3/32" Deep.
Note: 1 Indication < 1/8" long Revealed in Above Area

., \ " I
10 "" I I Ii
TURBINE END f Ii
c=:====-=====-. I I I
lo---1
\0
0\
MILLSTONE #1 6TA WHEEL CRACKING 5/91
Jay Ely
c
I
Hue FACE ANP

HUB 0.0.
c RAe 1:.$'
I
197
1-1
\0
ao
Millstone Point #1
Wheel: 6TA Hub Face Indications (ADM.)
335
0
-
-+-----:--- + l I J 91
0
I
Keyway 1

ts:::
t'11

MILLSTONE #1 WHEEL HUB FACE CRACKING
srA
........ I - - -
NOH. S6 Jl:SI
srRFSS FllcrOIl -= C. If
.. STEAM
FLOW
6 TA
C RAC.I< 3SS
(NOT IN-LINt!
I
.J
f-.l.
\0
\0
tr1
Millstone #1
170X346
6TA Wheel Hub (ADM)
Jay Ely
200
5.000"
-7.500" REF
L-3
Crack Sample To
Be Salvaged
For Analysis
.020..:.-.........-
-7.875" REF
L-2
.601 "
_______1 _
Keyway
HUB.MACHINING CONSIDERATIONS
i
I
I
I
...,..urt'"'.,
li i
. / / '
/ .. ' I //
" ,. j'': ..
~ / / II 1/ ~
~ . ..,
ORIGINAL
DESIGN
Jay Ely
AFTER
HUB MACHINING
AVERAGE BORE STRESS VS ALLOWABLE (\)
AVERAGE BURSTING STRESS VS ALLOWABLE (\)
LOSS OF SHRINK, OVERSPEED (%)
97.0
98.4
21.0
102.1
102.2
16.5
201
HUb
.. Face
N
o
N
Millstone Point #1
Wheel: 6TA Hub at 91
L' ........ _.7"".-'":.::-.;:i....... ... .
. ".j '. '.
_. I
Primary I:'
------.. ...
I
.... .. ;.. ,.:..... ,
F tj4Q t rHlE'!.
Possible Separate Cracks Ct.' . .'
1/111/111/11'1111/11 '/'11/ II11 II
123
45
Approximately 1x to 2x Magnification
Primary Cracking to Hub View: Hub and Bore Surfaces
Surface and Possible
Separate Cracks
<..=::
t'r:l
N
a
CJJ
Millstone Point #1
Wheel: 6TA Keyways Cracking
; ~ ~
/-
.. I. '. I
~ 1 " ,;i, - l'
III tIII JlIIIIIIIIIII"1l1111' 11111111111 III J' II II "1
1
" 1
111
1' 111
111
1'
1 234 5
Cracks Initiating From View: From Hub Face (Adm. Side)
Keyway Roof and Cracks
Possibly Initiating From Hub Face
1::;"
~
tT1
~
N
52
Millstone Point #1
Wheel: 6TA Keyway Cracking
, "
IIII'I/ /1/ '/ III /III f" IJ1I'I " II I ~ I \' I' \III \ \ \
I 2 3 4 5
Cracks Initiating From View: Keyway Roof and Hub Face
Keyway Roof and Cracks
Possibly Initiating From Hub Face
~
~
t't:l
'<'
IV
o
en
Millstone Point #1
Wheel: 6TA Keyway Cracking
J"
t. _y
c ttl
I . I ~ . J II!! II I II I1 1111 !I", ,1 1111" III i I.I.!! I ,! !! : I' II, !r I, ,,I ! ! I!' , ,f , , , !
Cracks Initiating From View: Machined Face
Keyway Roof (Opposite Hub Face)
~
"<
tT1
~
N
a
0\
'i;"<
tT1

Millstone Point #1
0
.... }
.
. ,
,.,
... " .

l
'','
,'. i
.' I
..L,":, , ** ._-, Bor.
jllIlllll"1111 11"1/ 1111"111111111111111111111 jll II 11111111111111
1 2 3 4
Approximately 1x to 2x
Primary Crack Extending View: From Hub Face and
to Bore Surface and Bore Surface
Possible Separate Cracks
N
o
'1
Millstone Point #1
0
\ \\, 'Ill \1'1' lill ' 111111-1111 t111I11111111111 Jill J1111 11 '[111 J II J
234 5
Primary Cracking Extending View: From Bore Surface
to Bore Surface and
Possible Separate Cracks

t't.l

IV
a
00
Millstone Point #1
Turbine #: 170X346 Serial #: FD2152
. Wheel: 6TA Hub Face Crack at 355
0
4x Magnification
View: Perpendicular to Hub Face

I..l:::
t"t:1

tv
o
\0
Millstone Point #1
Turbine #: 170X346 Serial #: FD2152
Wheel: 6TA Hub Face Crack at 355
0
4
4x Magnification
View: Perpendicular From Hub Face

m

N
t---l
o
Millstone Point #1
Wheel: 6TA Hub Face at 355
0

t'rJ

Possible Separate
Cracks
Possible Separate
Cracks
1
-secondary
Crack
+,- .
50x Magnification
Fracture Surface and View: Hub Face
Accompanying -Secondary Cracks
and Possible Separate Cracks
MILLSTONE #1 FACE CRACK - GROWTH RATE VERSUS EXPERIENCE CURVE
18 =t I I I I I
.
611,)<,,(/' 01-1.- H.
. /" 'k."' '-/: ") /
L .. /"'1-'/ to'-' (...I" (.,/
;78"2- <;...... 97
480
o IVA
6.PVA
a IVR Af[HSP
*M RflHSP
358
o
o
o
o
o
308
-e
258
,,----
MllLSTJoNE # I
6TA FAlcE C RAe K
o
Q)
200
1
'"" cr
>
"
Z
H
'-"
W
.-
([ 1"0-
1
0::
:i:
t-
:Jt
a
0::
(!)
10-
2
-1/ .
0
0
1
0
10-
3
I
158
WHEEL OPERATING TEMP (OEG F)
N
CRACK GfV)l!TlI R!\TE UPPEn
t'r:1
212
Turbine Stress Corrosion Cracking
- Duke Power's Perspective
Presented By
TomAlley
Duke Power Company
213
214
TURBINE STRESS CORROSION CRACKING
DUKE POWER'S PERSPECTIVE
c. Thomas Alley Jr.
Senior Engineer, Duke Power Company
ABSTRACT
The past decade has seen a significant increase in the occurrence rate and impact of stress
corrosion cracking (SeC) of power plant components. Turbine components have had a
particularly significant history involving both major US manufacturers and most major
components. Rotor forgings, blades, generator retaining rings, and a variety of smaller
components have all experienced SCC. This paper describes one utilities experiences with
SCC of turbine components. also discussed are remedial actions taken. and ideas of future
developments that may be explored as means to prevent SCC.
BACKGROUND
Stress corrosion cracking of turbine components emerged as a concern in the late 1960s
when Nuclear turbine owners and manufacturers were alarmed at the possibility of
cracking of built-up or disc machines. In 1969 a disc at the Hinkley Point Station in the
United Kingdom suddenly burst while the unit was operating.(1) The cause of the burst
was determined to be Stress Corrosion Cracking (SCC). This one incident served as a
land mark failure event eventually impacting a majority of the U.S. nuclear fleet. Some
years later this same mode of failure is haunting Fossil Turbine owners as well. The failure
at Hinkley point resulted in both major U. S. manufacturers issuing infonnation letters
recommending inspection of turbine discs or wheels. (2,3,4,5) The U. S. Nuclear
Regulatory Commission also became involved by issuing several bulletins and chartering
studies into probabilities of failure and the potential impact to plant operations and safety
as a result of disc failures (6,7.8). Since Hinkley Point, numerous other machines have
experiences significant stress corrosion" cracking resulting in the spending of millions of
dollars on replacement components, inspections and outage extensions.
Data and experiences suggest that SCC is possibly the most serious active failure
mechanism affecting turbines of all manufacturers and designs. Both Nuclear and Fossil
units are affected although the cooler regions of the Nuclear units make them more prone
to the conditions required to initiate and propagate stress corrosion cracks. Of primary
concern is the integrity of the rotor forgings themselves however other components such
as blades and associated hardware are ~ s subject to stress corrosion cracking.
More recently, stress corrosion cracking of blade or bucket attachments is emerging as
serious concern with numerous cases reported by most major utilities. Unlike disc
cracking where models, formula, and inspection techniques are established. the rim
attachment cracking inspections are valid for detection only, and models and formula do
not exist to adequately predict the behavior of these cracks. EPRI has sponsored several
215
216
programs to assist in this situation but development has been slow and results are painfully
difficult to implement.(9,lO) The major OEM manufacturers are well aware of the
magnitude of this issue are continuing to provide technical recommendations and offer
inspections as required
Most of the problems discussed so far are concentrated among the Nuclear turbines,
however, similar events are documented for the Fossil industry. G. E. turbines have
experienced cracking in the wheels of several once through type boiler units and have
issued additional recommendations to inspect Fossil wheels similar to the Nuclear
machines. (11) The cracking of Fossil wheels is somewhat unique in that for many of
these machines the material properties are poor and essentially no crack growth is
acceptable. This situation was the subject of an EPRI workshop in 1990 (12) at which
discussions were held to gain additional understandings of the risk associated with the
operation of older Fossil built-up machines. The Fossil cases however, seems to show a
trend among several units which may lead to the conclusion that perhaps these cases are
isolated incidents and the results of poor material, operating practice, manufacturing
phenomena or perhaps maintenance practice. History will indicate the true extent of the
Fossil Wheel cracking problem as inspections results areaccumulate and risks better
defined. One point is certain, the Fossil units have accumulated significantly more
operating hours than the Nuclear units without widespread cases of rotor SCC and since
the materials are very similar it maybe concluded that the higher back stage temperatures
or some other operational parameter might separate these two situations.
MECHANISM
Research has concluded that it takes three primary influences to initiate and propagate a
stress corrosion crack. These being:
MATERIAL SUSCEPTABILITY
CHEMICAL ENVIRONMENT
STRESS
With each influence there exist a series of secondary factors that may prove significant in
the initiation or propagation of sec. The elimination of any of the three influences and
lor the altering of some of the secondary factors may mitigate the presence of sce.
Secondary factors include the material yield strength, t m p r t u r ~ slight refmement in
chemistry, refinement of material constituents etc.
The presence of sustained tensile stress is one influence that has a well documented role in
the initiation and propagation of SCC. Figure 1 is used to understand the role of stress
and to show that there exist a critical value stress intensity (K
scc
) at which see cracking
will initiate given a specific environment. Above the K
scc
cracking will initiate and
propagate while below that level cracking will not occur. Since lowering stress intensity
to a level below Kscc eliminates sec it has become the basis for many of the solutions
proposed for turbine components. Theoretically, this is why increased radii in blade
Tom Alley
attachments, and shot peening are proposed as solutions for turbines already suffering
see damage. Although these actions appear promising, experience has shown that the
remedial action of remachining or shot peening has a very low probability of success
because it is very difficult to make significant reductions in stress for these highly loaded
areas. In general, key-ways and bucket attachments are subjected to loads well above
~ c c and small reductions in stress due to geometry changes or peening are ineffective.
The greatest benefit from machining may lie in the restoration of material to new
conditions and the restarting of sec incubation periods. The difficulty in proposing or
evaluating these techniques lies in the inability to accurately predict to value of Kscc and
to accurately project the effects expect from the remedial actions. These types.of
modification although very common are most likely not going to prevent future cracking.
K1cr--......----.-------------....---
K - -------------------
Iscc
--... time to fa i lure
Figure 1. Critical Stress Intensity for SCC
Influences such as chemical environment are more complicated and difficult to quantify
than the stress but can have a more profound effect on crack initiation and propagation.
Figure 2 shows data taken as the result of ERPI research in the area of see (13). The
data is widely scattered and dependent of the environment, crack growth rates can vary
significantly. Issues with the accumulation of this data typically show up as ignorance of
the actual conditions at the site of concern or error associated with the means of
measuring crack depth. Since for the most part units operate with-in the manufacturers
recommended bulk chemical requirements, solutions involving chemistry are not easily
pursued Chemistry excursions however, may playa significant role in cracking case
histories but insufficient data exists to understand or prove this postulate. This seems
particularly tme in the area of Fossil rotor wheel cracking where the cracking experienced
is confined to a relative few number of units. This conclusion is easily reached but if the
Fossil experiences follows that of nuclear, the units that experienced early cases of
cracking may be ftrst to initiate damage because of a particular forcing function with the
217
TomAlley
balance of the fleet to follow displaced by some finite time lag. Time will tell if the Fossil
experience is more unique or generic in nature.
rtflnr" "O'Ctl2
t
fj
swn,
11,. NIIOH
Y.$. =184 - la ICII
"
E

&
,5; t
e
1I

M t:
*

0
i

/t
i t
..

hI1
a::
PURE WAYEI
u
,.a. I'f .. fll Kit
IrClflM JftIlt1J
Figure 2. SCC Crack Growth Rate
source: EPRI NP-6444 / southwest research
The fmal influence is one of susceptibility of a given material to SCC. Alternating
material composition can certainly have a profound effect of the corrosion resistance,
initiation and propagation of SCC damage. This was certainly true for the 18Mn -SCr
non-magnetic generator retaining rings (14,15) where the remanufacture of retaining rings
from 18Mn-18Cr material is presented by the major manufacturers as a solution for sec.
For the short term this appears to be a good solution for this type of cracking, although
the long tenn integrity of this solution is currently being questioned (16).
The substitution of materials to eliminate see is by far the most successful long tenn
solution to this problem. Consideration is also given for those solutions that lower yield
strength and subsequently hardness. Substitution of materials has been very successful in
the resolution of blading sec. Examples include the reheat treatment of 422 stainless steel
buckets to lower the yield strength (with associated hardness reduction) eliminating
failures experienced by G. E. in some of the larger last stage buckets. Subsequent repair
218
TomAlley
of these same buckets resulted in SCC failure in the repair area that was corrected by
tempering the welded area to a lower yield strength. (17) Westinghouse too experienced
failure of LP turbine blading that was corrected by substitution of 17-4 ph material for the
traditional 410 stainless steel. Similar widespread success has been experienced with many
turbine components by altering or conditioning materials. Direct substitution of materials,
or altering material yield strength ( reheat treatment) of turbine rotors to eliminate SCC
cracking in key-ways and dovetails however, is not attractive because of the time and
expense involved or the adverse effects of heat treatment on mechanical properties. For
these components coatings or weld overlays may be needed to isolate the base material
from the corrosive environment.
TIlE OCONEE EXPERIENCE
Experience with rotor SCC cracking for Oconee Nuclear Station started with the
detection of a crack like indication in the rotor wheel key-way sections of Unit #1 in 1986.
Since that initial experience all 9 rotors on all three units have some degree of key-way
cracking. Two observations can be made from this particular event. First, SCC may have
occurred on all these rotors over a relative short time after a long incubation period and
secondly, during the 6 years from 1986 to 1992 the increased sensitivity of equipment and
the experience level of the NDE operators may have resulted in the detection of
indications not previously reported. Perhaps both these situations exist. Recent
inspections certainly support the fact that the indications exist, and are real, but-they also
show that the crack propagation is less than originally estimated from the NDE data. Did
the indications exist prior to the initial detection? Was there a very lengthy incubation time
to initiate a detectable crack? These questions will most likely never be answered but will
continue to playa significant role in the analysis and understanding of SCC damage. The
Oconee experience did prove that the maturity of the inspection methods, equipment, and
operators, along with the development of good fracture mechanics models makes this
situation as it exist today, manageable. In other words, inspections are petfonned and
evaluations made in a manner that has a small impact our current operation. The biggest
threat this situation presents is the potential impact on unit availability and outage should a
critical flaw be detected during any futUre examination. The risk presented by this situation
is typically reason for great concern and was critical in the decision to rebuild these rotors.
In addition to key-way cracking dovetail cracking became an issue for the Oconee rotors
late in the 1980's with the implementation of the 5 MHz transducer technology by General
Electric Company (GE). Prior to the 5 MHz testing, the Oconee rotors were inspected for
attachment cracking with no reportable indications. The fITSt signs of trouble were
indications discovered during the 1988-1989 inspection of the Oconee #2 rotors. The 5
MHz inspection detected crack like indications that were unconfrrmed during an additional
2.25 MHz test. Since the unit was well into the outage window the decision was reached
to rebuild and operate the turbine without any corrective maintenance or exploratory
disassembly. Actions as a result of the 5 MHz examination would be postponed until the
next scheduled outage to allow for engineering evaluation, adequate planning, and the
scheduling of addition work. Significant to this decision were the facts that no indications
219
TomAlley
were discovered during a retest utilizing the 2.25 MHz transducer, and the experience
with the 5 MHz testing at Millstone that showed a high false call rate for the 5 MHz
examination. It was also assumed that the low frequency examination was sufficient in
preventing failure since it was known that no dovetails had failed after testing with the
lower frequency. Engineering data also suggested that the G. E. bucket attachment has an
extremely conservative design with substantial margin against failure. Three years later
during the January 1992 outage the notch groups were removed from the L-3 Generator
end and the L-2 Generator end rows, and SCC cracking was confirmed. Extensive
exploratory disassemble and probe grinding was performed to quantify the extent of the
damage. The rotor was placed back in operation with a light notched piece and several
buckets removed.
The unit #2 "c" LP rotor has a very similar history and it too is operating with buckets
removed and modified notch configurations. The Oconee Unit #1 history is similar to that
of Unit 2 with the with the "B" LP and "c" LP rotors missing buckets due to SCC
damage and exploratory grinding. Ironically Unit #3 has to date no history of cracking in
the hook fits. The fact that this rotor has atypical results compared to the others is yet
unexplained.
Remedial action for the Oconee units consist of the rebuilding of all 9 LP rotors with new
wheels. Initially it was felt that the rotors would be replaced with Fully Integral ( PI )
rotors but increased financial pressure and the potential effects of torsional testing as the
new rotors were phased in service prompted the Oconee Rotor Task Force to recommend
rotor rebuilding. Wheels that contain either attachment cracking or hub and key-way
cracking will be replaced in a ordered sequential manner beginning in 1998. The project is
already underway with the ordering of a "seed" rotor which will replace the first rotor to
be rebuilt. The entire project is scheduled to take approximately 3 years and will cost a
fraction of the estimated costs of new FI rotors. Enhancements in the new wheels will
include tab designs and newer low stress hook fits. The rebuild will reuse existing buckets
and hardware where practical.
The lessons learned from the Oconee experience includes several significant items
e G. E. LP Rotors are Prone to SCC in the Wheel Key-Ways of the L-O, L-l,
L-2, L-3 Rows and in the Bucket Attachments of the L-2. L-3 Rows.
e The is No Logical Pattern to the Attachment Cracking i.e. Admission or
Discharge Side, L-3, L-2 Rows etc.
NDE Techniques are Inadequate to Provide Conclusive Data on the Size
and Depth of the Indications in the Attachments
e Tools do not Exist to Analyze the Risk Associated with Attachment
Cracking
e Reactionary Mode of Decisions Leaves the Utility with Few Options
Attachment Cracking will become a Life Issue before Key-Way Cracking
220
TomAlley
THE CATAWBA EXPERIENCE
Catawba Nuclear station went on line in 1983, 1984 and has operated without significant
turbine SCC damage since the initial start-up. Knowing that the Oconee rotors were
suffering a variety of problems and having Catawba personnel participating on the Oconee
Rotor Review Team the Catawba Turbine Engineering staff focused on ways to eliminate,
retard., or prevent the occurrence of SCC on their machines. Of primary concern was the
bucket attachment cracking since this was identified as the forcing function for Oconee
rotor rebuilds. The desire to be proactive in addressing potential sec of the rotor rim
attachments resulted in commissioning MPR to evaluate various hook modifications that
would reduce the stress concentration in the radii at the roof of the attachment tang. The
goal of this work was to evaluate options commonly promoted by the OEMs and to
closely study the potential benefits and risks of such modification. The study also included
the preferred geometry to leave in the hook fit radius should crack removal become an
issue. To obtain the rotor dimensions field measurements were taken along with reverse
dimensional data derived from a bucket. At this point only point reflectors had been
detected in the Catawba rotor attachment regions The study concluded that removal of
approximately 0.040 inch of material at the back of the hook fit as shown in Figure 3
would effectively reduce the stress concentration. The reduction in stress concentration
was anticipated to have a similar effect on the initiation and propagation of SCC. It was
concluded that machining of the Catawba rotors on rows L-2 and L-3 would be perfonn
during the next overhaul period. Plans were made" to incorporate this work into the
outage schedule along with shot peening the root area in an attempt to lower the operating
stress below K
scc
.
E'x1etJng Surface
MachTne to
F'ull Radtua
ExlwUng SUrfaee
1tl1lohiM to
F'ull
2 Places
.040
Full
Surfac:e
Machlhe to
FuI Radlu.
2 F'Iaces
FUll
Radius
F'ulJ
Rodlue
.040
Figure 3a. MPR Study Results on Hook Fit Skim Cutting
221
TomAlley
fun Radius
Figure 3b.
.130
Machining Details with Crack Removal
In the mean time, testing concluded that the Catawba rotor contained discrete indications
in some of the hook fit areas making crack removal an issue. Machiriing of the Catawba
Unit #1 "B" LP rotor was schedule and subsequently performed in July 1996. The goal of
the machining was to remove existing cracks, place a preferred geometry in the hook fit,
and to shot peen the radii. It was generally felt that the proposed work would provide a
high level of confidence that with these modification the unit could be expected to operate
free of indications for at least the same number of hour as that already accumulated .
Whether the remedial action taken for Catawba is successful will take some time to
conclude.
The Catawba Unit #1 "B" LP machining wasn't performed without issue. During the
hook fit machining a error was introduced in the rotor that resulted in mismachining and
oversizing several of the hook fit radii. This new configuration resulted in significantly
higher stress on that hook which was contrary to the goals of this project. Significant
delays and concern over the long term integrity of the rotors was again an issue. After
much debate it was decide that lightning the buckets by machining a "lightning groove"
could significantly reduce the mass of the bucket thereby reducing the stress an equivalent
amount as that introduced by the machining error. The bucket machining was accomplish
and the rotor was reinstalled and is currently operating. Inspections of the machined area
after completion of the reassemble still showed several point type reflectors. The presence
222
Tom Alley
of these indications came as a surprise to all those involved in the project and once again
focused attention on the accuracy or error of the NDE methods.. Future examinations will
be compared to this base line test to detennine if the modifications were successful.
Despite the problems associated with the rotor machining current plans are to modify the
rotors on the Catawba Unit #2 as a proactive measure to prevent or delay cracking due to
SCC.
Lessons learned during the Catawba experience includes:
Nothing Much has Changed from the Oconee Experience
Machining the Hooks is a Risk
Ultrasonic Results are Still Questionable
Possibility of Lower Stress by "Lightening Grooves"
THE McGUIRE EXPERIENCE
Among the first domestic units to repOlt rotor disk cracking were those manufactured by
Westinghouse. Since the McGuire Nuclear Station had Westinghouse rotors concern was
expresses over the potential risk associated with these rotOI'S. InItial studles by
Westinghouse resulted in identifying several discs of the McGuire rotors that were
evaluated as having poor material properties and inspection intervals for these discs were
set for 2 years. Because of the potential for cracking and other commercial issue the
McGuire rotors were replaced in 1985 by fully integral rotors. Although this action will
eliminate the potential for key-way cracking the blade attachments are still at risk without
some additional action. Integral with the Westinghouse issues is the Fossil supercritical
units at Belews Creek. These Fossil units contain Westinghouse BB 281 LP rotors and
are subject to disk inspections during turbine overhauls. Multiple tests petformed on these
rotors over the years have detected no indications. The issues with the Belews Creek
units are that the current inspection frequency does not meet our goals of extending unit
run times. Westinghouse currently recommends a disc inspection every 5 years. This
reinspection interval is not acceptable and internal investigations are beginning to analyze
and document the extension of the Belews Creek rotor inspection intervals well beyond
the 5 years currently recommended.
Although the Westinghouse Units at McGuire and Belews Creek haven't experienced
problems of the same magnitude as those at Oconee and Catawba (G.E. Units) there exist
a high probability of potential SCC problems associated with Blade attachments and fIrst
stage Nuclear lIP rotor and/or blades. Because of the Westinghouse design with the axial
entry. inspection is extremely difficult and progress towards refmement of this inspection
technique lags that of the straddle mounted blading. Significant progress needs to be
made in this area since industry data and experiences validate that this type of cracking is
occurring. The actual extent of the damage present in the industry due axial attachment
cracking is most likely underestimated because of the lack of sufficient inspection
223
TomAlley
techniques and the fact that inspections are not being perlormed as regularly as on the
G.E. type machines.
Lessons learned from the Westinghouse Experience Include:
sec Damage Typically is Not Confined to One Manufacturer
Westinghouse has some History of Cracking in the HP Blade Attachments
andlor Blades
Inspection of Westinghouse Attachments is Extremely Difficult
Patent Issues Exist with the Inspection of Westinghouse Blade
Attachments
PI Rotors May Eliminate Key-Way Cracking but Attachment Cracking
Remains an Issue
THE FOSSIL EXPERIENCE
Data and experience suggest that Fossil rotors have been almost immune to the SCC of
rotor forgings typical for the Nuclear industry. Experience with Disk Cracking and blade
cracking is scattered and seems to be concentrated among several units. The potential for
damage and the smaller estimated critical crack sizes for many of the Fossll machines
continues to place pressure on the development of newer inspection techniques and
analysis tools to assist utility engineers in detennining the relative risk of operation and
risk of outage and equipment delays. As Fossil units strive for longer operating time
between intervals disc SCC issues will remain high on the list of items that must be
addressed.
LIFE PROJECTIONS
Theoretically, the life analysis and projected growth of sec damage fits analytical models
extremely well. In the laboratory under controlled conditions the crack growth rate
associated with sec can be easily defmed. Knowing the crack growth rate, and the
critical flaw size, life can be estimated as below:
Growth Rate daJdt = and
hrs
Life
daldt
where Clc is the critical cmck size
indicated crack size
hrs since last inspection
224
Tom Alley
Unfortunately. the uncertainty associated the actual field applications of these equations
make life estimates extremely difficult. Material property uncertainty specifically,
fracture toughness. and the scatter associated with inspection sensitivity and sizing. lead to
unrealistic estimates of component life. Cle itself is estimated based upon stress,
temperature.. fracture toughness.. and crack tip parameters all of which are not accurately
known. Critical crack size may easily vary from 0.25 to 4.0 inches based solely upon
material property scatter. Add to that scatter variations in stress, temperature, and crack
tip parameters and results can very pessimistic or very optimistic depending on the
assumption made by the analyst. Knowing that a; is a function of a given test system, the
physics of sound energy, the geometry of the component.. the orientation of the flaw and
the consistency of the personnel fonning the inspection team.. large error maybe associated
with the reporting of flaws. Some data exist that demonstrates the repeatability of a given
system however for many of the techniques utilized in the utility industry, this data is rare
or incomplete. As an be seen in the repeatability on a variety of Boresonic test systems,
large scatter in the test results may be present even under controlled conditions. Given
that this data was collected under very controlled conditions.. field data would have an
even wider scatter. Therefore the use of field data to either assess the severity of a flaws
or supply the numerator in a growth rate equation should be treated with a high degree of
caution and uncertainty.
Crack initiation is another separate component of the life estimate of see that is not fully
understood. Data often suggest that initiation is a major portion of the component life
while the actual propagation and growth, once detected.. occurs rapidly. Incubation time
for crack initiation is thought to be as much as 80% of the component life. This is
certainly a common theory in the case of turbine disc cracking and rim attachment
cracking where years of successful operation was accumulated prior to the detection of
any significant damage. One may assume that during the early years of operation
initiation was under way. If crack initiation is a major life contributor it must be
recognized that the industry has no acceptable models for predicting this phenomena and
factoring this into life estimates. EPRI some years ago sponsored some impressive
research on corrosion damage and the coalescence of pits to cracks but the end effect of
the research as not lead to significant development of tools to assist the utility engineer in
assessing sec damage. (12)
REMEDIAL ACTIONS
The solution for stress corrosion cracking lies with the reduction or elimination of one of
the three primary influences associated w.ith crack initiation and propagation. Examination
of the influences on a large scale may offer some clues as to the potential solutions to
SCC. Table 1 shows the influences and some of the major secondary factors.
Consideration is given for each of these influences and the possibilities associated with
each.
225
TomAlley
Table I
INFLUENCE
ENVIRONMENTAL
INFLUENCES
STRESS
INFLUENCES
sec CORRECTIVE
ACTION
SECONDARY EFFFECT
-TEMPERATURE
-CHEMISTRY
-EXCURSIONS
-STEAD STATE OPERATING STRESS
-PEAK OPERATING STRESS
-SHOT PEENING
-DIMENSIONAL NOTCH EFFECTS
-MACHINING
-RESIDUAL STRESS
CHANCE OF
SUCCESS
low/medium
low
MATERIAL
INFLUENCES
-MATERIAL SUBSTITUTIONS
-COATINGS high
-CLADDING
. -REDUCED HARDNESS/YIELD STRENGTH
The best opportunity for the elimination of ~ problems lies with material substitutions
or coatings/cladding that may insulate the original material from the corrosive
environment. The expense and magnitude of direct replacement makes those options cost
prohibitive while other options have less chances of success.
CONCLUSION
Duke Power Company continues to deal with SCC through comprehensive programs
concentrated on advanced NDE testing and reactionary component replacement. These
programs however, are recognized as insufficient in delivering solutions to eliminate the
root causes of SCC and decisions resulting from these programs doesn't fit well into our
newer culture of proactive preventative maintenance, quicker outages, root cause, and
cost control. In fact, they are typically in conflict with these initiatives. In an effort to
advance our understanding of SCC and the impact it has on our equipment we are
sponsoring programs such as EPRI TC "Life Assessment Code For Rotor Rim
Cracking", the MPR study on rim stresses and machining, EPRI Turbine Outage
Extension project, and other funded and co-funded projects. The common interest in all
these efforts is to push for the development of tools, inspections," and remedial actions that
would effectively mitigate SCC as a known root cause for major components, or allow us
to evaluate see potential and damage in such a manner to fully understand the risk and
potential cost. Economic models exist that we can use to evaluate options and detennine
the best solution for us. The success of these programs is key to our future. Simply
replacement of damaged component in kind, without significantly improving the SCC
problems is viewed as unsuccessful.
226
TomAlley
References
1. Kalderon. D., "Steam Turbine Failure at Hinkley Point' A', Proc. Inst. Mechanical
Engineering., London, UK. 186 ( 1972), 341-77
2. General Electric Company Technical Information Letter (TIT...) # 857-3. "Inspection
of 1800 RPM Nuclear Turbine Rotors", March 31, 1978.
3. General Electric Company Technical Information Letter ( TIT..) # 1024-3. " In-Service
Inspection of Built-up Fossil Rotors", June 1988.
4. Westinghouse Operation & Maintenance Memo (OMM) # 111 "NucIear Disc
Detenninistic Inspection Intervals" June 1990
5. Westinghouse Electric Corporation Notes on the Presentation by Westinghouse
Electric Corporation to Electric Utility Executives" Charlotte, N. C. October 30,
1979
6. S. Nuclear Regulatory Commission "Safety Evaluation Report related to the Operation
of Hope Creek Generating Station" NUREG - 1048 Supplement #6, Docket # 50-
354. Public SelVice Electric and Gas Company, Atlantic City Electric Company. July
1986
7. S. Nuclear Regulatory Commission "Protection Against Low-Trajectory Turbine
Missiles" Regulatory Guide 1.115 March 1976.
8. Bush, S. H. "Probability of Damage To Nuclear Components Due to Turbine Failure"
Nuclear Safety, Vol. 14, #3, May-June 1973
9. EPRI "Life Assessment Code for Rotor Rim Cracking" TC Project #4597-001-
10759, Jan. 1997
10. EPRI, Dec. 1996 minutes from Utility Advisory Meeting
11. EPRI TR-100456 "Influences of Stress Corrosion Crack Merging on Remaining Life
Predictions" Prepared by University of Newcastle Upon Tyne, Newcastle England
April 1992
12. General Electric Company Technical Infannation Letter (TIL) # 1024-3. "In-Service
Inspection of Built-up Fossil R o t o r s ~ December 31, 1988.
13. EPRI "EPRI Fossil Steam Turbine Disc Cracking Workshop", proceedings Oct. lO-
11, 1990. Charlotte, N. C.
14. EPRI " Guidelines for Predicting the Life of Steam Turbine Disks Exhibiting Stress
Corrosion Cracking. Vol. 1 EPRI NP-6444 Final Report July 1989
15. Zawoysky, R. J. , others" Generator Rotor Retaining Rings: An Updated GE
Perspective", G. E. Power Generation Publication GER-3549A Aug. 1992.
16. Kilpatrick, N. L., Schneider. M., "Update on Experience with In-Service Examination
of Nonmagnetic Rings on Generator Rotors", Technical paper presented at EPRI
Fossil Plant Inspections Workshop. San Antonio, Texas. September. 1986.
17. Ctfudeli, R.,Pisano, A., others" Case History of a 18Cr18Mn Retaining Crack
Affected by Stress Corrosion Cracks After 33500 Operating Hours" Technical Paper
to be presented.
18. Alley, C. T. " The Weld Repair and Failure of General Electric L-O Turbine Buckets"
Technical Paper Presented at EEl Prime Movers Committee Meeting, Charleston S. C.
January 29-31,1990.
227
228
Arresting Stress Corrosion Cracks in Steam Turbine Rotors
Presented By
Russell Chetwynd
Southern California Edison
229
230
ARRESTING STRESS CORROSION CRACKS IN
STEAM TURBINE ROTORS
Russell J. Chetwynd
Riyad Qashu
Nabil EI Akily
EDISON INTERNATIONAL
SOUTHERN CALIFORNIA EDISON
5000 Pacific Coast Highway
San Clemente, California 92672
Ralph J. Ortolano
Turbine RESCUE
INTRODUCTION
In 1994 SeE became concerned that the
San Onofre Nuclear Generating Station (SONGS)
low pressure (LP) turbines may be suffering from
sec at the blade attachment areas and steam
balance holes. This concern was initiated by the
original equipment manufacturer (OEM)
recommendations and reinforced by actual findings
in similar equipment in Korea and Michigan.
The SONGS Units 2 and 3 original LP
turbine rotors are double flow, 8 stage, 6 disc,
shrunk-on, impulse construction with button drives
or dowels as disc anti-rotation devices. The
material of construction is 3.5%NiCrMoV steel and
operating speed is 1800 rpm. The first 6 stages
are straddle root type attachment with the closing
blades pinned through the wheel, and a riveted
shroud. The last two stages have an axial entry
blade attachment with continuous tie lacing
arrangements. The units were built in the late
1970's and put into commercial operation in 1983
& 1984 and to date have each operated close to
90,000 hours.
An inspection plan was put together for the
Unit 2 cycle 8 refueling outage with contingencies
for blade removal, crack excavation, skim cutting,
shot peening and blade replacement. The original
plan called for removal of the Low Pressure rotor
#2 (LP2) and #3 (LP3) rotors, Ultrasonic Testing
(UT) of the stage 1-6 disc rims, Magnetic Particle
ABSTRACT
Stress corrosion cracking (SCC) in steam
turbine rotors has been occurring for many years,
largely restricted to shrunk-on disc LP rotors.
Cracking has been found in many of the high
stressed' locations, and feedback from nuclear
turbine owners indicates that some designs are
more prone to SCC distress than others. Many of
the see discoveries have ultimately led to
purchasing replacement rotors, some of which may
still be susceptible to the problem. In Southern
California Edison's two 1127 MW nuclear units,
employing three LP rotors each, sec was
discovered after less than fifteen years of service,
in the dovetails of the discs, entry slots and steam
balance holes in several stages. Cracking prone
dovetail fillets were skim cut and polished,
extensive dress-out of defects was then effected
and steam balance holes were enlarged in many
areas. This was followed by shot peening. In the
second unit examined, titanium blades were
selectively used to reduce centrifugal stresses in
dressed out locations. Since the blades were
carefully removed to gain access to the disc rim, it
was possible to reassemble these parts with
minimal use of new parts. The rotors were
reassembled in the unit and operated for more
than 18 months at fuJI load. During the subsequent
Unit 2 refueling outage, the rows containing the
deepest dress-outs were de-bladed and re-
examined. This paper discusses the findings
related to the success of this work, as well as a
brief summary of the analysis work performed to
gain confidence in the approach used.
NOMENCLATURE
a =
K
1
=
K
1C
crack size
stress intensity factor
critical stress intensity
231
a) Straddle Root Attachment Repairs
Table - SONGS 2 - Stage 4 Disc Rim
Excavations
SCE's Schenk HTII slow speed rotor
balancing machine was converted into a lathe to
perform the skim cutting of the dovetail fillets. The
machining was performed by SeE personnel with
machine tooling and supervision provided by the
OEM. A nominal .010" was removed with a
tolerance of +/- ,005" required to achieve a full
circumferential cut. .
The next step involved polishing of the
surface and MT NDE to identify all of the defects,
followed by mapping and numerical identification of
each indication on a chart. This actual MT work
was performed by a local contractor with
supervision and documentation provided by
SONGS in house Quality Control personnel.
Each indication was removed by local
machining in both directions along the crack length
and depth. This was necessary because, although
the cracks were initiated at the fillet root surfaces,
they propagated circumferentially around the disk
and across the thickness. Edison personnel
performed this work under strict maximum depth
guidelines provided by engineering. The
machinists worked in teams with an inspector who
performed MT examinations between subsequent
machining operations until each and every
indication was cleared. Approximately 300
excavations were performed, the majority located
on the bottom serration and almost uniformly
distributed among the three LP turbines (Table 1).
The dimensions of each excavation were
measured and recorded on a computer generated
drawing in support of the engineering evaluation.
In some of the extreme cases plastic molds were
made to improve the accuracy of the data
collected.
Russell Chetwynd
Testing (M1) of the stage 7 & 8 slots and MT of the
steam balance holes. Plans had previously been
made to remove the stage 7 &8 blades for shot
peening in order to enhance their fatigue life.
Ultimately all three rotors were removed and 14
rows of blades disassembled for inspection and
repair. SCC was detected by UT on all of the stage
4 disc rims and by MT on all stage 7 rotor disc rim
blade slots. All stages 1, 2, 3, 5 & 6 disc rims were
clear of defects by UT performed manually by the
OEM's inspectors. Similarly, all stage 8 rotor disc
rim blade slots were clear of defects by MT
performed by a local NDE contractor.
All stage 4 blades were disassembled by
first carefully cutting through the shroud bands in
order to save the blades for reuse as supplements
to spares on hand. Disassembly was performed
and supervised by SCE personnel and the blade
refurbishment, or tipping for reuse, performed at
SCE's Mechanical Services Shop. The bare disc
rims were then glass bead cleaned by a local shot
peening contractor prior to MT to confirm the
presence of defects detected by the UT. At this
point defects were also detected on the stage 4
closing blade pin attachment holes and steam
balance holes on stages 4 and 5.
An "Engineering Team" was convened and
a repair plan was formulated with the "game plan"
to remove all indications and return the turbine to
unrestricted operation for the next fuel cycle.
SCE's nuclear engineers were tasked with
performing an engineering evaluation in support of
this goal and received exceptional cooperation
from our OEM. This evaluation included several
finite element models (FEM), stress analyses, and
fracture mechanics assessments to establish the
technical basis for the repairs and justification for
run time with the repaired discs. Further, Edison's
Nuclear Oversight Department was assigned to
perform a root cause assessment in order to better
understand the nature of the problem and help
evaluate measures to prevent reoccurrence.
REPAIR METHODOLOGY
The SONGS maintenance staff prepared
contingency plans prior to the outage to skim cut
the dovetail fillets, excavate indications and shot
peen on all stages as required. These repair plans
were then expanded upon, depending on the
results of the inspection, to include major rework of
three areas, the straddle root attachment, the
closing blade pin holes and steam balance holes.
Less extensive repairs were required for the stage
5 balance holes and the stage 7 rotor slots.
232
Rotor
LP1
LP2
LP3
# Of Excavations/Max. Depth
Top Middle Bottom Total
7 1 94 103
(.034") (.055") (.437")
1 2 79 82
(.062") (.098") (.313")
32 7 7 118
(.562") (.160") (.560")
Russell Chetwynd
Figure 1 Unit 2 Lockup Blade Region Cracks,
LP34F
The last phase of the disc repair work prior
to reblading was shot peening to produce a
beneficial surface compressive stress at the
highest stressed locations. The shot peening
procedures were prepared by the local contractor
working closely with the OEM and SCE's
engineering and maintenance staff. An automatic
process was developed for the fillets themselves
and a manual process was used for the
excavations. For the SONGS discs, the surface
compressive stress after shot peening are
estimated at 80-90 ksi with a penetration of .004 -
.013".
The general repair methodology for the
less extensive closing blade pin hole cracking was
enlargement by machining followed by shot
peening. The pin hole enlargement resulted in
unloading the holes. The LP3 turbine stage 4 front
disc pin hole cracking was left as is with the
exception of the crack extending from the bottom
pin hole towards the disc rim. This crack was
excavated consistent with the diameter of the pin
hole and shot peened. The remainder of the pin
holes were enlarged and shot peened. The cracks
through the ligament between the pin holes were
left without change, since they were deemed as
harmless and could not propagate.
A test coupon was removed from the rim at
the point where the uppermost pin hole crack had
propagated. This was used for metallurgical
analysis in support of the root cause investigation.
The stage 4 blades were reassembled with
a clearance between the pin holes and the pin to
eliminate high local contact stress. Crosskeys
installed between the adjacent blade platforms,
above the wheel rim, were used to assist in support
of the closing blade. Note: As discussed later,
similar crosskeys were installed between blade
platforms at areas of the disc rim with extensive
dovetail excavations.
c) Steam Balance Hole Repairs
Steam balance hole cracking was
discovered on both stages 4 and 5, with stage 4
being the most severe. The cracking was difficult
to detect by MT NDE technique without first
reaming and polishing the holes in order to achieve
a smooth surface finish and UT inspection
techniques was also used in parallel to help
quantify the extent of each defect discovered. The
cracks were found to originate at the surface and
were oriented radially at the 6 and 12 o'clock
positions with respect to the disc center line. They
were found to initiate both at the corner locations
and in some instance at the mid section of the disc.
In all cases they propagated towards the disc bore
and the disc rim. (Figure 2)
The cracking was repaired by a
combination of grinding and machining with the
machinists and inspectors working in a team in a
similar manner to the dovetail repairs. Typically,
cracks were first removed by local milling until
cleared by MT inspection, with maximum
excavation depths provided by Engineering as an
upper bound value. The holes were then
machined to a larger diameter, with an offset
center jf necessary, to remove the ground out
areas. In some holes, where the cracking had not
extended through the thickness of the disc. the
y
o
r
0
Outtee Side
OiJeRim
o
Inlet Si,Je
b) Closing Blade Pin Holes Repairs
The cracking in this area was surprising to
SCE as it had not been reported by others
operating similar turbine designs. The cracking
however, was consistent with cracking reported on
nuclear machines with pin root attachments and
similar rotor construction technology. It was not
discovered until the LP3 turbine front stage 4 disc,
closing blades were removed for investigation of
the dovetail cracking detected by UT NDE. This
disc had the most extensive, through wall, cracking
(Figure 1) which was radial and transverse to the
ligaments between the three pin holes and also
through to the disc rim. The most alarming crack
however, extended from one of the bottom pin
holes towards the disc bore.
1
233
Russell Chetwynd
o
/
/
Balance
Hcle
Rim
Figure 2 Typical Steam Balance Hole Cracks
cracks were partially removed by boring a larger
hole and the remainder 0.1" deep) removed by
grinding and blending. Following the boring
operations a generous radius was machined on the
edge of the holes. Finally, the holes were shot
peened with a manual procedure developed jointly.
by the OEM, the shot peening contractor
and SeE.
d) Stage 7 (L-1) Rotor Slot Repairs
Minor cracking was discovered on the
majority of the stage 7 axial entry rotor disc slot,
dovetail radii by MT NDE following glass bead
cleaning. This cracking was estimated as being
less than .010" deep and was removed by hand
machining which turned out to be a painfully slow
task, however, attempts to develop an automated
process were unsuccessful. Following confirmatory
MT the slots were shot peened using an automated
procedure, again developed jointly by the OEM, the
shot peening contractor and SeE.
e) Stage 4 Blade Reassembly
Several further enhancements were
employed in stage 4 blade reassembly to help
mitigate against a resumption of the cracking:
o Cross keys were installed to join adjacent
blades where the disc rim dovetail
excavations had resulted in significant
reduction in the disc rim sections remaining
to support blade loads. They were used
whenever excavation depths exceeded
0.18", on any of the serration fillet radii, so
234
that adjacent blade loads could be shared
by distributing them over a larger section of
the disc rim.
o The notch or entry gate region was
modified by machining corners at
intersecting surfaces to reduce stress
concentrations.
o Lockup area, dowel pin holes were
machined to elongate them in the radial
direction by.060",
o Caulking grooves were introduced at the
base of the two blades on either side of
the closing blade to ensure a tight build.
f) Second Unit Experience
The refueling outage on SONGS Unit 3
followed a few months after the return of Unit 2 to
operation. The time between the outages was
used to procure a supply of stage 4 titanium blades
and replenish the stage 4 stainless steel blade
spares with a mixture of new and refurbished
components. The competitive bid process was
used and two different after-market vendors were
selected. As a further contingency, pressure plate
material was also purchased for stage 4, which
was then machined at SeE's Westminster repair
facility to the OEM's design.
For schedule considerations, the Unit 3
outage plan was changed based on the Unit 2
experience. in the following manner:
o All 3 rotors were immediately removed
a All stage 4 blades were removed for MT
inspection of the dovetails without prior UT
inspection
a Water blast cleaning of the rotor discs was
used instead of glass bead cleaning.
Inspection results differed on Unit 3 in the
following way:
a Stage 4 disc dovetail cracking was much
less severe and detected on only 1 of the
6 disc rims
o Dovetail cracking was detected by UT on
2 of the 6 stage 5 disc rims
o Dovetail cracking was detected by UT on
3 of the 6 stage 6 disc rims
a Steam balance hole cracking was much
less severe, confined to the LP2 and LP3
turbines and more extensive on stage 5
than stage 4.
o Closing blade pin hole cracking was much
less severe, less than 0.25" in depth and
evident on stages 4,5 and 6.
Repair procedures differed on Unit 3 in the
following way:
o Titanium blades were used on stage 4 at
the lock up area, sections of the rim with
excessive dovetail excavations and
opposite for balance.
a Air cooling was used during crack
excavation in an effort to minimize
propagation of the crack during the
grinding process.
a Steam balance holes were over-bored
without prior excavation of cracks using UT
crack depth estimations as a guide.
ENGfNEERlNG EVALUATION
Analysis covered two areas:
1) Structural analysis of the discs to
assess the impact of the steam balance hole
repairs,
2) Evaluation of the straddle root disc head
for both ductile and brittle failure modes. Finite
element-based analyses were used to perform
both the structural and the fracture mechanics
parts of the LP turbine evaluation. A description of
the models used, and some sample results are
given below.
1. Structural Analysis of the Discs
All steam balance hole cracks were
removed by machining the impacted holes to a
larger diameter, combined with local grinding and
blending. To assess the effect of these repairs on
the structural integrity of the disc under operating
loads, three-dimensional finite element analysis
techniques were used to perform the disc
evaluation.
Russell Chetwynd
Figure 3 shows a closeup of the finite
element model, representing a segment of the disc
containing one steam balance hole, used to
evaluate stage 5 disc repairs. The figure shows
the steam balance holes, the disc head and part of
the disc hub.
The following loads applied to the model:
a) Rotating disc loads.
b) Blade loads due to rotation.
c) Load due to the disc hub-shaft shrink fit.
This was mOdeled by applying thermal
expansion to the shaft.
hoop stress (ksii
70.--------------------,
60
50
40
so
.20
0'-----'---.........__1--_......J.__.....l-__:....-_-.I
16.6 2t5 26.5 3tS 36.5 41.5 46.5 51.5
r (inch)
Figure 4 Hoop Stress Distribution
Steady state structural analysis was
performed to calculate the distribution of the stress
components in the steam balance hole region.
Analysis results show that the dominant stress'
component acts in the tangential (hoop) direction of
the disc, as expected. Figure 4 shows the
distribution of the hoop stress in the disc as a
function of the radial distance. The stress
concentration effect afthe steam balance hole can
be clearly seen. The stress increase factor, due to
the steam balance hole, is about 2.5. This explains
why all steam balance hole cracks were in the
radial direction.
235
Russell Chetwynd
Figure 5 shows the hoop stress
distribution across the thickness of the disc along
the edge of the steam balance hole. The variation
is less than 1DOlo, with the stresses slightly higher in
the middle of the disc.
Bounding steam balance holes size,
enveloping the largest crack found, was considered
in the analysis. Results showed that increasing the
hole size, to remove cracks, does not have
significant impact on the maximum hoop stresses
in the disc. Results also showed that a circular
hole has a smaller stress concentration effect than
an elongated hole.
stress (ksl)
10 .---------------------,
60
50
40
30
20
10
~
~
configurations were modeled including top and
bottom hook horizontal and inclined cracks were
modeled, and the stresses in the disc head
shoulders were calculated to determine the limiting
crack (excavation) depth, and the available margins
for the repaired discs.
Figure 6 Stage 5 Disc Head
2
ol...-_..l..-_.....i...._-..l._----l__-"--_--'-_---l._--'
-2 -1.5 -1 -0.5 0 0.5 1 . 1.5
distance from disc center, z (inch)
- hoop stress -+- stress Intensity
Figure 5 Hoop Stress Versus Disc Thickness
2. Disc Head Evaluation
Disc head cracks were excavated, the
excavated surfaces were shot peened to
enhance resistance to crack initiation. To assess
the impact of the excavation on the structural
integrity of the discs, two analysis were
performed:
a) Structural Evaluation
A finite element-based structural
analysis of all the discs impacted by the
excavation repairs. Figure 6 shows a typical
axisymmetric model generated stage 5 of the LP
turbine. Centrifugal loads, blade loads and
loads due to the disc-shaft shrink fit were
applied. Excavations were conservatively
modeled as cracks emanating from the hook
fillets into the disc head material. Different crack
236
b) Fracture Mechanics Evaluation
Fracture mechanics evaluations were performed to
assess the potential for brittle fracture. Cracks were
postulated at the locations of the disc head excavations.
Finite eleme"nt models of the impacted disc heads were
generated, as shown in Figure 7. The figure shows the
applied loads and the boundary conditions. The finite
element-based fracture mechanics program FRANCXT was
used to calculate the stress intensity factor, K
I
, as function of
the crack depth. Figure 8 shows a sample of the results
obtained for stage 5 disc heads.
Results show that ductile failure is limiting, i.e., the
disc head shoulder will fail plastically before K
1
reached the
critical stress intensity factor K,c.
Root Cause Determination
Fracture analysis and metallurgical evaluation of the
test coupon removed from Unit 2 LP3 turbine, 4th stage disc
revealed stress corrosion cracking (SCC) as the failure
mechanism. Cracks in the steam balance holes, blade
lockup area and dovetail fillets were also concluded to be
sec based upon the branching appearance.
Russell Chetwynd
STRATEGIC PLAN
A team was put together following the SONGS 3
cycle 2 refueling outage with the objective of exploring the
repair options available in order to restore the remaining life
of the LP turbines to 20 years, in a timely and economical
manner.
This study identified four options (A, 8, C & 0) which
included: (A) inspecting and repairing, (8) refurbishment
using two universal rotors as maintenance tools, (C)
purchasing three new rotors of an upgraded design, or (Q)
purchasing six new rotors. Of these options, Option 0
provided the least risk that cracking would not resume.
Option C addressed the cracking problem but assumed some
risk as three of the rotors would be old. refurbished rotors.
Option Bwas not considered practical as it was maintenance
intensive and would take too long to fUlly implement. Finally,
Option A assumed a great deal of risk as none of the rotors
would be replaced and all of the repair work was planned to
occur during refueling outages. It was felt that this option
could not guarantee that cracking would not eventually
resume which would have significant economic impact on the
plant if worse than predicted.
All of the above options were evaluated using a
present worth cost evaluation. The major economic inputs
included equipment costs, MWe gains or losses and outage
length impact. The IV1We gained for new equipment was
based upon average values provided by the vendors and
became the reason why the purchase of replacement
equipment is economically viable. The results of the present
worth cost evaluation indicated that both Options C and 0
offered a strategic solution for addressing the LP turbine
cracking issue.
Consequently, in July of 1996, SCE solicited bids for
replacement LP turbines from five suppliers with the optian to
supply three or six elements. The OEM, GEe Alsthom, was
the successful bidder with SCE accepting their bid for the
supply and installation of six replacement LP turbines.
0.6 0.5 0.2 0.3 0.4
crack length (inch)
0.1
Figure 7 Disc Head Model Used in the Analysis
V
/
/
/

/
o
o
60
20
80
40
Kl ksi sqrt(lnch)
100
Figure 8 K
1
versus a
Corrosion analysis of the debris samples
removed from crack locations on the discs of
both units revealed the presence of traces of
chlorides, molybdenum and sulfides. Disc
material evaluation revealed that the most
severe cracking occurred where material
ultimate strength exceeded 140 ksi, indicating a
correlation between disc mechanical properties
and crack intensity. The above analysis
supported SCE's determination of the primary
root cause of faHure as a combination of
susceptible material (3-S.S%NiCrMoV) and the
presence of corrodents in wet steam
environments under relatively high stresses.
SUBSEQUENT INSPECTIONS
SONGS Unit 2 returned to service from refueling on
May 23 of 1995, for fuel cycle number 8 and operated base
loaded, with only two very brief outages, until December 1996
when it was again shut down for refueling. The engineering
evaluation specified follow up inspections for verification of
the cycle 8 stage 4 repairs.
Inspection Scope
The cycle 9 refueling outage HP double flow and LP
double flow turbine rotor disc inspection scope was as
follows:
o HP stages 1-7 dovetail UT in-situ inspection.
a HP stages 6 and 7 dovetail MT inspection following
blade removal.
o LP1 stage 4 rear blade removal, doveta!l
closing blade pin hole MT inspection.
237
Russell Chetwynd
0 LP3 stage 4 front blade removal.
dovetail and closing blade pin hole MT
inspection.
0 LP1 stages 1-6 dovetail UT in-situ
inspection
inspection
inspection (contingency)
0 LP1 stages 2-5 steam balance hole
MT/UT Inspection
MT/UT inspection (contingency)
MTJUT inspection
Inspection Results
a No defects on HP stages 1-7 dovetails
from UT inspection.
o No defects on HP stages 6 and 7
dovetails from MT inspection.
a One minor defect on LP1 stage 4 rear
dovetail, from MT inspection
o Two minor defects on LP3 stage 5 rear
steam balance holes #s 5 &6, from MT
inspection
o Significant defects on 5 of the 6 stage 6
dovetails on all three LP turbines, from
UT inspection
o No defects on LP3 stage 4 front
dovetails, from MT inspection
o No defects on remaining four stage 4
wheel dovetails from UT inspection
o No defects on any stage 5 wheel
dovetails from UT inspection
o No defects on any stages 1, 2 or 3
wheel dovetails from UT inspection
Repairs Required
o HP stages 6 and 7 dovetail fillets were
skimmed shot peened (proactive)
a The minor defect on the LP1 rotor stage
4 dovetail was repaired by excavation -of
no significant depth. A full row of new
titanium blades was then assembled on
this wheel, with selective location
crosskeys installed, as additional
enhancement.
o The minor defects on the LP3 rotor
stage 5 balance holes were repaired by
excavation of no significant depth.
o The defected stage 6 wheels on LP1
rotor front LP3 rotor front and rear were
enhanced by the selective installation of
pins.
238
a The defective stage 6 wheels on LP1 rotor rear and
LP2 rotor rear were accepted as is for one more fuel
cycle of operation.
CONCLUSIONS
In conclusion, the work described in this paper has
produced significant benefits to the owners of the turbines
and their customers. The repairs performed have allowed
unrestricted unit operation without the expense of forced
outages or reduction in output due to the use of pressure
plates. lhe December 1996 (cycle 9) refueling outage
inspections on SONGS Unit 2 have provided confirmation
that, at least in the short term, sec does not resume in
excavated and shot peened locations. It is noted that no
cracking was found on any of the deepest excavations, even
though significant loss of strength in the affected section had
occurred.
Additionally, the success of the LP turbine repairs
confirmed the benefit of shot peening for application on the
HP turbine stages 6 and 7 dovetail fillets. SeE considers that
SCC has been eliminated as a concern for the remaining life
of this rotor by the skim cutting, polishing and shot peening
performed during the cycle 9 refueling outage.
Finally, it can be concluded that sec can be
effectively managed, in the short term. by the type of repair
described in this paper. However, in the long term the
expense of continued maintenance and inspection dictates
that replacement of the equipment with more modern and
efficient apparatus is the most cost effective solution.
REFERENCES
1. ANSYS User's Manual
2. FRANCXT User's Manual
ACKNOWLEDGMENTS
The work described in this paper was only achieved due to
the substantial efforts of a large team of talented people, with
special consideration to the foHowing personnel. From SCE:
Mehrdad Hojati, Tom Yackle, Dan Breig, John Hirsch,
Ransey Clark, Mostafa Mostafa, Mike Lister, Jim Rick, Mike
Love, Ken Ball and Pete Watkin. From GEC Alsthom: Alan
Holmes, Terry Cain, Bruce Roberts and John Bolton.
From Nakkashian Enterprises: Joe Nakkashian. From Metal
Improvement Co.: Bob Kleppe and Gary McFarland. From
Mobile Inspection Services: Cam Marsden. From APTECH
!=ngineering Services: Russ Cipolla and from EPRI Tom
McCloskey and Ramaswamy Viswanathan.
LP Turbine Dovetail see Management at BGE
Presented By
Dr. Mark Richter
Baltimore Gas & Electric
239
240
LP TURBINE DOVETAIL sec MANAGEMENT AT BGE
M. A. Richter, R. B. O'Hara, P. A. Klein, and P. R. Nagle
Baltimore Gas and Electric Company
1000 Brandon Shores Road
Baltimore, Maryland 21226
ABSTRACT
When the problem of stress corrosion cracking in LP turbine disk blade attachments emerged within
our utility, Baltimore Gas and Electric initially relied solely on the OEM to support the examination
and disposition of susceptible wheels. However, the desire to retain final control and disposition of
examination results, led BGE in 1992, to start developing a strategy for managing the issue in-house.
The approach BGE decided to use includes -applying in-house NOE and Materials Engineering review
together with independent condition assessment. The Calvert Cliffs Unit No.1 LP turbine
inspection process in 1996 is discussed. The unit, which contains blade attachment cracks, has been
returned to service without restriction based, in part, on BGE's assessment of the examination results.
241
242
LP Turbine Dovetail see Management at BGE
Introduction
Calvert Cliffs Unit 1 is a General Electric Code N2R turbine-generator rated at 835Mw's net, that was
initially operated in 1974 and commenced commercial operation in 1975. The unit consists of 1 high
pressure turbine and three double flow low pressure turbines in a tandem compound arrangement. The
three double flow low pressure rotors are the 'built up' or disk type rotors that consist of 14 wheels (or
disks) shrunk onto a shaft. Each wheel contains a row of buckets at its perimeter retained by a dovetail
attachment and each wheel is keyed at wheelbore to the shaft. The rotor wheels are manufactured from
NiCrMoV steel, having specifications similar to ASTM A471. Intergranular Stress Corrosion
Cracking (IGSeC) of these wheel keyways has been recognized in the industry since the early 9 ~ O s
and see of the wheel dovetails has been recognized since the early 1990's.
Non-measurable wheelbore keyway cracks were first noted during the 1980 GE UT examination on LP
"C", and measurable cracking was first noted on LP "A" and "B" during the 1992 GE UT
examination. Indications of possible dovetail cracks were detected during the first dovetail UT
examination in 1992 On LP "A" and LP "B". The cracking was confirmed by magnetic particle
examination in the notch area on both L-3 wheels of both rotors. However, the cracking identified by
MT did not correlate well with the UT results. Although only the notch group buckets were removed,
there were no MT indications extending under the adjacent bucket groups. LP "c" was examined by
BGE NDE staff in 1994. Although dovetail cracking was expected, none was detected ultrasonically.
This was confirmed by subsequent MT and PT exams of the notch group area.
In preparation for the 1996 RFO, numerous discussions were held with GE as to the limits of their
ability to detect and size dovetail indications and to provide an engineering approach on how they
243
,
Mark Richter
arrive at their recommendations concerning dovetail cracks. Although it had been noted that GE had
been fine-tuning their procedures as they gained more experience, GE maintained that all dovetail
cracks should be removed due to the uncertainties surrounding crack propagation rates and dovetail
failure modes. It was determined that as a contingency plan, BGE should utilize its existing
Data Recording and frocessing System (UDRPS) equipment for UT inspection and develop unit
specific finite element analysis (FEA) for use in evaluating dovetail cracks. Stress Technology
Incorporated (ST!) was chosen to perform the FEA and fracture mechanics models.
This contingency plan was developed and instituted in an effort to not rely solely on the OEM for data
acquisition and analysis. The OEM is motivated to pursue the most. conservative approach, from an
equipment integrity and reliability standpoint. In doing so, the OEM weighs all the data, applies
conservative boundaries to the data interpretation, and attempts to eliminate all risk. The OEM will
answer all of the customer questions, to the extent that the OEM feels is necessary and does not
transmit proprietary information. Inherent in this process, is a feeling of powerlessness on the part of
the customer. The customer is unable to perform a sound engineering analysis due to the lack of data
that the OEM describes as proprietary.
This tenable position was determined to be unsuitable for the purpose of developing short term and
long term plans for the repair or the replacement of the LP rotors. It was determined that a greater
understanding of the metallurgical issues and failure mechanisms was necessary for the owner to
develop sound engineering and fiscal decisions.
244
Mark Richter
Materials Considerations
Requirements for sec are often presented in a Venn Diagram showing tensile stress, and
environment. For an alloy, the conjoint action of tensile stress and the appropriate environment will
produce sec. The cracking process can be divided into crack initiation and crack propagation. In a
given there is a threshold stress below which sec will not Time to see crack
initiation is a strong function of stress. dovetail cracking is expected to initiate earliest in
the areas of highest stress. These high stress areas may be a result of design (e.g. notch group of
buckets), variations in bucket fit, or a result of service degradation (e.g. pitting). We have observed
dovetail region pitting in both our nuclear unit that has experienced sec and in a younger fossil unit
that has no cracking. In our the pitting provides the additional stress concentration necessary
to exceed the threshold stress to initiate sec. Figure No.1 shows dovetail cracks on the L-2 stage of
the ealvert Cliffs LP "A" rotor during the 1996 refueling outage. For LP rotor dovetail cracking, the
stages at the Wilson Line appear to be the most susceptible to cracking probably due to the
concentration of contaminants in the dovetail crevices. Since blade attachment NDE was not
performed early in the LP turbine's service life, time to crack initiation is unknown.
Successful sec crack mitigation is achieved through modification of one or more key parameters:
material, environment and stress. Examples could include a change in alloy tensile stress reduction,
or service environment modification. For the LP turbine disc owner experiencing there are no
.I
low cost modifications available to control cracking. Utilities can not readily change the disc material.
The service environment is likewise difficult to change unless coatings can be developed for the
dovetail region. Tensile stress reduction through shot peening andlor dovetail geometry redesign is the
remaining available field modification. if pitting or cracking has extended beyond the
maximum depth for a skim an expensive long shank bucket modification becomes the least
expensive option for crack mitigation.
245
Mark Richter
Once sec has initiated, crack propagation is assumed to initially occur at a constant rate. sec crack
growth rates are uncertain due to unknown initiation time and lack of NDE re-inspection data. Lab
data and industry crack depth observations indicate propagation rate increases with increasing
temperature. The crack propagation process is dependent upon the number and depth of cracks on the
adjacent hooks in a given location. If cracking were to initiate concurrently on all hooks and propagate
at the rates suggested in the literature, we would expect there to be a significant number of catastrophic
disc rim failures. This is not consistent with industry experience; no catastrophic failures within the
nuclear industry and few failures at fossil plants. Clearly the multiple hook design provides some
redundancy to catastrophic failure.. Isolated cracks that initiate and grow by sec reach a depth at
which load redistribution occurs among the remaining hooks and the crack arrests.
Having previously ground cracks in our LPA and LPB rotors in 1992 to a 7/16 inch maximum depth,
we had no attractive sec countermeasure options heading into the 1996 refueling outage. We
expected NDE to detect additional cracks during the outage. Our short term objective was to provide
an engineering basis for flaw acceptance to support one cycle operation. This would minimize the
probability of either a costly unplanned modification on the outage critical path or loss of megawatts by
pressure plate installation. We selected STI to perform analysis with the following
Profile the dovetail stress magnitude and distribution for the original design and with cracks.
Perform sec crack propagation studies.
Based on crack extent, assess the margin against failure from overload or other fracture modes.
Evaluate if titanium notch group spacers would extend life.
246
,
Mark Richter
Crack growth analysis evaluated three cases: isolated short cracks, isolated deep cracks, and
simultaneous cracks in all hooks. The analysis considered several scenarios capable of producing a
catastrophic event, including cracks growing to either a critical size to produce an overload failure or
to a size at which rapid crack growth occurred from high cycle fatigue.
Analysis of Results
The following summarizes key findings from the analysis:
Maximum equivalent stress occurs in the notch group.
The limiting condition for flaw acceptance resulted from the high cycle fatigue consideration.
Isolated cracks were not of concern since load redistribution would limit crack growth.
Widespread cracks of greater than 0.30 inches depth may result in resonance and high cycle fatigue
damage.
The equivalent stress with notch spacers (buckets removed due to SCC) was considerably less than
the maximum stress elsewhere on the disk. There was no significant benefit associated with the
use of titanium notch group spacers.
Analysis results were used to set an acceptance criteria for flaws during the 1996 refueling outage.
The acceptance criteria for flaws considered the widespread crack depths that could produce high cycle
fatigue damage and crack propagation during the next operating cycle. ~ o a materials engineering
perspective, the NnE needed to be capable of detecting widespread cracking that would challenge disc
integrity during the next operating cycle.
247
Mark Richter
NDE Background
The nondestructive examination of LP turbine blade attachments has been performed routinely on
nuclear LP wheels for about 15 years, primarily because of their susceptibility to stress corrosion
cracking (SCC). More recently, see has been found in the blade attachment areas of some fossil LP
wheels. This development has provided the motivation to also inspect the blade attachments on some
fossil LP wheels.
BGE has been actively performing NDE on our LP turbine blade attachments since 1993. Prior to
1993, BGE utilized OEM or "3rd party" vendors to perform this NOE. BGE made the decision in late
1992 to develop internal capability for ultrasonic examination of blade attachments and keyways,
complementing existing capability in rotor boresonics.
The motivation to pursue this objective is based on a number of things. During the Spring 1992
outage at Calvert Cliffs, the. OEM failed to ultrasonically detect significant cracks. The subsequent
interactions with the OEM brought to light the difficulty that a utility customer has in retaining control
of both OEM NnE results and component disposition recommendations. By developing internal
capability in this area we maintain ownership of the component disposition, and utiliZe our existing
automated ultrasonic system and Materials Engineering Unit staff. This approach has allowed BGE to
realize significant cost savings compared to any other outside services.
The BGE ultrasonic examination system consists of its automated ultrasonic system (UDRPS 2), an
automated scanner and specially designed end effectors to manipulate the transducer. The UDRPS 2,
manufactured by WesDyne International, Inc., is a commercially available ultrasonic data recording
- and processing system that provides digital data acquisition, signal processing, display, and storage
capabilities. The system has multi-ehannel capabilities, but, for this application, was configured in a
single channel mode. The system will accept an analog signal from virtually any ultrasonic instrument
248
,
Mark Richter
as long as it is biased not to exceed 0.5 volts in amplitude. The UDRPS 2 "front end" performs the
real time signal processing operations including digitization, peak detect and hold, re-eirculations and
filtering. The processed data is stored, manipulated and displayed on a Hewlett-Packard workstation.
Long-term archival storage of data is provided by 5 1/4 inch laser optical disks. Stored data may be
recalled and displayed in a variety of formats to facilitate analysis.
The scanner initially utilized by BGE and shown in Figure No.2, was developed under EPRI contract
by Tecnatom, s.a. That scanner featured two identical but independently operable arms that can be
positioned on either side of a disk or group of disks to conduct simultaneous but independent
inspections of two regions operating in a pulse-echo mode. Alternatively, the arms can be used to
precisely position and coordinate the motions of two transducers operating jointly in a pitch-catch
mode from opposite sides of a disk. A series of manually controlled motions (including movement of
the entire scanner along the rotor, separation of the two arms, rotation of the arms into radial
alignment with the shaft center, and independent extension of each arm to engage the rotor) provide the
preliminary positioning and setup of the arms. Once they have been positioned using these manual
motions, each transducer can be independently indexed via three automated, computer-controlled
motions: radial position along the scan arm, axial offset from the arm to engage the disk web, and
transducer rotation or skew. Although originally designed for disk bore/keyway inspections, the
scanner was modified at the EPRI NnE Center to accommodate disk blade attachment inspection as
well. Although the EPRIrrecnatom scanner proved to be an effective tool, it was also very
cumbersome to use. Prior to the Spring 1996 outage at Calvert Cliffs, BGE NDE staff designed and
built a LP disk scanner, as shown in Figure No.3, that met all examination performance
requirements, but at the same time was much lighter and easier to use than the EPRIrrecnatom
scanner. An important feature of the BGE scanner is its ability to examine disks "in-place", thus
minimizing costs associated with removing the rotor.
249
Mark Richter
In addition to the scanner developments, DGE personnel obtained custom built "end effectors" that
were attached to the end of the scanning arm. These devices, providing five axes of motion driven by
DC servo motors and controlled remotely, were needed to provide adequate positioning of the
transducers on the disk.
BGE NDE Staff qualified their UT procedures system and personnel with both scanners at the EPRI
NDE Center. The qualification consisted of acquisition and analysis of data generated from EDM
notches in the blade attachments of retired disks.
In the tangential entry blade attachment design shown in Figure No. 4 and existing on Calvert Cliffs
Unit 1 LP rotors, the fir-tree attachment is machined around the periphery of the disk rim and the
blades straddle and engage the disk hooks. Two or three hooks are common, and the sides of the disks
are generally flat and parallel, or near so, and amenable to ultrasonic inspection. The inspection is
normally o ~ u t e by directing an ultrasonic beam from tQe disk face below the attachment radially
outward into the attachment and across to the hooks on the opposite side of the disk. Beginning at the
bottom hook, the inspection is accomplished by selecting a refracted shear wave (S-wave) angle,
generally near 45 degrees. Where possible, other hooks are inspected by merely moving the
transducer radially outward to redirect the beam at them. However, because of interference from the
bottom hook on the transducer side of the attachment, this is not always a workable solution. Most
often, the upper hooks can only be inspected by moving the transducer radially away from the rim and
increasing the refracted angle of the beam. The approach described here, and one slightly modified for
axial entry blade attachments has been utilized by BGE since 1993. Blade attachments have been
ultrasonically examined on 10 LP rotors, including nuclear and fossil units, as well as two customers
external to BGE..
250
Mark Richter
Examination Results
A UT examination was performed by BGE on the L-2 and L-3 dovetails of LP "c" during the Calvert
Cliffs Spring 1994 outage. This inspection revealed evidence of but no cracking. These results
were very surprising to our engineering staff that expected to find cracking similar to LP "A" and LP
"B". To the notch group buckets were removed, and fluorescent MT and PT examinations
were performed of the area. The inspections revealed no cracking, only thus confirming our UT
results.
During the Calvert Cliffs Spring 1996 outage, UT examinations of LP "A" and LP "B" rotof, L-
3 stage dovetails were performed. The results are summarized below:
LP "A" Results
The examination of the L-2 dovetails revealed evidence of random pitting located on all the hooks
around the circumference of the disks. The depth of the pits was on the order of 0.03-0.04 inches.
The examination also revealed evidence of more significant indications.
L-2G DIS
L-2G DIS
L-2TDIS
L-2TDIS
L-2TDIS
L-2T ADM
L-2T ADM
Bottom
Middle
Middle
Bottom
Top
Bottom
Top
Pitting at Dotch corners
Heavy pitting or crack at notch comer
Crack at notch
Heavy at notch corners
Crack at notch
Crack at notch
Pitting at notch comers
The examination of the L-3 dovetails revealed no evidence of new cracking in areas ground out in
1992. This was confirmed by fluorescent MT.
LP "B" Results
The examination of the L-2 dovetails revealed no evidence of cracking. There was, however, evidence
of random pitting located on all the hooks around the circumference of the disks. The depth of the pits
was on the order of 0.03-0.04 inches.
251
,
Mark Richter
The examination of the L-3 dovetails revealed evidence of random pitting similar to the L-2 dovetails.
Also, more significant indications of pitting were detected

L-3T ADM Middle Pitting at notch comer
L-3T DIS Bottom Heavv Dittin2 0.25" from notch. d = 0.05". 1 =0.70"
L-3G ADM Top Pitted area on top side of hook, 1.70" from notch.
Light pitting on underside of hook.
The examination of the L-3 dovetails revealed no evidence of new cracking in areas ground out in
1992. This was confirmed by fluorescent MT.
These examinations generated much useful information for BGE. The NDE results were utilized with
an assessment of critical flaw sizes for LP blade attachments performed prior to the outage, to provide
a strong technical basis for avoiding further maintenance activities during the outage. The OEM
performed an inspection on the same LP blade attachments during this outage. They also found
cracking on some blade attachments, but was unwilling to use their NDE results in a fracture
mechanics based assessment as did BGE. The insp,ection highlighted some other important differences
between the OEM and the BGE inspection. The BGE scanner was built to meet the original
requirements of "in-place" inspection. The OEM scanner was not capable of performing that type of
inspection. The BGE scanner performed tbe inspection from the floor, whereas the OEM scanner
required scaffolding. Also, the BGE scanner was relatively insensitive to rotor rotation speed during
inspection, while the OEM scanner required close control of rotation speed.
252
Mark Richter
Summary
The future actions concerning the CCNPP-l LP rotor dovetails and wheelbores are uncertain.
Corporate decisions concerning plant life, non-OEM replacement rotor issues such as missile
probability analysis and torsional vibration analysis, NDE advancements in the detection and sizing of
sec, and the increase of industry experience gained through inspections are all variables in the "Risk
Management" equation.
It is incumbent upon the equipment owner and operator, not the OEM, to successfully manage the
repair/replacement decision process concerning these rotors. The ability to independently acquire and
analyze NDE data and failure mechanisms provides the owner with the necessary tools set sound
engineering parameters on some of the variables in the equation.
253
Mark Richter
Figure1. Wet Fluorescent MT indication of a dovetail crack on the L-2
stage in a CCNPP LP rotor.
254

.' ..., ...
FIGURE NO.2 EPRIlfecnotom Turbine Disk Scanner.
Mark Richter
FIGURE NO.3: BGE Turbine Disk Scanner
255
Mark Richter
256
FIGURE NO.4: Tangential - Mount Blade Attachment Design
Low Pressure Turbine Stress Corrosion Cracking Investigation
at the Navajo Generating Station
Presented By
Eugene Nowak
Salt River Project
257
258
LOW PRESSURE TURBINE
STRESS CORROSION CRACKING
INVESTIGATION
AT THE
NAVAJO GENERATING STATION
Eugene J. Nowak
SRP
Navajo Generating Station
PO Box 850
Page, Arizona 86040
259
260
LOW PRESSURE TURBINE
STRESS CORROSION CRACKING
INVESTIGATION
AT THE
NAVAJO GENERATING STATION
Eugene J. Nowak
SRP
PO Box 850
Page, Arizona 86040
Abstract
The Navajo Generating Station (;LTl.e) operates three General Electric (OEM) G3
tandem compound reheat four flow low pressure turbines each rated at 805 megawatts.
Each unit is fueled by coal and operates at supercritical conditions. The three units
went into operation between the years 1974 and 1976. Significant stress corrosion
cracking was found in Unit 1 LPB L-1 finger attachment during an April 1995 L-1
bucket replacement outage. This paper provides a 3 unit approach to the: history of
discovery, remedial action alternative evaluation, inspection, inspection results,
evaluation criteria, chemistry history, and remedial actions chosen.
Discovery
Initiating Event
._._-
c,

___ -.J
Unit 1 was removed from service Apri115, 1995, to replace a damaged low pressure
(LP) turbine generator end (GE) L-l bucket group (1 row upstream of the last stage).
The damaged group was visually identified during a forced outage on March 22, 1995.
The failed bucket had broken off at the tiewire hole and it was the leading bucket in a
four bucket group. The group bucket cover was also shed. The tiewire was still
contained within the group and bent outward along the convex side of the second
bucket of the group. Because the group was still contained by the tiewire, the
261
Eugene Nowak
replacement plan was to remove the entire four bucket group and replace it with a
covered and assembled group from warehouse stock.
Area
Tie Wire
'---
"p::>eo.r ;oe
>11:6'
j'''W;'>Q" plh' ..-d
1: 9"'"' 1:C
Figure 1- Typical L-1 Four Bucket Group & Damaged Group (140
0
rotor position)
A total of three bucket groups were removed from the LPB rotor GE during the April
outage: wo groups (100
0
and 310
0
rotor position) were removed after tie-wire hole
inspection revealed cracks in excess of 1/4
11
; the group at 140
0
rotor position was removed
as planned from the rotor on April I?, 1995. With the rotor "in place," the rotor bucket
attachment area of the removed bucket groups was hand cleaned and wet fluorescent
magnetic particle inspection of the attachment areas was performed. The wet magnetic
particle inspection of the rotor bucket attachment area revealed numerous llcrack like"
indications in the "fingers." -
262
I
,I
Bucket
Attachment
"Pins
- Rotor'Mleel
L-1 (18th Stage)
II-
I.
Figure 2 - L-1 "Finger" Style Bucket Dovetail (Left) &
L-2/L-3 IIPine Tree" Dovetail (Right)
-; Rotor Wheel
r--' l-2 J L-3
Il
Eugene Nowak
Plan of Action
During the inspections and investigations of the finger dovetail cracking, several
conference calls were held between the OEM and ii-@' personnel. The purpose of
these calls was to formulate a recommended plan of action for both short term and long
term operation. The requested goal of -life was to be able to operate for 6 months in
order to get through the peak summer season.
1
The formulated plan proposed by the
OEM consisted of three items:
removal and inspection of the entire GE & TE L-l rows
removal and inspection of a sampling of GE L-O buckets
ultrasonic inspection of the L-2 & possibly the L-3 rows
A i i : ~ ~
...., Action
_WS1N'
removed & inspected the entire L-l GE buckets
machined side of one UPinger
U
to verify validity of indication
surface replicated selected area of the L-l GE wheel
removed & inspected one 5 bucket L-l TE group
removed & inspected a sampling of L-O GE buckets
ultrasonic inspection of the L-2 TE & GE dovetails
ultrasonic inspection of the L-3 TE & GE dovetail admission side from the exhaust
side of the dovetails
Initial Findings
L-l GE - numerous crack-like indications in wheel
L-l GE wheel replication - stress corrosion cracking
L-l TE group - indications similar to, but fewer than GE
L-O GE (8 buckets removed) - no indications found
L-2 TE - "holding" indications (repeatable indications with length) away from notch
L-2 GE. - "holding" indications near notch
L-3 TE & GE admission side - "pit like" indications
L-O GE & L-l TE were reinstalled after completion of the inspections. The L-1 GE
buckets were reinstalled after some additional exploratory confirming machining on #4
finger. Unit 1 was returned to service April 26, 1995.
263
Eugene Nowak
Apri/1995 Replications
i :
Area 2, 200X
Figure 3 - Replication of LPB GE L-l Finger #3
(Fingers numbered 1 thru 6, in accordance with steam flow)2
Apri/1995 NDE Results
Inspection of the LPB L-1 GE finger dovetails found numerous crack indications which
were beginning the process of looping from pin hole to pin hole. Other indications
were scattered and in a radial direction. Although no blast cleaning was conducted,
there is no doubt as to the identification and severity of the cracks. The cracks are deep
into the wheel finger dovetails both axially and radiaIly.3 Figure 4 provides a typical
representation of indications found on the LPB L-l fingers. The lower land on finger #
4 was machined to the top of the lower pin holes to verify the indications were not
artifacts but indeed cracks.
,-, ! . _ ~
ViewA-A
Figure 4 - Typical NDE Indication Locations
The results of the LPB GE L-l attachment NDE were plotted by unfolding the wheel,
identifying each bucket & bucket grouping, showing location of pin holes &transition
lands (removed for clarity), denoting admission (0) or discharge side (X) indication
location, and labeling each finger. A sample of the plot is shown in Figure 5, indications
are noted on both sides of the fingers by the overlapping of the location indicators ~ .
Location of the March 1995 and other relevant information were also noted on the plot.
264
Eugene Nowak
f- inger .1:14
Lifting

Group 11 Group 14 Group l5
o {) o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
C 0
o 0
o 0
o
o
o
o 0
o 0
o 0
o 0
o 0
o 0
o
o
c
o
c'
o
o 0
o 0
01
0
o
o
o
o
o
o
o
o
o
o
Finger #1 0
o
o
Finger #-=' 0
c
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o
o 0
o 0
o 0
o I 0
010
o I 0
o
o
(1
o
o
c
o 0
o 0
o ()
o
o
o
000
000
0"0 0
o
o
o
o
o
Finger 1'*3 0
o
000
o 0 0
0
o 0 0 0
o 0 0 0
o : 0 -0 0
o
o
o
o 0 Q
o i 0 0
o I 0
o
o
000 0
o (; - 0-0
o 0 n 0
o
o
o
o
o
()
000
0(>0
.. 0 -C)
"8
o
Finger #4 c
o
o
o
()
o
o
o
o 0; 0 -0 0
o o' 0 0 0
0-0-10-0- "0
o 0
o 0
0-.
o
o
o
o 9
o 0
b' t!
o
Finger #5 ()
o
o
o
o
_1
o
o
o
.e
o
o
(}
o ! 0
o : 0
(:- (;
()
o
o
o
(}
o
o
o
,
t 55
(i
o 0
() 0
o 0
0---0
o 0
(l
o
o
I
be ' 63
o 0
o 0
o 0
o
o
o
65 ""
o 0
C D
0-0
.6
o 0
finger #6 0 0
o 0
o
o
o
000
000
000
o 0
o 0
o 0
o 0
o 0
o 0
o
o
o
o
o
o
57
o
o
o
o 0
o 0
o : 0
o
o
o
61
o
o
o
I
I

I
o : 0
I
....
o
o
o
o
o
o
o
(";
o
Figure 5 - 4/95 L-1 NDE Map (Sample)
Rock-n-roll pivoting of a straight edge about the #4 finger in the area of the March 1995
bucket failure caused concern of suspected localized sectional yielding. This area was
measured and noted on the plot for future reference.
Condition Appraisal
The need to return Unit 1 to service, the nature and location of the indications, bucket
attachment evaluation, fleet maintenance history, pressure plate lead
time 8 weeks) and !!.B.! water chemistry were all used to assess the risk of continued
unmitigated unit operation. As desire.was continued operation of
Unit 1 until the fall of 1995. A fall outage would allow sufficient planning and order
time to address pressure plates for the damaged L-1 stages and to also formulate a plan
for the remaining 2 units at !!..@.
265
Eugene Nowak
~ had access to a finite element model of the L-l attachment as a member of the GE
20.9" L-l User Group.4 The availability of this model provided a fast turnaround of
information to assist in the operation decision. A"Coarse Mesh" analysis
S
of the L-l
finger style attachment confirmed an ;:::;; 50% turbine overspeed capability of a non-
damaged attachment. Overspeed capability is reduced to ;:::;; 20% with the loss of one
attachment finger, and further reduced to;:::;; 5% with the loss of 2 fingers. The coarse
mesh analysis also revealed that fingers #3, #4, #5 carried higher stresses.
6
Possibly
shedding some light on the heavier SCC damaged in these fingers.
The members of the GE 20.9" L-l User Group, and other 20.9" L-l owners were
contacted and queried regarding experience of inspection of the L-l attachment.
7
Those
members of the L-l User Group who had inspected the L-l attachment inaccordance
with recommendations found in an OEM Technical Information Letter had not found
indications of SCC. A second utility was found to have experienced sec in a 20.9" L-l
attachment. Eventually it was found that a third utility had documented the existence
of see in the attachment fingers of 20.9" L-l buckets.
8
The previous ~ LP rotor Units 1,2, & 3 inspection records (1993, 1994,1992
respectively) including the rotor BoreSonic inspection (1984, 1985, 1986/1989
respectively) were searched for any previous indications in this area. No confirmed
indications of sce were found in these reports. But it was noted that in 1988 an
inspection of the L-l finger attachments during bucket changeouts had found "linear"
indications at the base of three fingers (1 between fingers 1 &2, 6 between fingers 3 &4,
and numerous between fingers 5 & 6). These indications were not replicated because of
location, therefore the nature of the indication could not be confirmed as sec. The
general consensus was the indications were artifacts leftover from the initial rotor
attachment machining. The indications were removed by machining the radius at the
base of the finger and the rotor was returned to service.
9
The extensive damage found in Unit 1 L-1 attachment is described as intergranular
stress corrosion cracking with the primary contaminant being caustic (NaOH).10 The
investigation of each unit's water chemistry and condenser tube leak history indicated a
strong possibility that the SCC damage found at iLt/..,. would be limited to only Unit 1
and related to the effects of condenser tube leaks.
Unit 1 original condenser tubing was arsenical copper, which had a Significant history
of tube leaks caused by shell side ammonia attack. The condenser was retubed in 1980
with CuNi (90/10 & 70/30) because of excessive tube plugs. Unit 2 & 3 condenser tubes
(90/10 & 70/30 original) were replaced with AL6XN in 1986 & 1988 after a history of
tube leaks caused by Microbiological Induced Corrosion (MIC). The Unit 1 CuNi tubes
were replaced in 1993 with AL6XN. The history of Unit 1 tube leaks, the Unit 21988 L-1
attachment inspection (with no other indications found on the fingers), and in 1989
266
Eugene Nowak
changing to a higher performance resin for the Unit 3 demineralizers all pointed toward
Unit 1 being an isolated incident.
l1
The unknowns which could not be entered into the condition appraisal formulas
centered in tvvo areas:
1. Complete inspection and condition assessment (lacking information on Unit 1 LPA,
Unit 2, Unit 3)
2. Thorough engineering analysis to determine remaining strength, crack growth rate,
critical crack size, and time to failure.
Plan
Economic Evaluation
The 1995 summer challenge was to determine a comprehensive plan for all 3 units at
&:e with physical information only obtained from Unit 1. An economic evaluation
model was created to compare the alternatives for Unit 1 and for all 3 units combined.
The evaluation model included 25+ combinations of alternative evaluation schemes.
Ingredients into the spreadsheet models included: new rotor cost, refurbished rotor
costs, outage duration & cost,. performance changes, reduced generation capability
(from pressure plates),. fuel pricing and replacement energy costs. Each of these major
ingredients included a series of options such as existing schedule outages. Figure 6 is a
graphic representation of the economic alternative selection range. The abbreviations
used in Figure 6 are:
AA - II Advanced Aero" rotor steam path design to improve unit heat rate by 1.2%
RIK - "'Replacement In Kind" rotor steam path
PP - Pressure Plate(s)
Ul,2,3 - Units, 1, 2, 3
AA R-I-K PP& PP& Refurbish
Rotor Rotor AA R-I-K Outage
, , , r
$10,000,000
Unit 1 - Only
$20,000,000
$$$ Driven By:
Net Present Value
$$$ Driven By:
Component $$
Better ... .. Worse
Outage Time
$35,000,000
Units 1, 2, &3
$50,000,000
J ..
U1 New R-I-K U1 NewAA Refurbish U1 NewAA
& U2/3 & U2/3 . Rotors & U2/3
Refurbish R-I-K Refurbish R-I-K Refurbish AA
Figure 6 - Plant Approach Economic Range
267
Eugene Nowak
Creating an outage to refurbish rotors and the rotor refurbishment cost was offset by
replacement energy cost and the corresponding outage time. Therefore, two new fully
bucketed LP rotors were ordered from the OEM during August of 1995. The new rotors
had a 15 month delivery lead time, therefore a November 1996 outage schedule was
s e l e t ~ for Unit 1 LP rotor replacement. The removed rotors could then become seed
rotors for refurbishment outside of outages. Reduction in pressure plate delivery time
was also aggressively pursued with the OEM. The reduction in pressure plate delivery
time :vvas accomplished by pre-engineering the pressure plates, risk shared advanced
material purchase, and short lead time material sourcing were all completed prior to the
Unit 1 outage. Unit 1 was planned to be reinspected in early 1996, and if only two
pressure plates were required the lead time was reduced to 14 days. If additional plates
were required, the lead time was 21 days for the third plate and 28 days for the fourth
plate.
Schedule
Table 1 depicts the composite alternative selection plan for each unit during the Spring
of 1996.
Outage Outage If Acceptable If Unacceptable
Dates Type Condition Found Condition Found
Unit 1 1/6-2/8/96 Rescheduled Reinstall L-1 Buckets Remove L-1 Buckets,
Extended & Replace Rotors 11/9 Install Pressure Plates,
Valve Outage -12/18/96 & Replace Rotors 11/9
~ 12/18/96
Unit 2 1/29-3/7/96 Extended Continue Operation, Remove L-1 Buckets,
Scheduled Reinspection 1998 Install Pressure Plates,
Valve Outage & Install Refurbished
Rotors 1/31 - 3/27/98
Unit 3 3/9-4/9/96 Extended Continue Operation, Remove L-l Buckets,
Scheduled Reinspection 1997 Install Pressure Plates,
Valve Outage & Install Refurbished
Rotors 2/22 - 4/18/97
Table 1 - Plant Plan
Inspection & Acceptance Criteria
Inspection and acceptance criteria to be applied during the spring 1996 inspections were
developed during the fall of 1995. An inspection plan was developed with the OEM to
provide rapid information to assist in the evaluation of the condition of the L-1
attachments. The inspection plan included the following:
Lift Check L-1 Buckets - the gap between the buckets and wheel was
measured in an effort to observe any U gross" area of lifting, see Figure 7
268
Eugene Nowak
Pinch Check of the bucket fingers - outer fingers were flexed 0.060" in an
effort to expose any tight indications in the bucket fingers. The fingers were
then Wet Florescent Magnetic Particle Tested (WFMT)
NDE L-l Wheel Fingers - same methods which were used 4/95
Rim Check Fingers with "Bridge Mic" - measure any relative attachment
finger movement and to recheck "rock-n-roll" area of Unit 1 LPB GE, see
Figure 7
Local Excavation Investigation - if required
The L-l buckets were to be removed with the rotors in-place, local hand cleaning of the
L-l attachment, and the inspections performed.
-I,
;:.. 1'-' -" i
i-i---
i
,--
---1
i 1:
\
i I I I
I I
i
\
I I i
.:\'-
i
I
I i I i
I
;
I I
i
I:
i
I
I
I
I
I [ [
I
I
I
I I
\
I
i
I
I
\
I
f
1
''';

I
n J
l
I-
i '\ [
!
I
I I
1
I
.L
I I
i ,I,
I
I
IT'
Figure 7 - Lift (Left) & Rim Check (Right)
Predetermined acceptance would be applied to the inspection results. An
expedited decision was required to trigger the order for pressure plates to minimize
delivery cycle. The following were the acceptance criteria:
Lift Check Fails = Install Pressure Plate
Pinch Check Fails =Install Pressure Plat'2
'tSignificant
lt
NOE Change Fails = Install Pressure Plate
Evaluation of Inspection Data (Lift Check, Rim Check, & NDE) Fails = Install
Pressure Plate
No "Fails" = Reinstall Buckets
269
Eugene Nowak
1996 Inspection Results
Table 2 is an all unit summary of the 1996 inspection results.
Checks NDE
Lift Rim Pinch Map
I Unitl I LPB GE okay okay okay no good
TE okay okay okay no good
I LPA GE okay okay okay acceptable
TE okay okay okay acceptable
I Unit2 I LPB GE okay okay okay no good
I LPA GE okay okay okay no good
I Unit3 I LPB GE okay okay okay no good
I LPA GE okay okay okay acceptable
Table 2 - 1996 Inspection Results
1996 NDE Plot
Figure 8 is included to graphically represent the wide spread extent of sec at i ~ It
is interesting to note the magnitude, pattern, and location of the sec. The following
observations can be made:
1. The most severely cracked are all four Unit 2 LP rotors ends (2 rotors) and Unit 3
LPB, both ends. Next in severity is Unit 1 LPB - both end and Units 3 LPA TE. Unit
1 LPA has a few indications, more at TE. Unit 3 LPA also has more indications at
TE.
2. In the heavily cracked rotors, the most severe cracking is in Fingers 3, 4, and 5.
There is no cracking in Fingers 1 and 6, with one exception being Unit 1 LPB GE.
There are a few indications of cracks in the upper pinholes in Unit2 LPA, both
ends, and in Unit 1 LPB TE.
3. In Fingers 3 and 4, there is a similar amount of cracking from the" admission and
discharge sides. In Finger 5, there seems to be more cracking at the admission side.
12
270
Eugene Nowak
Group 16 GrOt.lP 17 Group 18 Gr"'O\,.lP 19 Group 20
1t
.
7>
'"
,..
I
n
,.. ,..
III
I
""
8) ... e!l ...
I
.,
... ..,
I
0
(; 0 0
i
0 0 0 0 0 0 0 0 0 0 0 0 0 0 (1 0 0 0 \) (l
Finger ill 0 0 0 0 0
;
0 0 0 0 0 0 0 0 0 0 0 0 0
c'
0 () 0 0 0
,
l
f I I !
0 0 0
(> 0
I
0 0 0 0 0 0 0 0 0 0 c
0 0 0 (> 0 0 0 C'
6fl ro
"
72 7J r.; x-
l
n 76
,..
IlO 81 : B.? &:l &t ~ 116 eJ eEl v, 'Xl ,;>,
0 0 0 0 0
I
0 0 0 0 0

0 0_0
!
0 0 0 0 0 I c- o 0 c- o
.
-
!
FiY1ge'- 1t2 0 0 0 Co C 0 C 0 (") 0 0 0
~
(,") 0 0 0 0 0 C
I
(") 0 c C'
c.\
-
M
(,) 0-
0 0
.
-
O() 0 ..... (;.
S
C}
0 0 0 0
I
0 (\
I
0 0 (I
I
() 0
I
()
0 0
!
:
<., 6e 7:> 72 ~
'"
I
n 7lI llC 81
I
112 8J 8' ~ ...
!
k7
_9_ <) _0 ..0_0 1...!J_O_O_c_o_' o 0_ 0_0_0 1.:ii...-2......0 .... 0_ ~ i ~ c- o 0 ?_C
Finger *13 o 0 0 0 oj 0 G co 0 0 1 o 0 0 0 oj (; 0 0 0 0.J 0
-0-8
4
0 ::)
-0-0-0-0-0 : C ()-O-O-O: 0--0-0-0-0 I o-O-O-O-() 1 o-V-o--'(;-()
, ,
Finger- #4
,"
o 0 0 0
FInger #5 0 0 0 0
-0-0-0- OM
o I 0
o i 0
I
COO
o 0 0
0-0-0
o 000
Finger ~ 0 0 0 C
c} 0 0 0
C , 0
o f (.'
o (.>
n 1"l
o 0 0
000
o () 0
7<. ! 11
~ I ~
o 0
o
p
o
o
o
o
o
o
l
., I ...
Q ; c-
010
o 0
I
o
o
o
o
o
o
o
o
J .!
~ ! ~
0:0
o 0 C C'
C' 0 0 l')
o 0 ,.) 0
Figure 8 - Unit 2 LPA GE Representative 1996 NDE Plot
Pressure Plate Summary
A total of 9 pressure plates (Figure 9), were installed out of a possible12 locations. Each
of the 12 locations had some level of SCC damage to the L-1 attachment fingers. The
following is a listing of the pressure plate locations installed and unit capability
reduction.
Unit 1
LPB TE & GE (2 Plates)
Approx. 25 Mw Unit Capability Reduction
Unit 2
LPA & LPB TE & GE (4 Plates)
Unit 3
LPB TE & GE & LPA GE (3 Plates)
271
Eugene Nowak
Figure 9 - Unit 1 LPB GE Pressure Plate
Finger Section Samples &Analysis
Finger samples were removed from both Unit 1 LPB rotor and Unit 2 LPA rotor.
Samples of the fingers were distributed for analysis as shown in Figure 10.
:O"'C-oI"< .... 'i)"
, .. y b' 7..':
.. I C.....(;;",t.\o" >,'-1' / ""

r,,-.o....p I c, ........." -. :;"'oup 2B
.".. "
0-000 ll:":P \ STlSlmple "'MoperOnly """,." '0
-'C&aUnt d

1: :;HcdtCftckI3F.r
/HoIeC<oct"'F,,'lIO'
.1t. SRP s.m.-
il...
'i;9"J \
\ ;01 IJn;.....,c :6
j;I .....I')If..('1 r,..,.-
-
:n "".r
::;i
:t
-..O...r'!; .....
til ARB:
PIJHl f,kl."::if'\i::e-.t;ior'l

" :Ta"",'("'
<"I'" .... ' -:;iJ""",I",,
<J
Figure 10 - Finger Sample Distribution Map
272
Eugene Nowak
Figure 11 - Unit 1 LPB GE
Laboratory investigation of the sample removed from Unit 1 LPB GE U conclusively
showed that the cracking in the L-l rotor wheel is due to stress corrosion cracking.
Although the presence of sodium was not found in detectable quantities on the crack
surfaces, the intergranular and highly branched nature of the cracking indicates that
caustic (sodium hydroxide) was responsible for this condition."13
Samples from both Unit 1 LPB and Unit 2 LPA rotors were analyzed for chemical
composition (Table 3), and mechanical properties (Table 4), and compared to ASTM
A470 Class 7 material.
Element ASTM Unitl Unit 2
0/0 A470 Class 7 LPB GE14 LPAGE15
C 0.28 max 0.24 0.21
Ni 3.25-4.00 3.84 3.39
Cr 1.25-2.00 1.74 1.78
Mo 0.25-0.60 0.38 0.39
V 0.005-0.15 0.12 0.12
Table 3 - Chemical Composition
273
Eugene Nowak
ASTM Unit 1 Unit 2
A470 Class 7 LPB GE16 LPA GE17
Tensile 120 ksi 126 ksi 127 ksi
Yield 95ksi 111 ksi 114 ksi
Elongation 18% / 17% 26% 24%
Area Reduction ?2% / 50% 68% 70%
Hardness
-
24.5 Rc 25 Rc
Table 4 - Material Properties
Samples removed from Unit 2 LPA GE were mounted as shown in Figure 12. The
typical cracking was clearly intergranular, highly branched and oxide filled as seen in
Figure 13. The majority of the cracking occurred at the lower ledge area and
propagated towards the row of pin holes closest to the rotor. Cracking is "through-
wall" where both sides meet. The microstructure consisted of tempered martensite and
ferrite which is normal for high strength, low alloy steel.
18
Figure 12 - Unit 2 LPB GE L-1 Attachment
Finger 4 Metallographic Location
274
Eugene Nowak
Figure 13 - Typical Oxide Filled Crack
Water Chemistry
The limited"industry experience
lr
, the wide spread SCC damage at !l..G.e, and the
metallographic investigation seem to point in the direction of water chemistry as being
a "significant contributor.
n
I a ~ was being converted from AVT (All Volatile
Treatment) water treatment to OT (Oxygenated Treatment) water treatment about the
time of discovery of the seC. The OT conversion, by itself, brought significant changes
to the water chemistry which would reduce the potential for SCC.
The OT changes being incorporated have an impact on the performance and operation
of the mixed bed ion exchange demineralizers (polishers) used at !l..4. Polishers are
used to retain the "harmful" ionic elements from the once-through water cycle. As part
of AVT, ammonium hydroxide was injected to raise pH and hydrazine (later replaced
by Eliminox) to scavenge residual oxygen. Morpholine was also added to enhance the
corrosion protection in the initial phases of condensation.
19
Each of these items are
cations and compete for anionic sites within the polisher bed. Competition for these ion
sites result in polishers reaching ammonium break through (break through: process
which chemical ions make their way through an ion exchange demineralizer
20
) early in
the operation cycle of the polishers. Ammonium break occurred within approximately
3 days of operation of a newly regenerated demineralizer bed. The polishers were kept
in service for a total of 30 days. The changes involved with conversion to OT impact
each item mentioned and combined will have a significant impact on polisher
performance. Included in these changes is the discontinued use of oxygen scavengers
and Morpholine. The pH is also reduced from 9.3 to 8.3 which will yield a significant
reduction in addition of ammonium hydroxide.
275
Eugene Nowak
Figure 14 reflects a summary of the history of water chemistry changes and the other
contributors to sec (Material, Stress, & Environment).
'741 '75 i '761'771 '781 '791 '80 ! '81 I ' 82 1 '83 j'84 I '85! '86 1'87 ! '88 i '89 ! '90 f '91 ! '92 j '93 j '94! '95] '96
IMaterIal Umt 1, 2, & 3
I Suess Umt 1, 2, & 3
Environment
Enit2XO
Figure 14, sec Contributors History
Conclusions
The LP rotor inspections of 1996 conclusively revealed the Unit 1 see damage found in
1995 was not an isolated case. SCC damage at ~ extended through all 6 LP rotors.
The decision made to purchase 2 new fully bucketed LP rotors and then use the exi.sting
rotors as seed rotors for refurbishment were the most appropriate choices.
Chemical additives, polisher operation, and high ,centrifugal stress all contributed to the
complex interaction which caused the'SCC found in the LP rotors. The actions taken in
the conversion to OT, while beneficial, are not considered complete with regard to sec.
Therefore on March 21,1996, a Rotor Remedial Action Team (RRAT) was assembled to
address LP rotor sec at ~ The team was charged with root cause investigation
(Material, Stress, & Environment), new rotor installation, LP rotor refurbishment &
installation, and water chemistry system improvement recommendations. The
following is the charter for the RRATs: '
276
Eugene Nowak
1. Investigate the cause and determine remedial action(s) for sec in LP rotors
especially the L-l bucket attachment
To be completed by 1/1/97
2. Review water chemistry history @ NGS & determine role in SCC damage
3. Coordination of new LP rotor installation
4.. Coordination of LP rotor refurbishment and game plan of installation
Refurbishment of Rotors removed from Unit 1 for installation into Unit 3
to be completed by 2/15/97
Refurbishment of Rotors removed from Unit 3 for installation into Unit 2
to be completed by 12/31/97
Acknowledgments
John Alice, SRP - Technical and metallurgical support throughout the investigation and
editorial review.
Tony Lam, STI - Countless answers to stress questions.
Sue Nowak, Wife - Patience.
Rob Peterson, NGS - A demineralizer, and water chemistry education.
Mike Woods, NGS - Keen editorial eyes.
277
Eugene Nowak
1 C. Bran, "'LP-B L-l Finger Dovetail Pin Hole Cracking" (General Electric Co. report
4/25/95), page 5.
2 J.A. Alice, "Evaluation of Replicas, Deposits, Buckets, and a Broken Pin from the B
Low Pressure Turbine at NGS Unit I" (Salt River Project Metallurgical/Materials
Engineering Report No. MEN-3303, May 8, 1995), Figure 2.
3 Bron, page 5.
4 "Phase I Failure Investigation for GE L-l Single Boss Blading" (Stress Teclmology,
Inc. Technical Report PA823, GE L-l User's Group, October 1993).
5 H. Yu, "GE L-l Single Boss Bucket Disk Steady Stress Analysis" (Stress Technology,
Inc. May 9, 1995).
6 ibid.
7 E.J. Nowak, "Vnit 1 Maintenance Outage, LPB L-l Wheel Investigation & Repair
Recommendation" (July 25, 1996), page 6.
8 B. J. Piatt, Pennsylvania Power and Light's Experience with LP Dovetail Cracking (PWR-
Vol. 21, ASME 1993), page 4.
9 "Salt River Project Navajo Generating Station Unit #2 SIN 170x473 1988" (General
Electric Co Outage Report), LPB Rotor Tab 4.
10 J.A. Alice, "Evaluation of Replicas, Deposits, Buckets, and a Broken Pin from the B
Low Pressure Turbine at NGS Unit I" (Salt River Project Metallurgical/Materials
Engineering Report No. MEN-3303, May 8, 1995), Page 3.
11 R. B. Peterson, "History of Cycle Chemistry at the Navajo Generating Station,
Miscellaneous Chemistry Considerations" (March 3, 1996).
12 O. Jonas, "Stress Corrosion Cracking of Navajo LP Rotor" (January 1997), page 2-1 &
2-2.
13 J. A. Alice, "Evaluation of Rotor Wheel Fingers Removed from the L-1 Row of the B
Low Pressure Turbine at NGS Unit I" (SRP Memorandum MEN-3480, February 28,
1996), page 3.
14 ibid.
15 J. A. Alice, "Evaluation of Rotor Wheel Fingers Removed from the L-l Row of the A
Low Pressure Turbine at NGS Unit 2" (SRP Memorandum MEN-3500, April 4, 1996),
page 3.
16 Alice, MEN-3480, page 3.
17 Alice, MEN-3500, page 3.
18 ibid.
19 Peterson, #1 & #3
20 Peterson
278
Design, Qualification and Testing
of RotorlDisc Weld Repairs
Presented By
David Gandy
EPRIRRAC
279
280
DESIGN, QUALIFICATION AND TESTING
OF ROTORIDISC WELD REPAIRS
David W. Gandy
Shane J. Findlan
R. Viswanathan
R.Munson
Radian International
W. F. Newell, Jr.
William F. Newell & Associates
281
282
DESIGN, QUALIFICATION AND TESTING
OF ROTOR/DISC WELD REPAIRS
David W. Gandy
Shane J. Findlan
R. Viswanathan
R. Munson
Radian International
W. F. Newell, Jr.
William F. Newell & Associates
Abstract
One of the least understood types of weld repair, yet one of the most costly, is that
associated with a damaged steam turbine rotor and disc. During 1995-96,
comprehensive weld repair guidelines were assembled by EPRI, under EPRI program
RP3963-01, to assist utilities with repairs of these components. The guidelines prOVided
methodical approaches for each of the seven forms of weld repairs which are associated
with major rotating machinery, along with detailed information" on: repair evaluations,
selection of weld filler metals, postweld heat treatment criteria, benefits, risk and risk
management selected repair case histories, metallurgical testing, etc.
This paper is based upon information presented in the repair guideline and is focused
primarily on four key topics: materials properties characterization, design of a weld
repair, weld metal qualification, and weld procedure qualifications and mockups.
Detailed decision approaches in the foon of flowcharts are presented and reviewed.
Introduction
In 1995-96, under the sponsorship of twelve utilities (both domestic and foreign), EPRI
initiated a program (RP3963-01) to address steam turbine rotor, disc, and blade repair
issues and concerns. Under this program, a comprehensive Repair Guideline (TR-
107021, Volume 1) was assembled to assist utilities specifically with repair of high,
intermediate, and low pressure steam turbine rotors and discs. Volume 2 of the
Guideline was aimed at repair of steam turbine blades. The rotor and disc repair
guideline provided information concerning:
283
David Gandy
Rotor, Disc, and Weld Filler Metal Specifications
Turbine Repair Evaluation
Nuclear and Low Pressure Rotor Repair
Intermediate- and High-Pressure Rotor Repair
Postweld Heat Treatment
Repair Welding
Protective Coatings for Rotors
The focus of the Guideline was to assemble state-of-the-art information and technology
so that utilities can make better informed technical and more cost effective decisions
associated with rotor and disc repairs. Furthennore, the Guideline provides utilities
with background and information so that they can more closely interface with original
equipment manufacturers (OEMs) or repair vendors, thus improving the resulting weld
repair. The following discussion is based upon information contained within the Repair
Guideline and will concentrate on a few key issues surrounding rotor / disc repair
including: 1) materials properties. characterization, 2) design of a weld repair, 3) weld
metal qualification, and 4) welding procedure qualifications and mockups.
One of the more important sections of the Guideline document was targeted at Turbine
Rotor Repair Evaluations. Under this section, detailed Rotor Repair Technical Decision
Flowcharts were assembled to provide utility engineers with a step-by-step decision
analysis approach. A total of six decision flowcharts were assembled. Comprehensive
discussions were provided along with the flowcharts which described key
considerations such as:
1. Formulation of a Repair Plan (rotor usage/limitations, vendor selection, proposal
evaluations)
2. Metallurgical Considerations (evaluation of cleanliness, chemistry, properties,
identification of failure cause)
3. Technical Considerations (weld performance, repair impact on dynamics and
operation, performance improvement/reliability, influence of operating
environment)
4. Commercial Considerations (repair insurability, inspection protocol, warranty,
future serviceability, cost/schedule impact)
284
David Gandy
5. Quality Assurance Considerations (Vendor QA, post-repair inspection/ outage
planning)
The first flowchart as shown in Figure 1 served as the master chart from which the
remaining five flowcharts were derived. Key issues highlighted in the flowcharts
included:
Flaw Analysis
Root Cause Analysis
Materials Properties
Determination of Repair Options
Design and Qualification of a Repair
In this report, the third and fifth items will be exhaustively reviewed.
Materials Properties
Before an OEM or Repair Vendor will consider attempting a repair, the current
metallurgical and mechanical properties of the rotor must be determined. At a
minimum, one should expect the rotor to meet or exceed the specification requirements
to which it was constructed. A rotor that no longer meets its original properties because
of in-service degradation must be evaluated critically and, unless the properties can be
restored by thermal or mechanical treatment, it is a good candidate for the scrap pile.
An important limiting factor in determining rotor properties is often the availability of
test material. A steam turbine rotor has very little extra material available that can be
removed for testing. Test material can often be removed from the rotor bore by either
trepanning or boring. This material actually provides the "poorest" representation of
true rotor properties. The near-bore region has high inclusion density and less strength
than the rotor periphery. Occasionally a sample can be removed from the coupling.
This sample is very limited in size but will represent the rotor strength quite well.
If a major repair is to be made, especially if the rotor has sustained an in-service failure,
larger quantities of test material are available. Occasionally a rotor end may be
removed or a wheel rim parted, which supplies larger quantities of material. If little
material is available and the need for testing is great, one alternative is to locate and
sample a same vintage rotor with similar chemistry and tensile properties. This
approach is especially useful in making a weld repair mockup.
285
David Gandy
Several alternatives or levels of effort can be expended to measure rotor properties. The
selection of the properties to measure will depend on many factors, including but not
limited to the previous failure mode of the component, future service, and severity of
the weld repair. Figure 2 provides a road map for evaluating materials properties of a
rotor prior to finalizing a repair decision. As shown in the upper half of the flowchart,
at a minimum five tests are recoriunended to detennine if a rotor should be considered
for repair. If properties obtained through these five tests suggest the repairability of the
rotor is marginal for its intended service application, then additional tests shown in the
lower half of the flowchart should be considered.
Hardness is one of the easiest properties which can be measured without damaging the
rotor. Hardness can provide a good indication of the strength of the rotor. Samples can
also be removed from the rotor for testing of mechanical properties such as ductility,
yield strength, and toughness. A sample can be cut from the bore or a coupling face.
This sample should be used to measure tensile properties and toughness and establish
the Fracture Appearance Transition Temperature (FAIT). The FAIT will define the
ductile/brittle temperature for the rotor and indicate the toughness. Knowing the
FAIT is critical in establishing the weld preheat and interpass temperatures and in
specifying heating and cooling ramp rates. For the chromium-molybdenum-vanadium
alloys (primarily intermediate pressure (IF) and high pressure (HP) rotors), it is also
.beneficial to determine whether the rotor has been embrittled in service. This
embrittlement can lead to unexpected brittle fracture of the rotor in service or cracking
of the rotor during welding. Embrittlement can also lead to a reduction in creep
strength (1). The embrittlement of the rotor can be successfully overcome by a
deembrittling heat treatment (2) or a simple change in repair practice or, for in-service
failures, an alteration in operating practice. Embrittlement in a rotor can be measured
by impact testing of specimens from the rotor. Either full-size Charpy samples or
punch-test specimens have been used (3,4). An approach that is commonly used is to
remove a small ring sample from the rotor bore (Figure 3) (4). If the specimens are too
small for mechanical testing, coupons can be carefully electron beam welded to the
specimens to allow testing with limited effect on measured properties (Figures 4 and 5)
(5). There are few limitations (one being sample size) on the types of tests that can be
performed on samples trepanned from the rotors, and there is seldom enough material
to perform every test that an engineer would like.
Metallographic samples can also be examined to verify microstructure, characterize
nonmetallic inclusion content or type, or confirm fracture mode. If the rotor has pitting
or corrosion damage, a metallographic examination can usually suggest the likely
causative agent for the corrosion. There are also metallographic procedures to check for
embrittlement. If there is defomation of an IP or :HP rotor, it is always wise to check
for creep damage. Metallographic examination can be performed using in-place
techniques and documented using replication. The in-place procedures are essentially
nondestructive.
286
David Gandy
Design of a Weld Repair
The mechanical properties of a weld in a component generally do not exactly match the
mechanical properties of the rotor forging. The full implications of this must be
considered in the repair plan. For example, the weld strength can either overmatch
(have higher tensile strength) or undermatch (have lower tensile strength) the rotor
material. If the weld metal is undermatched, it will have lower fatigue strength than
the rotor, and this must be considered in the location of the weld. One does not want to
locate the weld in a highly stressed area. A solution may appear to be to use an
overmatched weld metal in all cases; however, this also can create difficulties.
Performance of a weld repair produces a heat-affected zone (HAZ). The properties of
the rotor are altered within this HAZ, and generally contain areas that are both stronger
and weaker than the rotor steel. The stronger areas are hardened and tempered by the
welding heat and self-quenching. These hardened areas are near the fusion line. The
softer (weaker) areas are remote from the fusion line by 1-5 mm and represent
overtempered areas. The soft areas have lesser fatigue strength. Thus, even using an
overmatched weld filler to combat fatigue does not overcome the soft (low fatigue
strength) areas in the weld HAZ. A good approach to this dilemma is to always locate
the weld in low-stress areas where the effects of this low fatigue strength are minimal
and not likely to cause a repeat failure. A finite elemental analysis, even if only a
parametric one, showing sensitivity of location is useful in optimally locating this weld.
Another approach is to minimize the size of the softened HAZ by using a low-heat
input welding process such as GTAW or LASER welding.
Another, more difficult problem is locating and engineering a weld in a high-
temperature section of an HP or IP rotor. In addition to the problem of
overmatch/undermatch and variable weld HAZ strength, there is an additional
problem of creep strain mismatch. In simple terms, this means that the weld and rotor
steels will plastically "stretch" at different rates. The accumulated strain differential can
cause additional stresses that lead to accelerated creep damage. This problem is not
well understood. There are two solutions generally used and accepted by the industry.
One is to locate all welds in rotor sections with operating temperatures below 700
o
P;
this is below the lower creep temperature for low alloy steels. The second approach is
to use a higher alloyed material, such as 120/0 chromium steel, for the weld filler. The
120/0 chromium has a better creep strength than the low alloy rotor steel and a favorable
creep strain differential.
Weld Filler Metal Qualification
The selection of the appropriate weld filler and weld process is one of the more
important steps in developing a commercial weld repair capability. Generally, the weld
vendor will have completed the selection before entering the weld repair market. A
287
David Gandy
reasonable request that a utility can make of the weld vendor may be to see the vendor's
weld metal filler qualification plan. This will immediately verify whether the vendor
indeed has established a weld qualification program. Figure 6 provides a generic
approach for design and weld metal qualification in a flowchart format.
As shown in the flowchart, qualification of a filler metal demands rigorous testing and
verification. At a minimum, qualification includes: 1) selection of a filler metal and/or
flux combination, 2) preparation of a test coupon from which mechanical test specimens
can be removed, 3) measurement of deposited composition, hardness, and tensile
properties, and metallography to establish inclusion content. If the measured
properties are deemed acceptable at this point, consideration must be given to the
intended service of the rotor. If the service is nuclear (LP, IF, or liP) or fossil LP,
acceptable Charpy impact or FAIT properties should be verified for the weld metal
deposit.
If the intended service is fossil HP or IP, the testing becomes somewhat more
complicated. Creep or stress rupture testing should be considered if temperatures
(>lOOOF) are such that creep damage is a concern for the intended repair. On the other
hand, if the expected temperatures are below lOOOF in the temperature range where'
temper embrittlement may be of concern, a step heat treatment and subsequent testing
may be of use to evaluate the weld metal's susceptibility to temper embrittlement.
Welding filler metal examples for selected applications are included in Table 1.
Welding Procedure Qualifications and Mockups
Once a reasonable design approach has been established by the Repair Vendor or OEM
and a filler metal has been selected for a particular repair application, specific welding
procedures must be developed and a weld qualification performed. The weld
procedure qualification should be performed on a sample coupon which has been
removed from the existing disci rotor in question. In the case of a shrunk-on disc
repair, such material may be difficult to obtain. A successful approach employed by
several vendors has been to section a ring as shown in Figure 7 (6). The "parting-off" of
the ring:
1. removes the cracked (damaged) material entirely,
2. provides adequate material for the procedure qualification and/or mockup,
3. allows the base metal/repair interface to be located in a lower stressed region of
disc.
If it is not possible to remove sufficient material for qualification testing, the alternative
is to obtain accurate material chemistry and mechanical property data (hardness, FATT,
etc.) for the disc and acquire a sample of material that closely matches these properties.
288
David Gandy
Generic coupons/mockups for welding procedure qualifications have been provided in
Figures 8 and 9). In the case of disc repairs, the quantity of base material may be
somewhat limited (unless a ring can be obtained) and may require ingenuity on the part
of the engineer. In any case, the coupons shown provide some indication of the
qualification coupon's proportions sufficient for removal of test material.
Upon completion of the welding qualification (and/or mockup welding) and PWHT,
the coupon will be tested in a similar manner to that shown earlier in Figure 2. The
only difference in the approach employed is that now the testing is focused on the HAZ
and on the deposited weld metal as opposed to the base metal only. Test coupons
should be oriented to maximize the amount of information that can be obtained (i.e., the
center of each test specimen should be located in the center of the HAZ). Each of the .
five properties described in the upper half of Figure 2 should be evaluated and
compared to that measured for the service-removed base metal. Also, depending upon
the intended service, supplemental testing may be required (these tests were outlined in
Figure 2). Again, it should be emphasized that both HAZ and allweld metal properties
should be tested and examined.
Summary and Conclusions
The establishment of rotor/disc weld repair capabilities by a number of repair vendors
and OEMs over the past decade has presented utilities with a host of potential repair
options and choices. Along with these options and choices, the need for guidance and
uniformity in repair approaches has surfaced a key issue for utilities. In 1995, EPRI
initiated an effort to develop repair guidelines to assist utilities in working more closely
with OEMs and repair vendors and to increase the level of understanding surrounding
rotor / disc repair. This paper has taken several important excerpts from the guidelines
and expounded on key issues surrounding materials characterization, weld design,
filler metal qualification, and procedure qualification and mockups. Detailed
discussions and flowcharts have been presented to assist in presenting step-by-step
approaches to materials testing and filler metal qualification.
Specific guidance has been provided for rotor/ disc testing to determine its repairability.
Furthermore, recommended approaches have been reviewed for test coupon removal
and weld qualification.
References
1. R. L. Bodnar, S. S. Hansen, J. R. Michael, and R. E. Steigerwalt. Improved Superclean
NiCrMoV Rotor Steel. ER-6887 Research Projects 2426-4,2741-4.. Palo Alto, California,
July 1990, pp. 6-1 - 6-37.
289
David Gandy
2. H. G. Gayley, T. Roginski, D. Macdonald, and C. McMahon. Relationship Between
Turbine Rotor and Disc Metallurgical Characteristics and Stress Corrosion Cracking
Behavior. Vol. 1: Test Materials and Laboratory Test Program, NP-4695SP, Vol. 1,
Research Project 1929-6, September 1986.
3. 1. Roman, C.A. Rau, Jr., A.S. Tetelman, and K. Ono. Fracture and Fatigue Properties of
lCr-Mo-V Bainitic Turbine Rotor Steels. NP-1023, March 1979.
4. J. R. Foulds, C. W. Jewett, and R. Viswanathan. Miniature Specimen Test Technique for
FAIT. EPRI Workshop on Steam Turbines and Generators, Charlotte, NC, July 1991.
5. Provided by Donald McCarm, Manager Engineering Materials and Analysis,
Siemens Power Corporation, West Allis, Wisconsin.
6. R. R. Ballal. "Weld Refurbishment of Turbine Rotor Blade Fixings." Proceedings--
Second International EPRI Conference on Welding and Repair Technology for
Power Plants, May 21-24,1996, Daytona Beach, Florida.
Acknowledgements
This paper is based upon information extracted from IIVolume 1: Weld Repair of Steam
Turbine Discs and Rotors" which was created under the State-of-the-Art Weld Repair
Technology for Rotating Components program (RP-3963-01). The authors would like to
thank those utilities who have sponsored this effort including: Baltimore Gas &
Electric, CornEd, ENEL/CRAM, GPU Nuclear, Illinois Power, Korean Research
Institute of Standards and Science, Pennsylvania Power and Light, Potomac Electric
Power, Southern California Edison, Southern Company, and Taiwan Power.
290
David Gandy
Table 1
Selected Weld Filler Metal Examples
Component Material Process Filler Metal
LP Rotors & Disks A293, Cl. 2-3 GTAW ERIOOS-l (Mod.)*
A470, Cl. 2-4
A471, Cl. 1-3
SAW EF5*, EM2*
LP Rotors A293, Cl. 4-5 GTAW ERl20S-1 (Mod.)*
A470, Cl. 5-7
SAW EM2*, EF5*, EB6*
HPRotors A293, Cl. 6 GTAW ER1205-G*,ERS05*,
A470, Cl. 8 ER410*
SAW ER410*, EB3*, EB9*,.
*Modified typically indicates lower residual elements and special alloy additions to .
enhance strength.
291
David Gandy
Rotor Repair
Technical
Decision Matrix
Prior History
OEM Experience
Preliminary
Assessment
of Cause
Flaw
Analysis
(Figure 3.2)
OEM
IRV
Consultants
Material
Properties
Evaluation
(Figure 3-4)
Root
Cause
Performed
(Figure 3-3)
Yes
No
No
Yes
Derate,
Change Inspection
Interval,
Change Operation
Marginal
Perform
Repair
Design &
Qualify
Repair
(Figure 3-6)
Select Repair
Option
Additional
Sampting&
Testing
RequiTed
Figure 1
Teclmical Rotor Repair Decision Matrix
292
David Gandy
Mechanical Testing
for
Repairability Determination
r
Additional Assessment _
Fitness for Service
Marginal
Additional
Sampling 100-----(
&Testing
Return to
Figure 3-1
Return to
Figure 3-1
Figure 2
Decision Matrix That Guides the Evaluation of Materials Properties in a Repaired Rotor
293
David Gandy
294
Figure 3
Bore Ring Remoyed from a Rotor for Mechanical Property Testing (5)
David Gandy
/
3/4 in. dia. radial
trepan sample
-- ~
Attachment
~ - - - 7 EB weld lines
I
I
I
I
I
I
...
3-point bend test specimen geometry
showing location of trepan sample
obtained from rotor spindle
Test material
+-t--- Attachment
Figure 4
Schematic Showing Configuration of EB Welded Trepan Specimens
for Fracture Toughness (5)
295
David Gandy
Figure 5
Photo Showing Compact Tension Sample Removed from a Rotor Bore Ring Section.
Additional Stock Has Been Electron-beam Welded to Accommodate Testing (5)
296
David Gandy
Bounding PWHT
Parameters
for
Base Metal
(Figure 5.1)
No Select
Another Weld
Filler Metal
Fossil LP &Nuclear LP, IP & HP
Design &
Qualify
Repair
Select
candidate Weld
Filler Metal
Fossil IP/HP
Filler
Metal Selection
Database
Weld Filler Metal
Satisfactory
Use Step Heat
Treatment
Return to
Figure 3-1
Weld Filler
Metal
Satisfactory
Figure 6
Weld Filler Metal Qualification Flowchart
297
David Gandy
298
Figure 7
Cracked disc head parted off by machining (6)
....---4
11
---.......
Figure 8
Weld Qualification Coupon
David Gandy
299
David Gandy
1) Build up weld pad on disc prior
to repair to obtain test coupons
2) Sever at line just above where
new HAZ will be located per
design
3) Perform weld repair
..... ......
Ir
- , , ~ ~
-
c: c:
In situ wheel
Sever
line \
/
Location of
HAZ per
design of
weld repair
Figure 9
Alternate Weld Qualification Coupon
300
PANEL DISCUSSION
Reported by
Henny Hunter Gerard, RPR-RMR
Edited by
Dr. R. Viswanathan, Chairman
Panel Members
Dr.R. Viswanathan, EPRI, Chairman
Mr. T. McCloskey, EPRl, Co-Chairman
Mr. C. DeCesare, General Electric Co.
Prof. M. Speidel, Swiss Federal Institute of Technology
Dr. D. Rosario, Structural Integrity Associates, Inc.
Mr. T. Alley, Duke Power Company
Mr. D. Amos, Westinghouse Electric Corporation
Mr. D. Gandy, EPRI RRAC
Dr. M. Richter, Baltimore Gas & Electric Company
R. Viswanathan: We are reaching the best part of the meeting so far. We heard
many differing points of view on many issues but here is the time to recognize
some of these apparently differing view points and try to come up with some
answers. We have a panel of experts who kindly consented to help us out in this
task. We tried to pick different people to represent different points of view. First of
all, Tom Alley from Duke Power Company who has spent most of his career in NDE
and metallurgical type activities will give a metallurgical point of view. Dave
Gandy is our welding expert in the Charlotte Center. Darryl Rosario has been active
in the life assessment area. And then, of course, we have Professor Speidel. We are
very fortunate to have him here. He has been in the corrosion business many years,
is very well known world wide, and is here to tell us about the materials issues.
Then we have Carl DeCesare from General Electric Company and Dennis Amos
from Westinghouse who could address any of the issues that include materials,
environmental, and repair technology.
The way we would run this panel discussion is I will ask each one ot" these
gentlemen to make a few remarks briefly to stimulate discussion, describing what
they have heard, what they have learned, what they see in the crystal ball as to the
301
Panel Discussion
future developments in the industry and so on., We will go through each one of
them and after that is finished then we will start the floor discussions. I hope you
will all be very active and aggressive in pursuing these discussions. And I would
like to start this program with Professor Speidel.
M. Speidel: This morning we heard a number of impressive statements by
operators about Stress Corrosion Cracking (SeC) in Steam Turbine Rotors indicating
that they cannot be sure when cracks will occur, where they will occur, and why they
do occur. This is why we go to statistical approaches. They are necessary at this time
because we don't know what determines a crack, the nucleation or the growth.
Parallel to the statistical approach we can also try to learn more about what actually
determines a crack; what causes it and what makes it grow. It would be helpful to
the utilities if they knew where the cracks came from and where to expect them.
And looking at the present knowledge we have in this field, and what has been
presented here, there are several obvious gaps. One is the role of residual stresses
and that of the surface condition after machining when the rotors are put into
service. If we knew more about the surface condition both in terms of stresses and
cold work from machining we could more definitely expect to learn where cracks
come from and why they grow. The surface condition and the residual stresses have
not been addressed in this conference in the depth they deserve. There is another
point I would like to make, and that concerns the chemistry of the environment.
We have heard several times that maybe cracks start at one location or another
because that is where contaminants have access to the material. However, we don't
know which contaminants actually influence this nucleation and growth of stress
corrosion cracks. We need to know the effect of pH and potential and specific ions
on crack nucleation and crack growth. This knowledge, could be acquired in a
rational way by doing systematic studies in the field. Such studies would tell us
whether or not there is an influence of condensing steam. The third point that I
would like to make has to do with the wetness of the stages. I think that in this
room there is enough experience to find out whether stress corrosion cracks can
occur on stages that are dry. Of course, they can occur on stages that are wet. But is
there or is there not experience with stress corrosion cracks in steam turbines in
locations that can be considered dry or can be considered dry most of the time.
Those are my three points.
R. Ortolano: Dr. Speidel makes some very good points. The one that captures my
attention in particular is the possibility of stress corrosion cracking appearing in
stages which normally run dry. We have observed in fossil machines that the
heaviest corrosion pitting attack on the rotor blades is not in the normally wet part
of the turbine; it is in the dry part of the turbine. It is in the first few stages where
the corrodant is allowed to accumulate without any problems and the damage is
done when you shut the machine down. Then you keep it on turning gear and
glancial steam. The glancial steam eventually passes the dew point and then you get
condensation on these stages that had been allowed to collect the corrodant. The
latter stages which normally run wet get a certain amount of scrubbing and as you
look at each stage through the machine, you find this very heavy attack in the
302
Panel Discussion
earlier stages, then decreasing all the way down to the last stage which is usually
pretty clean. If you can get pitting on rotor blades then you can get pitting on the
rotor. It is just as wet as the blades are. I think one of the answers is to turn off your
glancial steam just as soon as you can and dry your machine out. People who
routinely install dehumidifiers and make attempts to dry out their machines are
doing a world of good for the machines because if it is not wet, then it does not get
corroded. .
M. Speidel: What you say is certainly true and a very valid view concerning pitting
corrosion. But pitting corrosion in such an environment where impurities are
allowed to accumulate can be fast while the cracks that we address here,. i.e. SCC, are
probably all very slow. We don't have yet information shOWing that they speed up
with impurity accumulation. And therefore for such cracks to form and to grow to
significant sizes, time is necessary. More than for pitting.
w. McNaughton: One of the items we heard a lot of discussion about yesterday was
the role of oxygen. I would like to hear a little bit more from you on some of the
tests you have run, in particular, the ranges of oxygen over which you did your
growth rate tests and any thoughts that you might have about the role of oxygen on
initiation of SCC, even though I realize that was not the main thrust of your
experiment and investigations.
M. Speidel: As you say, we did not really study initiation, but others in the audience
have and we should ask them for that aspect. As far as growth is concerned, I think
if you look at the combination of the data that we hC:ive presented and Stuart
Holdsworth has presented referring to oxygen, you see that at least within a very
wide range there is no measurable effect of oxygen on the growth rate of stress
corrosion cracks. So far we have not seen any investigation which shows clearly
what, if any, is the role of oxygen other than we find probably none.
J. Takhar: Dr. Speidel refers only to the oxygen effects on the basis of research from
Switzerland. These should be compared with American research. Maybe it is about
time the two research groups get together and resolve the differences. Dr. Speidel
seems to imply that oxygen does not matter.
R. Viswanathan: Dr. Speidel only said that it does not matter to the growth. He
didn't say that about initiation, he was simply referring to other studies. You have
to view stress corrosion initiation and propagation, as well as the final failure.
J. Takhar: Yes; the question of oxygen is yet to be resolved.
R. Viswanathan: I couldn't agree more that we need to do this. I wonder where the
belief that oxygen is detrimental has come about during the last 20 years.
C. DeCesare: This belief was stemming from the Key-Way issue. Starting in 1981,
we started seeing Key-Way cracking in our rotors. Around 1980 we started seeing
303
Panel Discussion
actual Key-Way indications. We had to figure out what the mechanism was for the
cracking. We knew it was stress corrosion cracking and the mechanism we found to
be condensation in the Key-Way. In certain geometries involved, condensation was
able to form more readily; we also noted a difference between BWR and PWR
environments which differ in oxygen contents.
R. Viswanathan: What is currently the OEM view if there is such a thing? Is
oxygen good? Is oxygen bad? Please tell us.
c. DeCesare: In the dovetails, we don't know.
P. Marsh: I would say that Fred Lyle's paper presented yesterday is probably as good a
study of the problem as there is and I would accept Fred Lyle's.work, Le. oxygen is.
detrimental even for crack initiation.
J. McCracken: I have two comments; one on the oxygen. Our belief and our
experience at Westinghouse has been that oxygen is detrimental for both phases, i.e.
initiation and propagation. And secondly, Dr. Speidel's initial question about dry
disc experience. In the Westinghouse fleet there have been several instances of so-
called dry disc cracking where we found Key-Way cracks in discs which are
traditionally thought to be dry. On each of those which we have investigated we
have found that for those particular units there were Moisture Separator Reheater
(MSR) problems. For a significant length of time, they were operating with severely
degraded MSR conditions or MSR out of service completely which made those
particular discs, which typically were dry, operate with steam for long periods of
time. So in our experience we have not seen dry area cracking, unless the MSR was
out of service and the disc was actually wet; or there was some severe chemical
upset which caused severe ingress of caustic or chlorides, leading to corrosion of
another sort. Cracking has always been traced to some moisture or severe chemical
problem. So far.
R. Viswanathan: Is your information on the adverse effects of oxygen on both
initiation and growth, as I heard you say, is it published, is it in the public domain
that we could look at the data? Is it available to this audience?
J. McCracken: I am the wrong guy to ask.
R. Viswanathan: Could you please find out? It seems to be a burning issue. Get back
to either Tom or me; we could disseminate the information. We would appreciate
that very much.
s. Holdsworth: The information I presented yesterday was for both crack growth
and crack initiation and those tests were performed in a pure condensing steam
over a period out to 40 to 50,000 hours. The oxygen concentrations that we achieved
were obtained by adding pure oxygen and we were checking oxygen concentrations
both in the free water and in the bulk condensate, so we were lucky we were
304
Panel Discussion
achieving the levels of oxygen that we wanted and they were either 10 ppb, 100 ppb,
and 10 and 20 ppm. We saw no effect at all of oxygen on crack initiation or growth.
We rationalized this by virtue of the fact we assumed that in the steam chamber
what was happening was that we were getting partition although the bulk steam
had these specific oxygen concentrations in the vapor phase the gas content was
going to be higher than it was in the liquid condensate on the metal surface equally.
I also said yesterday that although conductivity was probably one percent centimeter
in the feed water locally in the condensate fill we got partition and we were getting
higher salt concentration. So in the liquid phase we got higher salt concentration,
lower gas concentration. And the calculations that we performed suggested that the
oxygen concentrations in the liquid phase were actually more or less about the same.
So that because of partition what we were getting in the condense fill at 100 degrees,
say, was not the same variation that we were actually imposing into the feed water.
So there is obviously a lot of conflicting information. But I throw the information
that I presented yesterday as something that is significant enough to take into
account into the debate.
M. Speidel: We just heard a comment from General Electric that they thol:lght there
was a difference between boiling water reactor and pressurized water reactor
behavior with respect to Key-Way cracking, but I feel that in this conference on rim
cracking, we have seen such corrosion cracks in both types of steam generating
equipment. Therefore, I donlt see the difference here between the different quality
of steam.
G. Nowak: I am from the Navajo Generating Station. I don't know if I mentioned
during the presentation that our concern is with a fossil unit. We have recently
gone from all volatile treatment to oxygenated treatment, OT, within the last year
about the same time we are discussing the sec. I hear a lot of conflicting
information or information that I can't resolve with regard to oxygen. Our set point
is 1 ppb with maximum of 150. I understand European utilities have been running
OT for a fair amount of time. With the move towards OT in the US, I notice some
research was done which supports OT. I am curious as to what the European
experience has been with regard to SCC and all the information we are hearing now
about oxygen levels and the formation in growth of sec. Any of the European folks
with some experience here?
C. DeCesare: For dovetails, we are not seeing a difference between BWR and PWR
and that is what I said. For Key-Ways we did see a difference. We continue to see a
difference. We have done a lot of tests and re-tests and we continue to see a
difference with the Key-Ways. With the dovetails we don't see a difference. I know
there was data presented yesterday showing a higher percentage of PWRs that
showed cracking than BWRs. But I think Tom alluded to other issues that you have
to keep in mind when you see that data. The BWR fleet is older and the older rotors
in these units were starting to be replaced earlier due to Key-Way issues. Due to
retirement of those rotors out of the fleet they were no longer in the mix of cracked
rotors. The other issue is that the older rotors were tested earlier with a less
305
Panel Discussion
sensitive test. The test we use now is a 5 megahertz automated test and we are able
to detect things that we didn't detect back in the testing of those rotors. In other
words, because you see higher percentages don't try to draw conclusions on that
alone. You have to keep in mind all the other factors involved. And when we did
that, when we looked at that, we don't see a difference between BWR and PWR for
dovetails.
A. Mosqueda: In the samples that you guys analyzed, have you looked at or
accounted for the occlusions, the impurities of the mill, on the samples that you
have had? I believe that stress corrosion cracking may be initiated at the occlusions
that are on the surface or right below the surface as the metal corrodes.
R. Viswanathan: Another issue I want to discuss is steel cleanliness. We heard
comments which were in apparent contradiction but they- are in fact not. I want to
summarize my understanding of what I think I heard about the behavior of super-
clean steels and clean steels. You have to consider their effects in terms of three
stages of failure, namely, crack initiation, crack propagation and critical crack size.
What I heard people say was that with respect to crack growth, cleanliness made no
difference. With respect to crack initiation, it appears to make a difference under
pitting conditions, for example, at certain strength levels. I think that it is Dr.
Holdsworth's point of view that below a critical strength level pitting was a
mechanism of initiation. In those cases, super-clean steel did fare much better and
delayed initiation. Lastly, don't forget that the critical crack size, which is a function
of the toughness, is profoundly affected by the impurities. The super-clean steels are
known to be immune to long time temper embrittlement in service as well as
during heat treatment. Hence the critical crack size is larger. Thus, if you look at the
overall benefits from all three stages, one would conclude that dean steels are better.
M. Spiedel: There is no doubt that if you have a tougher material you can sustain
larger flaws and therefore the third point of the three you mentioned is certainly
correct. As far as crack growth is concerned, there is general agreement that there is
no effect of impurities in the steel. So the only question then remains whether you
have cracking initiation through pitting or from crevices and through the trenches
that Fred Lyle spoke of. And in one case you might really have those inclusions that
can start pits being important and in another case they might be not at all important.
S. Holdsworth: Obviously the answer depends very much on whether you believe
that cracking comes from pitting. ~ certainly we believe that it does as far as
materials that are exposed to pure oxygen condensing steam in the lowest regime.
There is evidence published from my company which shows that systematic
correlation of inclusions and stress cracking behavior. It is not so hard to believe
that the probability of crack initiation was higher in steels that had the high
proportion of manganous sulfide. In the high strength regime and in the sulfuric
water environments that Fred Lyle was talking about, this is not relevant. However,
in the conditions where cracking does occur from pitting I believe that there is an
306
Panel Discussion
advantage as far as super-clean steels are concerned for initiation and that would be
for steels with strength levels below about 800, 850 MPa.
R. Viswanathan: We have learned there is one Japanese steel company, Japan Steel
Works, who has made 36 super-clean rotors in the 700-1000 MW class. They do this
mainly from a stress corrosion point of view although in some instances to allow
high cross over temperatures in the LP section. There is a lot of interest in Japan
and they have made many super-clean rotors.
T. Khalid: Are there any actual testIng of coupon specimens in the steam paths to get
experience in the actual environment of the machine?
R. Viswanathan: Barry Dooley at EPRI is looking at specific compositions at the
locations where the condensation takes place and he is also talking about some
electric charges arising from the impingement of the high velocity steam that could
lead to electrochemical action in corrosion.
T. McCloskey: We knew going into this meeting that there would be some
controversy regarding the effects of oxygen. I was hoping for a little more resolution
of this issue, but there is actually more disagreement than there is agreement. I
think EPRI is going to have to step in here to try to resolve this since there are many
things going on at EPRI that are hanging on this issue. We may have to go to actual
field testing as TVA mentioned.
A. Manukian: To answer Tony's question, several years ago we (American Electric
Power Co.) installed some coupons inside a General Electric exhaust hood; samples
were made of different materials and they were pre-tensioned. We were going to
watch for stress corrosion cracking, but over time, relaxation was taking place and
the actual samples did not show any significant difference in stress corrosion
cracking. We did not get any conclusive results from the coupon tests in the
exhaust hood.
T. McCloskey: Ten years ago there was work done by EPRI and Southern California
Edison Co. with Ralph Ortolano on blade coatings that were developed. Coupon
samples were tested to show that coatings like nickel cadmium and IVD aluminum
were 'practical alternatives to prevent corrosion fatigue of blades. We probably ought
to do the same testing on rotor materials. There seems to be some encouragement
from some of the OEMs and some of the third party companies who are looking at
coatings regarding sacrificial type coatings. These could be a really practical,
relatively low-cost alternative to stress corrosion cracking as opposed to apparently
what is going on here. We would like to extend the type of work that we did in the
blade area to rotors. We also need to investigate the long-term effects of things like
shot peening .
T. Fox: We had 142 V-bend specimens installed in Rhedondo 7 that were coated
along with a few were uncoated for comparison. Later on we tested some more U-
307
Panel Discussion
bend specimens in a geothermal unit and I would consider that a time compressor.
You can learn in a very short time in a geothermal machine what would take you
years in a regular unit because of the very corrosive environment. It is wet. It even
has some abrasion and there is a lot that could be learned if you really want to do
that. And there are still a lot of people in Southern California and Central California
operating geothermal units that might be just as interested in having those
specimens in there. The specimens can be inserted and removed very quickly in
those machines. You can shut the unit down and take the specimens out almost
immediately. Whereas in the other machines you have to wait a little bit of cool
down time and the accessibility is also more limited.
R. Viswanathan: Considering the fact that they are more aggressive environments,
are they also able to simulate this type of see we are discussing? Do you think it
would be an applicable environment? Would you be able to relate back to real
experience?
T. McCloskey: It would be useful for rapid materials evaluation. If you want to take,
for example, a weldment that is made out of 12 Cr and check it to make sure that
you are not missing something in using it for rotor material it could be easily done.
G. Nowak: I am from the Navajo Generating Station. Our plan is that as we
refurbish rotors, we are taking coupon V-bend samples and putting them in, at an
R-6 extraction location between L-1and L-O. We are thus beginning to set the stage
for gaining information. We will have U-bends in place with the first rotor
refurbishments and then the next set and it sounds from this meeting what we
ought to do is put a couple other V-bends in there, some that are shot peened, some
that are not shot peened representing the weld rnateria!. It is going to take years for
us to see the results of these tests. A favorable factor is the location. We can access
this location by popping off a flange.
D. Amos: Westinghouse has been involved for many more years with some of the
major causes of sec than the academics. There is a lot of disagreement on what
causes SCC and on how to do the testing. Since about 1992 we have focused on how
to repair it for the long-term. No one here yesterday or today has disagreed on the
point that 12 Cr is an improvement. We are therefore focusing on the use of 12% Cr
filler. We have some U-bends being tested right now.
D. Rosario: Speaking for our company (Structural Integrity), we have been involved
with this problem on different levels. We have NDE techniques that we have been
using. We ha"ve been involved on a life assessment level, and also from a weld
repair standpoint. We have been working with Welding Services on weld repairs of
rotors. From life assessment stand point, it appears that we don't have a very good
understanding of the initiation and growth of sec that would support good life
prediction models or algorithms. One of the things that came up from our review of
the data was that a lot of the test data on initiation shows that you can get some very
short initiation time which could be ~ n y w r from 400 hours to a few years, just
308
Panel Discussion
one or two years. And when you compare that with actual experience in the field
we see that initiation typically occurs after 50,000 hours and there are units with a
hundred thousand hours that are inspected and they find some minor cracks.
Obviously, there is a big difference between the laboratory test data and the field
experience. The other concern I have is that the data that has been presented by the
European side has shown that there is a relatively small scatterband in the crack
growth rate and the comment was also made that the wide scatter we were alluding
to was not there. There appears to be a big difference between the U.S. data and the
European data on crack growth. The US data shows scatter of an order of magnitude
on either side of the mean.
c. DeCesare: I would like to share a couple of issues from a GE perspective. We are
making big strides in the NDE area and our goal is to continue with the latest
technology and try to size these things from a depth stand point. As far as life
assessment is concerned 1'd like to make some comments about Monte Carlo
simulations that somebody mentioned yesterday. Tom Alley gave us which was
basically a Monte Carlo simulation for four events. You could see not only the
scatter from each different variable but also interactively they affect the outcome.
The uncertainty in the outcome may be one to two orders of magnitude. You
should not be surprised at that because if somebody tells you there is an order of
magnitude in growth rate or initiation times you should not then be surprised that
there is order of magnitude in the life assessment. What that means is that you are
going to have to use some judgment. We look at about 150 to 200 rotors a year and
we know how to use judgment. I don't think we are to the point yet where you can
hang your hat on some sort of probabilistic analysis. At the risk of sounding
negative, I don't think we are there yet. I think the reason we have been successful
as an industry with the keyway cracking is because we have had a chance to go in,
understand the mechanism, we have had a chance to go in and test and retest
without destroying anything. By repeated testing and sizing we recalibrate our
assumptions on growth rate and initiation. We have units out there up to refueling
outage 8, 16, and they have been subjected to testing all along the way. 'We have not
yet built up a similar experience base with respect to dovetails. I think we have a lot
of confusion. Crack growth rate and crack initiation are a complex function of
environment, design, material. And for people to assert that there is not much
scatter in the data, basing it on tests of like 3 thousand hours in a laboratory or
varying one condition at a time, misses the complexity of the problem. Regarding
shot peening, there is considerable lab data and field data for other concerns as well
as for concerns specifically for sec. Our own experience, in terms of operating hours
is extensive. With dovetail pins, for instance, where we had experience with stress
corrosion and cracking, we employed shot peening in certain cases. In other cases
we were able to just employ a reduction in tensile strength of the material. And
both of those remedies, both of those fixes, you could say, have yielded basically no
cracking. You have to keep in mind that there is actual experience out there. We
have a lot of experience in the field. We can use laboratory data only to sort out
trends, because it does not factor in all of the complex variables.
309
Panel Discussion
T. Alley: Tom Alley with Duke Power Company. I would like to try to bring to light
here the utility perspective because the utilities are the ones living with these
problems right now. The practical question for us is are we going to go out and buy
all new turbines? No, I don't think there is a utility in this room that can afford to
go out and buy all new turbines. Use of 12% Cr steels may be an answer, but it is a
multi-million dollar answer for utilities and it is not going to happen. So I would
like to try to encourage folks right now to look into the practical side of the equation.
We have heard a'lot of discussion about oxygen. Let's assume that elimination of
oxygen would mitigate the problem by reducing the crack growth rate. Is it practical
to believe that we are going to run units with absolutely no oxygen? The answer is
"no". What is really important to us is what is to know the crack growth rate under
the conditions we have to live with out there in the field. I would like to encourage
folks in their research and in their development and in the tools they are
developing to look at ways for us to live with these problems and to continue to
focus on the practical side. The equipment we have out there in the field today is
the equipment that is really of highest concern to us right now. Since we are going
to be living with it for the next few decades. I don't'see anybody buying new units.
We are going to have to maintain the ones we have got. So I would like to conclude
by saying that we should continue to focus on the delivery of solutions. I see the
research out here now with no solution in mind for what we are haVing to do
today. So I would like to try to change focus a little bit and say what pieces of this
puzzle are we missing and now sponsor the research to try to support those pieces
that we are missing.
D. Gandy: Over the last day and a half we have heard quite a number of discussions
related to repair. In particular we heard two papers yesterday morning by GE and by
Westinghouse. In those presentations there were a number of options presented for
repairs ineluding re-machining of the dovetail or re-engineeringIre-machining of
the dovetail area and subsequent shot peening. We have also discussed long-shank
bucket repairs, and class 7 type repairs where you actually go in and put in new
forging, and class 4 repairs. There are a few other areas that I still think need to be
looked at such as the use of 12 Cr alloys ~ some of the higher strength materials.
There needs to be some investigation from the BWR side about the or the potential
for on-site repairs. At present, utilities are not able to send those units to the OEM
for repair. There also needs to be some investigation into the shrunk-on disc repair
certainly in-situ if possible or at least on the turbine deck if possible. The use of
coatings has already been discussed and deserves further investigation. Localized
repairs of individual steeples is not something that is recommended but there are
some techniques and technology that ought to be looked into in that arena.
T. McCloskey: This workshop reminds me of my own experience about 10 or 15
years ago when the fossil industry was faced with the generic problem with solid
particle erosion. EPRI in conjunction with ASME and EEl put together a task force
in which we got together the turbine manufacturers, the boiler manufacturers, EPRI
consultants and others. I think we did a pretty good job of resolving that problem, at
least getting it to the point where it is not a household word like it seems like stress
310
Panel Discussion
corrosion cracking is right now. We may need to follow the same approach here.
EEl, unfortunately, because of the deregulation, is not able to support this activity.
However, EPRI can take a leadership role in trying to resolve the major issues we
have right now before us and define the R&D that is going to be required to resolve
those issues. As Tom Alley points out we have got some potential solutions here,
some of them could be rather cost effective, but we have to do our homework to
make sure these things work and work properly. One of the reasons for convening
this meeting was to see where we are right now and to discuss what further R&D is
required.
R. Ortolano: I am going to take you up on that, Tom. Russ Chapman tells me that in
1998, the rotor that we described this morning will be available to be cut up and do
destructive analysis of the shot peened surfaces that will have been in service for
three years. That should provide you with some material for investigation.
J. Lewis: I am a little concerned after listening to the discussion the last few days. Do
we really know the turbine impact of the high oxygen water treatment system,
particularly on possible once-through with perhaps building blocks 73 from
Westinghouse or potentially susceptible play group areas or bucket root areas from
GE? I am a little concerned that we may be off doing something that is interesting
but is going to prove deleterious later on. During one of the panel discussions, I also
heard that we have been fairly successful in addressing the key-way issues. From a
utility perspective I disagree. We do have a situation that is manageable but the
industry has spent about $200 million managing it. That does not meet my
definition of success.
T. McCloskey: I share your concerns about the oxygen effects. I am not a chemist
and represent more the stress area. Vis represents more the materials area.
Unfortunately my colleague Barry Dooley, our chemist, is not able to be here. I am
sure he would have some very valuable opinions to offer on the oxygen effects.
Although we are going forward with the oxygenated water treatment from the
boiler stand point, we have to resolve once and for all the oxygen effect on the
turbine side. We need to go into the details of the technical work that has been done
both in the United States, Europe and in Asia and reach a consensus. Instead of
examining opinions, we have got to get down to the facts.
R. Viswartathan: Going back to the Key-Way cracking issue, I would like to ask
some of the utilities if they feel that the cracking has been contained or has not been
contained.
J. Takhar: Key-Way is the only place where cracking can lead to catastrophic failure.
Blade cracking is more easily managed; but Key-Way issue is more important
because of the old rotors in our system, which are not going to be replaced. We have
three big Westinghouse units producing power at 101 St. Clair. The discs are bolted.
We share Jim Lewis's concern about the serious consequences of keyway cracking.
311
Panel Discussion
T. Alley: With regard to key-ways, "management" is a very good description of
what we have achieved because we certainly have not solved the problem. It is a
management approach because we still have crack growth. The Fossil discs of
concern to us are oil quenched and have very poor ductility. There is no allowable
crack growth rate, or very little. Unfortunately we have not seen the recurrence of
problems in those machines as an industry so far but should we start seeing that,
that becomes an unmanageable problem. So there is some risk out there on the
horizon with regards to thefossil units.
T. Lam: Our job (at Stress Technologies) is to try to help the utility to manage, to
continue to operate 20, 30-year-old machines with stress corrosion problems.
Probabilistic methods will continue to be used but with refinements. It is a matter of
the completeness of the model and the quality of the statistical data. I think there is a
lot of work that needs to be done and standardize some of the methods and there is
a lot of work that needs to be done in the area.
R. Exner: I have a question for Westinghouse. My question is what work have you
done on shot peening as a way to prolong a disc life? What is your position on shot
peening?
D. Amos: We have done an awful lot of work at Westinghouse on shot peening.
B. Lay: At Westinghouse, our position with shot peening is that if we think it is
going to pit through the compressive layer, it is not a good idea. If you have
evidence it is going to pit then you are really worsening the situation. You also
include a very high tensile stress which could actually worsen the situation if you
pit through to that tensile layer. I have another comment to make as far as David
Gandy's comment about repair of radioactive rotors. We have repaired three BWR
radioactive rotors at our Charlotte facility. We have a license to do it there, and are
planning to do several more in the future.
w. Welsch: We at Metal Improvement Co. are the people that supply you with
compressive stresses in the form of shot peening and it is true you have very high
magnitude compressive stresses. But I have to caution you that the balancing
tensile stresses are relatively low. And when you eat away a layer of compressive
stresses by corrosion the balance of the low magnitude tensile stresses disappear.
The balancing tensHes are only there as long as you have the compressive stresses.
These are things we have learned in the industry.
M. Gull: I am with Nuclear Electric in the UK, one of the four companies that used
to be a CEGB. I would like to spend a few minutes on stress corrosion cracking and
what can we do, what we found how we approached the problem. We probably have
about 80 rotors designed and built in 1950, 1960 that now have 150 to 220,000 hours
of operation. We want to get a hundred thousand hours of some of these units at
relatively minimum cost to the company. We have got lots and lots of experience in
cracking in all kinds of regions on our rotors. We have monoblock LP rotors and
312
Panel Discussion
we have got built-up LP rotors and cracking in Key-Ways. Where we have got them,
plain bores even the central plain bores, dowel holes are put in driver excesses. I
don't think we have any button holes but we have them in 60 percent of disc rims
are cracked so we have got a fairly enormous problem in terms of dealing with it.
There has been a lot of discussion over the last three days about
probabilistic/materialistic type assessments. I agree with you we should be using our
assessments. What you should be doing with the answers you might get, what it all
means. There has also been a lot of discussion about some of the theory, some very
detailed aspects of lab work and the relevance of lab work to real situations. I would
like just to spend a few moments talking about the term probabilistic. We used to
make a great deal probabilistic assessments when we thought we didn't have a
massive problem the standard care of assessment appearing in the documents over
the last 8 to 10 years or so. Outside of the number of defects we found has increased
and our assessment work has continued, we found the probabilistic assessment is
very limited value for us. Principally because you can, according to what judgments
or assumptions you put in what bounds you find on your assessment, you can reach
any answer you want, which is as I say you can reach any answer you want is
virtually useless when you have an awful lot of defects in very many locations, you
end up with a table of safe lives or whatever you want to call them varying between
minus ten years and 400 years, you come to a particular area you are looking at and
you take that to your director where you want to make a SOD-pound investment
decision what the hell is this? We don't know what it means. How did you come
up with your assessment? What are your judgments? It is all very, very confusing.
So over the last three years or so we have been developing probabilistic which we
apply to all areas of our rotors which basically takes all equal fully probabilistic
distributions and there are rumors those distributions you define on the basis of
judgment and, you know, there has been talk about maybe if you are not good
probabilities to single data, that maybe the probabilistic route is not as valuable as it
should be. But in some respects, if you can use your judgment, distribution, it is
then transparent to all exactly what your judgment means and certain distribution
or initiation time or a certain distribution for growth rate and it is there and the
assessment people can see it. People can pick it up and you can actually get
something which you can actually use which we believe gives you fairly reasonably
levels of what the failure rates are and from a utility point of view it is bad enough
to give you a reasonable answer, in five years something which might give you a
slightly better answer cost of x-million pounds worth of research and development.
So that is one point really. I think judgment is extremely valuable but I think there
is room for judgment to give probabilistic assessment as well as materialistic type
assessment. In terms of what we actually do with the data, we generate failure rate
data for our machines considering all of our locations and we.spend a lot of time
trying to decide what we are going to do with these when we get them. Basically
because we put the assessment together to be a reasonable, what we consider are
reasonable, best estimate of failure rates, we can use them for commercial
investment decisions. We can also use them to produce safety ranges to continue in
all machines and we do run machines which are at risk or failure rates which we
call high use total be considered the norm, if there are defects, 3D-millimeter defects
313
Panel Discussion
and you have got to accept some risk unless you are going to spend 30 millions of
pounds doing something about it. So you can establish what your criteria or you can
use your probability assessment to put together something .which directors can
understand and can support your commercial decision-making, your safety cases,
and certainly the company is happy with the way we do things and the way we
present our assessments and it is something which I think all the utilities would do
well to follow in terms of using the probabilistic route. To touch on some of the
technical issues. Shot peening, with a bit of disagreement five, six years now, we
have been machining out cracks, welding disc rims or shrunk-on discs. We are
getting to the point where some of these units are being taken out of service and
new discs, better material being put up, cutting up. Things were bad a few years back
and we got very low expenses reinitiation of these discs at the moment but that is
something we are working on. That is all I want to say, really.
D. Amos: Let me comment on what you have put in your disc in terms of. weld
metal. It is of lower yield strength than what you had on that originally, is that true?
M. Gull: Yes.
D. Amos: I think for years and years Americans and we at Westinghouse have used
very high yield strength materials as opposed to Europeans. I think as sizes have
been enlarged, yield strengths have gone up to accommodate that. And now all of a
sudden you are seeing some of the things that we have seen for years, that is, higher
yield strengths cause more cracking.
T. Khalid: What. we need is the road map to achieve a better detection method for
the dovetail cracking. Right now for TVA, the most critical thing is the dovetail
cracking. And we have to find a better way to detect and size the cracks now. We
have two or three discs out already and they are in the shop now. So we are very,
very desperate to move on.
R. Viswanathan: My concern is that these sort of meetings end up open ended. We
have discussions but no follow-up. We have no plan of action. I would suggest to
the utility guys handling the problem, if we were to invite some of them to form a
committee and have them come up with a plan of action and recommendations,
Tom McCloskey and I would look for funds to carry out the work.
C. DeCesare: We at GE are already committed in the NDE development area to
spend some money this year. Once we improve our crack depth measurement
capability, we should leave cracks in place and periodically size the cracks." Crack
growth rates established this way wou1d be a lot more useful than lab data.
P. Sabourin: I would like to comment that, in fact, blade attachment inspections
pertaining particularly to GE rotors is continuing to evolve. There are a number of
utilities or vendors that currently perform the inspection.. Within the last two
weeks, we brought a tool to the EPRI NDE Center which appears to be rather helpful
314
Panel Discussion
in the detection and we believe it is going to be able to do some excellent sizing. It
has to do with phased array technology. There will be several papers presented on it
at our turbine workshop this spring. We talked to GE last night about perhaps
getting into some kind of a program with them. Of course, that is the kind of
program that may also require some funding as well.
R. Ortolano: I have a concern about the UT methods that have been typically used.
You have an accuracy of about 50 mils. Defects smaller than SO mils you typically
will not .detect. Our company several years ago developed an ultrasonic automatic
tenon inspection system that was able to reduce the detectable defect size down to
about 10 mils. I believe that some of that technology could be transferred to the
rotor inspections and improve the accuracy. Our company's position is that if you
don't mind the concept of replacing your rotor eventually, then there is nothing
wrong with UT inspection. If you really want to get to the bottom, get the defects out
and we have the confidence that you can find the defects with mag particles, but you
have to take the blades off, and you can dress them out and you can shot peen them,
then you stand the chance of putting the problem to bed over the long hauL Are
there any companies represented here that have had stress corrosion cracking
appear in their feed water pump drive trains? That is interesting because they run
on the same steam that the main turbines run on . But they are all integral rotors.
That means they don't have the high strain out at the rim of the wheel that the
built-up rotors have. I can't use our company's experience because we happen to
have 12 Cr rotors for our feed water pump drive turbines.
R. Viswanathan: One of the things you need for the LP rotors is the hardenability.
That is the reason for people are now using 12 Cr and stainless steels. These are
huge rotors several feet in diameter. NiCrMoV is the optimal candidate for through
thickness toughness and that is the reason for using it. What does Mr. Sabourin
think about the need to detect 10 mil cracks and whether we could adapt that for our
purposes?
P. Sabourin: With the grain sizes that you are talking about and the frequencies that
you would have to use to do that, it is nearly impossible to detect something that
small.
T. Lam: In addition to NDE blade vibration monitoring systems may be useful.
Now, when you are running the rotor with a known crack and the crack is going to
propagate, we know that when a crack propagates to some extent the characteristic of
the blade vibration will change, and that can be detected by a monitoring system. I
know Westinghouse had this PBM system and it may be a good idea to put on the L-
2 or L-3 stage of nuclear unit. The financial pay back may be better than to stop the
machine and inspect it or replace it.
T. Khalid: Your technique seems applicable to free standing blades. How about if
you have a group of blades?
315-
Panel Discussion
Voice: That can be handled. Westinghouse has used techniques to get around it.
P. Klein: I am interested in hearing from people on sacrificial coatings and what
their experience has been. Ralph Ortolano mentioned about coatings this morning.
Do you apply the coating outside the crevice area or is the coating applied in the
crevice area? Could the OEMs comment on the on going research in the use of the
coatings and their philosophy on the use of coatings?
R. Ortolano: Although we at SeE have talked at length about coating rotors, the
complications of trying to coat a rotor with any of the processes that we have used
on rotor blades makes rotor coating impractical. Therefore, when we talk about
coating to protect the slots on the rotor, we are talking about putting the coating on
the rotor blade. This means you have to use a sacrificial coating. A barrier coating
on a rotor blade isn't going to help the rotor any. In fact, it may hurt it. And we
have taken some Ni-Cad coated compressor blades off and examined the slots and
were able to determine that you get sacrificial protection for about a 16th of an inch
from the point of contact where the two pieces of metal come in contact. This is
good information because it says that the typical groove that has not been l t ~ r e
will get protection. Now, if you go into these very large cut-aways where your blade
is not near the rotor anymore, you may want to look at putting other sacrificial
material in there to make sure that you get the effect that you are looking for. The
steam balance hole we talked about coating with a paint. We talked about using an
aluminum sleeve that could be swaged on both ends to keep it in place. These were
abandoned as we approached the final decision that the company would replace the
rotors. So there are some things that others can try and I would be certainly willing
to assist in any way I can. .
C. DeCesare: We at GE are spending some money on NDE aspect for depth sizing,
and on coatings development as well. We are looking at both barrier type coatings
and sacrificial coatings and combination of the two. We are also looking at making
that process be applicable on-site. I can't really say when that is going to be finished.
I know it has been in progress for maybe roughly a year and a half. We can provide
more details upon request.
D. Amos: We have a very extensive program on both with some off-the-shelf
coatings and some proprietary coatings. It will still be some time before the results
become available. The work is being funded by our Materials Engineering group in
Orlando.
G. Willard: We have been talking about see a lot. I wonder how prevailing the
problem of corrosion fatigue was in comparison to SeC? do you see corrosion
fatigue at all or is it just a minor problem?
D. Amos: We see that phenomenon more in the boiler feed pump turbines.
316
Panel Discussion
MR. DeCESARE: As far as corrosion fatigue for the dovetails, that is not the main
problem. The issue with corrosion fatigue has been more of concern with low
pressure blades and buckets.
A. Manukian: I have a question for Dennis Amos regarding corrosion fatigue and
boiler feed pump turbine rotors. Any idea about 12 Cr material being better suited
for that application as a cladding compared to original base soft metal?
D. Amos: We would have to look at that on a case-by-case basis. I tell you why. It is
going to be a function of the yield strength of the original rotor. Generally the boiler
feed pump turbine rotors are of lower yield strength, in the range 80-90 ksi, although
occasionally you see some that is over a hundred. We will just have to look at the
test reports of your rotor and the phenomenon of failure and make a
recommendation. In general, we think that the long-term solution for SCCC is
going to be 12 chrome steels.
s. R. Holdsworth: From GEC-Alsthom Point of View
I very much enjoyed this conference, and thank EPR! for generating what has been
an excellent forum for discussion on stress corrosion cracking. It is very easy to
become focused on ones own specific objectives/interests. The Baltimore meeting
has served to place the far wider issues into perspective, in particular those faced by
the utilities.
It was clear that there is still a lot to do as far as see initiation is concerned,
although I believe that there is more knowledge available than was given credit for
by some participants, in particular in the wind-up debate in the final afternoon. I
guess the downside for me is that information was placed on the screen which is
either being missed or simply ignored.
For example, the GEC ALSTHOM paper had initiation data in pure steam out to
almost lOO,OOOh, and yet it was repeatedly said that there was only short term SCCI
data available. Admittedly, the lower bound of the databand for O"app-RPo.2 is a few
thousands of hours, but unfortunately this is a fact of life substantiated by service
experience. Clearly, the lower bounds at lower stresses go out to much longer
durations, but the OEMs who have this data are unlikely to part with such
information, except on a confidential basis to prospective customers.
On the oxygen question, our 40/000h data gave a clear indication of the effect of
oxygen concentration on SCCr. There was no added complication of other
parameters being varied such as the NaCI addition in the case of the 100ppb O
2
SWRI data. However, this information was effectively dismissed as being simply
/I another opinion".
317
Panel Discussion
The main point is that there is still a lot to be done to fully understand the situation.
However, part of that activity must include a "serious and informed" critical
appraisal of the existing knowledge from Europe and Asia, as well as the U.S.
The panel discussions were closed with some concluding remarks from R.
Viswanathan, P. Sabourin and M. Richter.
318
A Short Note on
The Phase Transition Zone of Low Pressure Steam Turbines:
Failure Mechanisms and Oxygen
R.B. Dooley
Superheated steam entering a turbine typically contains low levels of
impurities<l-3). As the steam expands through a turbine, the subsequent
precipitation of impurities onto the surfaces of the steam path components
leads to a variety of problems in the HP, IP or LP turbine. Thus the purity of
boiler water, feedwater and steam is one of the most important criteria for
ensuring the availability and reliability of components in power plants.
By far the most common and serious problems caused by the effect of
impurities, such as corrosion fatigue and stress corrosion cracking, occur in
the low pressure turbine. Historically there has been much confusion over
the last twenty years of the part which oxygen plays in these failure
mechanisms.This has been based almost exclusively on the results of
laboratory testing, and not on an understanding of the exact environment in
the region where the mechanisms take place.
To correct this situation and to address questions with regard to the use of
oxygenated treatment" much work has been performed over the last five years
in measuring the processes of transport" nucleation of droplets, formation of
liquid films, and concentration of impurities. The phase transition zone
(PTZ)" where the expansion and cooling of the steam leads to condensation, is
the area where a number of processes take place, such as precipitation of
chemical compounds from superheated steam, deposition, evaporation" and
drying of liquid films on hot surfaces. These processes lead to the formation
of potentially corrosive surface deposits.
Steam Chemistry and the Phase Transition Zone (PTZ)
Failure of blades and disks of LP turbines ranks in the top five of the
industry's availability problems; when combined with deposition and
performance they emphasize the importance of understanding the
environment and processes within the PTZ (4). International collaborative
work conducted over the last four years (and presented openly for the first
time at the Fifth International Conference on Cycle Chemistry for Fossil
Plants(S has advanced the state-of-knowledge by about 40 years. It is now
possible to determine and monitor, as a function of the inlet steam
composition:
The moisture and droplet size between turbine stages.
The composition of these droplets (first or early condensate).
319
Panel Discussion
The thickness and composition of liquid films on blades.
The composition of deposits formed on blades.
Some of the key results from this work(6,7) include:
All steam chemistries, however pure, produce deposition of particles that
range in size from 300A to 1 ~ m with a typical particle density of over
10
7
/mm
2
The concentration ratio of impurities in the first condensate compared to

steam for sodium, chloride and sulfate typically range from 2 to 150.
The concentration of chloride and sulfate in the first condensate appears
higher with phosphate and all-volatile treatments than with oxygenated
treatment (OT).
Liquid films form on blade surfaces at moisture levels below 1%, and the
thickness appears less with OT than AVT. As mentioned below, oxygen
doesn't concentrate in the condensate, even with up to 200 ppb of oxygen
in the steam inlet.
With this massive increase in knowledge of the environment in the PTZ, the
major challenge now involves developing more representative steam limits
and a better understanding of the failure mechanisms of blades and disks.
Oxygenated Treatment (OT)(2)
The introduction of OT into the U.S. has been a tremendous success story.
Since 1991, more than 90 once-through units have been converted to OT. It
was expected that the penetration for drum units would be slower and .
directed toward specific problems such as flow-accelerated corrosion. The
majority of oxygenated units easily achieve much less than 1 ppb iron at t!le
economizer inlet during normal operation. Only slightly elevated levels
occur during startup periods, compared to many hundreds of ppb when
operating on AVT.
One item with OT is the question of the effect of oxygen in steam on possible
blade failure mechanisms in the LP turbine. The question was raised by US
turbine manufacturers in the late 1980s, when the OT guidelines were first
being assembled. At that time, and as recently as up to two years ago, the
evidence was circumstantial and pragmatic: there had never been, and still
hasn't been, any failure influenced by the steam environment of a turbine
component on a unit operating on QT. Currently it is estimate there are more
than 350 units worldwide on OT. .
Today, however, the situation is drastically different as the result of two
major pieces of research mentioned in the previous section. Work conducted
by an international collaborative group(6), outlined above indicates clearly the
superior performance of OT over all other cycle chemistries in terms of blade
deposition phenomena. Work conducted on a model turbine in Moscow(7),
320
Panel Discussion
has quantified the understanding. Measurements have clearly shown OT to
produce the cleanest phase transition zone environment:
Confirming the theoretical understanding of oxygen distribution between
water and steam, and of Henryis law(8), the early condensate drops that
form 0.1 )lID) contain no detectable oxygen when the inlet steam
contains between 100 and 200 ppb.
The use of OT (pH 8.0, oxygen 100 - 200 ppb) reduces the concentrating
effect of chloride and sulfate in these drops.
The liquid films that form on blade surfaces and provide the environment
for the failure mechanisms are thinner with OT than with other
chemistries (AVT, PT).
Together, these results have started to quantify the excellent turbine
performance with OT. As oxygen in steam now appears to play no part in the
failure mechanisms (corrosion fatigue, stress corrosion cracking), the research
and development focus is moving to other explanations such as the presence
of the positive charge of the liquid films on blade surfaces.
References
1. R.B. Dooley, A. Aschoff and F.]. Pocock, Cycle Chemistry Guidelines for
Fossil Plants: Phosphate Treatment for Drum Units. Electric Power
Research Institute, Palo Alto, California. EPRI Report TR-103665,
December 1994.
2. A. Bursik, R.B. Dooley, D. Gunn, B.A Larkin, 1. Oliker, F.J. Pocock, D.
Ryan, and L.C. Webb, Cycle Chemistry Guidelines for Fossil Plants:
Oxygenated Treatment. Electric Power Research Institute, Palo Alto,
California. EPRI Report TR-I0228.5, December 1994.
3. R.B. Dooley, A. Aschoff and F.J. Pocock, Cycle Chemistry Guidelines for
Fossil Plants: All-Volatile Treatment. Electric Power Research Institute,
Palo Alto, California. EPRI Report TR-I05041, April 1996.
4. Turbine Steam, Chemistry, and Corrosion. Electric Power Research
Institute, Palo Alto, California. EPR! Report TR-I03738, August 1994
5. Fifth International Conference on Fossil Plant Cycle Chemistry.
Charlotte, North Carolina, June 10 - 12, 1997.
6. Steam, Chemistry, and Corrosion in the Phase Transition Zone of Steam
Turbines. Electric Power Research Institute, Palo Alto, California. EPRI
Report TR-I08184, to be published September 1997.
321
Panel Discussion
7. Turbine Steam, Chemistry, and Corrosion: Experimental Turbine Tests,
Electric Power Research Institute, Palo Alto, California. EPRI Report TR-
108185, to be published September 1997.
8. R. Fernandez-Prini, Solubility of Oxygen and Other Gases in Water
under Power Cycle Conditions. In Proc. of International Conference on
the Interaction of Iron-Based Materials with Water and Steam." Edited by
R.B. Dooley and A. Bursik, Electric Power Research Institute, Palo Alto,
California. TR-10210L 1993. pp 2-1 to 2-11.
322
EPRI STEAM TURBINE STRESS
CORROSION CRACKING CONFE'RENCE
Baltimore, Maryland
March 19-20, 1997
ATTENDANCE LIST
Matthew Radcliff
ABB Power Generation
1200 Willis Road
Richmond, VA 23237
804/743-4748
804/743-4721 fax
Robert Scott
ABB Power Generation
5309 Commonwealth Center Parkway
Midlothian, VA 23112 '
804/763-3023
804/763-2183 fax
David Carr
ABB TRC (UK) Division
Unit 26, Brookside Business Park,
Cold Meece
Stone, Staffordshire STI50RZ,
ENGLAND,UK
44 1785 761626
44 1785 761571 fax
Bob Sarber
Allegheny Power
Cabin Hill Drive
Greensburg, PA 15601
412/830-5923
412/838-6464 fax
V. Alex Manukian, P.E.
American Electric Power Service Corp.
Mechanical Engineering Division
1 Riverside Plaza
Columbus, OH 43215-2373
614/223-3229
614/223-3204 fax
Joseph F. Lesiuk
Aptech Engineering Services, Inc.
6925 Oak Brook Way
Cumming, GA 30130
770/781-3756
770/781-3756 fax
Richard G. Clisham
Baltimore, .MD 21226
410/787-5240
410/787-5299 fax
John Ecke
Baltimore, :rvID 21226
410/787-5321
41 OJ787-6623 fax
Robert Guanti
Baltimore, NIl) 21226
410/787-5461
4101787-6408 fax
323
Attendance List
Joe Guntner
Baltimore, 1vID 21226
410/787-6623
410/787-5483 fax
Bob Hardies
Calvert Cliffs
1650 Calvert Cliffs Parkway
Lusby, MD 20657
410/495-6577
410/495-6937 fax
Paul Klein
Baltimore, rvID 21226
410/787-5460
410/787-6408 fax
John C. Limpert
Baltimore, :MD 21226
410/787-5511
410/787-6408 fax
Martin Murphy
Calvert Cliffs
Lusby, !vID 20657
410/495-2544
410/495-6937 fax
Paul Nagle
Baltimore, MD 21226
410/787-5473
410/787-6623 fax
324
Tom Phelan
Baltimore, 21226
410/787-5348
410/787-6623 fax
Jeff Poehler
Calvert Cliffs
Lusby, :MD 20657
410/495-6972
410/495-6937 fax
Mark A. Richter
Baltimore, :MD 21226
410/787-5510
410/787-6408 fax
David Snyder
Baltimore, :rvm 21226
410/787-5111
410/787-5508 fax
Don Wright
Baltimore Gas & Electnc
Baltimore, 21226
410/787-5519
410/787-6408 fax
Robert L. Hadley
Carolina Power & Light
P. O. Box 1551
Raleigh, NC 27602
919/546-4616
919/546-3688 fax
Mark R. Miller
Cleveland Electric Illuminating
Perry Nuclear Power Plant
10 Center Road
Perry, OH 44081
216/280-7523
216/280-8035 fax
Alex Bar
ConEdison
Indian Point Station
Broadway & Bleakley Avenues
Buchanan, NY 10511
914/734-5496
914/739-8098 fax
Neal J. Wikert
CONMEC, Inc.
1480 Valley Center Parkway
Bethlehem, PA 18017
610/758-7433
610/758-7501 fax
James B. Lewis
Consumers Energy
Room P13-313A
1945 W. Parnall Road
Jackson, MI 49201
517/788-5810
517/788-7086 fax
517/788-1115 fax
Warren McNaughton
Cornice Engineering, Inc.
P. O. Box 2542
Durango, CO 81301
970/247-9548
970/247-9577 fax
Attendance List
Paul Hudson
Detroit Edison
Nuclear Generation - Technical
Mail Code 145 AlB
Newport, MI 48166
313/586-5254
313/586-1431 fax
Dr. Richard 1. Lynch
Detroit Edison
2000 Second Avenue
Detroit, :MI 48226-1279
313/897-1369
313/897-0940 fax
Jag S. Takhar
Detroit Edison
Power Generation
2000 Second Avenue
Detroit, l\1I 48226-1279
313/897-0136
313/897-0177 fax
Tom Alley
Duke Power Company
Mail Code Ee tOE
P. O. Box 1006
Charlotte, NC 28201-1006
704/382-9338
704/382-9769 fax
Laurent Capponi
Eleetricite de France (EDF)
Quartier Michelet
13-27 Esplanade Charles de Gaulle
92060 Paris La Defense, FRANCE
33 1 49 02 05 99
33 1 49 02 06 84 fax
Sam Christy
Elliott Company
901 N. Fourth Street
Jeannette, PA 15644
412/527-8451
412/527-8372 fax
325
Attendance List
Tom
EPRI
Palo Alto, CA 95130
415/855-2655
415/855-8759 fax
Dr. R. (Vis) Viswanathan
EPRI
Palo Alto, CA 95130
415/855-2450
415/855-2287 fax
David W. Gandy
EPRI NDE Center
P. O. Box 217097
Charlotte, NC 28221
704/547-6198
704/547-6109 fax
Paul Sabourin
EPRI NOE Center
P. O. Box 217097
Charlotte, NC 28221
704/547-6155
704/547-6168 fax
Harold Presson
EPRI NDE Center
P. O. Box 217097
Charlotte, NC 28221
704/547-6098
704/547-6028 fax
Sanjeev Srivastav, Ph. D.
Failure Analysis Associates
149 Conunonwealth Drive
Menlo Park, CA 94025
415/688-7239
415/328-2990 fax
326
Carl DeCesare
GE
1 River Road
NY 12345
518/385-0376
518/385-2438 fax
Dennis Lessard
GE
Life Extension Services
1 River Road
Schenectady, NY 12345
518/385-4795
518/385-3178 fax
John Nolan
GE
Bldg.2-5B
1 River Road
518/385-8622
518/385-0441 fax
Robert E. Seeley
GE
Bldg. 37-400
1 River Road
518/385-5919
518/385-3896 fax
Pierre Aubry
GEC Alsthom - LCA
141, Rue Rateau
93126 La Courneuve, FRANCE
33 1 49 92 46 20
33 1 49924692 fax
Richard Cox
GEC Alsthom
4 Skyline Drive
Hawthorne, NY 10532
914/345-5158
914/347-5432 fax
Claude Enzer
GEC Alsthom
4 Skyline Drive
Hawthorne, NY 10532
914/345-5159
914/347-5432 fax
Dr. Stuart R. Holdsworth
GEC A1sthom Large Steam Turbines
Willans Works, Newbold Road
Rugby, Warwickshire CV21 2NH,
ENGLAND, UK
441788531138
44 1788 531469 fax
Michael l. Barbaretta
GPU Nuclear
Route 441 South
Middletown, PA 17057
717/948-8416
717/948-8714 fax
Mark Carroll
GPU Nuclear
P. O. Box 388
Forked River,'NJ 08731-0388
609/971-2309
609/9714651 fax
Thomas J. Fox
GPUNuclear
System Materials Laboratory
P. O. Box 15152
Reading, PA 19612-5152
610/375-5149
610/375-5820 fax
Salvatore Simonetti
GPUNuclear
P. O. Box 388
Forked River, NJ 08731-0388
609/971-4688
609/971-4651 fax
Attendance List
Dr. Shigeo Sakurai
Hitachi, Ltd.
Mechaniqal Engineering Research
Laboratory
3-1-1 Saiwai, Hitachi, Ibaraki, 317,
JAPAN
81 294235790
81 294 236959 fax
Sam Dasgupta
Illinois Power
Clinton Power Station
P. O. Box 678
Clinton, IL 61821
217/935-8881 x 3459
217/935-8460 fax
Dr. Ruth Magdowski
Swiss Federal Institute of Technology,
ETH
CH 8092 Zurich, SWITZERLAND
41 1 6322565
41 1 6321101 fax
Swiss Federal Institute of Technology,
ETH
CH 8092 Zurich, SWITZERLAND
41 1 6322590 '
41 1 6321101 fax
Osamu Matsumoto
KOBE Steel, Ltd.
Steel Casting & Forging Plant
2-3-1, Shinhama, Arai-Cho
Tokosago 676, lAPAN
81 794 457196
81 794 457233 fax
327
Attendance List
Michael J. Gull
Magnox Electric
Berkeley Centre
Berkeley, GloucestershireGLI39PB,
ENGLAND
44 1453 810451 x 3216
44 1453 8]294] fax
Robert M. Curran
Mechanical Dynamics & Analysis
152 Tanin Place
Latham, NY 12]]0
518/399-3616
518/399-3929 fax
Denis J. Graves
Mechanical Dynamics & Analysis
Suite B-1
9507 Hull Street Road
Ftichnlond, \fA 23236
804/276-2557
804/276-2241 fax
Win Welsch
Metal Improvement Co., Inc.
472 Barell Avenue
Carlstadt, NJ 07072
201/935-1800
201/935-8670 fax
Gary Willard
Murray Turbomachinery
1106 Washington Street
Burlington, IA 52601
319/753-5431
319/752-1616 fax
Marcos Burotto Forman
Norgener, S.A.
Miraflores 222, Floor 5
Santiago, CHILE
56-2-6326291
56-2-6868810 fax
328
Jay F. Ely
Northeast Utilities
P. O. Box 128
Waterford, CT 06385
860/440-2023
860/440-2025 fax
Richard Stewart
Nuclear Electric
Barnett Way, Barnwood
Gloucester, GL4 3RS ENGLAND, UK
44 1452 654091
44 1452653097 fax
Dr. Diana Vincent
Nuclear Electric
Barnett Way, Barnwood
Gloucester, GL4 3RS ENGLAND, UK
44 1452 652048
44 1452653097 fax
Lloyd Kadden
Ontario Hydro
P. O. Box 4000
Tiverton, Ontario NOG 2TO, CANADA
519/361-2673 x 5058
519/361-7017 fax
Ming Lau
Ontario Hydro (HI3)
700 University Avenue
Toronto, Ontario M5G lX6
CANADA
416/592-5937
416/592-4483 fax
Paul Myers
Ontario Hydro (B06)
P. O. Box 3000
Tiverton, Ontario N2Z166, CANADA
519/361-2673 x 3780
519/361-64]0 fax
Gary Smith
Ontario Hydro
P. O. Box 4000
Tiverton, Ontario NOG 2TD, CANADA
519/361-4794
519/361-4998 fax
Robert Exner
Pacific Gas & Electric
P. O. Box 56
Avila Beach, CA 93424
805/545-4302
805/545-6992 fax
Abel Mosqueda
Pacific Gas & Electric
P. O. Box 56
Avila Beach, CA 93424
805/5453387
805/545-6992 fax
David Barone
PEeD Energy
Mail Code S:MB2-1
1848 Lay Road
Delta, PA 17314
717/456-3031
717/456-4845 fax
Stephen R. Roberts
PEeD Energy Company
Cromby Generating Station
Township Line & Cromby Roads
Phoenixville, PA 19460
610/983-4704
610/983-4841 fax
Stefan Todorovski
PECO Energy Company
Mail Code 63B-3
965 Chesterbrook Blvd.
Wayne, PA 19087-5691
610/640-6431
610/640-6582 fax
Attendance List
Bruce M. Swoyer
Pennsylvania Power & Light
Mail Code A6-2
Two North Ninth Street
Allentown, PA 18101
610/774-7643
610/774-7830 fax
Richard A. Masten
Public Service Electric & Gas
Mail Code H-18
P. O. Box 236
Hancocks Bridge, NJ 08038
609/339-3604
609/339-3416 fax
Dr. Edgar 1. Gunter, Jf.
Rodyn Vibration, Inc.
Suite 223
1932 Arlington Blvd.
Charlottesville, VA 22903-1560
804/296-3175
804/971-2695 fax
John A. Alice
Salt River Project
Mail Code PAB356
P. O. Box 52025
Phoenix, AZ 85072
602/236-5766
602/236-5477 fax
Eugene Nowak
Salt River Project
P. O. Box 850
Page, AZ 86040
520/645-6478
520/645-6234 fax
329
Attendance List
Michael A. Woods
Salt River Project
P. O. Box 850
Page, AZ 86040
520/645-6242
520/645-7298 fax
Donald R. Oehar
Sermatech International, Inc.
507 Industrial Way
Boynton Beach, FL 33426
561/582-6080
561/582-6505 fax
Blaise Sincavage
Sermatech International, Inc.
155 S. Limerick Road
i m e r i ~ PA 19468
610/948-2812
610/948-5246 fax
Walter David
Siemens/KWU
45466 Mulheim/Ruhv
GERMANY
49208 4562020
49 208 4562922 fax
Dr. H. R. (John) Jhansale
Siemens Power Corp.
1040 S. 70th Street
Milwaukee, WI 53214
414/475-3054
414/475-2657 fax
William E. Meade
Siemens Power Corp.
1040 S. 70th Street
Milwaukee, WI 53214-3103
414/475-3606
414/475-2682 fax:
330
Russell Chetwynd
Station Technical
Bldg. D2H
P. O. Box 128
San Clemente, CA 92674-0128
714/368-9703
714/368-6249 fax
Ralph J. Ortolano
Shop Services and Instr. Dept.
7300 Fenwick Lane
Westminster, CA 92683
714/895-0423
714/895-0150 fax
Fred F. Lyle, Jr.
Southwest Research Institute
P. O. Drawer 28510
San Antonio, TX 78240
210/522-2508
210/522-5122 fax
Robert P. Dewey
Stress Technology, Inc-.
Rochester, NY 14623
716/424-2010
716/272-7201 fax
Tony Lam
Rochester, NY 14623
716/424-2010
716/272-7201 fax
Avinash Sarlashkar
Rochester, NY 14623
716/424-2010
716/272-7201 fax
Darryl Rosario
Suite 24
3315 Almaden Expressway
San Jose, CA 95118-1557
408/978-8200
408/978-8964 fax
Clifford H. Wells
Structural Integrity Assoc.
Suite 24
3315 Almaden Expressway
San Jose, CA 951]8
408/978-8200
408/978-8964 fax
Marta Alvaro
Tecnatom, S.A.
Integrity Analysis Group
Avda Montes de Dca, NO.1
San Sebastian de Los Reyes
Madrid, 28709 SPAIN
34 1 6516700
34 1 6541531 fax
Tony Khalid
Mail Code BR-3B
1101 Market Street
Chattanooga, TN 37402-2801
423/751-6996
423/751-8194 fax
Dave Dreier
Toledo Edison
Davis-Besse Station
Mail Stop 1056
550] N. State Route 2
Oak Harbor, OH 43449
419/321-7343
419/249-2340 fax
Attendance List
Shuichi Inagaki
Toshiba Corporation
Heavy Apparatus Eng. Laboratory
2-4, Suehiro-cho
Tsurumi-ku, Yokohama, 230 JAPAN
81 45 5106682
81 45 500 2542 fax
Robert F. Hayes
Turbine Blading, (USA) Inc.
980 Old Colony Road
Meriden, CT 06451
203/238-0] 60
203/238-9407 fax
.Michael T. Milewski
Turbine Blading (USA), Inc.
980 Old Colony Road
Meriden, CT 06451
2031238-0160
203/238-9407 fax
Brian Biackall
WesDyne International
Westmoreland Industrial Park III
6011 Enterprise Drive
Export, PA 15632
412/733-6017
412/733-8127 fax
Dr. Peter G. Marsh
Westinghouse Electric Corp.
Mail Code 303
4400 Alafaya Trail
Orlando, FL 32826-2399
407/281-3458
407/281-2334 fax
331
Attendance List
James (Jay) R. McCracken
Westinghouse Electric Corp.
Mail Code 200
4400 Alafaya Trail
Orlando, FL 32826
407/281-3191
407/2812452 fax
Dennis R. Amos
Westinghouse Turbine
P. O. Box 7002
Charlotte, NC 28241
704/551-5426
704/551-5377 fax
Elizabeth Lay
Westinghouse Turbine
P. O. Box 7002
Charlotte, NC 28241 .
704/551-5162
704/551-5175 fax
332
About EPRI
EPRI creates science and technology
solutions for the global energy and energy
services industry. U.S. electric utilities
established the Electric Power Research
Institute in 1973 as a nonprofit research
consortium for the benefit of utility members,
their customers, and society. Now known
simply as EPRI, the company provides a wide
range of innovative products and services to
more than 1000 energy-related organizations
in 40 countries. EPRI's multidisciplinary team
of scientists and engineers draws on a
worldwide network of technical and business
expertise to help solve today's toughest
energy and environmental problems.
EPRI. Powering Progress
e 1999 Electric Power Research Institute (EPRI), Inc. All
rights reselVed. Electric Power Research Institute and EPRI
are registered selVice marks of the Electric Power Research
Institute. Inc. EPRI. POWERING PROGRESS is a service
mark of the Electric Power Research Institute, Inc.
* Printed on recycled paper in the United States
of America
EPRI 3412 Hillview Avenue. Palo Alto. California 94304 PO Box 10412, Palo Alto, California 94303 USA
800.313.3774 650.855.2121 askepri@epri.comwww.epri.com

Artículos Sobre SCC en Turbinas de Vapor

Încărcat de

Informații document

Descriere originală:

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Artículos Sobre SCC en Turbinas de Vapor

Încărcat de

Drepturi de autor:

Formate disponibile

EPf21

Ten Disk Rotor DesIgn

10000boun:iJor pure lowU2 steam

1.000 '- --'

datahandfor pure lowO

3-3.5o/oNiCrMoV steels, condensing steam, 95C

3-3.5o/oNiCr:MoVsteels, condensing steam, 9SOC, (japp RJJo2

... Cyclic Load en 120-

'0 0 > Static Load 00-

"Clean" steels focus on the reduction/elimination oftrace

FREQUE<NT PROBABLE OCCASIONAL REMOTE IMPROBABLE

The ultrasonic inspection of LP turbine shrunk-on disk bores and

deaerated steam Rm =830M?a

deionized H20, Ap < BOOM?a

in high temperature water

9200 (1f2h)lair/8500 (112h)/oillT (2h)/air

annealing temperature, T, rOC] annealing temperature, T, [DC]

stress corrosion cracking

yield strength, MPa

9200 (1/2h)/air/8500 (1/2h)/oilrr (2h)/air

see of HP steam turbine

Northeast Nuclear Energy Company

CRACK GfV)l!TlI R!\TE UPPEn

The concentration ratio of impurities in the first condensate compared to

S-ar putea să vă placă și