You are on page 1of 9

Overview

Life-Assessment Technology for


Power-Plant Components
R. Viswanathan and S.M. Gehl

The nation's aging infrastructure, and the refining the assessment procedures un- different sets of stress-temperature con-
costs of maintaining it, have heightened in- der each of these levels, with special ditions add up to unity, failure is pre-
terest in technologies that can help predict emphasis on results of work sponsored sumed to occur. Hart} Woodford,3 and
the time to failure of critical components. In by the Electric Power Research Institute Bolton et a1. 4 have demonstrated thatthe
the power-generation industry, where com- (EPR!). A more complete review of the LFR is valid for temperature changes
ponent failure can cause injury, societal extensive body of relevant literature may butnotforstress changes. Ina multiclient
hardships, and economic losses, life-assess- be found in Reference 1. project recently completed at ERA Tech-
ment technology has been a major focus of CALCULATION TECHNIQUES
nology, an extensive body of long-time
research. Because of the high costs of re- data has been generated, with the specific
peated inspections, however, one concern of In the calculational procedures, plant intent of verifying the LFR for varying
researchers is to create assessment techniques records and the time-temperature his- temperature conditions. s Two heats of a
that provide more accurate but less conser- tory of the component are reviewed. The 0.5Cr-0.5Mo-0.25V steel were evaluated:
vative estimates of remaining life. This ar- creep or creep-fatigue life fraction con- a pipe-grade steel with a predominantly
ticle reviews improvements made over the sumed is calculated using assumed ferritic microstructure and relatively
last five years in various life-assessment material properties and damage rules. high ductility (15% elongation) under
methodologies, including calculational tech- This procedure is usually inaccurate due long~term service conditions, designated
niques, nondestructive evaluation, and de- to errors in assumed history, in the ma- MG, and a cast turbine casing steel With
structive testing of samples removed from terial properties, and in the damage rules. a bainitic microstructure and low duc-
installed components. The temperature-history information tility (3.5% elongation), designated FW.
may be refined by supplemental nonde- Specimens from heat MG were sub-
INTRODUCTION
structive or destructive examinations jected to creep damage fractions of 0.25,
Components operating at elevated such as microstructural studies, hardness 0.45,0.63, and 0.82 at 575°C at a stress of
temperatures can fail due to excessive measurements, and oxide-scale mea- 69.5 MPa; specimens from heat FW were
creep deformation, bursting under pres- surements. The uncertainties in material also subjected to similar levels of initial
sure, or cracking. Failure by cracking properties can be reduced by building a creep damage by testing at 550°C and
may further be viewed as consisting of database, particularly with respect to 84.9 MPa. The predamaged samples were
three stages: crack initiation, crack service-retrieved components taking into then subjected to accelerated rupture
propagation, and final failure of the account the environmental effects. The testing at higher temperatures, holding
component once the crack reaches a limitations of the damage rules have the stress at the same level as in the
critical size. Life-assessment techniques also been explored based on extensive initial creep damage tests.
are, therefore, aimed at quantifying uni- laboratory studies. Results of the accelerated stress rup-
form or localized incipient damage prior ture tests on heat MG at 69.5 MPa for the
Life Fraction Rule for Creep
to crack initiation, the rate at which cracks virgin condition and for the predamaged
grow, and the critical crack size that will The Life Fraction Rule (LFR) states conditions are shown in Figure 1. The
lead to final failure. Eventual failure of that at failure target values of remaining life at 575°C
the component can occur either by (1)
based on the LFR are also indicated. If
leakage or rupture at the operating Ltjtr=l the LFR is valid, the temperature vs.
temperature or by rapid brittle fracture where tj is the time spent at a given stress time-to-rupture lines should be parallel
at lower temperatures during start-up I and temperature, and t, is the rupture to the virgin material line and should
shutdown transients. In the latter case, life for the same test conditions. When pass through the respective target val-
the critical size of the crack is defined by the damage fractions incurred under ues. This is clearly not the case; linear
the fracture toughness of the material. extrapolation of the accelerated-test data
Since embrittlement phenomena occur- predicted remaining lives at 575°C in
ring in service can reduce the toughness 700 excess of the target value in every case.
of the material, knowledge of fracture 680
For the 0.82 predamaged samples, the
toughness in the service-exposed con- accelerated-test results grossly over-
dition is necessary. E660 predicted the remaining life compared
Based on worldwide experience dur- !! to the calculated value based on the LFR.
.; 640
ing the last five years, a three-level ap- ! In the case of the brittle heat FW, the
proach for life assessment has evolved. ...~ 620 trend was exactly the reverse of that
In this approach, simple calculational observed for heat MG. With the excep-
600
techniques are used in the first level, tion of the 0.45 predamaged sample, for
followed by nondestructive and de- 580 all other samples, the predictions from
structive tests in the next two levels. The 560 L - ' - -_ _---'_ _ _- ' -_ _----:-'
the accelerated tests fell short of the tar-
focus of research has been mainly to get values calculated from the LFR. A
reduce the uncertainties and hence the Time-to-Rupture (hr) correlation between the expected life
conservatism in each of the levels of Figure 1. Stress-rupture curves for heat MG in based on the LFR and the observed life
assessment. This article is an overview the virgin condition and after prior creep dam- for the two heats (Figure 2) shows that
of advances during the last five years in age to various life fractions. for the brittle material, the actual life is

34 JOM • February 1992


always shorter than the expected life, 100.000
Iiminary results from one heat of 2.2SCr-
while the opposite case obtains for the 1Mo steel (heat BSH) have shown that
ductile material. In either case, the LFR the rupture life in air is increased by
predictions are accurate within a factor increasing the specimen size from 2.5
10.000
of two. .c mm to 9 mm with a further increase of
The difference in behavior between
the brittle and ductile cases is believed to ..~
~
life in argon. The effect is increasingly
pronounced at time-temperature com-
result from differences in damage ~ 1000
binations resulting in large values of the
mechanisms. Prior creep damage in the Larson-Miller parameter (LMP)-the
brittle case consisted of "mechanical o Ductile cast MG different curves converge at low values
• Brinle casl FW
damage" with extensive formation of of the LMP. Limited data available sug-
creep cavities. In the ductile case, dam- lOO~L- __- J______ ~ ______ ~

gest that the rupture curves for 2.5 mm


100 1000 10.000 100.000
age was "metallurgical" damage con- e.pectea Ufe Based on LFR
and 9 mm specimens tested in argon and
sisting of microstructural changes. In for the 25 mm diameter specimens tested
Figure 2. The correlation between actual life
the brittle steel, subsequent testing by in accelerated tests and expected life for two
in air may superimpose, indicating that
temperature acceleration presumably heats of Cr-Mo steels. the argon test results reflect component
resulted in rapid linkup of cavities, lead- behavior. The oxidation effect seems to
ing to premature failure; in the ductile 1.2 ~
saturate for specimens at about 25 mm
steel, on the other hand, temperature o diameter . Heat-to-heat variations in
O-MG
acceleration presumably led to recovery 1.0 ; o e-FW terms of specimen-size-dependent OCF
of some of the metallurgical damage, have also been observed.
resulting in increased life. , The rupture data for various speci-
The correlation between prior ex- g 0.8 men sizes can be plotted as a ratio of the
] I
pended life fraction and the remaining rupture lives using the 2.5 mm diameter
life fraction measured in the accelerated ,; 0.6 specimen tested in air as the reference
Theor.tical
tests shown in Figure 3 may be used as a condition. This ratio, referred to as the
modified basis for LFR. For a ductile
.;.~ OCF for different stress levels, is illus-
E 0.4
material, the LFR may be used for a t! trated in Figure 4. The stress depen-
conservative prediction of remaining life. 0.2
dence of the OCF is not due to the effect
For a brittle material, a value of <1 as of stress per se, but to the time-tempera-
defined by the lower-bound envelope in ture combination (the LMP) for failure
Figure 3 may be used as a conservative 0.2 0.4 O.S 0.8 1.0 1.2 corresponding to the different stress
failure criterion. Expended LH. Fraction levels. The use of OCF values in remain-
For a ductile material, the implica- Figure 3. Expended life fraction VS. remaining- ing-life calculations should yield more
tions of the above results to actual sit- life traction in accelerated test for two heats of realistic and less conservative predic-
uations in which the LFR is invoked are Cr-Mosteel. tion of heavy-section component life.
as follows: When LFR is used to calcu-
NONDESTRUCTIVE
late cumulative damage under condi- (NRIM) databases, are based on test data
TECHNIQUES
tions where the average metal tempera- generated on small specimens with di-
ture increases with service time (e.g., ameters of 6-10 mm. When these data Conventional nondestructive eval-
superheater tubes), failure will always are utilized in remaining-life-fraction uation (NDE) methods fail to detect in-
occur at L t/ tr > 1. Hence, L t/ tr =1 should calculations, conservative results are cipient damage, which can be a precur-
be used as a conservative failure criterion. obtained. In addition, when standard sor to crack initiation and subsequent
The same reasoning may be applied to ASTM specimens of 10 mm or less diam- rapid failure. Several improved NDE
situations where periodic temperature eter extracted from heavy-walled pip- techniques have recently been developed
excursions occur. ing are tested in air for remaining-life for estimating life consumption. These
In all life-fraction calculations, the vir- assessment, the resulting values are likely include strain-monitoring techniques,
gin material behavior (minimum or mean to be conservative. Results of isostress microstructural techniques, hardness-
of the International Organization for rupture tests conducted at elevated based techniques, and oxide-scale mea-
Standardization's [ISO's) Larson-Miller temperatures may need to be corrected surement techniques for superheater /
data} is used as the reference condition for oxidation effects. reheater tubes.
for calculating the tr. In reality, as the An extensive study of oxidation cor-
Strain Monitoring
component softens, the applicable Lar- rection factors (OCFs) for 2.25Cr-lMo
son-Miller curve is continuously chang- steels has been published by Nakashiro Strain measurements are often em-
ing in such a way that the actual cumu- et a1. 6 Oxidation correction factors were ployed to detect creep damage. Gross
lative damage can be more than that developed based on oxidation kinetics
calculated by using the virgin material derived from literature and based on ...-c}--- 34 MPa
curve. Application of the LFR could be loss of diameter actually observed on
improved by taking the softening reac- samples. Using the correction factor, the ~ 3
u.
tions into account. original NRIM Larson-Miller plot based c
.2
Oxidation and Specimen Size
on 6 mm diameter specimens was re-
vised to reflect the behavior of standard i
Correction
10 mm specimens. This study was, how-
8c 2 _---.~-- 69 MPa
o
The effect of oxidation on rupture life ever, confined to tubing applications; .~

:2 103 MPa
is relevant in several aspects of remain- specimen sizes beyond 10 mm were not o
ing-life assessment, particularly for addressed.
heavy-section components with wall An alternate approach for developing
thickness values exceeding 25 mm. Rup- OCFs consisted of rupture testing speci- 12.7 25.4 38.1
ture life databases used in original de- mens of varying size (2.5-50 mm).7 Se- Specimen Thickness (mm)
sign, such as the ISO, ASTM, and Na- lective tests were also carried out on Figure 4. Oxidation correction factors with
tional Research Institute for Metals' 9 mm-diameter specimens in argon. Pre- reference to a 2.5 mm thick specimen .

1992 February • JaM 35


changes such as header swelling and ness) and the Larson-Miller rupture re- They have quantified the effect of stress
rotor bore expansion have been moni- lationships for low-alloy steels. These on the aging process so that by comparing
tored in the past. Frequently, due to correlations enable selection of the ap- the kinetics of hardness change in the
unknown variations in the original di- propriate Larson-Miller plot corre- rotor with that of thermally aged
mensions, changes in dimensions can- sponding to a given hardness level, samples, the local stress can be deter-
not be determined with confidence. Di- which can be used to calculate the re- mined. This value of stress and the known
mensional measurements fail to provide maining life. value of service temperature are used in
indications of highly damaging and lo- Current procedures do not recognize conjunction with the Larson-Miller rup-
calized creep strains such as those in the that the Larson-Miller parameter for ture plot to estimate t, for the material in
heat-affected zones of welds and regions rupture decreases with in-service aging. its current damage state.
of stress concentrations in the base metal. A model that explicitly takes into ac- In contrast, McGuire and Gooch show
Cracking can frequently occur without count the decreased rupture life due to that the magnitude of the stress is unim-
manifest overall strain. Furthermore, the softening has recently been developed portant l6 once a "threshold stress level"
critical strain accumulation preceding for 2.25Cr-1Mo steels by Grunloh and has been exceeded. For all stress levels in
fracture can vary widely with a variety Ryder.l1 The effect of aging on creep the range 70-240 MPa, the hardness
of operational material parameters, and rupture endurance is modeled by the change in the stressed condition was
with stress state. To enable measurement incorporation of the ultimate tensile found to be 21 % higher compared to the
of localized strains, an "off-line" condi- strength at room temperature (UTS) into change in the unstressed condition.
tion surveillance system has been devel- the second-degree polynomial shown in A creep model incorporating struc-
oped. The system uses the replication Equation 2, which was derived by analy- tural degradation as monitored by hard-
principle to evaluate localized strains sis of a large body of data. ness changes has been proposed by Cane
and life consumption. 8 A surface grid is and Bissall.8 By equating the kinetics of
LMP(ruptu,e) = 40,975 + 57(UTS)
scribed at the region of interest and pre- hardness change to the kinetics of in-
served by means of an oxidation-resistant - 5,225(log cr) - 2,450(log cr)2 (2) terparticle spacing changes, the decrease
coating. A hard replica of high stability In this equation, time, temperature, and in the threshold back stress for creep is
is used to duplicate the grid. Biaxial stress are in units of hours, R, and ksi, modeled. Substitution of the threshold
strain assessment is then made by high- respectively. back stress in the Norton creep law (Le.,
resolution measurement of the replicas The effect of aging is taken into ac- f. = Acr n ) yields an expression for sec-
taken at successive plant inspection count by applying an aging correction ondary creep rate f. in terms of hardness
shutdowns. A predictive strain-rate factor to the original UTS, based on changes. By integrating the f. between t
lifetime model approach is used to es- known kinetics of aging for the steel. = 0 and t = t" where t, is the time to failure
tablish "fitness for service." No field Using temperature, applied stress, and defined in terms of the time to reach an
experience with this technique has been starting UTS as input, a computer code arbitrarily chosen critical strain, the re-
reported. calculates the LMP for rupture in the maining rupture life is predicted. The
aged condition (and hence t,) at 1,000 model is currently based on limited data
Hardness-Based Techniques for and involves numerous assumptions that
hour intervals. The incremental damage
Creep Damage Assessment
for each time step, and hence the cumu- can only be justified by further research.
The changes of hardness in low-alloy lative damage tit" can be calculated. Hardness and Low-Cycle
steels as a function of time and tempera- Kimura et al. developed a similar ap-
Fatigue Life
ture have been extensively quantified, proach to calculate creep life consump-
so that hardness changes can be used to tion of CrMoV rotors l2 and casings. J3 Considerable work has been carried
estimate the operating temperature. 9•10 Goto et al. have proposed using hard- outl3-l5 in applying hardness for calcula-
Correlations have also been established ness as a stress indicator in the remain- tion of fatigue-life consumption in
between tensile strength (hence hard- ing-life assessment of CrMoV rotors. 14•l5 CrMoV rotor grooves. It has been ob-
served that low-cycle fatigue damage
10 results in strain softening and can be
measured as a hardness decrease. The
premise, therefore, is that if the fatigue
curve corresponding to the current
hardness (in service) could be defined,
the fatigue-life fraction consumed could
be calculated by entering the appropri-
ate total strain range ~£t versus number
of cycles to crack initiation (Nf ) curve, as
shown in Figure 5. These relationships
have been quantified:
(3)
logkl = 2.59x10- H v -0.9
3

k2 = 22.0 (4)
(Xl = 1.64 X 10-4 Hv - 0.09
(Xz =0.59 (5)
Hv = 203
cryd = 1.07Hv- 40.6 (6)
where kl and (Xl are hardness-dependent
0.1 ~----------~~------------~----------~------------~ constants, and cr d is the cyclic yield
10 strength. For any given hardness of the
Number of Cycles to Crack Initiation steel, Equations 3-6 can be used to calcu-
Figure 5. Estimation of low-cycle fatigue properties by hardness for a Cr-Mo-V rotor forging at late fatigue-life fraction consumed un-
566°C. H, = Vickers hardness number. der different transients, and these frac-

36 JOM • February 1992


tions can be summed up to calculate the
total fatigue-life expenditures using the
procedures detailed in Reference 12. D 0 0 fiO 0
Creep Cavitation Model
Metallographic methods have been c:
C 0 fi 0000
0
developed that can correlate either cavi- =
~
tation evolution or changes in carbide
spacings with creep-life expenditure. It ~
C3
B 1000 o om
has been observed that, in boiler piping, Q)
CD

cavitation is the principal damage mech- '"f;l 1- Undamaged


anismat brittle zones, weld heat-affected 0
A 0 crrmO((X) 0 A - Isolated Damage
zones, and high-stress regions in the base B - Oriented Cavities
metal.!? In the other cases, carbide C - Microcracks
coarsening was found to be a better in- 0 0 D- Macrocracks
dicator of life consumption. A pre-
liminary model based on carbide coars-
ening has also been developed .9
Neubauer and Wedel characterized o 0.2 0.4 0.6 0.8
cavity evolution in steels at four stages- Expended Life Fraction, til,
isolated ca vities, oriented cavities, linked Figure 6. The correlation between damage classification and expended creep-life fraction for
cavities (microcracks), and macro- 1.2SCr-O.SMo steels.
cracks-and formulated recommenda-
tions corresponding to the four stages of each class (see Figure 6). The Wedel- increased inspection intervals. Several
cavitation.18 To provide a theoretical and Neubauer classifications of material utilities that have adopted the approach
quantitative basis for cavity evolution, condition (undamaged, class A, class B, have realized significant savings in
Cane et al. used a constrained-cavity- class C, and class D) correspond roughly inspection costS. 21 An illustration of the
growth model and proposed a relation to expended-life-fraction (tit,> values of inspection intervals according to the old
between the number fraction of cavi- 0.27,0.46,0.65,0.84, and 1, respectively, and the new procedures for a plant that
tated boundaries ("An parameter) and using the conservative lower-bound has been in service for 30 years is shown
the life fraction consumed using heat- curve shown in Figure 6. Since the ex- in Table I.
specific constants. 19 Values of these pended life fraction corresponding to The field replication data forming the
constants either had to be assumed or various damage classifications is known, basis of the Wedel-Neubauer recom-
determined experimentally for each heat, the remaining life can be calculated us- mendations have recently been re-
thus restricting the usefulness of the ing the relationship viewed. 22 When these limited data were
model. Based on interrupted-creep tests analyzed, no consistent trends of cavita-
on simulated heat-affected zone 1Cr- tion evolution with operating time or
O.5Mo steels, Ellis et a1. concluded that t rem =:t('!""-l)
t
(8) with calculated creep life fraction con-
the data had too much scatter to verify sumed could be discerned. The labora-
the life-prediction model of Cane. The where t is the service life expended. The tory data generated and analyzed by
data could nevertheless be used empiri- remaining life corresponding to various Ellis et a1. appear to be the only available
cally, by plotting all the data together in damage classifications can be calculated quantitative basis for setting inspection
the form of a scatterband whose lower to be 2.7t, 1.17t, O.54t, and 0.19t for un- intervals.
limits are defined bylO damaged material, isolated cavitation, The progress of creep cavitation as a
oriented cavitation, and microcracking, function of creep life fraction expended
A=: 0.517(t/t,> - 0.816 (7)
respectively. By applying a safety factor in CrMoV rotor steels has been investi-
In using the A parameter method, the of three to the above values, the safe
specific procedure used to measure A is reinspection intervals can be established 30
crucial. The A parameter is defined as to be 0.9t, 0.4t, 0.18t, and O.06t, respec- - - APTECH/EPRI scheme

25 - - ~c~~: Neubauer
the number fraction of cavitating grain tively. This approach has been devel-
boundaries encountered in a line paral- oped and applied by Paterson as shown ~ 20
lel to the direction of maximum princi- in Figure 7. 20 The figure clearly shows ~
.~
pal stress. To measure A reproducibly, that the reinspection interval is not only 15

the procedure needs to be standardized. a function of the current damage level, I 10


Since Equation 7 was developed us- but also the service life elapsed, unlike in ~
ing two steels of widely different impu- the Wedel and Neubauer approach,
rity concentrations, it is anticipated that where the reinspection interval pre- o~~~~~
o 5 10 15 20 25 30 35 40 45
the scatterband encompasses most 1Cr- scribed is independent of elapsed ser- Service life Expended (yr)

O.5Mo steels in service. Limited data on vice life. For all combinations of damage Figure 7. Service life VS. reinspection interval
heat-affected zone samples from 2.25Cr- classifica tion and service life in regime B for the APTECH/EPRI remaining useful life
1Mo steels also have shown that the in the figure, the new approach results in method and the Wedel-Neubauer method.
cavitation behavior can be described by
Equation 7. Extension of this correlation Table I. Suggested Reinspection Intervals for a Plant with 30 Years of Prior Service
to weld metal, to steels with fine grain Inspection Interval (Years)
size, and to other steels, however, would
Damage Classification Wedel-Neubauer EPRI-APTECH
lead to errors.
An alternate, easier-to-use method has Undamaged 5 27
also been proposed by Ellis et al. The A. Isolated Cavities 3 12
damage classifications have been corre- B. Oriented Cavities 1.5 5.4
C. Linked Cavities (Microcracks) 0.5 1.8
lated with life fractions, and thus a life- D. Macrocracks Repair immediately Based on fracture mechanics
fraction range has been established for

1992 February • JOM 37


o carbides at the grain boundaries. The
carbides act as sites for crack nucleation,
promoting transgranular cleavage cracks
at low temperatures. Carbide formation
C
also results in softening of the steel since
it depletes the matrix of the strengthen-
ing precipitate M02C and the molybde-
g> B num in solid solution. The softening re-
"16 action is also manifested as decreased
a: creep strength and rupture strength.
CD
16' Carbide evolution in 2.25Cr-1Mo steel
fij
C A has been investigated by Stevens et al.
1- Undamaged and Munson. 28-30 In both studies, the
A - Isolated Cavities amount of M6C carbides as a percentage
B- Oriented Cavities of the total weight percent of carbides
C - Microcracks increased with time and temperature.
o- Macrocracks Plotted in terms of a Larson-Miller time-
temperature parameter, the results (Fig-
ure 9) show the family of curves from the
Stevens and Flewitt study to be shifted
o 0.2 0.4 0.6 0.8 1.0 laterally by a significant amount com-
pared to that of Munson.
Creep Life Fraction, Vt, These studies suggest that although
Figure 8. The correlation between damage classification and creep-life fraction for CrMoV rotor the evolution of M6C may be used as a
steels. qualitative index of service temperature,
wide variations may occur due to dif-
gated by Goto. Additional data by curve. The life fractions consumed are ferences in initial microstructure and
Carlton et al. show wide variations in estimated to be 0.3, 0.6, 0.85, and 1, re- composition. Data in Figure 9 also show
behavior due to differences in grain size, spectively, for undamaged and classes that the kinetics of M6C formation are
ductility, and impurity contentsP The A, B, and C damage. These values are considerably accelerated by phosphorus.
results of Goto agree reasonably well larger compared to CrMo steels, indicat- Improved procedures for carbide ex-
with those of McGuire and Gooch for ing that for a given damage classifica- traction as well as a larger database on
expended life fractions up to about 0.75, tion, life consumption is higher in CrMoV samples with various initial composi-
but show larger values of A at higher life steels, compared to CrMo steels. Blade tions and heat treatments are needed
fractions. Comparison of the behavior of root fixings after blade removal may be before the microstructural techniques can
CrMoV rotor steels with CrMo piping amenable to the use of replication, and be used for plant-life assessment. Stevens
steels suggests that the CrMo steels ca vi- cavitation has been observed in rotors at and Flewitt also found that the evolution
tate more readily and reach much larger these locations. More complex equip- of the M6C phase was unaffected by
values of the A parameter at failure com- ment required for replication of rotor stress. Nakatani et al. have developed
pared to the CrMoV steels. It has also bores has also been developed. 14.15 correlations between the percent M6C
been noted that in CrMoV steels, cavi- However, creep cavitation in rotor bores and rupture-life reduction so that the
tation is observed only at the end of has not been reported to date. percent M6C in a service-exposed com-
second-stage and beginning of third- ponentcould be used to deduce the creep-
stage creep, unlike in CrMo steels where Analysis of Carbides life fraction consumed.31
cavitation occurs during the second- Askins et al. have attempted to utilize The evolution of carbide phases in
stage creep.24 The relative insensitivity parameters such as matrix-solute com- CrMoV rotor steels has been investi-
of the A parameter to expended creep positions, ratios of carbide phases, and gated on a single heat aged at 650 a C by
life for rotor steels makes accurate matrix-lattice parameters as quantitative McGuire and Gooch. 16 The percentages
determination of the A parameter cru- indices of temperature during aging of of M23C6, M7Cy and VC were found to
cial. For this reason, the damage classifi- 1Cr-0.5Mo steels.9 Their data indicate a increase appreciably with aging, with a
cation as a function of expended life is rapid initial decline in matrix-solute decrease in M 3C. Most significantly, the
shown in Figure 8 using the lower bound content and an increase in theM23C6/M3C average Fe:Cr ratio in the carbides de-
ratio with time and temperature of ag- creased systematically with aging. Ex-
a 0 .04 'loP } ing. Changes in the matrix-solute con- pansion of this database to more heats
60 t:,. 0.02%P
• 0.005 %P Stevens and Flewllt tent were also found to be reflected in and aging conditions could open the
50
® Aged under stress changes in lattice-parameter values. Al- door to the widespread use of the Fe:Cr
o Munson
though these general trends could be ratio in carbides as an index of degrada-
40 confirmed, excessive scatter in the re- tion in these steels.
sults did not permit quantitative corre-
DESTRUCTIVE TESTS
lation of any of these parameters with
aging history. Destructive tests provide a direct
20 In 2.25Cr-1 Mo steels, there have been measure of the current damage state of
several investigations showing that pro- the material. They are the last resort in
10 longed exposure at about 280 a C causes the assessment route, since the tests re-
an irreversible form of embrittlement quire removal of samples from operat-
o~-J __ oL~~ __J -__ ~~~-J
known as "carbide-induced embrittle- ing components. There are limitations
34 35 36 37 38 39 40 41
LMP = (T + 460 ) (20 + logl ) ment."25-27 The embrittlement is mani- on the number of available samples and
Figure 9. Evolution of MsC in 2.2SCr-1 Mosteel fested as an increase in the fracture-ap- locations from which they can be taken.
as a function of aging time and temperature. pearance transition temperature (FATT) Periodic assessment of the remaining
(In the equation for LMP, temperature is in and is generally associated with the life is not possible. The costs of cutting
degrees Fahrenheit and time is in hours.) formation of M6C molybdenum-rich out material, machining samples, and

38 JOM • February 1992


conducting tests can add up to a signifi- fected zones in welds, and fusion lines in
Creep Crack Growth
cant expenditure. These costs are further welds may fit under the "brittle" cat-
compounded by the plant outage dur- egory, while the base metal in forgings, Extensive creep crack growth data
ing the extended period of evaluation piping, and tubes may fitin the "ductile" pertaining to CrMo piping steels and
and decision making. Despite these category. These observations are consis- CrMoV rotor steels have been collected,
limitations, removal of core plug samples tent with the recent results of Masuyama, analyzed, and consolidated.38,39 It has
from headers and ring samples from Nishimura, and Igari, indicating differ- been observed that a crack-tip driving
bore or periphery of rotors is becoming ences in behavior between ductile base force parameter termed Ct , which takes
increasingly common. Destructive test- metal and brittle heat-affected zones in time-dependent creep deformation into
ing generally includes tensile, impact, 2.25 Cr-1 Mo steePS They further ob- account, correlates much better with
and creep! stress-rupture tests. Since the served thatin brittle heat-affected zones, crack growth rates (da! dt) than the
purpose and limitations of the tensile both stress-accelerated and temperature- traditionally used elastic stress intensity
and impact tests are obvious, only the accelerated tests gave equivalent results factor K. The relation between da! dt
creep life evaluations are described here. that closely conformed to the LFR, while and Ct can be expressed as
in the ductile base metal, only the tem-
Creep Tests
perature-accelerated isostress test results da = bCrn
A procedure for estimating the cur- were in agreement with the LFR. (9)
dt t
rent damage state by conducting stepped Since isostress tests are normally con-
creep tests using a single specimen has ducted in air, oxidation correction fac- To perform a remaining-life as-
been described in detail by Voorhees. 32 tors are also important. The extrapolated sessment of a component under creep-
The procedure defines which stage of remaining-life values obtained from the crack-growth conditions, two principal
creep the material is in under service accelerated tests should be corrected for ingredients are needed: an appropriate
conditions but does not provide esti- oxidation effects. If tests are conducted expression for relating the driving force
mates of remaining life. The Monkman- in inert environments, no correction fac- Ct to the nominal stress, crack size, mate-
Grant relationship for steels has also tor would be needed. A further refine- rial constants, and geometry of the com-
been in wide use and enables estimation ment to the technique consists of using ponent being analyzed; and a correla-
of remaining life by determining the sec- miniature specimens to minimize sample tion between this driving force and the
ond-stage creep rate from creep tests. requirements. It has been demonstrated crack-growth rate in the material, which
Improvements to this procedure have that miniature specimens (10 mm long, 3 has been established on the basis of prior
also been suggested by Castillo and mm diameter) tested in argon yield re- data or by laboratory testing of samples
KhOUP3 The most common procedure, sults representative of large specimens from the component. Once these two
however, continues to be the use of tested in air. 36,37 ingredients are available, they can be
isostress rupture tests. combined to derive the crack size as a
CRACK·GROWTH STUDIES
The isostress rupture test procedure function of time, The general methodol-
consists of conducting accelerated rup- All of the techniques described so far ogy for doing this is illustrated below,
ture tests at a temperature above the relate to life prediction from a crack- assuming Ct to be the driving crack-tip
service temperature. The stress is kept as initiation point of view. For heavy-wall parameter.
close as possible to the service stress components, the initiation criteria have The general expression for Ct given in
value. The results are plotted as log tr to be combined with crack-growth data Equation 9 essentially reduces to the
versus T and extrapolated linearly to the to perform a fracture mechanics analysis form
service temperature to estimate the re- of remaining life. Fatigue crack growth
maining life. Results from recent research analysis procedures are well established. Ct = at(A,n)aH (geometry,n) (10)
have identified the limitations of and For creep crack growth and creep-fa- where a is the stress far from the crack
possible refinements to the technique. tigue crack growth, however, method- tip, obtained by stress analysis; t is the
Results described earlier suggest that ologies and data needed for analysis strain rate farfrom the crack tip, which is
linear extrapolation of results from tem- have been gathered only during the last a function of the constants A and n in the
perature-accelerated tests may yield few years. Norton relation; a is the crack depth
optimistic predictions of remaining life
under service conditions in the case of
ductile materials; the reverse would be 1.25Cr·0.5Mo Steel, 538'C
true for brittle materials. In either case,

--------
12 Cold Region Properties
the predicted values are expected to be
--,
Hot Region Properties
in agreement with the actual value within
a factor of ±2. 10 ""
This conclusion is in agreement with E "" ""
.s "" ""
that of Melton. 34 Unfortunately, it is "", ""
", , ", , ,
Q)

generally difficult to define a "brittle"


N
8
,,
iJj

and "ductile" class of materials since


~

~ ,,
ductility can vary from heat to heat and <.)
,, ,,
with the test conditions. For our pur-
(ij
.+=
:s
6
,, ,,
poses, we can define a "brittle" steel as ,, ,,
one prone to creep damage by cavita-
4 ,,
p = 13.79 MPa', ,, ,,
tion, and a ductile material as one in
,, ,,
which creep damage occurs primarily
by softening. This is consistent with the
2 ,, ,,
observation by the ERA researchers that
the ductile cast MG showed no evidence 2 4 6 8 10' 2 4 2 4 6 8
of cavitation even at a life-fraction con-
sumption of 0.82, while the brittle cast Remaining Life (h)
FW cavitated profusely. Using this defi- Figure 10. Remaining life as a function of initial crack size for an internally pressurized cylinder,
nition, castings, coarse-grained heat-af- illustrating a typical output from crack-growth analysis.

1992 February • JOM 39


obtained from NDE measurements; and ria and inspection intervals. certainty in the remaining-life
H is a tabulated function of geometry Ainsworth et al. have recently de- analysis.
and the creep exponent n. The values of scribed a unified approach for struc- • Crack growth rates at welds, fusion
A and n are either assumed from prior tures containing defects.4o This ap- lines, and heat-affected zone mate-
data or generated by creep testing of proach incorporates structural failure by rials are at least a factor offive higher
samples. By assembling all the constants rupture, incubation behavior preceding compared to base metal.
needed, thevalueofC, can be calculated. crack growth, and creep crack growth in • The presence of localized chains of
Once Ct is known, it can be correlated a single framework. Service life is gov- inclusion, further assisted by segre-
to the crack-growth rate through the erned by a combination of time to rup- gation of impurities to interfaces
constants band m in Equation 9. Com- ture, time of incubation, and time of such as grain boundaries and fu-
bining Equations 9 and 10 provides a crack growth. All of these quantities are sion lines, causes significant in-
first-order differential equation for crack calculated using a reference stress that is creases in creep crack growth rates.
depth (a) as a function of time (t). Theo- specifically applicable to the geometry • The presence of large amounts of
retically, this equation can be solved by of the component and is derived analyti- impurities in the steel accelerates
separating variables and integrating. cally or based on scale-model tests. If the crack growth by increasing m.
However, the procedure is complicated desired service life exceeds the calcu- • All material and experimental vari-
by the time-dependency of Ct and the lated rupture time, retirement may be ables that reduce creep ductility re-
crack size dependence of the term H in necessary. sult in higher crack growth rates.
Equation 10. To circumvent this, crack- In the opposite situation, further • Temperature can have mixed effects
growth calculations are performed with analysis is carried out to calculate the on crack growth. In cases where the
the current values of a and the corre- incubation time during which no crack effect of temperature is merely to
sponding values of da/ d t (or el), to deter- growth is expected to occur. If the cal- increase creep rate, the da/ dt in-
mine the time increment required for culation indicates that the incubation creases with increase in tempera-
incrementing the crack size by a small time ti is less than the desired service life, ture due to increase in Ct' On the
amount fia (Le., f1t = fia/ a). This pro- then a crack-growth analysis is per- other hand, if a transition from a
vides new values of a, t, and Ct, and the formed to calculate the crack-growth life brittle to ductile condition is in-
process is then repeated. When the value t . If the total life, t. + t , is less than the volved, increase in temperature may
of a reaches the critical size ac as defined desired service lif~, sife operation be- actually decrease the crack growth
by KIe , JIe, wall thickness, remaining yond that point would be considered rates.
ligament thickness, or any other appro- undesirable. This approach seems prom- • Crack-tip constraint has a pro-
priate failure parameter, failure is ising and deserves further exploration. nounced effect on crack growth.
deemed to have occurred. A number of variables affect the crack Assumptions regarding plane-stress
Although this procedure appears growth rate by modifying b, Ct, or m. The or plane-strain conditions can have
complex at first sight, the calculations effect of these variables is summarized a pronounced effect on da/ dt.
are relatively easy, once the principles below. 38,39 • Inclusion of primary creep, in addi-
are understood. Computer programs • In-service degradation increases tion to the secondary creep in calcu-
have been developed to perform the en- crack growth rate. In ductile CrMo lating C" results in larger value of
tire analysis on personal computers. steel base metal, the effect was at- da/ dt and reduced remaining life.
Among these, the Steamline Inspection tributable to increased Ct, while in
Code and PCPIPE are in wide use in the Creep-Fatigue Crack Growth
brittle CrMoV rotor steels, the effect
United States. The only judgement in- was attributable to an increased Major advances have been made in
volved is in selecting proper values for value of m. Widely divergent esti- developing methodologies and data
the constants A, n, b, and m, because mates of remaining life arise due to needed to treat crack growth under the
large scatter in creep and crack-growth uncertainties in the crack growth combined effects of creep and fatigue at
data necessitates subjective choices. If coefficients band m and in the elevated temperatures. The loading con-
actual creep and/or crack-growth tests Norton law coefficients A and n. ditions in elevated-temperature power
could be performed, more accurate re- Determination of component-spe- generation components can often be
sults could be obtained. Even if only the cific A and n by miniature specimen simply represented by a trapezoidal
values of A and n in the Norton law creep tests on samples, and deter- wave shape consisting of a loading pe-
could be pinned down by running creep mination of band m by creep crack riod, a hold time and an unloading pe-
tests in argon using the miniature growth, can greatly reduce the un- riod. Typically, the hold period is
samples described earlier, uncertainties significantly larger than the loading and
could be reduced. A sample output may unloading periods. For predicting crack
be in the form of crack depth versus time O-CCG growth during such loading, it is con-
or a plot of crack size vs. remaining life, /::,. - CFCG, 98 s Hold venient to divide the crack growth dur-
as illustrated in Figure 10. This plot was
o - CFCG, 10 min. Hold ing each cycle (da/ dN) into two parts.
10-'
generated for a thick-wall cylinder un- The first part is due to loading and un-
der internal pressure containing a longi- loading, (da/ dN)o' which is determined
tudinal crack. The outside radius and by the magnitude of fiK, the cyclic stress
wall thickness of the cylinder were as- intensity parameter. The second part is
sumed to be 45.7 cm and 7.62 cm, re- the crack growth during hold time, which
spectively, and the hoop stresses were is also called the time-dependent crack
calculated for internal pressures of 8.96 growth, (da/ dN)t' If the hold time (t h) is
and 13.79 MPa. Materials properties in considerably less than the transition time
the degraded condition (hot region) as (tT ), small-scale creep conditions prevail
well as in the undegraded condition were in the specimen. 41 ,42 Therefore, it is logi-
considered. The results show that the 10-2 10-' 10° 10' 102 cal to attempt to correlate the crack
remaining life is a function of the stresses (Ct)ayg [kJ/(m2. h)j growth rate during this period with Ct'
as well as of prior degradation. Plots of Figure 11 . A comparison of creep crack growth However, it must be recognized that the
this type could be used to determine (CCG) and creep-fatigue crack growth (CFCG) magnitude of Ctchanges during the hold
remaining life or to set inspection crite- data for 1.2SCr-O.SMo steel at S38°C. period as stress relaxation due to creep

40 JaM • February 1992


800
defonnation proceeds. This implies that • GErotorCOl4lb.ilrs
(lj) ' 00 tors and laboratory aged samples.47 These
""
Al olh«~;'ic.Jllio\15.lIeeRn. 41

the time rate of crack growth must also 700 00 grade _orors
Degrade rom's
samples had been temper embrittled to
decrease with increasing time during 600
[jl)""
GIlCiil tE varying degrees, the worst embrittlement
JOO

the hold period. 500


occurring in the exposure temperature
(j)
Saxena and coworkers4~5 have sug- @ range of 370-425°C. The FAIT in the
gested that the average time-dependent
. 00
0
"" ® 200
embrittled condition could be correlated
crackgrowthrate,(da/dt)av ,duringhold ''''' ~ . 08
•••n~~
with the phosphorus content of the steel,
time can be correlated wit~ the average 200 @ •
0(j)®
~@ .
100 as shown in Figure 12. The D-grade rotor
value of C" (C,)avs: during the hold time. samples, which had been austenitized at
These are definea as 1,01 DoC and therefore had a coarser grain
oL-~-L-L~

o
_ _~~-L~~~
size compared to the C-grade rotor
(da / dt>.vg = 50 100 150 200 250 300
Poosptlorus COMlent (ppm)
350 400 450 500

samples, exhibited slightly higher ten-


1
- [da/dN-(da/dN)ol (11) Figure 12. The variation of post-exposure dency to embrittle. Within limits of scat-
th FATT (corrected for exposure temperature ter, the post-service FAIT could be esti-
and and location variations) with phosphorous. mated based on a knowledge of the
f
1 '.
(C, )avg =-- c, dt
th 0
(12) mum stress considerably higher than
phosphorus content alone.
Small Punch Testing
the stress during hold. As yet, no meth-
In test specimens, (C,)avg can be deter- ods are available for considering such In the small punch test, a finished
mined from the measurements of load- loading. Analytical expressions for esti- small punch (SP) specimen is placed
line deflection as a function of time dur- mating (C,).v need further numerical between a 2.54 mm diameter hemi-
ing the hold period. The value of (C,).vg verification. Xlso, new expressions that spherical punch and a die. The punch
can be estimated by the following consider primary creep deformation are displacement rate is controlled during
equation: needed. The crack growth data for veri- the test, and a load versus deflection
fication of these concepts are currently curve is thus developed; the area under
(C) = ~(8V) (F'/F) (13) limited to a single specimen geometry. the curve denotes the energy absorbed
" vg BWt h Therefore, data on a different specimen during the test. By conducting tests at
geometry must be obtained. different temperatures, a curve of ab-
where 8 Vc is the change in load-line de- sorbed energy versus temperature can
NONDESTRUCTIVE
flection during the hold period due to be developed thatis very similar in shape
EVALUATION OF TOUGHNESS
creep deformation, 8P is the applied cy- to that of a Charpy energy versus tem-
clic load, Band Ware the thickness and In heavy-section components, severe perature curve, with the difference that
width of the specimen, respectively, F = thennal stresses occurring during start- the SP test curve is shifted laterally to
K- a calibration factor = (K/P)BWI /2, and up/shutdown transients can occasion- lower temperatures along the tempera-
F' =dF/d(a/W). ally cause failures by rapid brittle frac- ture axis compared to the Charpy curve.
Figure 11 shows a plot of (da/dt),vg ture at relatively low temperatures. The The midpoint of the energy curve can be
with (C,),vg for 1.25Cr-0.5Mo steels for crucial last step in remaining-life analy- used to define a ductile-to-brittle transi-
538°C. The data include test results for a sis of cracked components is, therefore, tion temperature (Ts ) . Foulds et a1. have
98-second hold time, ten-minute hold the determination of the critical crack successfully applied the test procedure
time, and also the creep crack growth size, based on a knowledge of the current to detennine the T for 12 retired rotor
rate data. 45When plotted as a function of toughness of the materials at the critical samples.48.49 The Tsp values obtained
(C,).v ' the time rates of crack growth for location. This toughness, generally ex- correlated well with the Charpy FAIT
theseg very different conditions fall on pressed in terms ofthe plane-strain stress values.
the same trend. The significance of the intensity for fracture (KI) determines the The specimen needed for an SP test is
above trend with regard to predicting critical crack size for failure. The usual only 0.508 mm thick with a diameter of
the hold-time effect in engineering com- procedure involves estimation of the K1c 6.35 mm. Specimens of this size can be
ponents is obvious. It must be mentioned at low temperatures based on well-es- easily removed from critical locations in
thatifda/ dN were correlated with&(,as tablished correlations between the 50% the rotor bar without affecting the rotor
has been done in the past,46 the data for ductile-to-brittle FAIT and K1c . Consid- integrity by using a recently developed
each hold time will follow a different erable progress has been made in "scoop" device. If bore samples from the
trend. In order to use this approach for estimating the FAIT of steels using rela- critical locations were unavailable,
predicting crack growth during hold tively nondestructive techniques. samples from other locations could be
time, it is necessary to estimate the mag- In Cr-Mo steels, operating at or above tested and the FAIT values obtained
nitude of (Ct).vg for components. An 538°C, the embrittlement mechanism of could be corrected for location depen-
equation has been proposed to estimate concern has been the carbide-induced dence using the approximations de-
the (C,).vgfor any geometry for a material embrittlement described earlier. An scribed in Reference 47.
defonning by elastic-cyclic plasticity and electrochemical polarization test has
power-law creep.45 The details of this LlFE·ASSESSMENT FOR
been developed by Watanabe and Shoji
SUPERHEATER TUBES
equation are given in the above refer- for measuring the increase in FATT due
ence and are outside of the scope of this to embrittlementin such cases. In CrMoV The EPR! has developed a methodol-
overview. rotor steels temper embrittlement is more ogy that uses measurements of steam-
There are additional concerns that of concern. The most promising tech- side oxide-scale thickness to predict the
must be addressed in order to develop niques in this case involve the use of remaining creep life of high-tempera-
total confidence in this approach. Cur- phosphorous versus FAIT correlation, ture superheater / reheater tubing.5o,51 In
rently, there is no provision to consider picric acid etch (PAE) tests, small punch addition, the validity of measuring the
the interaction between loading/un- tests, and electrochemical tests. oxide-scale thickness ultrasonically has
loading rates and the crack growth rate also been demonstrated by establishing
Phosphorus-FAIT Correlations
during hold time. During start-up, large a good correlation between ultrasonic
transient thennal stresses often develop. Viswanathan and Gehl have per- measurements and direct measurements
Therefore, the hold period may be pre- formed an in-depth review of FAIT data on tube samples removed from super-
ceded by a fatigue cycle with a maxi- available on several retired CrMoV ro- heater assemblies from several fossil

1992 February • JOM 41


boilers. Comparison of uniaxial test re- tension (1988), p. 116. Defects in Plant Operating in the Creep Range," Fatigue Frac.
17. F.V. Ellis et aI., Remaining Life Estimation of Boiler Pressure Eng. Mater. Struct., !O (2) (1987).
sults with tube burst tests results has Parts, vol. 1, Identification of Relevant Damage Mechanisms, 41. A. Saxena, "Limits of LEFM in the Characterization of
EPR! ReportCS-5588 (Palo Alto, CA: EPRI, November 1988). High-Temperature Fatigue Crack Growth," Basic Questions
also helped identify the correct formulas 18. B. Neubauer and V. Wedel, "Restlife EstimationofCreep- in Fatigue, vol. 2, STP 924, ed. RP. Wei and R Gangloff
for computing the reference stress in ing Components by Means of Replicas," Advances in Life (Philadelphia, PA: ASTM, 1989), pp. 27-40.
Prediction Methods, ed. D.A. Woodford and J.R. Whitehead 42. H. Riedel, "Crack-Tip Stress Fields and Crack Growth
tubing, replacing a more conservative (New York: ASME, 1983), pp. 307-314. Under Creep-Fatigue Conditions," Elastic-Plastic Fracture:
formula used in the past. The kinetics of 19. B.J. Cane and M. Shammas, "A Method for Remanent Life Second Symposium, vol. I,InelasticCrackAnalysis,STP 803,ed.
Estimation By Quantitative Assessment of Creep Cavitation C.F. Shih and J.P. Gudas (Philadelphia, PA: ASTM, 1983), pp.
steam-side oxide-scale growth in heat- on Plant," Report TPRD/L2645/N84 (U.K.: Central Electric- 1-505-1-220.
flux and nonheat-flux areas are also be- ity Generating Board, June 1984). 43. A. Saxena and B. Gieseke, "Transients in Elevated Tem-
20. S.R. Paterson, APTECH Engineering, Mountain View, perature Crack Growth," Proceedings of MECAMAT-Inter-
ing measured in boilers. Cumulatively, CA, private communication (1989). national Seminar on High Temperature Fracture Mechanism and
these results have provided industry 21. "Field Metallography Research Leads to Improved Re- Mechanics, vol. 3 (1987), pp. 19-36.
Examination Interval For Creep Damaged Steampipes," EPR! 44. B. Gieseke and A. Saxena, "Correlation of Creep-Fatigue
with needed confidence in its application First Use Report B197 (Palo Alto, CA: EPR!, December 1990). Crack Growth Rates Using Crack-Tip Parameters," Advances
of a more cost-effective and nonde- 22. "Development of a Replica Database For Life Assessment
of Balance-of-Plant Components," RTUV and Failure
in Fracture Research, ed. K Salama et aI., (Elmsford, NY:
Pergamon Press, 1989), pp. 189--196.
structive method for assessing the con- Analysis Associates, Interim Report, EPR! Project R.P2253-1 0 45. KB. Yoon, A. Saxena, and D.L. McDowell, "Influence of
dition of tubing. (Palo Alto, CA: EPRI, January 1991).
23. RG. Carlton, D.J. Gooch, and E.M. Hawkes, 'The Central
Crack Tip Cyclic Plasticityon Creep-Fatigue Crack Growth,"
22nd ASTM National Symposium on Fracture Mechanics
Electricity Generating Board Approach to The Determina- (Philadelphia, PA: ASTM, in press).
References tion of Remanent Life of High Temperature Turbine Rotors," 46. A. Saxena and J.L. Bassani, "Time-Dependent Fatigue
I Mech E paper C300/87 (1987). Crack Growth Behavior at Elevated Temperature," Fracture
1. R Viswanathan, Damage Mechanisms and Life Assessment of 24. W. Lempp, N. Kasik, and U. Feller (Paperpresentedatthe Interaction of Microstructure, Mechanisms and Mechanics
High Temperature Components (Metals Park, Ohio: ASM, 1989). Second International Symposium on the Nondestructive (Warrendale, PA: TMS, 1984), pp. 357-383.
2. RV. Hart, Metals Technology, 3 (1976), p. 1. Characterization of Materials, Montreal, Canada, 1986). 47. R. Viswanathan and S. Gebl, "A Method for Estimation of
3. D.A. Woodford, Trans. ASME, JEMT, 101 (1979), p. 331. 25. N.5. Cheruvu, Met. Trans. A, 20A (1981), pp. 87-97. the Fracture Toughness of CrMoV Rotor Steels Based on
4. e.J. Bolton, B.F. Dyson, and KR Williams, Matis. Sci. and 26. Z. Quand KH. Kuo,Met. Trans A, 12A (1981), pp. 1333- Composition," ASME J. Eng. Mat. & Tech., 113 (Aprlll991), p.
Eng.,46 (1980), p. 231. 1337. 263.
5. J.M. Brear and P. McCarthy, ERA Report 89-0230, EPR! 27. Y. Watanabe and T. Shoji, Met. Trans., in press. 48. J.R Foulds, e.W. Jewett, and R. Viswanathan, "Miniature
Project R.P2253-2. 28. RA. Stevens and D. Lonsdale, "Isolation and Quantifica- Specimen Test Technique For FAIT," Joint Power Generation
6. M. Nakashiro et aI., "Evaluation of Long Term Creep tion of Various Carbide Phases in 2.25Cr-lMo Steel," SER/ Conference (New York: ASME, in press).
Strength of 2.25 Cr-l Mo Heat Transfer Tube in Actual SSD/84/0046/N (U.K.: Central Electricity Generating Board, 49. J.R Foulds and e.W. Jewett, "Miniature Specimen Test
Service Stress Level Range," ISIJ International, 30 (10) (1991), June 1984). Technique For Toughness Assessment," Final Report, EPR!
pp. 823-S28. 29. RA. Stevens and P.E.]. Aewitt, "The EffeetofPhosphorus R.P1957-1O (Palo Alto, CA: EPR!, in press).
7. J.M. Brear, 'The Effect of Oxidation Upon Long-Term on the Microstructure and Creep Strength of 2.25 Cr-l Mo 50. S.R. Paterson and T.W. Rettig, "Remaining Life Assess-
Stress Rupture Data 011 CrMo and 2.25 Cr-l Mo Steels," ERA Steel," SER/SSD /85/0020/R (U.K: Central Electricity Gen- ment of Superheater and Reheater Tubes," EPR! Report C5-
Report 89-0662, EPR! project R.P2253-4. erating Board, March 1985). 5564 (Palo Alto, CA: EPR!, 1988).
8. B.J. Cane and A.M. Bissall, "Predictive Assessment of 30. R Munson, Radian Corporation, Austin, TX, private 51. R Viswanathan, S.R. Patterson, H. Grunlow, and S.N.
Damage in Elevated Temperature Weldment" (Paper pre- communication (1990). Gehl, "Life Assessment of Superheater /Reheater Tubes,"
sented at the EPR! Plant Maintenance Technology Confer- 31. H. Nakatani et aI., "Metallurgical Damage Detection and Proceedings of the International Conference on Boiler Tube Fail-
ence, Houston, TX, 14-16 November 1986). Life Evaluation System For Boiler Pressure Parts" (Paper ures (Palo Alto, CA: EPR!, in press).
9. M.e. Askins, Remaining Life Estimation of Boiler Pressure presented at the EPR! Conference on Predictive Mainte-
Parts, vol. 3, Base Metal Mode, EPR! Report C5-5588 (Palo Alto, nance of Fossil Plant Components, Boston, MA, October ABOUT THE AUTHORS _ _ _ __
CA: EPR!, November 1989). 1990).
10. F.v. Ellis et al. Remaining Life Estimation of Boiler Pressure 32. H. Voorhees, Metals Handbook, Ninth Edition, vol. 8, Me-
Parts, vol. 4, Metal/ographic Models For Weld Heat Affected chanical Testing (Metals Park, Ohio: ASM, 1985), p. 339. R. Viswanathan earned his Ph.D. in materi-
Zones, EPR! Report C5-5588 (Palo Alto, CA: EPR!, November 33. R Castillo, A.K Koul, and E.N. Toscano, "Lifetime Pre- als science and engineering at Carnegie
1989). diction under Constant Load Creep Condition for a Cast Ni-
11. H. Grunloh and RH. Ryder, General Atomics, San Diego, base Superalloy" (ASME paper 85, GT241 presented at the Mellon University in 1972. He is currently
CA, private communication of results from EPRI projeet International Gas TurbineConference, Diisseldorf,Germany, manager of materials applications technol-
R.P2253-1O (1989). 8-12 June 1986). ogy at the Electric Power Research Institute in
12. K Kimura, KF. Fujiyama, and M. Muramatsu, "Creep 34. K.N. Melton, "The Isostress Extrapolation of Creep Rup-
and Fatigue Ufe Prediction," High-Temperature Creep-Fat, ed. ture Data," Mat. Sci. and Eng., 59 (1983), pp. 143-149. Palo Alto, California.
R Ohtani, M. Ohnami, and T. Inow (Elsevier, 1988; Japan: 35. F. Masuyama, N. Nishimura, and T. Igari, "Creep Dam-
Society of Materials Science, 1988). age Simulation Test and Creep Life Assessment System For S.M_ Gehl earned his Ph.D. in materials sci-
13. K Fujiyama et al., "Evaluation of Creep Properties of Elevated Temperature Plant Components," ISIJ Interna- ence and engineering at the University of
Degraded CrMoV Turbine Cost Steels Through Hardness tional, 30 (10) (1990), pp. 885-895.
Measurement and Its Application to Life Assessment," ISIJ 36. Y. Kadoya et aI., "Assessment of Remaining Life of Fossil Florida in 1977. He is currently program man-
International, 30 (10) (1990), pp. 869-875. Power Plant Parts By Means of Miniature Creep Rupture ager for fossil plant performance at the Elec-
14. T. Goto, "Study on Residual Creep Life Estimation Using Tes!."
Nondestructive Material Property Tests," Mitsubishi Tech- 37. M.e. Askins and KD. Marchant, Remaining Life Estima-
tric Power Research Institute in Palo Alto,
nical Bulletin, no. 169 (Mitsubishi, April 1985). tion of Boiler Pressure Parts,vol. 2, Miniature Specimen Testing, California.
15. Y. Kadoya et al., "Material Characteristics NDE System EPR! Report C5-5588 (Palo Alto, CA: EPR!, January 1988).
For High Temperature Rotors" (Paper no. 85-JPGC, PWR-lO 38. A. Saxena, J. Han, and K Banerji, "Creep Crack Growth
presented at the ASME/IEEE Joint Power Generation Con- in Boiler and Steam Pipe Steels," EPR! Report CS-5583 (Palo
If you want more information on this subject,
ference, 1985). Alto, CA: EPRI, January 1988). please circle reader service card number 58.
16. J. McGuire and D.J. Gooch, "Metallographic Techniques 39. A. Saxena, "Creep Crack Growth in CrMoV Rotor Steels,"
For Residual Life Assessment of lCrMoV Rotor EPR! R.P2481-5 report (Palo Alto, CA: EPR!), in press.
Forgings,"International Conference on Life Assessment and Ex- 40. RA. Ainsworth et aI., "CEGB Assessment Procedure For

Joint TMS/ASM Student Members


are you planning
a move?

To notify both TMS and ASM International of your new address,


just telephone (216) 338-5151 and ask for the Member Service Center.

42 JOM • February 1992

You might also like