DEVELOPMENT OF FATIGUE CRITERIA FOR REMAINING

DEVELOPMENT OF FATIGUE CRITERIA FOR REMAININGLIFE ASSESSMENT OF SHELL STRUCTURES

J. S. Porowski, W.J. O'Donnell, and M. L. BadlaniO'Donnell and Associates, Inc.

Pittsburgh, Pennsylvania

S. ChattopadhyayUniversity of VermontBurlington, Vermont

S. S. PalusamyWestinghouse Electric Corporation

Pittsburgh, Pennsylvania

ABSTnACTA technical approach is presented for developing

improved fatigue life evaluation criteria forextended life of shell structures. The object is todevelop S-N curves which include aging anden vi ro nme nta 1 effects on ductil ity. strength. crackinitiation and crack propagation properties. The useof J- integral approach is proposed for crack growthand high strain cycling along with specialconsideration for short cracks. The procedure alsoinclu:des an approach for developing fatigue lifeevaluation curves for aged weldments based on crackinitiation. crack growth and fracture.

INTRODUCTION

Environmental and aging effects are addressed inthe design fatigue curves of the ASME Code byassigning arbitrary factors to cycles and strainranges in the fatigue data for materials tested in anair environment. The proposed approach is toseparate the crack initiation and propagation phasesof fatigue failure. Fracture mechanics concepts havebeen extensively used in recent years for thequantHication of crack growth rates incl udingenvironmental effects. Problems in quantifying crackiniti'ltion. crack propagation threshold conditionsan d the i r dependence on environmenta 1 interacti ons.critic:al crack sizes for various materials aftertherRla 1 and strain aging. and other complex issuesremairl to be resolved. The proposed procedure is tocombine the S-N and fracture mechanics approaches inorder to make a major improvement in the methods ofquanti fyi ng the remaining safe life of aged nuclearplants:.

l'here are very significant economic benefits tobe gained by extending the life of existing nuclearplants. This can help maintain energy supplies untilnew plants, based on improved technology aresimul taneously designed and constructed. Thepropos~d technique combines the existing S-N fatiguedata base and experience in design for fatigue. withmajor elements of crack propagation technology. Itwi 11 al so provide continuity and logical transition

'115

to the eventual inclusion of environmental and agi ngeffects in design criteria for new plants.

USE OF ELASTIC-PLASTIC FRACTUREMECHANICS

We propose to use elastic-plastiC fracturemechanics (J-integral) CRef.1) concepts to obta i nimproved S-N curves that incorporate more generalcl'ack propagation solutions applicable to the growthof stable fatigue cracks in low-cycle fatiguespecimens including the strain hardening exhibited byboth ferritic and austenitic steel s. Apparentanomalies which have been reported for short fatiguecracks based on linear elastic fracture mechanics canpotentially be resolv~d by J-integral-based methodsfor the materials and conditions of interest.

The general elastic-plastic crack propagationtechnology proposed herein will make it possible toaccurately evaluate crack sizes in the unnotched low-cyc 1e fa t i g ue specimens starti ng with the measuredfailure cycles at each alternating strain level. Theeffects of thermal and strain aging and irradiationare accounted for by decreased ductil ity in the S-Ncurve approach. The low-cycle fatigue end of the S-Nfailure curve is controlled by the true strain atfracture which is directly related to the reductionof area measured in the tensile test. Changes in theultimate strength of the material affect the high-cycle fatigue strength.

The stress intensity parameter t.K is based onlinear elastic fracture mechanics, and has been shownto be quite useful in correlating Mode I fatiguecrack growth rates where nominal stress ranges do notexceed yield. However for situations involving grossplasticity. such as the low-cycle fatigue tests ofinterest herein. t.K has little physical meaning. Amore general parameter capable of accounting forlarge scale plasticity e.ffects is needed for theevaluation of low-cycle fatigue specimens.

In order to more accurately analyze the fractureconditions in a component undergoing nonlinearelastic deformatfon, the energy available to drivethe crack per unit extension. J. has been developedby Ri ce (~efs. 1 and 2]. For linear elastic Mode I

(1)

where P and 0 are the associated loads andden ecti ons.

(3)10/" 0i .d!:. ,J 1J

where the strain energy density. W. is:

0i' and !:i" are the stress and strain tensors. s isJ J .-the arc length alo~ r. T i~ the traction vector forthe outward normal n, and u is the displacementvector. .

For the case of a closed contour, the lineintegral of Equation (2) is equal to zero. The linecontour begins on one surface of the crack and endson the opposite surface. For the deformation theoryof plasticity. the value of the 1 i ne integra 1 hasbeen shown to be independent of the path (Ref. 1].Thus, the line integral can be taken sufficientlyremote from the crack tip to obtain reliable stressesand strains for use in the solution, even thoughconsiderable yielding occurs. This techniqueprovides a means of extending fracture mechanicsconcepts from linear elastic (K) behavior to elastic-plastic behavior.

From Equations (1. 2 and 3). J can be evaluatedfrom load vs. displacement records for test specimenscontaining slightly different crack lengths. Begleyand Landes [Refs. 3 and 4) have done considerablework developing the J-integral as an analytical toolfor elastic-plastic cracks usi ng thi s compl i ancecharacteristic. For the power law stress - plasticstrain relationship, Hutchinson (Ref. 5), Rice andRosengren (Ref. 6] showed that crack tip stress andstrain singularities are functions of J.

Dowling and Begley [Ref. 7] used the range of 6Jto characterize fatigue crack growth in compact-type(Cn speci mens of A533B steel at room temperatureunder conditions of gross cyclic plastic deformation.Dowling later extended this crack propagation work toother specimen configurations and material s (Refs. 8and 9J and to the growth of small cracks in low-cyclefatigue specimens (Ref. 10]. Mowbray (Ref. 11Jshowed that under certain simplifying assumptions,the J-integral approach leads to low-cycle fatiguerelationshi ps of the type used in Sections III andVIII of the ASMECode. El-Haddad. Dowl i ng. Topperand Smith (Ref. 12] have shown that 6J could be usedto characterize the growth behavior of short cracksin low-cycle fatigue.

Figure 3 shows the load deformation and stress-strain hysteresis loops in low-cycle, constant strainamplitude fatigue testing. Cyclic hardening orsofteni ng. depending on the metallurgical state ofthe material, occurs very early in life. Normally, anear-constant steady-state cycl ic response isachieved after a number of cycles which is smallcompared to the number of cycles-to-failure. This

(2)

P = constant

-n

v

y

a

V = constant

J " - j~ doa oa 0

p

behavior, J is equal to the energy release rate peruni,t crack extension, G. For nonlinear elastic

. conditions, J is the potential energy difference perunit of cract extension between two identicallylo.aded bodies possessing slightl y di fferent cracklengths. The difference in strain energy, 6Uassociated with slfghtly different crack lengths 6ais shown in Figure 1 and is equal to ~a, so thatJ •dU/da where:

J ,,' r~ dPtI oa P

Fig. 1 Determination of J-Integral Based on Fixed

Displacement Vo and Fixed Load Po

Thus for either lfnear or nonlinear elasticbeUlavior, J is the energy at the crack tip per unitcra.ck extension, or the crack driving force. Underirl"eversible plastic straining. however, J is nolonger equa 1 to the energy ava i 1ab 1e for crackex·tension. By defining J in the same way fornonlinear elastic and elastic-plastic conditions, Jrernains a measure of the intensity of the entireelastic-plastic stress-strain field surround i ng thecrack ti p.

As illustrated in Figure 2, J is a line integraltaken counterclockwise about an arbitr~ry contour raround the crack tip. ' ,

J " /WdY - T ~fds

x {J ~a

-1V . V

1a tJ.V

l---tJ.V p

Fig. 2 Coordinate System and ArbitraryLine Contour at Crack Tip

Fig. 3 Strain and Deflection Measurementsin Low cycle Fatigue Tests

116

n • cyclic strain-hardening exponentA • material constant

steady-state response is called the cycl ic stress-st.rain curve and is the locus of the stabilizedstress and deformation (converted to strain) valuesfrom a series of tests at different strain amplitudes(Figure 4). The plastic strains in the cyclicstlress-strain curve can be represented by a power lawso that the total strains are:

4

104 105

A5JJB STEa

102 103N. CYCLES

10

1/0.16S

a. (mO.).Cl •• ~+

with 0. in ksi

-.J 4V " 0.0506 Inches~ 0.0316

0.01780.0118

:-.':'0.0090 -

-0.006~

.

,.

,...§ 60...-J

~ ~o

••'4 aa

'"'"...~ 20'"

0.1

~ 0.002

0.001 1

-J.~ 0.005:=

-::1 80

00

0.01 0.02 O.OJ

,. STRAIN AMPLITUDE

Fig. 5 Cyclic and Monotonic Stress Strain Curves

(4)

(5)

to the elastic andstrain:

The two terms correspondplastic components of the total

£ •• alE, £ • (a/A)ne p

a ( n£ •• !' + a/A)

The cyclic stress strain curves can be used toapproximate the J-integral for use in evaluation ofexp<erimental data. The use of the J-integralapproach for the cyclic plastic conditionsexpl~rienced in low-cycle fatigue testing, in fatiguecra,ck propagation testing, and in operating nuclearpOWl!rcomponents is on a sound technical bash. Itis therefore proposed to include reactorenvironmental effects into the fatigue design lifeassessment methods of Section III of the ASMECodeusing the Section XI crack propaga ti on technologyextended by the J-integral approach to include thelarge-scale plasticity effects encountered in10101-

cycle fatigue testing.a, STI€SS

lI

• HALF-LIFE

• OTHER THAN HALF-LIFE

0.1

0.001

...!::0:

~t;; 0.01

Fig. 7 Proportionality Between Plastic Strainand Plastic Deflection

0.001 0.01 0.1

4.p' PLASTIC OEFLECTIOPI RANGE. INCHES

Fig. 6 Variation of Strain Range with CyclesUnder Constant-Amplitude DeflectionControl

correlation is quite encouraging.The upper three curves in Figure 10 takenfrom

[Ref. 13) give the ratio of K(equivalent) from J toK(elastic). It can be seen that the ratios are quitelarge for large strain amplitudes. The lower twocurves give the plastic zone correction, which isquite inadequate. and is close to that used in theQ

fac~or of Section XI of the ASMECode.

CTCLl C a - £ CURvt

Fig. 4 Hysteresis Loops and Cycl1cStress-Strain Curves

We take A533B as a representative pressurevessel steel characterized by intermediate strengthfor which the cyclic stress-stra i n properti es aretake~ from [Ref. 10] and shown in Figure 5. Constantamplitude deflection control across the fatiguespecimen ends results in the strain amplitude in thetest section being very nearly constant during mostof the fatigue life. Figure 6 shows the strainrange, A£, vs. cycles from the A533B tests of [Ref.10]. Also the plastic strain range A£ , was shown to

, p

be nearly proportional to the plastic deformation,Au , over the entire range of measurement, as shown

in PFi!iUre 7. This makes it possib 1e to re 1a te theamount of strain energy in continuum elements nearthe cl'ack to the overall deformation conditions.Fi gur'e 8 shows the fatigue properties including theelast1lc, plastic and total strain amplitudes vs.cycle!i-to-failure. Figure 9 shows the compilation ofthe cf'ack growth data plotted vs. 6J for large cyclicdefor'mations in the center-cracked specimen. The

IJ

II,

If

t

I

117

Low-Cycle Fatigue Properties - Elastic,Plastic, and Total Strain AmplitudesVersue Life

E • Z07 CPo

A • lZ~~ P<P.

" • !.Z5

1. J I"'. PI•••• PI ••Z. J. CI ••cul",' C'-let

J. "'uttc lon" PI, ••. Plasttc Zone, Pt, c

5. J, C1,.c"I .••. Cr'Ck

NO. I( '''0Ie- ...•-/

.-.-

/.-

.-.-

.-.-

.-.-

.-// .'/ .'

1// ..••//: 5' l///.'

f •• 'ye ., I .•'!I."

/.: l .•.•. -••/ .-.-~~:.'=::=::::::::::~::.._ 4." ••........

..)Z

10ruu HE E1.

10--~-103 104 iOS

". CYCLES -TO..fA Il.uRE

o.OOZ

O.OS c;' 0.3%

0.001

?!~:;; O.OOS

wg O.OZ::

Fig. 8

10'l

ell' STRA Ul A1'1Pl.I TuOE

For reactor pressure vessel steel, n • 1. 587.

For failure in 1000 cycles, ta • 0.74: in air, and

K q /K • 2. The error in life is therefore 21. 587 •Ie

3, which means that LEFM overestimates life by afactor of 3. On the steep part of the reactor wa tercurves of Section XI, n • 2.975, and a factor of 2el'ror in K results in overestimating the life by a',actor of about 8. This demonstrates the importance0'1 using elastic-plastic fracture mechanics (J-integral) for crad propagation in unnotched fatiguespecimens.

Crack growth rates are given by:

Plasticity Modified K Based on J andon Plastic Zone Corrections, For Cracksin Infinite Bodies Remotely StressedNormal to the Crack Plane.

Fig. 10

CONSIOERATIONSFOR SMALLFATIGUE CRACKS

2 2

J •• (1 - v ) ( 1. 12) It t12 a + () (7 )£ Q 1 h n,g t11tpa

The initiation and growth of small cracks is ofmajor importance. The small crack question becomescritical as it represents a regime where the re is apossible "breakdown" in linear elastic fracturemechanics analyses. At the same time, soundsolutions for this region are very much needed sinceinitiation and growth of small crack s prov i de amissing link between the S-N classical approachdesign, and damage tolerant methodologies based ondefined tolerant technologies which consider theimperfections and inspection results. This topic isof special importance in assessment of residual lifesince the ti~e spent when the cracks are incubated orrema i n sma 11 accounts for the large majori ty of thecomponent life in the high-cycle regime.Subthreshold extension of small cracks may lead tooverprediction of life. However, their initialenhanced growth rates often decay to arrest. or inother cases. propagate to merge with the long-termcrack behavior. Figure 11 from (Ref. 14) illustratesthe complexities of developing growth correlationsvs. the range of stress intensities, 6K.

In studying the available data on sma 1 1 cracksit was observed that in order to obtain measurablecrack growths, the tests were often conduc ted atr e 1a t iv e 1y hi g h nom i na 1 s t res s e san din eludedconsiderable plasticity. Thi s suggests that theactual driving force was significantly higher thanwhat the LEFM principles predict. The elastic-plastic J value for a small crack in plane strain isgiven by:

(6)

(lIJII.Sl7

102 .10J

t.J '"-1.8I1N2

ASJJ8 STErr

Fatigue Crack Growth Rate Versus Cyclic6J Data For Elastic-Plastic Tests onCenter Cracked Specimens

~ • Cl(6J)ndN

Fig. 9

118

Fig. 12 Crack Growth Data for Short and Very ShortCracks Compared with "Best Fit" Line forLong Cracks

measurements, generalized by crack initiation models,to include such effects in the total fatigue lifeevaluations. Progress in the understanding andquantification of crack initiation can then be usedto make further improvements in the S-N curves and inthe related inspection program requirements.

As far as environmental effects on crack growthare concerned, the process is primarilyelectrochemically related. As such, the propagationrate is controlled by reaction rates on the crack tipand how these are affected by the passivation rate,liquid diffusion rate, and strain rate at or near thecrack ti p. Knowl edge of the interaction betweenthese controlling processes hel ps to quanti fy thetesting conditions which will produce the mostmeaningful and useful results, and applications of amechanistic understanding allows the data to beextended to a broader range of conditions than tho setested.

Corrosion-assisted crack growth rates depend notonly on the material, temperature and coolantchemistry, but on the strain rates, loading waveform,stress intensity range, temperature and flowconditions, maximum stresses and sequence of loading.The exi sti ng ASME Code Section XI Appendix A crackgrowth curves for reactor water environments doexplicitly account for the mean stress by the R ratiodepe~dence. These are the upper two curves in Figure13, which are shown with the air environment curvepreviously described. In the S-N fatigue lifetechnology of Sections III and VIII of the ASMECode,the fatigue curves are adjusted for the maximumeffect of mean stress which is potentially moreconservative than the Section XI treatment.

Stra i n ra te is another important variable notexplicitly included in the Section XI crack growtheva 1 ua t i on method. The Appendix A curves are basedon test frequencies between 0.1 and 1 cycle perminute which were found to produce high growth ratesin A533B. The rate of rupture of a passive film atthe crack tip or the rate of metal surface exposureat the crack tip is related to the crack ti p stra inrate. Although it is recognized, it is not feasibleto use crack tip strain rates directly in a life

GROWTH OF SHORT CRACKS

AS33B

ACT

• 0 0.04

• 0 0.02

••• C> 0.012

+ 0 0.009

.0 0.005OPEN SYM80lS ARE FOR

CRAn OEPTHS lESS

THAN 2.3 MILS

1 , ,I ! ,

10 100 LOOOt.J, in, Ib~./in2

* .2.1J l 10-' (6Jl"'"

AGING ANDENVIRONMENTALEFFECTS

\,III ~ONG CRACK THRESHOLD 6KTH

LOG 6K, a

Fig. 11 Schematic Representation of the TypicalVariation in Fatigue Crack Growth Ratesda/dN, with the Nominal Cyclic StressIntensity Factor AK, or Crack Length a,for "Long" and "Small" Cracks. AKTHisthe Nomlnal Threshold Stress IntensityRange Below Which Long Cracks RemainDormant.

Crystalline defect theory and the use ofelectron microscopy have contributed immensely towardthl! understanding of the factors involved in theinitiation of fatigue cracks. For practicalpurposes, it is useful to consider three phases ofcruck initiation:

SHORT CRACK

da FROM NOTCH

LOG em '",~

1) the generation of persistent slip bands,2) the initiation- of permanent damage, and3) the coalescence of micro-voids to form a

crack.

Fatigue damage appears to begin as a sequence ofsmall cavities oriented on the persistent slip bands.Such cavities increase in number and size, ultimatelyjo'ining to form either a continuous microcrack orshallow surface intrusion. The stress and stra i namplitudes and the number of active slip systems doesnot change the basic process of crack initiation, butonly the rate and mode of cavity coalescence.

Environmental and aging effects on crackinitiation are not well understood from a amechanistic point of view, nor have they been wellqua,nti fi ed by tests. We therefore take a pragmaticapproach of using high-cycle fatigue life

CONSTANT-M1PLITUOE LOADING

R = CONSTANT

Dowling [Ref. 9], measured the growth of smallcl'ads in low-cycle fatigue specimens of the A533Bsteel. Figure 12 shows the crack growth data whichmakes use of Equation (7) to calculate t.J for eachte-st. Considering the difficulty measuring accuratesizes in smoothed unnotched low-cycle fatigue testspecimens, the correlation of Figure 12 is consideredqui te good. These results confirm the val idity ofthe t.J correlation for crack lengths 0.002 inches orlarger in A533B pressure vessel steel.

While the use of elastic-plastic fracturemechanics resolves some of the anomalous crack growthbehavior found using LEFM, the short crack problemar.d related crack initiation issues need furtherstudy. Short cracks in various material s ha ve beenobserved to propagate below the thresholdt.K levelswhich have been found for long cracks.

119

~ 10"5.•....a io,"7

ASJl8

A5"£ HCT IOH 11aUCTOO WATER

R 1. 0.6S

a ~ O. 2s -........

.CC.W-l-• CT. \I .z·

LINEAl ElASTIC TESTS

• CC. ~ • \"

• (T. ~ • 2'

: ii: ~ : ~:s·(PAris n. a1.)

CC • Center Cracked Seedmen .•C1 • Ccr.::IICt Specilllltns .

QUI fro" ZO IttHS

.•..

10·' 10-' I .StO lQ so ICO S to :a so 100

,I((W,",,,",, ll«(N".V"'ii'1

Fig. 14 J. D. Gilman (EPRI) Theoretical Time-FrequencyDependent Crack Growth Predictions for A533-1Steel in Reactor Water Environment

10 102 103

6J. IHAI.IIH.2

Fi g. 13 Compari son of Fati gue Crack Growth RatesFor Reactor Water Environment from ASMESection XI with Rates for Air Used HereinBased on 6J

assessment evaluation. Gilman [Ref. 15] and othershave worked with rate dependent models whichtecognize that the corrosion-cracking component ofcrack growth is time-dependent rather than cycle-dependent. Such models are quite useful inpredicting long-term behavior from short-termlaboratory measurements of crack growth rates. [Ref.15] results for A533B ma teri a 1 for BWR and PWRe'nvironments are illustrated in Figure 14 along witht.he Section XI curves. These results indicate thatthe Section XI curves are sensible but there may bemuch higher crack growth rates at low-cyclicfrequenci es.

LIFE EVALUATIONCURVESFOR \oIELDMENTS

The fatigue life of aged weldments may bee,xpected to control the useful safe life of manyc:omponents. Our approach is to treat the imperfec-tions inherent in production welds of nuclear gradequa 1 i ty. i nc1 uding the effects of residual weldingstresses. An extensive literature survey concerningweld discontinuities was completed for PVRC by Lundin[Ref. 16].

The elastic-plastic crack propagation methodsdeveloped for uMotched fatigue specimens can be~xtended to include cracks growing from imperfectionssuch as porosity and inclusions in weldments. Low-(:yc1e fatigue behavior .of weldments will bequantified by considering the nucleation and growthof fatigue cracks from the imperfection intoplastically deformed material. The role ofimperfections in the failure response of weldments is,omplicated by the nonuniform nature of the residualstress distribution. In the case of residual stressf'etai ned tens i1 e resid ua 1 s tre s s mus t be as s umed

unless .there is evidence justifying a lessconservatlve approach.

The total fatigue life of a weldment can bec:onsidered to be equal to the number of cyclesrequired to initiate a micro-crack at an imperfection

120

to a size which reduces the remaining cross-sectionsuch that it will no longer support the cyclic load.The weldment then fails by ductile or brittlefracture. The low-cycle life of a welc1ment in athick section is dominated by the crack growth phase.The high-cycle fatigue endurance strength is usuallydominated by crack initiation and the early phase ofcrack growth where the crack is reorientingperpendicular to the maximum stress direction.

The complicating issue for weldments is thetreatment of crack nucleation and short-crack growthIn the presence of gross plasticity effects. To dealwith this. we propose to use the J-integralformulation as applied to cyclic plasticity. TheJ-integral formulation wi 11 be used to study thegrowth of short and long cracks with cyclicplasticity. The needed ingredient required to extendthe procedure to deal with weldments is theJ-integral solution for cracks emanating from bl untnotches whi ch characterize weldment imperfections.Solutions for such models will be used in conjunctionwith crack-growth rate correlations to compute thefatigue life S-N curves for air and waterenvironments. TheS-N ~ndurancedata will anchor thecurve at the high-cycle end. Design curves forweldments can then be generated by applying theappropriate factors. Metallurgical notch effects andgeometric fatigue strength reduction factors areapplied to the nominal stresses before entering thelife evaluation curve.

ADDITIONAL ISSUES IN DEVELOPMENTOF IMPROVEDFATIGUELIFE EVALUATIONTECHNOLOGY

There are a series of technical issues which canhave a significant effect on the assessment Of theresi dua 1 1ife:

1. Factors used to Account for Scatter in FatiaueData. Surrace FlnlSh Effects. SHe Effects. andEnvironmental Effects.Envlronmenta 1 effects such as reactor wa teI' ca nbe expl i citly taken into account using S-N dataobtained in the appropriate environme nt. or byusing their known effects on crack initiation andgrowth rates to adjust the S-N curves us i ng theJ-integral approach proposed herein. Theexisting Code Design fatigue curve isba sed or

.,t ••

3.

4..

5.

6.

7.

applying a factor of 2 on the total strain range,and a factor of 20 on cycles to each point on the"best fi t" fa i 1 ure curve to account for datascatter, surface finish effects, size effects andenvironmental effects. n~ factor of 2 on strainrange governs in the high-cycle regime and thefac tor of 20 on cyc 1e s governs the low-cycleregime. Since the data were obtained usingcompletely reversed deflections, corrections weremade in order to include "worst case" mean stresseffects. Use of more involved methods in thedevelopment of fati gue curves shoul d justi fyreevaluation of safety margins in order toeliminate undue conservatism.

Loading Seguence EffectsThe ASMECode design criteria uses a linearcumulative damage rule, the inaccuracies of whichhave long been recognized. The sequence ofload i ng is known to have a major effect on thedamage accumulation. Of particu1 ar concern isthe possible early crack initiation which couldoccur under large strain ranges occurring ~uringshakedown early in life. Such cracks couldpropagate under the "endurance 1 i mi t" of theunnotched material which was used as the basisfor the design curves currently in theCode.

Extension to Very High CyclesAn effort 1S under way to extend the fatiguedes i gn curves for carbon and low-alloy ferriticsteels beyond one million cycles in order toincl ude flolo/-induced high-cycle thermal fatigueand mechanical vibrations. Mean stress andenvironmental effects are quite significant inthe high-cycle regime. little information isavailable and the test data for very high cyclesare difficult to generate.

Crack Initiation TechnoloayAnalyt1cal methods of treating crack initiationare needed to formulate more effective inspectionprograms. Such methods will serve to i denti fythe location of critical areas which are expectedto include stress raisers in addition towe1dments. The predicted cycles to crackinitiation will provide a sound basis forspecifying inspection locations and intervals.

Propagation of Multiple Cracksr t 1S not u nus U a 1 to f 1 n d a s e r i e s0 f sma11cracks during in-service inspections. Moreaccurate methods of accounting for interactioneffects on their Propagation behavior are needed.

Methods of Accountina for Severe ThermalmnS'1entsRap1d thermal transients involving largetemperature variations and complex heat transferconditions may induce early crack initiation andsignificant fatigue damage. More accurate meansof quantifying such damage would improve theresidual life assessments.

Safety Marains for Instability for CracksThe stab11lty or faor1cat10n, stress corrosionand fatigue induced cracks raise safety marginissues, particularly under complex conditionsinvolving weldments, stress raisers, aging and/orirradiation effects, and leVel 0 loading.

121

8.

CLOSURE

The Pressure Vessel Research Committee incooperation with ASMECode Committees has been quiteeffective in provi di ng industry wi th improvedcri teri a for the design, inspection and fabricationof structural components. Our approach is intendedto focus research and development efforts in theemerging teChnology of Plant Life Extension in orderto develop the desired life extension criteria usinglimited available national resources. The focus ison development of fatigue life evaluation methodsbased on use of S-N curves, modified in accordancewith advanced fracture mechanics solutions andextensive stable crack growth rate test resul ts no';/available. Although results obtained by traditionalfatigue testing of smooth specimens has provided asound techn i ca 1 design basis for shell structures,improved accuracy and reliabil ity are needed forresidual life assessments. Our approach makesmaximum use of available world-wide data quantifyingthe effects of aging and environmental influences onthe properties of the relevant structural materials.Effects on material strength, ductility, notchsensitivity, fatigue resistance will be integra tedinto a quantified criteria for evaluating theresidual operating life which remains within AS:-IECode safety margins. Program to develop improvedfatigue design criteria is given in the Appendi x.The program has been approved by ASMESubgroup onFatigue Strength and the Subcommittee on Design.

REFERENCES

(1] J.R. Rice, "A Path Independent Integral and theApproximate Analysis of Strain Concentration by

Notches and Cracks," Journal of A~pli ed Mec ha n i c s ,Transactions ASME,l2., 1968, p. 37 •

(2] J.R. Rice, Treatise on Fracture, H. liebowitz,Ed., 2, Academic Press, New York, 1968, p. 19l.

0) J.A. 8eg1ey and J.D. Landes, "The J Integral asa Fracture Criterion," ASTMSTP 514, 1972, pp. 1-20.

[4] J.D. Landes and J.A. Begley, "The Effect of

Specimen Geometry on J1c'" ASTMSTP 514, 1972,pp. 24-39.

(5] J.W. Hutchinson, "Singu1ar Behavior at the Eneof a Tensile Crack in a Hardening Material," Journa 1of the Mechanics and Physics of Solids,li, 1968,

pp. 13-31.

(6) J.R. Rice and G.F. Rosengren, "P1ane StrainDeformation Near a Crack Tip in a Power-Law HardeningMateri a 1, II Journal of the Mechanics and Physics ofSolids, li, 1968, pp. 1-12.

[7] N. E. 0010111ng and J.A. Begley, "Fatigue CrackGrowth During Gross Plasticity and the J Integra 1, " .ASTMSTP 590, 1976, pp. 82-103.

Part 1:

Part 2:

(8] N.E. Dowl ing, "Geometry Effects and the JIntegral Approach of Elastic Plastic Fatigue CrackGrowth," ASTMSTP 601, 1976, pp. 19-32.

(9]N.E. Dowling, "Fatigue Crack Growth Rate Testingat High Stress Intensi ti es," ASTMSTP 637, 1977, .pp. 139-158.

[10] N.E. Dowling, "Crack Growth During Low CycleFatigue of Smooth Axial Specimens," ASTMSTP 637,1977, pp. 97-121.

[11] D.F. Mowbray, "Derivation of a Low CycleFatigue Relationship Employing the J-IntegralApproach to Crack Growth," ASTMSTP 601, 1976,pp. 33-46.

[12] M.H. El Haddad, N.E. Dowling, T.H. Topper andK.N. Smith , "J I n t e gI" a 1 Ap p 1 i cat ion s for Sh0I" tFatigue Cracks at Notches," International Journal of,Fracture, .l§.. 1980, pp. 15-30.

[13] N. E. Oowli ng, "Growth of Short Fa ti gue Cracksin an Alloy Steel," ASMEPaper 83-PVP-94. 1983.

(14] R.O. Ritchie and J. Lankford, "Small FatigueCracks: A Statement of the Problem and PotentialSolutions," Small Fatigue Cracks, TMS-AIME,Warrendale, PA,1986, p. 1.

[15] J.D. Gilman, "Application of a Model forPredicting Corrosion Fatigue Crack Growth in ReactorI>ressure Vessel Steels in LWREnvironments," ASME

I'ublication PVP Vol. 99, Predictive Ca~abi1itieS inJnvironmentally Assisted Cracking, 198 •

[16] C.D. Lundin, "Bibl iography on Fatigue oflieldments and Literature Review on Fatigue CrackXnitiation from Weld Discontinuities," WRCBulletin.

FIROGRAMTO DEVELOPIMPROVEDCUMULATIVEFATIGUE DESIGNCRITERIA IN ASMEBOILER AND PRESSURE VESSEL CODE

Background- The ASMEBoiler and Pressure Vessel Committee is/Ittempting to develop improved cumulative fatiguedesign curves and criteria which take advantage ofthe major developments which have occurred in thetechnology since the current criteri a were wri tteninto the Code over twenty-five years ago.

The Seven Part Program Plan which has beendeveloped for this purpose is described herein. Thefirst two Parts are in Category I, Work In Progressby ASME Code commi ttees. The last five Parts inCategory II, Future Work, include environmental andagi ng effects.

In addition to the need for improved fatiguedesign criteria which include aging and environmentaleffects, the Code needs criteria for evaluating thesafe residual life of all hardware. Such rules wouldmake it possible to evaluate the operative lifeextension which continues to meet Code fatigue safetymargins. In spite of recent advances, inspectiont~chnology is not yet totally reliable, and these areareas not accessible for inspection. S-N technologyprovides a proven means of evaluating residual safeoperating 1ffe without depending on inspectionresults.

Dltegory I: Work In Progress

122

Updating of fatigue design curves for carbon,low alloy and high tensile steels to incl udeupdated fatigue data. (This requiresextensive data eva1uations,parameteranalyses and sensitivity studies.)

Extension of carbon, low alloy and hightensile steel fatigue curves beyong 1QOcycles to include mechanical vibrations,thermal striations, thermal striping and meanstress effects studies and consideration ofthreshold crack initiation teChnology.)

Category II: Future WorkPart 3: Upgrade cumulative fatigue usage factor

Quantification methods. Include loadingsequence, si ze, and surface finish effectswhere cracks initiated in low-cycle regimecould propagate well below the endurancelimit in the high-cycle regime. (Thisrequires sensitivity, studies of crackinitiation and subsequent propagation undervarying stress range loading conditions.)

Part 4: Inclusion of reactor water environmentaleffects in the fatigue design curves forcarbon, low alloy and high tens; 1e steel s.(Thi s requires extensive elastic-plasticcrack propagation analyses using availableworldwide data for ferritic steel crackgrowth rates in reactor water.)

Part 5: Inclusion of reactor water environmentaleffects in the fatigue design curves foraustenitic steels and nickel chromium, iron,and copper alloys. (This requires evaluationof wor1 dwi de data for austenitic stainlesssteel and crack growth rates in reactorwater, and extensive elastic-plastic crackpropagation analyses.)

Part 6: Development of fatigue design curves for.ferriti c and austenitic we1dments in c 1 u din gthe effects of welding residual stresses,acceptable imperfections and metallurgicalnotch effects. (This requires extensiveweldment fatigue data analyses, considerationof e aI' 1y CI" a ck i nit i at ion i nth e he a t-affected zone (HAZ) due to metallurgicalnotch effects and residual stress effects.Extensive elastic-plastic crack propagationanalyses must be carried out for cracksgrowing from acceptable imperfections withmean stress effects.)

Part 7: Development of fatigue life evaluation curveswhich include aging as well as reactor waterenvironment effects for ferritic andaustenitic steels and weldments. Such curveswill provide a criterion for nuclear plant1 i fe extens i on beyond forty years. (Thiswork builds on all six of the first six Partsof the effort described above. The effectsof aging on the crack initiation andpropagation elements of fatigue failure willbe included in the curves, accounting for anyreduced toughness and ductility, and theincreased notch sensitivity of the materialsand weldments.l

Part 8: Development of improved analytical proceduresfor performing fatigue analysis. (This