Corrosion Fatiga.pdf

f*l¡g

(ot/,o9toil

Environmentol CrockingCorrosion Fotigue Richard P. Gangloff'

CoRRosIoN FATrcuE (cF) is

an important but complex mode

of failure for high-performance structu¡al metals operating in deleterious en\,ironments. This view is based on the likelihood of cyclically varying loads and chemical enüronments in service, the need for predictable longJife component performance and lifc exte¡sion, the universal susceptibiliry of pure metals and alloys to CF damage, and the time-dependent multivariable characte¡ of corosion fatigue. For example, stress corrosion cracking (SCC) immune alloys are susceptible to CF. Corrosion fatigue has aFtected nuclear po\\,er systems, steam and gas turbines, aircraft, marinc stn¡ctu¡es, pipelines, and bridges; CF issues are ce¡tral to the behaüor of many aging systems [1-3]. The objective of this chapter is to highlight moderD laboratory methods for characterizing the cor¡osion fatigue behaüor of metals in aqueous electrol¡es. The principles and mechanisms of CF are summarized in the first section of this chapter, follorved by discussions of expeúmental methods in the second section. Specimen dcsign and loading,

enüronment control, stmin and crack size measurement, and computer automation are discussed. The emphasis throughout is on exemplary experimental methods and results, as rvell as on CF data anahsis and interpretation. The third section of this chapter cices applications o[ CF data to ser-vice, the advantages and limitations of the experimental methods, and directions for research on CF experimentation. St rnbols and terms are defined in the Nomenclature. This chapter, with extensive references, extends previous rrcüervs of corosion fatigue test techniques [4,5]. This chapter rvas published originally in 1995, the following literature in the BibliogEphy reflects new devclopmenrs in Corrosion Fatigue.

No¡rrenclature

^(' A€, Aer € €¿

applied engineering üue stress range in a fatigue load cycle, óm- - onir where o is load/initial crosssectional a¡.ea fi-ue plastic axial strain mngc in a fatigue load cycle, rpmax - €pmin true total axial strain range in a fatigue load cycle, €¡-.. - E¡-;n

t,d a't,b E'l,c

gage length and specimen diameter, rcspcctively

Basquin relationship material proper'ty pammeters Coffin-Manson relalionship material property parameteIs number ofload cycles for specimen failutc by fatigue number of load cycles for fatigue crack initiation

tmnsition fatigue life, or number of load

c¡'cles

where the magnitudes of the elastic and plastic axial

slrain ranges are equal d¿ldN macroscopically averaged fatigue crack growth mte fatiguc crack length fatigue load c."-cle count Ar appljed slres\ intensitt fa.¡or range, ¡(.- - K , Paris Law material propert.v constants A, t1t aKnr thrcshold stress intensitv range R strcss ratio, r(-in/K,** p notch tip radius Ioading frequency in cycles per s or Hz 1td|' time-po¡tion of the load c-vcle rvhere CF damage occurs Kraaa threshold stress intensity for monotonic load SCC üLldl velocit], o[ monotonic load SCC Ec free corosion potential E modulus of elasticity Ar("¡ effectivc stress intensity ¡angc, K.,.. - (.1 (.r strcss intensit!,value in a fatigue load c.vcle rvhcre cmck-surface closure contacf is expe¡imentally i!solved and operationallSr defined stress intensitv ¡?nge at a fixed crack lcngth, ao ^(o c constant in the equation \üth units of mm-l ^I(-contiol ü/dtr timc-based crack gror.vth rate in fatigue, often approximated by (d¿ldN)(0 CF co¡'r¡sion fatigue SCC stress conosion cmcking HCF high-cycle fatigue LCF low-cycle fatigue FCP fátigue crack propagation HEE h¡dr-ogen enüronment cmbrittlement H atomic hydrogcn

f

HCP

hcxagonal-closed-packcd

LEFM linear e!astic fiacturc mechanics SN strcss range versus life CP cathodic polarization

lr'ue axial strain,ln (1/lo)

cpm

true diamctral st¡ain, ln (d/do)

Mode

I

Mode

II applied load parallel to the crack plane and the

lP.of.ssor and Chair, Depar(menr ol Maleria¡s Scicncc and Engineering, UniveNity ol Vireinia, Charlorlesli¡¡e, VA 22903.

cycles per minute

applied load perpendicular to the cmck plane and grorvth direction [ó-8]

clack gro*th direction

2

CORROSION TESTS AND STANDARDS MANUAL

Mode

I1I applied load palallel to the crack plane and pcryendicdar to the gro\r4h direction

v" LVDT EPD

Poisson's ratio for isotr-opic elastic deformation, often taken as 0.33 Lineal Variable Differential Transformer

electricalpotentialdifference

BASIC PRINCIPLES Fundamentals of Corrosion Fatigue Delinition Conosion fatigue is deñned as the sequential stages of metal damage that evolve rvith acct¡mulated load cycling, in an aggrcssive environment compared to inert or benign surror.rndings, and resulting f¡on] the interaction of ireversible cyclic plastic defolfnation \l,ith localized chemical or elcctrochemical reactions. Environment-enhanced fatigue is a modern tem; however, corosion fatigue is traditionally used when emphasizing electr r¡chemical environments. Mechaoical fatigue cxperiments and anal¡'ses, detailed in recent textbooks [6-8], providc the basis for understanding CF. Stages of

Corosion Fatigue

CF damagc accr¡mulates with increasiog load cycle count (N) and in four stages: (l) cyclic plastic deformation, (2)

microcmck initiation, (3) small crack grorvth ro linkup and coalescence, and (4) macrocrack propagation. A cardinal principle is to design a CF experiment to isolate and quantitativelv characterize one of these four stages. The methods i¡ this chapter are organized as follorvs: (f ) smooth specimen life for high rycle fatiguc (HCF) described bt, A6 versus NÍ d^ta, (2) smooth unia\ial or notched specimen life for low cvcle fat¡gue íLCF) describcd b\ Acp ver.sus \ or All Vp vcEus N,, respcctivclv, and (3) [arjgue crack propagalion (FCP) kinetics described by da/dN vercus rhe 6.acture mechanics AK.

Meclunisms It is important to underctand damage mechanisms in

order to co¡'rectly inte¡p¡et and extEpolate laboratorv CF data. Similar to SCC, rhe mechanism for CF mav involve

h\drogen embri lcmenr;

hlm rupture, djssolu;ion

and

¡epassivation; enhanced localized p!asticitv; interactions of dislocations rvith surface dissolutior¡, 6lms or adsorbcd atoms; and complex combinations of rhese processes [9-ló]. The contribution of each mechanism is co¡¡oversial and rlepends on metallurgical and cnüronmenr (thermal and chcmical) variables. While providing significant insighr, existing mechanism-based models are generally not capable of accl¡¡ately predictiDg CF behavior beyond the range of labomrora data. Hydrogen enüronmcnt embrittlement (HEE) is an important mechanism for CF crack propagation in felritic and martensitic steels, as well as aluminum, titanium, and nicltel-bascd alloys exposed to gases and elecrrol).les within

ordcr r¡f 100"C of ambient tempemture U5-2l1. This hv_ polhesis is supporrcd bv exrensivc bul cir'cum"r¡ntial eü, dence, and is mosr readilv acceprcd lbr high-strcngrh alloys

in strong hldrogcn-producing enüronments. In HEE atomic

hldrogen chcmically adsorüs on straincd-clean initiation sites or cmck surfaces as the result of electrochemical ¡'eduction of hldrogen ions or rvater- (Adsorbed hydrogen is also produced by the reactions of H2, HrO, C2H2, or I-l2S molecules rvith metal surfaces.) Hydrogen production follo\\'s mass tmnsport within the occluded crack (pit or creüce) solution, crack tip dissolution, and h¡dr.olysis of cations for local acidiñcation; and precedes hldrogen diffusion by latiice, interface, or dislocatio¡l processes in the initiation-vohlme or clack tip plastic zone. Fatrgue damage is promoted by hydrogen-affected lattice bond decohesion, grain or dislocation ccll boundary decohesion, enhanced localized plasticilv, or metal hydride fomation (in materials such as HCP titaDium-based alloys). Hydrogen-enhaaced CF cracking is either i[tergmnular or transgmnular, with the latter iñ'olüng dislocation subso.l-rc' turc, low index crystallographic planes, or interfaces. A second mechanism for CF is bascd o¡ damage b], passive film rupture and ransient anodic dissolution at a sur_ face initiation site or crack tip. This modcl was developed with several necessary empirical elements to prcdict CF propagation in carbon and stainless steels in high-tempemture pure tr¡ater [22,23], ard is sometimes applied to titanium and aluminum alloys in aqueous chloride solutions. Localized platic straining, described by continuufir mechanics or dislocation plasticily, ruptures the protectivc ñlm. C¡ack advance occurs by transient anodic dissolution of metal at the breached film, and at a decrcasing rate rvhile the surface repassivates pending rcpetition

of this

sequence. The in-

crement of CF gro$.th depends Faradaically on the a¡odic charge ([ansient current-time integial) passed per load cycle. Charge is governed by clean-repassivating surfacc reaction kinetics for the CF-sensitive alloy microstnrcture in occluded cr:ack solution, and by the time between n¡ptures given by local strain rate and film ductiliry- As with the hydrogcn mechanism, film rupture modeling is complex and controversial; confirming data exisr 122,23), bu1 other rcsearch shows the model ro be untcnable for specific alloy/ en\'ironment systems [24]. Several CF mechanisms were proposed based on inte¡?ctions betrveen dislocations aDd environment-based processes at initiation sites or cmck tip surfaces. For example, in-situ tmnsmission electron microscopy and dislocation rñodeling show that adsorüed hydrogen localizes plastic defomation in several pure metals and allo¡,s [25]. Second, reactionproduct films are not capable of extensive plastic deformation rclative to the under\¡ing mctal, and may cause CF damage by one or more proccsses, \,iz: (l) i¡lterference u.ith the rcversibility of slip, (2) localization of persistent slip bands, (3) reductio¡ of nea¡-surface plasticitv Ieading to reduced or enhanced CF depending on rhe oácking mechanism, (4) localization of Dear-sur-face dislocation structure and voids, or (5) film-induccd cleav^ge U4,15,2ó-28). Adso¡'bed caiions could similar!¡, affect farigue [29]. Finallv, anodic dissoltr¡ion may eliminale neat.surface uork harde;ing ald hcnce stimulare faligue damagc [30]. These mechanisms have not been developed and tested quantitatively.

Factors Controlling Corrosion Fatigue Two consideEtions are central Lo understaDding the effects of mechanical, metallurgical, and chemical variables

CHAPTER on CF- Thc infll¡ences of electrolyte composition, conductivity, pH, electrode potential, temperature, viscosiry, and biological acúvily are govemed by the mass t¡anspoñ and electrochemical reaction conditions within occluded pits, cr-evices, or cmcks, including the role of st¡ain io creating reactive ctean surfaces [3,/,32]. Second, CF can be timedependcnt. Crack growth is ohen rate-limited by one or more of the slow steps in the mass tmnsport and crack surface reaction sequence; slow loading rates enhance CF dalr. ge L171. Increased cmck tip strain rate is deleterious when the extent of per-cycle electrochemical r€action is prorñoted f22-21).2

Variables that affect CF wer€ rel,ielved elsewhere [.15--17]. lmportant factors are cited herc to illustrate important CF test methods and to guide data interpretation.

Mechanical Varíables An important issue is the influence of an electrochemical environment on the cyclic deformation behavior of metals U 1,33-351. As illustr¿ted by the data in Fig. I for a carbonmanganese steel in high-temperature water, enüronment does not tlpically affect the relationship between stresscs and st¡ains derived from the rñaximum tensile (or compressive) points of steady-state (saturation) hysteresis loops [36]. Such loops should rclate to elastic and plastic deformation prior to substantial CF microcracking. CF data of the so¡'t shown in Fig. I arc produced by either stress or total stmin controllcd uniaxial fatigue experimenrs, ide¡tical 1200

^c = -

289"C

o

to tlre mcthods developed for measuring purely mechanical c)'clic stress-stmin data [8,37]. Whilc macroscopic constitutive properties may not be environment sensitir,e, slip localization ca¡ be affected by electrochemical reactions [14,3J]. Meclnn ic al Drir ing Force s Considering smooth specimens, thc ranges

of

applied

stress or plastic strain control the fatigue or CF responses of metals for HCF and LCF conditions, respectively. For HCF, smooth specimcn CF life increases with decreasing elastic sfess mnge, at cycles in excess of thc hansition fatigue life, Nr, according to the Basquin equation

Ld

=o'r(Nt)1'

(l)

and due to decreasing globally plastic stmin at cycles less than N¡, according to the Coffin-Manson equation for LCF ae,

=ei(N¡)-'

(2)

Alternatively, Eq l, divided by E to relate N,. to elastic strain range (^6/E), is added 1l] Eq 2 in order to relate NIto total applied strain range, the surn of the elastic and plastic stmin ranges. The material propeny parameteñ for HCF and LCF (ó, c, dt, and t'¡) depend on metallurgical, environmental, and lime variables. Data in Fig. 2 shorv that the HCF life of A1SI 4140 steel is dcgraded by aerated neut¡al NaCl solution, compa¡ed to similar fatigue lives for dry and moist air as well as deaerated chloride [-]&J91. The data in Fig.3 shorv that distilled water and aqueous 3¿l¿ NaCl similarly degrade hardened aluminum alloy [35]. The Basquin and CoffinMarison relationships are generally obeyed for faligue in clectrochemical enüronments; howevcr', multiple porver larv segments may occur. Critically, the HCF endumnce limit or

eoo

o.^^^ É

ENVIRONMENTAL CMCK.NG 3

the LCF resistance of an unrecrystallized precipitation-

4533-E Siool 1mo

2ó-

^rr

{%)= {ad1965

o)* (^d§56)11

1

40o

tt> 2ñ

Ten.€olnp. Slrán Rale (%/s) Open Symbols: 0.4¡0.4 Closed Symbolsi 0.004rc.4

O A o

An Pure Waler

Fflfl

Wat6.

0L 0.0

05

10

15

Tolrl Stratn Range, a€r (%) FlG. l-The nil eflect o, env¡ronment on the cycllc stressstrain response ol a CMn sl€el ¡n moist a¡r and pressurlzed waler at 288"C [36¡. Data are represented as true stress-{rue total straln range, 2Considering HEE, ir is impoñant (o consider the p.¡maÍv htóogen sor.rrce when designing CF experimenls and in(erpre{ing ¡€-

sults.

In addition ro occluded crack tip

hydrogen produclion,

hydrogen can dittuse over Iong distances lrom ploduclion siles al mildly strained bulk solution-exposed specjmen sur[aces (o the

propagating CF c¡ack tip. Bulk surlace hydroscn production is important for tel-I us alloys at long exposure times in acidic or' H2S beadng solütions, and \\'ith cathodic polarization [21]. This hydrogen source is less impoúant for aluminum and titanium álloys in electrolytes that forrn passne surlace tilms capable of blocking hvd¡ogen uptake.

CvdBroi.iru

2-The déleter¡ous effect ol aerated aqueous chloride solution on the ¡lCF l¡lé of smooth specimens of tempered mart. ens¡tic AlSl 4140 steel. Symbols with horlzontal anows ¡ndicate ihat CF failure has not occu ed after I 07 load qcles [38]. FlG.

4

CORROSION IESIS AND STAND/ RDS MANUAL Ar,

10

Al.Zn-Mg-0.01


t0'

+ +

:; -,

Cu

Srres Inrensi\ nange lñt

?ol0¡0Ó0s

910 Ti-óar-6v-25n l¡

Dry Air

ll2

'n

ill

Añ.€al,

wR R"0.1 l.0.ur"

to'

Hao

-.G

Nacl

,o

l0'l = o.82 l0

-0

.e r0

2Nr FtG. 3-The deleterlous effect of aqueous chlo¡ide solutlon on the LCF tife of a pre{ipitation harden€d alu¡ninum alloy [e5]'

''

t

&tldN

=

A^K-

o . o

vanced aluminum'lithium-copper alloys (Fig. 5) [43,44). Note the complex of dald.Ay' for FCP in chloride solution, but ^¡(-dependence nol vacuüm or moist air. Environmental

on Paris regimc FCP have been

A(s¿r . A(rcc . AKrcc

chamcterized

broadly; however, data on near-thrcshold CF (da/dN <10-6 mm/cycle) are scarce [/ó,42]. Crack closure can strongly affcct fatigue and CF [45]. This

phenomenon is based o¡ crack surface contaci duüng unloading, critically at siress intensity levels above ze¡o and applied-positive K-,. values. Crack rvake contact is caused by corrosion debris, plasticitl,, crack path roughness, or phase tmnsformation products; each mechanism may be sensitive to aqueous cnüro¡mental reactions [ó]. To account for closure, ddldN is correlated with an effective stress inter¡sity range that is defined operationallJr as the

differcnce between applied K,""* and thc I(, level rvhere surface contact is resolved (see Data Analysis and Evaluation in this chapter).

'

.

¡0

Ax,

0.6MNaCt + IC'

14 ¡oH

,

6.¿,

ll

'

10, I t¡l|

rPH

10.5 IDB

/t-The eñect of soldion o¿/p

(DH'1,0'

l0 H¡

l llr

r05

¿

H¡r

, ó4. l0

Hzl

l0ó

. l.o, l0 ¡rl

§tr¿rs tnl¿¡s¡ty na¡9¿,

crack propagat¡on ¡n an

I

ó0

l0 40

20

8 ¡0

ti¿cl(Du ' 6.4 -10 Hr 0.6r' NaCl roH , ó.'¡l - l Hr 0.6M

0.6Mr\3cl l flcl (pH. t.0 -Ambie¡l Air (Loftr Lrmit,

AKs.c - l0 r¡tl

FIG"

pend on environmental a¡d time variables. Fo¡ metals in vacuum a¡d moist air, FCP is described by a single porver law and an apparent threshold stress intensity range below rvhich d¡ldN tends to zero [42]. Mor.e complex cracking behavior is obse¡r,ed for CF, as illustrated for aqueous chlo ride solution-enhanced FCP in titanium (Fig.4) and ad-

elfects

3

(3)

AK is limitcd to stress intensity changes above zero because compr.essive loads do not cause appreciable crack tip plastic stmin and damage. The material properties (A and ,?) dc-

e

l' 10

threshold stress mnge can be eliminated b)' the aclion of the electrolytc, as illustrated in Fig 2. A common explanation for this effect is pitti¡'¡g-based CF crack i¡itiation. Rates of CF cmck propagation are uniquely defined by the linear elastic fi"acture mechanics stress intensity factor range that combines the effects of applied load, cmck size, and geometry Í17,401. The similitude principle states that fatigue and CF cracks groN at equal rates when subjected to equa! values of [ó-8]. The d¿ldN versus AK telationship may be complex;^K howcver, an effective approach is based o¡ a poiver (or Paús) relationship of the foún [4/]

I

r¡r

'

100

lr0

in.l/?

pH and loading frequency on CF litanium alloy exposed lo aque

ous chloride [4r1. CF cmck formation in notched specimcns is most effectively characterized by the notch-root plastic sfain range calculated by Neuber's method, elastic-plastic finite elements, or fracture mcchanics approximation [7,8]. The latter approach is illusÍated in Fig. ó, showing the results of over 100 experiments rvith C-Mn and alloy steels in aqueous chloride solution compared to moist air [E4ó--48]. The load cycles to produce 1 mm of fatigue crack extension, mcasuied opticá[y, increase nith dccieasing AKlrfi, an estimatc of Dotch root AeP, for air and chloride solution.r At fixed Ni is reduccd b-v chloridc exposure at fiec corosion, ^K/Jp, to fatigue in moist air. An endur¿nce limit is obrelative sewed for moist air, but not this CF case. Cathodic polarization (CP) restores a polion of the moist air fatigue initiatioD/ early grorvth life, as discussed in an ensuing section.

lnading Frequency (I;requetlcy Domaín

Issues )-Slo'.v

fre-

quency CF experiments mav be necessar.¡ because of mass tlansport and elec¡rochemical reactio¡ ratc limitations or¡ r¡r1: is calculared assuming a shary crack ol leng(h ánd geometry equivalent to the not.h. This merhod is a reasonable alternative ro a linile elemenr oI Neuber ana,vsis of notch srmin, bur only for c¡acklike notches oF the sort shown by the insefl in Fig. ó f71.

CHAPTER

l0

-

100

ENVIRONMENTAL CMCKING 5

A333-6 sle€l (Higuchi)

AA2O9O TBT

LT.1/2. l

26-

Hz

K-",- l? MPaJm

O ^o

EPPrn 0 2po.n

a a

0.Ol lpm o.a ppm

0.ol wn A106-8 Sresr (ANL)

1o-4

l

ro-1

!

10 -

E

E

E

z

Carbon Steel 250-290.C

10-

10 -

to''

ro6

lo-?

UR Plate

UR Sheet

10

'

1o' 10Strsln Rate (%/8)

1o_'l

10'

100

7-The effect ol strain rate on lor,!, cycle CF ¡n the GMn steel/higtFtemperatue water system. The dissolved orygen conlent of the pure water eñv¡ronment was vaÍeq as indi' FlG.

cated by the we¡ght-parts.per-milllon values in the legend [361.

1

5 ^K

Flá FTtre efiea

l0

(MPaJrn)

of aqueous NaCl on lt|e CF crack

propagation response of un¡ecrystalllzed sheel and plate ot an advanced Al-Li-Cr.FZr alloy [44¡. damage, but are challenging becausc of prolongcd test timeThe generally deletcrious effect of decreasing / on smooth specimen CF life is illustrated in Fig. 7 for an LCF case in-

!'oh'ing a C-Mn p¡essu¡c vessel steel, corroding freely in high-temperatLlre watcr \üth varyi¡g dissolved oxygen levels bet$,een 0.01 and 8 ppm. (The ftee corrosion potential for these steel CF spccimens incrcases as the dissolved oxygen

concentlation increases.) For fixed Aep, the ratio of N} for faligue in ',vater to rhat lbr air, cach al 270'C, declines w¡¡h

decreasing average total strain rate (proponional to fre, quency) [36]. LCF lives are mte-independent for fatigue in laboratorv air at low to modemte tempcratures wherc creep

Frequency effecls on CF crack propagation have been chaEctcrized broadly and rnodcled bascd on the HEE and film rupture mechanisms Uó-18,22-24). FCP mtes are l independent for alloys in r¡oist air, inef gases such as N2 or A,, or vacuum at lo\v to moderate temperatures. For CF, there are th¡ee possible ftequency responses: (l) purely timc-dependent, rvhere d¿ldN increases with decrcasing frequency proportional to ( 1/ür, (2) cycle-time-dependent, rvhere d¿ldN i4cr€ases with decreasing flequency proportional to (l/<¡rp with P on the order o1 O.S,-and 1¡) cicledependent, where daldN is cnvironmentally e¡lhanced but frequency-independent. The parameter n gives the proportion of the load-cycle time that produces CF damage, and is olte¡ taken as 2 for a svmmetrical cvcle, since enüronmental cmcking may not occur during uoloading [./ó]. An altemative model of the Íiequency effect considerli rhat

is minimal.

7000-f6 Series Alumlnuñ Alloys (S-L) ¿.5.3.5%

N.cl

(pH 7), E"q,

E

:

Q

to"

z

10-

E E

Cycres to húratio¡ {r mm Crack)

6-The efiecr d chlor¡de on the CF crack formal¡on and early (1mm) growth rqs¡stañce of notched steel specimens: FlG.

sol¡d line fgr lour steels in NaCl, data po¡nts fo¡ seawater, ancl dashed l¡ne for lour steels in NaCI [46-411.

o

7075-T73,80.C

e

7475, PH 3, N¡2C,O.

o

7079,23% NaCl

FlG. 8-The varied efks of loading frequeñcy on CF crack propagatlon rate ¡n peak aged AA7g75,7017,7475, and lii,g exposed to aqueous cfiloride sola¡t¡on (free coroslon) at constant and R. The fatigt¡e crack is parallel to the plate roll^K ¡n the SCC sensitlve S-L orientation [50]. ¡ng plane

6

CORROS/ON ?ESTS AND STANDARDS MANUAL

0.5 HSLA

0_005

S1Oé

.9 ,o' E

/.)-/'

,/l

,.1 I

3.5% NaCr,-1000rvs.¿

lAr+ 0.01

I

r""o

;;;;;;

FlG. g-The efiect ol load¡ng frequency on CF propagation ¡n API-2H and A7l0 steels, at the 525 to 750 MPa yield strerigth le\rel, exposed to aqueous i¡acl solution w¡th CP to -1000 mVSCE at constant and R [5] l.

^K

electrochemical reactions occur throughout the entire load-

cycle [/7]. A single alloy/enüronment slr'stem can exhibit each drr/d-fv-f relationship, depending on A1(, f, and alloy composition, as illustrated for aged 7000 series aluminum alloys (AA) (Fig. 8) and steels (Fig. 9) [,¡8,49 5,¡]. These plots illustrate the two usual ways of presenting fiequencydependent CF daldN data, \\,ith the abscissa as either ]og

(l/{).

o

The / range where d¿ldN is most circle-time dependcnt varics rvith mass transpon and reaction kinetics that are material-environment sensitivc. Time-depcndent cracking is obscrved for high-strength SCC-prone alloys rvhen K.". in the fatigue c'ycle is above K¡5¡¡ and dalclr is rapid. This casc, illustrated in Fig. 8 for AA7O79 < 1 Hz and for AA7075 ar F< 0.001 Hz, is modlinear superposition of SCC daldt and inert enüeled by ^tf ronment d4ldN data f49,52,531. Cycle-time-dependent CF (or cyclic stress corrosion cracking) involves substantial CF at levcls that arc belorv (r... or rr,herc daldl is slo$,. This

or log

r€sponse is illustrated for AA7017 and M7475 (Fig. 8) in seawater and acidified NaCl, respectively, and for two stecls in ¡eutral NaCl \vith cathodic polarization (Fig. 9), each at fixed Á¡( and R U8,50,51).In both systems, p is 0.3 to 0.6. Note (Fig. 9) thal cycle-rime-dependent daldN achicves a plateau or saturation level at slorv frequencies belorv a critical value that depends on steel composition. The rhird case, findependent CF, is illustrated for AA7075 in 80"C chloride solution (Fig. 8) and for the t\\.o steels at high frequencies in NaCl at 25'C (Fig. 9) [a9,5/]. This Ésponse (true corosion fatigue) is oftc¡ obseffed at lolv at high loading Ire^1(, quencies, or for alloys that resist enüronmeDtal cracking. In some cases CF da./dN increases with incrcasing F. This behavior is illustrated in Fig. 4 for a titaniuú alloy at low A,K; in addition, note a crossover to time-dependent CF above 20 MPafi and time-cyclc-dependent CF at interr¡ediate [43]. Figure^i(l0 illustrates the minimal effect of loading ftequency on CF crack formation ald early growth for blunt notchcd steel specime¡s in aqueous chloride solution at free corrosion [4ó]. Thcse data are notable for the Ére combination of lorv f and high cyclic lives ( 175 days rvere required to obtain 3 x 106 cycles at 12 cpm), and for the lack of a frequency effect on the cvcles required to produce I mm of CF cracking. This result iridicates that one or more of the earlv

-o-

^

.-.--._

¡_

_

-f.11-6

l,¡0

=

:

I

t.8

^K

lo5 No

cycles lo c.aek

l

-mm depth.)

fr

fi1

urc

l'¡trotr,N

FIG. 1(FThe frequency independenc€ of CF llfe for blunt notch specimens ol A5l7 steel, lreely corroding in aqueous 3.5% NaCl as a funct¡on ol notch root strain range [46]. (lnitiation here ¡s in fact crack fo¡mation plus early growth to a f

roe

b.r of Cx l.r ro

Nx¡

rc¡

ol

C!.ls

¡o

frillrr

FlG. fl-The effect of env¡ronrnent on tensile and tors¡onal high-cycle corrosion fatigue in the To7s/NaCl sotution systern

[54,

CHAPTER

cF (cyclic pl¿stic defor'¡nation, pitting, crack nuclcation, and small cr:rck grorvth) are frequency-insensitive, consistent wiih the /:independence of long crack CF at lorv AK for many allo¡* including steels in NaCl [,]ó,5/], Results of the sol1 shorvn in Fig, 10 are limited and this behaüor has not been modeled.

stages of

Other Meclnnical Factors Applied load or stmin waveform, stress ratio, load spcctrum, and ovcrloads can CF [1ó,22,23,54-5ó]. This ^fÍect chapicr emphasizes uniaxial tensile loading of CF specimens because en\,ironmental effects on Iatigue under tor.sional or multiaxial loadiag have not been studied- An exception is illuslated in Fig- I l, showing fatigue lifc data for smooth spccimens of AA?075 in moist air and aqueous NaCl [57]. Note the strong e¡vironmental degradation of Nt fcrr uniaxial loading rcprcsented by

but a rcduced effect for torsion

¡ormal stress range (^o), aL a given applied shear

stress ránge. Such results can be explained based o¡ thc deleterious role of triaxial tensile stresses, and the associated high hydrostatic (mean) tension, in CF propagation by HEE; this stress state is present for uniaxially loaded Mode

I crack, but

not for torsional loading and Mode

II or III

cracks. Fatigue initiation may be similarl!, environment enhanced for unia\ial tensile and torsional loading, explaining the modest reduction in Nt for torsion (Fig. I 1).

Electrochemical Variables Ebctrode Potential Both anodic and cathodic polarization can affect CF, with different trends observcd for crack iniliation compared to propagation, and for steels comparcd to either aluminum o¡ titaniúm alloys. For ferritic and martensitic steels in aqueous chloride solution, high-cl,cle CF occurs at electrode pote¡tials near free corTosion for aerated solt¡tion (Ec = -650 mvscE), but is often reduced in severity or eliminated b¡' cathodic polarization to ncar -1000 mVSCE [58]. This beha\.ior is illust¡¿ted in Figs. ó and 12 for notched and

if a critical

=

200

cor'rosion and water rcduction at cathodic potentials [,]ó,ó0].

Cathodic polarization of aluminum and titanium alloys in chloride provides an interesting contrast to steels. Duquette

a

ln l% N¿cl,

z ii

O

rn l?¿ N¡Cr

t

v scl:

API-2H Steel 3% ^K=23 0.1 Hz

l\¡Pa\ñ

B=0.1

NaCl

I

ct

5 e o

f=3Hz

r50

In this

enüronme¡t-cnhanced fatigue crack initiation LI 1,38,39): hydrogen plays a secondary role for the fast loading f.equencies, near-thrcshold stress intensities, and uniaxial stress states t)?ical ofsmooth specimen studies. Slow loading frequencies and cmck tip hyd¡ostatic tension plomote crack growth by HEE. Here, impofant contúbutions to crack tip hydrogen producrion are from cmck acidification near [Tee

l0'1

E = Ec = -1.0

anodic cu¡rent was exceeded [3&39].

study, CF was cssentially eliminated by solution deaeration rvhich reduced the srcel corrosion rate cuEent and free corrosion potential (Fig. 2). Applied cathodic polarization similarly reduced CF of polished spccimens. Understanding the effect of applied polarization on CF propagation requires a description of crack tip electrochemistry, particularly local pH and potential, as aftecting rhe kinetics of passive film fomation, dissolution, hydrogen production, and hydrogen entr,y. Occluded cmck processes a¡e complex, as are the obserwcd dependencies of CF d¿ldl1 on electrode potential. For example, CF crack glowth rates for steels in chloride incrcase with incrcasi¡g cathodic polarization, rvith a modest minimum in daldN at about 200 mV active to the free corrosion potential [1ó,20]. Figure 13 illustratcs that thc ennironmental cnhancement of d¿ldN increases with the total mte of H production at thc cmck tip, raised to the -/4 power, for C-Mn steel in NaCl (Ec = -ó75 mvscE), polarized betu,een -750 and -1325 mVSCE [1óJ. Hvdrogen production was calculated f¡om a crack chemistry model [60]. For this system, solution deaemtion does not affect CF d¿ldN whe[ electrode potential is fixed potcntiostatically [ó1]. The oppositc effects of polarization o¡ smoolh specimen CF life and crack propagation in steel can be reconciled. Dissolutiorl and pitting probably govern

3

to3

ENVIRONMENTAL CRACK|NG 7

smooth-uniaxial fatigue specimens, r€spectively L47,591. CF of polished specimens of 1020 and 4140 steels, exposed to NaCI during high-frcquenc!¡ r'otating bending, occurred only

E

t(]2

2ó-

100

&

r00

5.0

Clacl

t05

I07

t06 Nu,nbcr

d

to8

109

Cyclos ro Fai urc

FIG. 12-The benef¡c¡al efiect

o,

cathodic potarization

on h¡gh-cAcle CF crack ¡n¡tiat¡on ¡n a C-Mn stoel ¡n 1% NaCt solution [591.

Ip

Hydrogen Produclion Barc (¡,/cm¿)

Fle l3--The effect ot electrode potential on relative CF crack propagatlon rate ¡n the C-Mn steeuchlor¡de system, as portrayed by the crack t¡p hydrogen production curent calculated tom a crack chemistry model. The equat¡on of the line lrom regression analysls is: y = 2.39 + 0.22 )q whe.e y ¡s the normalized (dimensionless) crack grcvrth rate and x is crack t¡p hydrogen productlon rate ln A,/cm2 [ ,6].

8

CORROSION TESTS AND STANDARDS MANUAL

and others reporLed that the fatigue lives of smooth specimcns of AA70?5 and Al-Mg-Li in NaCl sol!1tion wer-e ma\imum at potentials mildly active of free colTosion: both anodic and highly cathodic polarization degraded corosion fatigue life [15,57,58]. CF propagation occurs at the f¡ce corrosion potential, is exacerbated by either anodic polarization or extreme catbodic polarization, and is arrestcd by modest cathodic polarization. This trend was demonstrated tor AA7o79, M7075 and AA2090 in NaCl and explained based on HEE [19,ó2,ó31. At ncgati\.e potentials from cathodic polarization, crack tip hydrogen production may be rcdr,rced by the effect of alkaline occluded-cmck solution on the overpotential and exchange current density for hydrogen production on strain-bared surfaces, and hydrogen uptakc may be blocked by crack su¡-face films.

t

aK¡ tsi-3/2

2 a5338,

4 68¡0 20

40

60

40

60

A508

STÁGNAT¡T

PIVR.288'C

o0t6? Hr

. ¡ +,

o

O0l8 %S 0.0t3 %s O00S

rS

ASI¡E

í980) Rf

¡I

IIET

0.25

Sulfur lon Content Sulfur ions in electrolytes exacerbate CF crack propaga

tion in many alloy;. Various forms of sulfur are important in sour oil or gas well and geothermal brine enüronments, and are also ploduccd in unlikely ways. For example, the daldN versus data in Fig. 14 shorv the CF behavior of ^K steel in sterilc 3.5,/¿ NaCl solution with martensitic HY130 CP, an environment that enhances cracking by t\vo- to eight-fold relative to moist air ar¡d more compared to FCP

Postgate C

E iol

;tf (4,

NaCr (-1000 mvscEl

(,

o o 0_r H2 R=0.1

(ar MotsT atR

11050 STRESS ¡NTENSITY RANGE

(MPa/n)

FlG. f ¿t--CF crack propagaüon in HY'lg¡ stel ¡n aqueor.s (sterile) chlorlde and active sulfatereduc¡ng bacterla envi-

ronrE¡ú§, boh_úth CP to

-l(m

mVSCE, AK values

Jm r€lr applied at R of Ol and f of 0.1 tq wh¡le AK < 20 [4P a vm riras applid at corstant túmx ol aborre 20 MPa

MP a v m and f of 1 or 5 Fl¿ Ttle dashéd llne repr€sents l¡leratu¡re data lor FCP ¡n the sleewacuum syslem [64],

«l

4 68t0

20

AK, Ht¡ ñ-¡/2

FlG.

ls-The effect oI alloy ¡rpurity-sulfur content

on CF crack propagal¡on ¡n the GMn

steeuhlgh-

temperature pure waler system [221.

dissolved H2S. In a second example, CF is promoted by increased impuritysulfur in ferritic steels subjected to low-fi'equency loading in pressurized purc water at 288oC (Fig. 15) 122,231. MnS inclusions, which intersect clack flank su¡'faces, dissolve to enrich the occluded crack solution in sulfidc. These anions promote cmck advance by increasing the anodic charge that is passed per film rupture eventr or perhaps by the HEE mechanism. This effect of steel sulfur content is severe for a

.E

,o

2

in vacuum. Note the strong eovironmenial effect caused by a sulfate reducing bacteria (SRB; desulfrovibrio r.ulgaris) cultured in Postgate C food-medium and with cathodic polarization [ó4]. Thesc, and similar data fot both biologically actil€ a¡rd abiotic gaseous H2S saturated sterile chloride solutions, are interpreted based on sulfide-stimulated HEE 164,ó51. Bacteia metabolize the food-source and consume hydrogen to produce sulñde ions and other sulfur species. These anions poison recombinant hydrogen desorption to reduce H2 production and increase hydrogen entry into the metal at the crack tip, similar to the deleterious effect of

HY13O STEEL

I

I

stagnant environment \\.ithin the autoclave, and is eliminated by turbulent solution []ow rvhich reduces sulfide buildup within the cmck [óó].

Other Chetnical Factors Solution pH, ionic composition, conducti\¡ity, and temperature can affect CF [11 17,22, 23,43,58]. M e tallurgic

al V ariable s

Metallurgical factors ca¡ influence CF c¡'ack initiation and growth. Prominent in this iegard are se¡sitizatiotr of gmin boundaries in austenitic stainless steels [22,2J], locally inte¡se slip in aluminum alloys from dislocatio¡ interactions rvith shearable precipitates [ó7], and metalloid

CHAPTER

26-

ENVTRONMENTAL CRACKING 9

impurities segregated to giain boundaries in stecls [ó8], CF crack initiation often occurs at surface intersecting inclusions that concentrate strain and may dissolve to produce a locally aggressive e¡üronment [ó9]. CF crack propagation in the ferritic steel/aqueoLis chloride system is apparcntly not aifected by substantial changes in steel composition, microstnlcture, and yield strength [5/]. In another case, aluminum allo¡, processing route, and the resulting degr-ee oI recrystallization and gmin size, did nor substa.tialiy affect aqueous enüronmental LCF life and FCP kinetics [35,44,701. Increasing copper content in precipitation-

HCF to notched cases. CF experiments follorv direcdy from procedures for mechanical tests with bcnign enüronments; the latter are well-developed and standardized, while CF expeúments a¡e not. For example, ASTM Committee G-l has published 14 standards or practices for SCC experiments (see Volume 03.02, Section 3 of the Annual Book of ASTM Sta¡dards), but none for corrosion fatigue. CF experiments arc hindered by several commor¡ factor.s. Aggressive environments are difficult to contain at a con-

hardened Al-Zn-Mg alloys reduced the eDvironmental enhancement of da/dN, rvith the importance of the effect dcpending on aging condition [ó7].

enced by many interactive mechanical, chemical, and microstrl¡ctuÉl variables that must be factored into ex-

Relationship between CF and SCC Corosion fatiguc is related to, but uniquely distinct from, SCC [7,/]. Purely time-dependent CF crack propagation in SCC-prone alloys is govemed by the integrated amount of time-based caack extension per fatigue load cycle. In such cases CF and SCC occur by the same mechanism and are affectcd by the same variables, as modeled by simple linear superposition L49,52,53). SCC is discussed elsewhere in this

manual. Time-cycle- and cycle-dependenl CF are more complex, involve unique mcchan¡sms, anJ occur al slress intensitics rvhere SCC is insignificant. In these cases, the CF damagc mechanism is unique for reasons traceable to cyclic loading, and including: (a) increased crack tip stÉin rate [22241, (b) resharpening of the blunted crack geometry during (c) altered crack chemisunloading, panicularly at high ^K,evolution of persistent slip try by convective mixing [72], (d) band, slip step, and dislocation ccll structures into embryonic damage, often at low Aep or lou, LK Í73,74), ar,d, (e)

mobile dislocation transport of hydr.ogen in the cráck tip process zone [18,50]. These additional factors must be considered when inter?¡€ting and modeling CF data [,/ó].

Literature Sources for CF Data and Mechanisms In additior¡ to the examples prescnted in Figs. I through 15, extensive CF data have been published in seveml volnñes f l,

2, 9- I 2, 5

8,7 5-7 81.

TESTING TECHNIQUES Common Elements of a Corrosion Fatigr¡e Experiment

An experimert to characteúze the CF properties of a metal involves cyclic straining of a pl.ecisely machined specimer¡ in an electroll,te. (Precorrosio¡ effects on fátigue are not considered.) Environment containmcnt about the specimen must guarantee constaDt solution pürity and composition. Specimen potential should be monitored, often controlled potentiostatically, and not affected by galvanic coupling to the grips or test machine. The mechanical parameters that must be measured depend on the experime¡t, be it HCF, LCF, LEFM-crack propagaLion or notched, and are paogfessivcly more difficult to monitor ñ!m the

stanl condilion, and hinder precise measurements of

specimcn displacement, load, and crack size. CF is influ-

perimental design. It is often nccessarl, to investigate slow, rate deformation and cracking phenomena in a realistic time; experiments must be conducfed for one day to one year or r¡ore. CF damage is localized at surface slip struc, ture and near the crack tip; high resolution observations are not generally available and behaüor must be interpreted from indirect measurements.

Smooth Specimen Ao-Life Meüod

for High Cycle CF

Gooerning Stand.arcls Experiments to characterize high cycle CF life according to the Basquin Law (Eq l) follow from ASTM standards for metals in mojst air (see ASTM E 466, Practice for Conducting Constanl Amplilude Fatigue Tests of Metallic Marerials; and E 468, Practice for Presentation of Constant Amplitude Fatigue Test Results for Metallic Materials). Such methods werÉ detailed for steels and aluminum alloys in aqueous chloride solutions 138,39,57-59,ó9,79, 801. Tlpical data arc presented in Figs. 2, I

l, and

12.

Specimens and knding High-cyclc CF specimens focus failure in a careftilly pre-

pared reduced-unifo¡m or mild-blend-radius gage section, often of circular cross-section and with cnds for gripping in the fatigue machine. T)?ical specimen designs are shown elscwhere, including methods for lorr-damage gage machining and polishing (sce ASTM E 466 and Ref 8/). HCF specimens are loaded in uniaxial tension or bending (threepoint, four-point, or cantilevered) rvith electromechanical, scruoh¡draulic, or. Iotating wheel/mass machines, and grips o[ various designs (see ASTM E 466 and Ref 81). Elastic strainjng is load or displacement controlled; involves eithcr negative, zero, or positive mean stress; and varies with time in a sinusoidal or linear-mmp rvaveforrn. Since CF is dominated by electrochemical su¡'face damage, N¡ could decrease and variability may increase \üth increasing s!¡rfacc area that is stressed.

sl-mmetrically

High-cycle corrosion fatigue experimcnts are conductcd 105 to 109 cycles to failure, at a relatively high frequency of 25 to 100 Hz to conserve time. (Nt of 106 cycles requires 5.5 h, u,hile 109 c-vcles rcquire 230 days of continu-

for

ous loading at 50 Hz.) Multiple, reliable, and inexpensjvc

rotating-bend machines are often dedicated

to

these

experiments. Caution is r.equired rvhen extrapolating the results of relatively rapid frequency experiments, to lower / and/or the very tong life regime in excess of l0e cycles,

IO

CORROS/ON 7ES7S AND STANDARDS MANUAL

Ent ironr rcnt al Cell D es i gn The design of the enüronmental cell depends on electrolyte aod temperatr.u!. A tlpicat cell is illustrated in Fig. 16 for containing aqueous chloride solution at añbient tempemture [79]. Glass. Plexiglas@, Teflon@, or other plastics are adequate celt materials. The specimen is grippcd outside of the test solution to preclude galvanic effects, and the cell is sealed to the round or square/rectangular specimen with O-rings arvay foom the high-stress gage section. Solt¡tion can be circulated hom a large stoÉge volume and at a con-

stant flow ¡ate. The cell should include reference and counter electrodes to enable specimen (working electrode) polarization rvith siaodard pote¡tiostatic (or galvanostatic) procedures [38,39,59,79). When fixing potential or current, care should be taken to correctly ground or isolate the test specimen, to uniformly polarize the gage surface, to account for lR effects u'hen necessary, and lo isolate counterelectrode reaction products. Provided that potential is controlled, there is probably oo reason to control the oxygen content of thc solution. Tempemture is maintained betwecn subzero and boiling levels with a heater or cooling coil, a specimen thermocor.rple, contro! electronics, and a condensation appaÉtus. Cells for more complex environments are detaiied in ensuing subsectior¡s on E¡vironmental Cell

FIG- l6--Lrn¡ax¡ally loaded srnooth-gage specimen and environmental chamber for high.cycle CF experimentat¡on wilh aqueous chlor¡de solutions at near ambient

temperatures

fal.

Dcsign,

Paranteter Measurement, Cotltrol,

and Cotnputer

A ut

ot11.ttio

n

The maximum and minimum applied loads (or, alternately, displacements) are measured with a load cell (or remotely attached extensometer'/L\aDT) and controlled during a HCF expcriment. Ao is calculaied f¡om standard elastic solutions for bars under uniaxial tension, or beams subjected to bending. Elastic strain range is calculated f¡om and gage displacement is not twically measured.

^o/E Crack Delection Total load cl,cles to failure are measured, but crack initiation and growth are not monitored during an HCF experiment. IDfoÍnation on these siages of HCF is critical, but

difficult to obtain. Methods for CF crack detection

are

discussed in the subsection on Small Crack CF Methods,

Data Analysis and Ettaluation High-cycle CF data are presented in a logarithmic plot of stress range versus cycles to failure according to the Bas' quin Larv (Eq 1). Dara are also plotted as srress amplirude (Ao/2) 't ersus reversals to failure (2N¡ for simple wave,

forms). The Basquin Law is based on the initial applied elastic srress mnge and does not consider the complicating effect of a growing crack on Material property constants o; á, and the endurancc limit^o.(or fatigue strength at a given number of cycles) are environment-sensitive. The mechanical information that should be l-epoñed \ ¡ith HCF data is sta¡dardized (see ASTM E 468), and electrochemical factors should be cited as discussed in this manual. Design and alloy development studies require mean as well as minimum (lower bor¡nd) Basquin relationships and statistics on the distribution of versus Nl data. This distribution depends on the fatigue^odamage mechanism. Most

CF studies have not provided such information, often because of time-intensive slow frequency exfrcriments a¡d tbe man]¡ important variables. A sense of the variability in CF data is porlrayed by the extensive HCF results collected for C-Mn and alloy steels in manne environments [58]. The distribution of HCF lives at a given stress range $,as analyzed based on Weibull statistics applied to a data base of 30 000 points for stecls in benign environmcnts [82]. While instnrctivc, these statistical results are not necessarily relevant to high-cycle CF rvhere electrochemical reactions introduce additional vaúability. Futlrre work must address two areas to proüde the foundation fol. statistically based analyses of high-cycle CF (as rvell as enüronmental LCF and FCP). For simple laboraton, conditions, the Weibull analysis of mechanical HCF failurc probability [82] must be extended to include CF. Second, variable load, temperature, and environment chemistry histories are likely to be complcx in applications and significantly affect CF life. Such historv effects have not been studied. The scaling of Basquin relationship data to prcdict the life of a strucnrre is qualitative and uncertain. Either the local st¡ain approach to CF crack formatio¡/early growth lifc or the fracture mechanics analysis of CF propagation proüde a better foundation for life plediction and failurr: analysis.

Smooth Specimen Merhod for Low.Cycle CF ^€,(N) Goteming Standards Expeúmcnts to characterize low-cycle CF life according to the Coffin-Manson rclationship (Eq 2) follow f¡or¡ an ASTM standard for LCF of metals in air, and a classic ASTM manual on laboratory methods (see Ref 37 and ASTM E 606, Practicc for Strain-Controlled Fatieue

CHAPTER

26-

l ala

Testing). Lo\\-cycle CF experiments are detailed elsewhere 133 36, 74,831, and r\pical data are prcsented in Figs. 3 and 7.

ENVIRONMENTAL CMCKING

II

,qI

5

-

Specinrcns and Loading Unilbrrn-gage or mild-radius (lrourglass) round specimens are subjected to uniaxial tensile loading rvith rigid gripping

and a test machine that pnrüdes for either tensior)compression or tension-telsion snaio cycling. Square or rcctangular specimens have also been successfully cmployed. Specimen aligúnent is critical for this rigid gripping systcm [8,1]. Total axial displacement is generally controlled,

with regard to maximum and minimum limits as well

time depcndence, while load is measured, often with a computcr-automated servoelectric ol servohydraulic closed-loop test machine. Such machines proüde a variety of stÉin-time profiles. Loading frequencies are spically less than those employed for HCF experiments because fuilure occurs in less than lM cycles. Lorv cycle CF experinrents diffe¡ f¡'om the HCF case in that gage displacement must be measured with sufficient resolution to apply bctwee¡ 10r and lfr]. Env ironntental Cell D es ign

l. Wo.l,ñC

.ñ.ñ!..

3 6rip¡ 6 r.i!lo¡.! i.r. 6 R!l.r.nc. .r..rod. 9. iúb¡.r Dond¡ lO c.rni.. .r..ro¿.

.l..trod.

2. !uc¡r.

§. E¡tan.oñ.r.r

as

7

o-.iit

^sP

Cells for environment containment and control are i
tical tor LCF and HCF (see previous subsection rvith this title), LCF cells are, ho\\,ever, more complicated becalrse of the requircment to measurc specimen gage displacement, as illustrated in Fig. 17 [J3,3ó], For simple aqueous enürDn-

mcnts, diametral ol axial displacemcnt is measured by a contacting (but galvanically insulated) extensometer, perhaps cmplol.ing pointed glass o¡ ceramic añns extending from a bodv located outside oI the solution. A hcrn, etically sealed extensometer or LVDT can be submergcd in many electrolytes over a range of tempeÉtures and pressures. Alternately, the specimen can be gripped in a horizontally mountcd test ñachine, and half-submerged in the electrol],te, \vith the exte¡someter contacting the d¡y side of the gage (Fig. 17a) [J3]. For both simple and aggressive environments, grip displacement can be measurtd extemal to the cell-contained solution, as shown in Fig. l7b for hightemperature u¡ater in a pressurized autoclave [34,3ó]. It is necessary to conduct LCF experiments in air (at temperature), with an extensometer mounted directly on the specimcn gage, to rclate grip displacement and specimen strain.

Pofa Éfer Measurenrcnl, Conlrol, and

C o í1p

Lt

te

t Aut o

t

nat io n

Applied load and gage displacement are

measured

throughout an LCF experiment. Frorr¡ these data, it is possible to calculate true and A€P, where axial plastic ^d,-^eT, strain mnge equals (^Er For thc hourglass speci^o/Á). can bc measured at any men, tñ.re total diametral strain time, and converted to axial strain according to [37]

E=(o/E)(1,2v¿)-2ed

(4)

If or is mair¡tained constant by the test machine, ^s¿ then ^Er increases and decreases until the hystercsis loop ^d at a constant^Ép stabilizes form for a ryclically hardening material, and üce-versa for a softening alloy [237]. The stabilizedIoop value of axial is used in the Coffin-Manson Law to correlate N/, When^€p substantial crack glowth occurs, these

aAJú\n.,h:eh.¡.BurFun¡r

¡3. o¡l'sphst dkrd¡d ¿rÉ.¡

'nd.!

b

FlG. l7--€ñvlrorünental chambers for low.cycle CF in: (a) an elecholyte at aÉ¡erú ternpemtut€ 1331, ar¡d (b) h¡gh-lentperature water [361.

a¡d stmin ¡angc calculations are not accu¡?te. E\periments are often terminated and failure is defined for a pcrcent decline in Tota! shain mte, average plastic stmin rate, or ^o. of displacemcnt cycling should be maintained the frequency constant during 1]rc lo\uc,vcle corrosion fatigue cxpeúment. Personal computer programs are ar,ailable to control the stress

closed-loop test machine during an LCF expcriment and to analyze load-displacement-time data. Crack Detection, Data Analysis, and EvaluLtion Thcse procedures are ide¡tical for LCF and HCF, as dis cussed in the preüous subsections, Crack Detection and Data Analysis and Evaluation.

Fracture Mechanics Methods for Corrosion Fatigue Crack Propagation Gotenting Standatds Experiments to char¿cterize CF cÉck propagation kinetics,

in terms of LEFM Paris relationships (Eq 3) and near-AK1¡¡

12

CORROSION ?ES?S AND STANDARDS MANUAL

data, are guided by an ASTM standard for metals in air (ASTM E 647, Test Mcthod for Measuremcnt of Fatigue C¡?ck Growlh Rates), a compilation o{ laboratory experience \vi1h this standard [84], and a review of inert enüronment FCP testing [85]. The ASTM standard contains an appendir specific to CF crack growth in marine environments (ASTM E 647 and Ref86). Procedu¡es for CF in other enüro¡ments are not standardized; horvever, methods have evolved for specific technologies [2,4]. CF d¿ldN data are presented in Figs.4,5, 8,9, and l3 to 15,

Displaceñenl transducer

Specimens and loading CF crack propagation data are obtained with a variety of notched and precracked specimen geometries that are q'ellcharactenzed in terms of stress iñtensity and compliance functions [85,87]. The Mode I compact tension, single-edge cmck, three- or four-point bend, altd center-cracked tension specimens are used commonly, and are prepared based on standardized procedures (ASTM E 647). Specimen in-plane dimensions are selected to guamntee elastic deformatio¡)

with limited small-scale crack tip plasticity.

Specimen

thickness, and the plane stress or plane st?in deformatio¡ state, is a variable. Closed-loop servohydraulic test machines are most effective lor CF propagation experiments, ñith the control parameter being load. (Load-line displace-

ment or crack mouth opening displacemeñt cor¡trol are

somctimes used.) Such machines proüde a range of load or

displacement waveforms (sinusoidally time-dependent is q¡pical) and frequencies (10-4 to 100 Hz arc tl?ical !alues). Env i ro nn tent al

C ell D es

ign

Cells fo¡ CF crack propagation expenments are designed to control solulion composition and specimen electrochemistrl analogous to HCF and LCF (sce previous subsections with this title), while allowing for measurements of applied

load, crack opening displacement, and crack length. A variety of approaches is reüerued elseu,here, including electrochemical details for seveml enüronments [4]. Two cells for ambient temperature electrolytcs are shoi!,n in Fig- 18. If the test machine and loading axis are mounted in a horjzontal plane (90' to the normal orientation), thcn the cornpact tensio¡ crack tip can be dipped into solution r,vhich is exposed to air without complex sealing (Fig, l8a). The alternate approach in Fig. 18b employs a vertical loading axis \üth a segmentcd Plexiglas or Teflon cell, sealed to the faccs of the compact tension specimen i!,ith O-rings and including a short segment in thc notch mouth. Solution flolv is

through the notch, parallel to the cmck f¡ont. Small cells clamped to specimen surfaces (Fig, l8b) arc equally effeclive for center-through-cracked plate specimens. Specimens with sur{ace cracks ot single-edge cracks are effectively contained in environmental cells that are large and sealed to the round or flat ends inboard from the grippjng points 188]. Compact tension and surface crack specimens have been contained in pressurized metal autoclaves \vith electrolytes or steam at elelated tempemture (Fig. l9) [89]. In this particutar stud¡r, a yoke rvas employed to simultaneously tesi two spccimens in one autoclave and load ñame. Refere¡ce and counter electrodcs are readilv employed in the various cells (e.g., Fig. l8b), perhaps with salt bridges to

FlG. l8-+nv¡ronmental chambeE for CF

in

electrolytes

where: (a) the CT spec¡men is loaded horizontally and dlpped ¡nto solution [4], and (b) the CT crack is enclosed by a clamped cell and polarized potent¡ostat¡cally. ln parl b load (Pld¡splacemed (^) data are recorded by computer and autographically for coÍrpliance measurement of crack length and lGr,

remote containers and including high pressure and tempeÉture compensation capabilities. Procedurcs to eliminaie galvanic couples and to maintai¡ solution puritv arid composition are identical to those cmplo.ved in HCF and LCF experime¡ts. Crack opening displacement is measured with a remote or immerscd extensometer or L\¡DT- The cell must not iüterfere with specimen loading or displacement. For example, in Fig. l8b, thc shaded midportion is a {lexible membrane designed so that cell clamping forces do not alter specimen compliancc. This is importanf for aco.rrate measurements of ctack closu¡e. Typically, metal-belloNs/gasket sealed HIGH VACUUM chamberc are utilized for environmental fatigue experiments in inert or deleterious gaseous en\,i¡ro[ments L4]. Such an apparatus is outside the scope of this review.

CHAPTER

26-

ENVIRONMENTAL CRACKING 13

due to this current distribution in the mctal, arc small (less than 1 ñV) and probably unimportant. Nonetheless, pm-

de¡ce tlictates thal the influence of applied cuúent tested

by employing an independent monitor of

be

cmck

length in limited qualification and calibration experiments. Third, the electrical potential signal olten increases \\¡ith increasing load in the fatigue cycle becausc electrically conductive crack surfaces are increasingly parted. Accordingly, potential should be measured at maximum load, and errors in crack length due to residual "crack surface-shorting contact" should be corrected based on post-test crack length measu¡ements [94]. Finally, solutio¡ flow or temperature changes can upset the stability of the electrical potential signal dtre to thermally-induced voltages and material resistance changes-

GroundJoops betr,veen potentiostat and electrical potendegrade the quality of

tial crack monitoring electronics

polarization as well as CF cmck length and gro\-,"th mte

FlG, l$-Two compact tension spec¡mens mounted in a h¡gh pressure,temperatule sia¡nle§s steel autoclave capable of contain¡ng water, steam, or acldified |üS bearing b.ine environments [489].

Ctack

l-e ngt

lt

M e axre

n

rcnt

CF propagation experiments are unique in the need to monitor crack length, often over a long time. Much has been $ritten on two videly employed methods, dircct or altemating current electrical poter¡tial difference and compliance (see ASTM E 647, andRets U,90,91), as r.vell as on less-tried methods such as eddy current, acoustic wa!e, optical, or

crack su¡'face marking [92]. Compliance and EPD are broadly caliblated and well-suited fo. precision (f25 ¡rm or better), Iong-teIm-slable monitoring of crack length in many common fractu¡re mechanics specimens and aggressive CF enüronments. Thcse methods can be computer-automated for test machinc and stress intensity contrcl (sec follorving subsection). The principles and instrumentation for each mcthod arc detailed elsewhere (ASTM E ó47 and Rcf 8-7). Scvei'al points are pertinent 1o CF. First, compliance alIo$§ monitoring of both crack length and crack closure [9-3]. The advantage of the EPD method is simplicity; it is not necessarJ to imme.se a displacement gage in the electroiyte. (Thc compliance scheme illustrated in Fig. 18b avoids this complication.) The EPD method is best-suited for monitoring the g¡owth of surface fatigue cracks. Second, therc is no eüdence that the I to 50 A dircct or alternating current used in the potential diffcrence method affects CF kinetics. Prcsumably, the resistance to current flow through the metal is orders of magr¡itude lower com-

pared to that oI the most conductive soluiions in an oc

cluded crack. Voltage differences along a crack surface, and

measurements. A successful approach to this problem is based on a specimen (rvorking electrode) that is grounded commonly \üth the test machine, constant-ct¡üent polver supply and EPD amplifier, coupled with a potentiostat that uses a grounded working electrodc. In this case the specimcn should not bc grounded virtually through an operational a¡nplifier. With this procedure, and the seveml-ohm (or larger) altematc clrrent path through thc t]?ical loading system and test machine, it is not necessar,.r to electrically isolate the CF specimen from the grips. As a test of goundJoop inteF"ity in CF, guarantee that specimcn elect¡ochemical current does not change, at fixed electrode potential, upon revesing the polariry of thc cmck monitoring current. Giren the success of the direct-EPD method, with cu¡rent applied tbrough the bulk of the CF specimen, there is generally no need to employ an indirect method based on a foil gage bonded to the specimen surface and solely carrying the applied currcnt. This latter method eliminates the effect of applied current on cmck electrochemistry (if such an effect exists); however, long-term creüce and galvanic co¡rosion associated with the attacl¡ed foil may be important. CF cmck grorvth in the midregion of a specimen is reasonably indicated by compliance or direct EPD, but not by the surface-mounted foil gage method. Pal,anxeter Measurenent, Cont rol, and C o111p Lú er Aut otl tat io n

Applied load and crack length ñom electrical poteltial or compliance are measured as a function of load cycles during a CF experiment. Aparl from simple constant load range (increasing loading, mode¡rr FCP and CF experimcnts ^K) io control the Al(-history, with the are computer-automated mode seleciion based on thc goal of the work, be it mechanism- or application-based. Load js computer-vaded, in real time and af ftequencies between 10-4 and 50 Hz, lo maintain the crack length dependence of stress intensity range according to (ASTM E 647) AK=a,Ko q¡r¡61. -

r.,,

(5)

changes. A R-value is typically maintained constant as ^( programmed C of 0 yields a constant AK ex'periment, §,hich

14

CORROSION TESTS AND STANDA RDS MANUAL

is useful for establishing transient and steady state CF grouth mtcs for mechanistic rescarch. Negative proüde for

a

C-values experiment for measuring near-

^K-decreasing CF. Positive C gives a experi¡¡ent that ^Krs data in a reasonable time ^K-increasing yields and confirms stress intensity govemed cracking when used in conjunction with á nega¡ive C experiment. Guidancc on those values of C that minimize the effect of prior load history is standardized (sec ASTM E ó47)-'

Data Analysis and. Evaluation Applied strcss intensity range and CF crack growth rate arc calculated at regular crack Iength intervals, based on standardized analytical procedures (ASTM E ó47). For continuously increasing or decreasing experiments, da/dll is ^.K difference (sccant) or calculated by either a point-io-point incr:emental polt'nomial method. For constant AK, daldN is dete¡mined by linear regression of cyclic crack length data. The simple secant calculation amplifies daldv variability, rvhile polynomial methods average the growth kinetics ovcr seven (typically) ¿ versus N points. Growth mte variations depend on the size of the crack groMh inter-val. It is particularly challenging to establish physically meaningful

variations from an average gro\lth rate law. FCP variability was considered in conjunction with an in terlaboratory test program that measured daldN versus ^K for a well-behaved high-strength steel in moist air [95]. For this best case and 14 laboratories, &¿/d¡r' variability frtm replicate standardized experiments rvithin a single laboratory equaled between lt3,/¿ ard x50o/o (i two rcsidual stándard deviations about the mean regression curve) at fixed Variability in CF experiments has not been ad^K.quantitatively; however, thc rcsults from the moist dressed air FCP prog¡¿m provide a lorver bound. The increased complexity and prolonged test times t,?ical of CF experiments, as u,ell as the FCP behaüor of more complex alloys, should lead to inc¡eased variability and uncertaintv. Two complications are notable for CF experimentation. data may be affectcd by a crack CF dald1v versus applicd closure mechanism that ^K depends on aqueous enüronment exposure (see subsection, Mechanical Variables) [ó,45]; such CF g¡owth kinetics are defined as an extrinsic propcrlv that may be test method-spccific. As a diagnostic, if CF daldN dcpcnds strongly on /1, if the environmental frevalue is high and increases \^,ith decreasing loading^KrH quency, or if crack arrest occus during CF propagation at constant applied then environment-sensitive crack clo^K, A bilinear specimen compliance sure should be suspected. tracc (o[ applied load vcrsLrs crack mouth or load line displacement) confirms the presence of crack closure. Closure is characterized and eliminatcd approximately for the difa given ü/dN bv reducing the applied AK to and Kñi., Kc! ference between K-o, and K.¡ rathcr than K.", ^¡<.fi, is detemined by a global compliance method that is nearing aDuring tatigue and CF crack propagalion, a südden larye decrease or incrcase in the maxiñum stress inrcnsily level of the load cvcle produces a slfong reducrion in subsequenl gro\t,(h mtes, or socalled delay rctardation. These g¡o\¡lh rates are important, bul arc not stead)-s(a{e and not simply govemed according to Eq 3 [ó-8]. Mosl CF experiments are desigüed and conducred to avoid such

standardization [96]. CF daldN versus AK"¡ data, as well as rrcsults obtained al high constant R (above about 0.7) or high constant K.,. (see Fig. 5 and subsection, Programmed Strcss IntensitJ, Experimentation), are reasonably closure-free and are an inirinsic property for a given materialenvironment system [ó2]. Closure-hee CF data are neces' sary for basic studies of crack tip process zone damage mechanisms, while crack closure phenomena may be important to applications. As an example, corTosion fatigue cracks in steels exposed to seawatcr at lorv R are arrested by cathodic polarization, because calcareous cor-rosion prod ucts precipitate within the growing crack and cause crack suface closure contact at K levels well above zero f20,97)A-K.ü is substantially less than both the applicd and the ^K intrinsic AKrH. Rough intergranular CF crack surfaces, coupled with local Mode II displacements, may also promote crack clost¡re. Closure benefits may, however, be limited to simply loaded laboratory specimens. For example, compression elements of a complex load history can crush corosion debris and crack surface roughness asperitics, and can thus rcduce crack closure. As a second complication, crack tip str.ess, strain, and stmin mte lvithin the proccss zone are more fundamental than Ar( or and gove¡¡t CF crack groyth kinetics. It is ^K.ff, possible to unambiguously calculate the st¡css not presently intensity dcpendence of these more fundamental parameters t16,22-24). Shoji and corvorkers argue that the timebascd mte of mechanical FCP (d¿ldry' in an inert environmeot is proportional to the rate of dislocation emission from thc cmck tip, or equivalently, to the crack tip strain rate [98]. The value of daldr¡ for FCP in vacuum, or more typicall]- moisi air, is therefore an indircct crack tip drjving force parametcr for correlating CF daldN dau which arc also often stated lvith ¡€spect to a time-ratc. In this approach, daldl¡ is the product of daldN and f. An example of this con'elation is shou.n in Fig. 20 which represents the CF enhancement in d¿ldlv rclative to the air case for three

í

1E-04

E

1E.05

ASTM A302-B Sieels At 243 C

E

'r'-""---.

ot

9,¿.0"

.1

.,91E-o7 L]J

! E

re-o¿

.D

E

i:

re¡s ,E-og a 1E-09

1E

0A

IE

07

o datA,0.2 < F < o.7 A HealB,B=0,333 É Heatc,B-07 I 1E-O6

1E¿5

1E,O9

1E-04

Time-Based Ak Rale, mm/second

FlG. 2G-TirE-based CF propagatlon rate tor several heats of A«12€ steel ln high-tempeJature water versus ¡r¡echanlcal da/dt for the sarne alloy ln molsi air and at several R values [991.

CHAPTER high-sulfur-content heats of a C-Mn stcel in elevatedtcmpcrature water [99]. The speculation is that this relationship between thc benign and CF time-based latcs of R, l, and loading cracking is followed independent of ^K, material-e¡vi¡on\\,avefofin, as demonstrated for severa! ment systems [98]. In FiC- 20 tlrcsc parameiers are varied u,idcly: for example, /between l0-4 Hz and l0-1 IIz, and R between 0.2 and 0.7, but a siogle CF crack g¡oüth law is observed. Based on the film mpture (and perhaps HEE) mechanism, environmental d¿ld¡ should increase with increasing crack tip strain rate L22-24), ancl henceTvith the mechanical daldti: as suggested in Fig. 20,

Specialized Corrosion Fatigue Experiments Several new CF experimental methods have evolved over the past decade.

Programoted Stress I ntensity ErpeiDrcntatiotl Real-time computer-control of stress intensity during a CF experime¡t provides impoftant beneñts. For cxample, CF experiments can be designed rvith a large negative Cvalue (Eq 5) to produce continuously decreasing at constant applied K-n" and increasing R [ó2, /O0]. This^K approach minimizes the complicating effect of crack closure aod proüdes near-AKrH data, albeit at high R. The CF results presented in Fig- 5 were obtained based on this method [44]. Since both and R change during this tlpe of experiment, the effects ^( of these trvo mechanical paramcte)'s must be charactcrized and understood lvhen using the constant K-.. method in CF [ró]. Second, variables such as elect¡ode po-

or fiequency are easily to probc subtle changed as the crack grows at constant ^,K growth mtc changes for basic research [ó/,ó4]. A constant segment can be conducted over a¡ interval of CF crack ^1< extc¡sion, then can be step-increased or decreased at tential, solution composition,

^( The data presented in Figs. 9, 13, constant r<;ax [51,ó2,64). and 14 werc obtained with this method.

Cyclic Strain- Induced D issolut ion The cyclic-mechanical depassivation meihod involves mcasuremenf of t¡ansient electrochemical current during cyclic plastic st¡aining of a smooth specimen in an electrolyte at 6xed potential [J4, /0,/]. A three-electrode cell, coupled with a fast-response potentiostat and tbe mechanical

26-

ENVIRONMENTAL CRACKING 15

and the magnilude of current transients depended on the sign of the plastic stmin, and the repassivation characteústics for this system varied with cycle count, demonstrating the complexitl¡ of CF. This method was employcd to rank rhe susceptibilis of alloys to CF, based on the stability of surface passive films U0l1.

S¡tall Crack

CF Methods An impo¡1ant goal of CF experimentation and modeling is to quantitatively couple smooth spccimcn and fracture me-

chanics approaches to unde¡stand tbe total life of compo-

nents u,ith microscopic defects. Studies of the so-called "small crack problem" have contributed in this regard [ó]. Small crack size can be a particularly important variable that affects CF propagation rates 1102,1031. For example, CF cracks sized betwcen 100 and 1000 Lrm grew up to 1000 fold faster than predicted compared to long crack compact tension specimen daldN data at fixed for the case of a ^,K,[,/04]. Such crack high-strength mafensitic stcel in NaCl geometry effccts are traced to differences in crack solution mass transport and crack surfacc electrochcmical reactions that govern HEE and film rupture processes fló,102,1031.

Both electrical potential and high magnification optical methods have bec¡ dcvcloped to monitor the formation and gro*.th of CF cracks smaller than 500 }rm 1105,10ó1. Each method is capable of micron-level resolution. The electrical potential approach monitors average short crack growth into the specimen bulk, while microscopic methods focus on su¡face crack interactions with spccific microstructuml features including inclusions and corosion pits. The data presented in Fig.2l ',vere obtained f¡om in situ monitoring of AM024 in aqueous chloride solution with a long focal length (15 to 40 cm) and high magnification (500x with I pm resolutio¡r) optjcal microscope interfaced rvith a servohydraulic test machine and video system [,10ó]. Micro' sfructLlrally small CF cracks initiated at co¡stitücnt particles aDd grew at mtes that were equal to values obtai¡red rvith the standard LEFM method discussed in the subsection on Fracture Mechanics Methods. While the Alloy 2024T3 to_¡

Oeaéraled NaCl C700 mVScE)

LCF procedures described in the subsection on the Smooih

Specimen Mcthod, are cmployed in this regard. Data i¡clude time-dependent applied plastic strain, stress, and an' odic current density. The phase difference between the mechanical and electrochemical parameters, the strain dependence of the cl¡r'rent density during repeated repassiva

tion rcpair of ruptured suface films, the anodic charge passed per fatigue cycle, and the char€e accumldation with increasing cycles and time are interpeted to probe CF damage mechanisms. For example, peak anodic current density and the cycle-cr,rmulative charge increased with increasing strain rate for LCF of a ferritic stainless steel in NaCl, consistent with the fil¡n rupture model [3,1]. (This current reflects metal oxidation to produce cations in solution and in formation of the passive film, or, collectively, metal removal.) Additionally, the time dependence of repassi\ation

to'5

10'7

110 §ress ¡ñlen§¡ly taclor range (Mp.

FlG.

2l--CF propagallon rate versus

m

'r2)

lor micrGstruc-

^K lo deaerated turally small cfticks in AA2O24 exposed aqueous chloride at f¡xed potential. CYack growth was monitored by in situ optlcal microscopy [106'1.

1ó

CORROSION TESTS AND STANDARDS MANUAL

chcmically small crack effect is not observed for the system h Fie. 2l L1061, data of this tlT)e are impo¡1ant and lacking for a range ofmaterials and enüronmcnts 1102-1051. The use of short crack specimens provides an important benefit for cor^rosion fatiguc experimentation. Since highresolution crack moDitoúng is employed, crack growth mtes arc quan¿itatively defined with small crack extensions and the associated rcduced N- It is possible to obtain accurate low growth raie CF data at low loadilg frequencies. For example, a short crack specimen rvas employed to measure a dald-¡V value o[ 0.5 lrm/cycle at constant A,< and f oI0.0 Hz during a CF test time of l0 days Uó,1051. A l0- to 2o-fold longer test time is required to obtain this measurement with a standard long crack method.

Measurentenl of CF Crack Solution pH and Potenrial It is important Lo measure occluded crack pH, potential, and solution composition because these factors govem CF by either HEE or film ruptu¡r mechanisms. Several experimental approaches have succeeded in this regard fo¡ simple ambient temperature and complex high-tempe¡ature pressurized water electrolyfes L|07,1081- Reference and pH electrodes were located in small holes drillcd in thc compact tension specimen to intersect the CF crack plane growing 6:om the notch, as shown i¡ Fig, 22 U081. This method provides information on the cmck size and position dependence oI local pH and potential, as well as on the cffects of AK, R, and I Additionall]¡, crack solutiori can be sampled ñrom cmck-intcrsecting holes for composilion analysis by ion chromatography and capillary electrophoresis [./09]. Impoftant data were obtained to test modcls of crack chemistw 172,108) aod, to understand solution-

and metal sulfide effects on civen the very small volume of a t)pical

dissolved oxygen [22,23,1091.

CF CF

crack, solution extraction methods are likely to upset crack electrochemistry and alter daldN.

APPLICATIONS OF CORROSION FATIGUE TEST RESULTS Modern Approach to CF Life Prediction A cardinal principle is to design the CF experiment to iso-

late and chamcterize quantitatively an¡r one of the four stages of fatiguc damagc defined in the De6r¡ition subsection of this chapter. The choice of stage depends on the problem, be it pitling-based crack nucleation in a polished and rigorously inspected medical implant o¡ steam turbine blade, or macrocmck propagation from a u,eld defect in a targe offshore marine strucarre. If a CF experimcnt measures total life, without quantifying thc four damage stages, then basic understanding and component life prediction are compromised. This is the situation for standardized highand low-cycle fatigue experiments that measure total Nfof a small laboratory specimen. The CF material properties eñbodied in the Basquil and Coffin-Manso¡ relationships (Eqs I and 2) are ¡ot directly scalable to predict the livcs of components of alternate geometries and pelhaps containing preexisiing flaws. These data cannot be used to test models of CF because Nf embodies cl.clic deformation, microcrack

FlG. 2Hompact iension specimen and electrode arrangement for in silu measurcment ol crack electode potent¡al, pH,

and solntion composit¡on during CF gotrrth [rA8]. @NACE lntehalional. All rights reserved by NACE; rep.¡nted w¡th permlss¡on.

nucleation and coalescencc, smali/short crack growth, a¡d long crack propagation.

Implementing the cor¡€ct method for CF life prediction and CF resistant alloy developmcnt is often controversial, as existing approaches (often based on smooth specimen HCF

data) are challenged by more modem approaches. A consensus is perhaps developing; an effective method couples the local straiü approach to CF crack formation and carly gro{,1h to a gi\€n detection threshold \vith the LEFM approach to propagation in order to calculate the summcd total component life. The recommendation is that the Coffin-Manson approach be employed, but that the number of cycles to f<¡rm a (reasonably) resolvable crack size (perhaps 0.1 to I mm) be measured in place of total cvcles to failure. Equation 2 then describes the CF crack formation rcsistance of a given allov-environment system, with the size of the initiated crack defined operationally. These materiale¡vironmcnt pr.operty data are coupled with Neuber's method or finite element calculations of local plastic strain range in a component to define sen icc initiation life [7]. The ñ'acture mechanics appr-oach should be employed to charactcrize CF crack propagation kinetics, with emphasis on both the microstrl¡cturally small, physically short, and conventional long crack ¡egimes [ó,8]. The Paris rclationship (Eq 3), or more complex formulations, are employed u¡ith stress intensity similitude and an analysis of thc stress and stress intensity condifions of the component to predict CF propagation life, integmted from the size of the cr¡ack formed in the initiation stage [4.1]. A variet¡, of desktop computer programs have been developed for the LEFM po¡tion o{ the fatigue lile prediction problem [/10]. If nonde-

stR¡ctive tesling so indicalcs, then the LCF initiation

portion of the problem can be equatcd to zclo, and thc fracture mechanics integr-ation stafled at the appropriate cxisting flaw size. The output of an integratcd CF prcdiction method is plots as a fu¡ction of applied strEss range, or

of total c¡clic life

CHAPTER crack length versus load cycles at constant applied

for

26-

ENVIRONMENTAL CRACKING T7

electrochemical rcactions, crack tip process zone damage processes, and microcrack advance. Fractographic anallses of CF must be improved, inclnding quantitati\,e measurement oI crack su¡'face crystallography [,/19], and computerized imagc analysis methods to characterize and reconstruct thc CF process [120]. 7.For mcchanistic modeling, CF results must be coupled with tmnsient electrocher¡ical reaction kiEetics, hydrogen permeation, and hydrogen trapping analyses descnbed elsewhere in this manual.

^d,varjspecific material, time, and enüronment chemistry ablcs- Coffin-Manson and Paris Law data depcnd on thc 6. rariables citcd in the section, Factors Controlling Corosion Fatigue. Since a legion of variables is important, a¡d sincc prolonged CF test times are often required, mechanistic modeling of the nucleation and cmck propagation processes is a critical tool to develop algorithms for extrapolating the results oflimited laboratory expcrimcnts [111].

Examples of Component Service Life Pr€diction

with Laboratory CF Data

Acknowledgments

Corrosion fatigue problems hale bcen attacked aggressively in several technologies over the past decadc. The cou-

This chapte. rvas sritten based on the support of the Office of Naral Research (Grant N00014-91-J-41ó4, with Dr. A. John Sedriks as the Scienlific Officer), the NASA Langley Research Center (Grant NAG-l-745, with Mr. D. L. Dicus as Grant Monitor), and the Virginia Center for Innovative Technologv (Technologv Development Center for Electrochemical Science and Engineering, with Professor G. E. Stoner as Director).

pled local stmin-forrnation and

LEFM-propagation approach has not, horvcver, been broadly employed. Early codified design predictions, using elaslic smooth-specimcn fatigue data (HCF-SN) adjusted empirically for deleterious time-depcndent enüronmental effects, are being replaced by LEFM predictions of crack propagation from an inspectionbased or estimated initial c¡ack size [.117]. Examples of this procedure rvcre reported for CF in weldcd offshore structures in the marine em,ironment L1,3,58,111-113f, in-core and out-of-corc components in commercial light water nuclear reactors l2,22,lll,ll3l, oil and gas pipelines U141, and aircraft [,115,1/6]. Emphases focused on the conflicting effect of cathodic polarization on CF cmck formation and

grouth (marjnc structures), the uni§.ing role of crack tip strain rate (nuclear reactors), the dcleterious effect of sulfur co[taminants (nuclear reaclor systems and pipelines), and the CF kinetics of small multiple fatigue cracks (aircraft). Tens of man-years are tlpically required to address a complex CF problem, and large databases for SCC and CF resulted kom these efforts [58,117].

BIBLIOGRAPIIY Fatigue 'o2, Anders Blom, Ed., Eneine€ring Materials Advisorl SeNices, West Midlands, UK (2002).

Gangloff, R. P., "Environment Sensitive Fa(igue Crack Tip Processes and Propagation in Aerospace Aluminum A¡¡oys," in Fati1ue 02, Anders Blom, Ed., Engineering Materials Adf,iso¡l Serviccs, West Midlands, UK, 2002, pp. 3401-3433.

casem, Z. and Ganglott, R. P., "Rate-Limiting Processes in Envi ronmental Fatigue Crack Propagarion in 7ooo-Series Aluminum Alloys," in Ch€ru¡l¡-'v and Electtochemístry of Corrosiotl and S/¡¿ss Corrosiorl Crackirg: A $nlposiunl Horlorinc the Contrib.t lions ol R.W. Slaehle, R. H. Jones, Ed., TMS, Warrendale, PA, 2001, pp.501-521.

Future Research Needs in CF Experimentation

In addition, the eme¡ging body ot literature on the ettect ot precorrosioÍ on látigue lite, particulañy ol áerospace allovs, is impor-

Enüronmental ef[ects have not been rigorcusly incorporated i¡1 fatigue life prcdiction proccdures [1/O]. The timc dependence of CF, the many important interacting variables, and seve¡al uncenainties confound the problem, From the experimental perspective, LCF and LEFM-based laboratory CF methods must be improved to add¡ess the

tanr. These include the tollou'ing: Barler, S. A., Sharp, P. K., Holden, G., and Cla&, C., "Iniliátion and Early Grc$,th ot Fatigue C¡acks in an Aerospace Aluminum Alloy," Fati9lte and Fractu-e of EkgúE¿ring Materiols and Srnc-

f ollowing

unceft ainties:

Methods must bc furthcr developed to pr-obe the growth of single small CF cracks sized below 500 pm, and the interaction and coalcscence of multiple small cracks must be characterized. Both LCF and LEFM approaches must be modified in this rcgard. 2. Near-threshold CF crack formation and propagation, and environmenrdependent cmck closure, must be chamcterized including the impoÍant effect of low loading fre 1.

quency [./,/B]. 3, 4.

Load- and cn\.ironment-spect¡um history cffects on CF crack formation and propagation must be characterized. The statistical distributio¡ of CF initiation/eady growth and crack propagation properties must bc defined.

5.

High-resolution probes must be developed to measure occluded crack chemistry, tmnsicnt cmck surface

ares, V ol. 25, 2002, pp.ltl 125. DuQuesnay, D, L., Underhill, P. R., and Brirt, H. J., "Fátigue C¡ack Gro\rth trom Corosion Damage in 7075-T6551I Aluminum Alloy Under Aitcra& Loading,' lntematíonal Joumal of Fatique, Vol. 25, 2003, pp.37l-377. Fawaz, F. A., "Equivalent Initial Flaw Size Testing and Analysis of TÉnsport Aircraft Skin Splices," Fatigüe and Fmcl re ol Etlgineeing Materíals and Sttuc tltre s , V ol. 26, 2OO3 , pp - 279-290. Sp€nce, S. H., Wilhams, N. M., Sronham, A. J., Bachc, M. R., Ward, A. R., Evans, W. J., Hay, D., Urbani, C., Cra$,ford, B. R., Loader, C-, and Clark, G., "Fatigue io the Presence of Corrosion Pitling in an Alumunum Alloy,' in Fatigue'02, Anclers ^erospace Materials Advisory Services, West MidBlom, Ed., Engineering lands, UK, 2002, pp. 701-708. Sha¡p, P. K., Mills, T., Russo, S., Clark, G.. and Liu, O., "Effect of Extoliation Con¡sion on the Faligue Lile of Two High-Strcngth

Aluninüm Alloys," Ag¡,?B 2r0r, DOD/FATNASA, 2000. Cmenbers, K. M., Crais, B. A., Hillbenf, B. M., Bucci, R. J., and Hinkle, A. J., "Prediciins Fatieue Lil¡ of Pre-Co¡roded 2024-T3 Aluminum from Breaking Load Tes|s," Intenatioftal Joumal ol F.LtiEue ,

Vol.

26

,

2OO4 , pp .

615-427

.

18

CORROS/ON TESIS AND S?ANDARDS MANUAL

Gruenbcrg, K- M., Craig, B. A., Hillberry, B. M., Bucci, R. J., and Hinkle, A. J., "Predicting Fatisue Life of P¡e Com)ded 2024-T3

[23] Ford, F. P., in Enlironment lnduced Cracking ot Meta]s, R. P. Gangloll and M. B. lves, Eds., NACE, Housron, TX, 1990,

Aluminum," lntemational .Ioünal of Fati\u¿, Vol. 26, 2004,

dissena(ion, LehiEh Universit!, Bethlehem,

[1] Prcceedings Inslitule ot Mechanical Engineers

t.ll

Conference on

Faligue and Crack GroNth in Olfshoie Struclures, Inslitüle of Mechanical Engineers, London, England, 1986. Proceedings ot the Second lnternalional Atomic Energy Agcncy Specialists Meeting on Subcrirical Crack Grouah, Vols. I and II, U.S. Nuclear Regulatol'v Conmission Document, NUREC CP-0067, Washin8ton, DC, 198ó. Hudak, S. J., Bumside, O. H., and Chan, K. S., Joumal ot En' elEy Resources Technolo8y, ASME Transactions, Vol. 107, 1985,

1988.

"4. ot Mctals, [25] Birnbaum, H. K., in Enlironmenr lnduced Cmckins

REFERENCES

[2]

pp. 139 166.

[24] Hudak, S- J., "Cor¡osion Fa{igue C¡ack Grou'th: The Role ot Cmck-Tip Deforúation and Film For_malion Kinetics," Ph.D

pP. 629 ó40.

pp.212)19.

t4l Ganglotf, R. P., in Me¡als Handbook: Mechanical

Tesrinc,

9th edirion, Vol.8, ASM International, Metals Park, OH, 1985, pp. 403-4! 0. tsl Turnbul¡, A., Test Methods lor Envi¡ronmenl-Assisted Cracking, National Physical Láboralory Report DMM(A)66, Teddingtoo, U.K., I992. [ó] Süresh, S., Faligue of Materials, C-ambridge Unilersily Press, Cambridsc, U.K., 1991. [7J Dorvling, N. E., Mechanica¡ Behavior of Materials, Prentice Hall Inc., Englcwood CIiffs, NJ, I993. [8] Fatigue and Fracturc Control in Stmclures,2nd ed., J. M. Barsom and S. T. Rolfe, Eds., Prentice-Ha¡¡, lnc., Engle*ood Cliffs, NJ, 1987. [9] Environment Induced Cracking of Metals, R. P. cangloft and M. B. Ives, Eds., NACE, Houston, TX, 1990. []01 Corrosion Fatigue, Chemistry, Mechanics and Microstructure, O. Devereux, A. J. McEvil!, and R. W. Staehle, Eds., NACE, Houston, TX, I972. L//l Srress Cor.osion Cracking and Hvdrogen Embrittiement ot lron Based Alloys, J. Hochmann, J. Slater, R. D. Mccrisht, and R. W. Stáehle, Eds., NACE, Houston, TX, 1977.

t2ól

R. P. Cangloffand M- B. I!es, Eds., NACE, Houslon, TX, 1990, pp- 2l-29. Sudarchan, T. S. and Loulhan, M. R. Jr., Inlernarional Metal

lursicat Revie(, Vol.32, 1987, pp. 21-151.

[27] Crinberg, N. M., In(emational Journal ol Faligue, April

1982,

pp.83-95. f28l Sieradzki, K., in Environment Induced Cracking ot Meta)s, R. P. Gangloff and M. B. Ii,es, Eds., NACE, Houslon, TX, 1990,

pp.l25

l3'7.

[29] Duquetre, D- J., in Environment Sensitile Fracture of Engi neerins Materials, Z. A. Foroulis, Ed., TMS AIME, WaFen dale, PA, 1979, pp. 521 537.

[30] Jones, D. A., in Environment Induccd Cmcking of Meta¡s, R. P- Ganglottand M. B. Ives, Eds., NACE, Houslon, TX, 199O,

pp.265-210.

[.].¡l Embrittlement by

úe

Localized Crack Envimnment, R.

P.

Gangloff, Ed., TMS-AIME, warrendale, PA, 1984. [32] Corrosion Chemisrry- wilhin Pits, Crevices and Cracks, A. Tumbull, Ed., National Physical LaboÉto¡f, Teddinglon,

u.K., 1987. [J3] Yan, B., Faninglon, G. C., and Laird, C., Acta Metallursica, Vol.33, 1985, pp. 1533-1545. [341 Magnin, T. and Coudrer¡se, L., Materials Science and Ensineerins. Vol. 72, 1985, pp. t25-134. lJSl Lin, F. S. and Sra*e, E. A. Jr., Marerials Sciencc ánd Engineering, Vol.39, 1979,

pp.2741.

U2l Hldrogen Effects on Material Behavior, N. R. Moodv

Chune, II. M., Chop¡a, O. K., Ruther, W. E., et al., E¡vironmcntal¡v Assisred CECking in Lighl Water Reactors, U.S. Nuclear Rcgulatory Commission Repoñ NUREG/CR-16ó7, Vot. 16. ANL-93/27, 1993. lJTl Manual on Lo\\.Cycle Fatigue Testing, ASTM STP 465, Ame.ican Society fo¡ Testing and Materials, Philadelphia, l9ó9. lJSl Lee, H. H. and Uhlis, H. H., Metallursical Transactions, Vol. 3,

U3l

[39] Duque(te,

U4l

and A. W. Thompson, Eds., TMS-AIME, Wanendale, PA, t990. Chemistry aud Physics ot Fracture, R. M. Latanision and R. H. Jones, Eds., Maftinus Nijhoft Publishers BV, Netherlands, 1987.

Duquerte, D. J.,

in Environmert tnduced Cracking ot Mekls,

R. P. Ganglotfand M. B. lves, Eds., NACE, Houston, TX, 1990,

pp.45-53. [./5] Duquelle, D. J., Mcchanisms ot Conosion Fatigue of num AIloys, AGARD Repoñ No. AGARD-CP-316,

198 |

.

^lrmi,

[,/ó] Gar¡glott, R. P., in Environment Induced C¡acking ot Metals, R. P. Cang¡oltand M. B- Ives, Eds., NACE, tIouslon, TX, 1990, pp.55-109. t./71 wei, R P. and Ganglotf, R. P., in FÉcturc Mechanics: Per§pecliles and Di¡ections (TNentielh Symposium), ASTM STP 1020, R. P. Wei and R. P. Gan-qloff, Eds., American Societv lor Testing and Mate¡ials, Philadelphia, 1989, pp.233-264. [,18] Iloltord, N. J. H. and Haldie, D., Corrosion Science, Vol.23, 1983, pp.527-546-

t19l Speidel, M. O., Blackbum, M. J., Beck, T. R., and Feenev, J. A., in Con_osion Fatigr¡e, Chemistry, Mechanics and Microstn¡c ture, O- Devcreux, A. J. McEvil), and R. W. Slaehle, Eds., NACE, Houston, TX, 1972, pp.324-345. [2¿]l Scou, P. M., Thorpe, T. W., and Silveste¡, D. R. V., Corrosion Science, Vol.23, 1983, pp.559 575. [21] Güffiths, A. J., Hutchings, R., and Tu.nbüll, A., Scripra Melal

lurgica, Vol. 29, 1993, pp. 623-626.

[22] Fotd, F. P., Joumal of P¡cssure Vessel Technologv, Transac tions ofASME, Vol. 110, 1988,pp. 113,128.

tjól

1972 , pp.

1968,

ll0l

2949)957

.

D. J. and Uh¡ig, H. H., T¡ansaclions ot ASM, Vol. 61,

pp.449 456.

4., McMiUan, J. C., and Wei, R. P., Metalluryical Transaclions, Vof . I, 19 0, pp. 17 41 -17 57. [4.¡] Paris, P. C., comez, M. P., and Anderson. W. E-, The Trend in Ensineerir¡g, Univcrsitv o[ Washingron, Seallle, Vol. ] 3. 1961, pp- 9-14. [42] Taylor, D., Fatisue Thrcsholds, EM S, Ltd., Warley Hearh, u.K., t982. BJI Dawson, D. B. and Pelloux, R. M., Metallurgical Transaclions Feenev, J.

1

A, Vol.5A, 1974, pp.723-731.

taal Slavik, D. C., "Envircnmental Eatigue Crack crou,(h and Micr-oscopic Damaee Mshanisms in AA2O9o," Ph.D. disserla(ion, University of Virginia, Char¡oltesville, VA, 1993.

[45] Mechanics ot Fatigue Crack Closure, ASTM STP 982, J. C. Newman and W. Elber, Eds., Amedcan Societr for Testing and Malerials, Philadclphia, 1988.

[46] Taylor, M. E- and Barsom, J. M., in FEcture Mechanics (ThiF tcenth Conl¡rence), ASTM STP 743, R. Rober6, Ed., American Sociely for Testing and Mate¡ials, Philadelphia, | 981, pp. 599522. [17] Rajpathak, S. S. and Harlt, W. H., in Environmenlally Assisted Cracking: Science and Engineering, ASTM STP 1049, W. B. Lisagor, T. W. Crooker, and B. N. Leis, Eds., American Socicty tor Testi¡g and Mate als, Philadclphia, 1990, pp. 425446. STP la8l Novak, S. R., in Corrosion-Fatiguc Tech¡ology, ^STM Eds., 642, H. L. Craig, Jr., T. W. Crooker, and D. W. Hoeppner,

CHAPTER American Society lor Testing and Malerials, Philadelphia, 1978,pp-26-Í3[a9] Speidel, M. O., in Stress Corrosion Research, H. Arup and R. N. Parkins, Eds., Sijthoit & Noordhoft, Alphen aan dcn Rijn, Netherlands, 1979, pp. 117 183. M. E. and Ganglotl, R. P., 'Modeling Time-Dependent Corrosion Fatigüe Crack Propagatioo in 7000 Senes Alüminum Allors," in Proceedings. FATNASA Internalional Svmposium on Advanced Slrucluül Inlegrhy Methods for Airlramc Durabilit! and Darnage Tolcrance, C. E. Hanis, Ed., NASA Conterence Publicarion 3724, Par-L l, NASA-La¡glev Rescarch Centcr, Hampton, VA, 1994, pp. 441-462. [5,¡] Krishnamu¡^thy, R., "Mic¡ostructure and Yield Strength Etfec15 on Hydrogen Environmen! Fatiguc ol Steets," Ph.D. disseÍátion, University of Viryinia, Charlottesville, VA, l99l. [52] Wei. R. P. and Landcs, J. D., Marerials Research and Stan dárds, Vol. 9, l9ó9, pp. 25-28. [53] Bucci, R. J., "Environment Enhanced Fatigue and Stress Conc sion Cr-acking o[ a Titanium Alloy Plus a Simple Model for (he Assessment of Environmenral tnlluence on Faiigue Behavior," Ph.D. dissertation, Lehigh Unive6ilv, Bethlehem, PA, 1970. [54] Selincs, R. J. a¡d Pelloux, R. M., Metallurgical Transac(ions, Vol. 3,,972, pp. 252s-2s31.

[.5o] Mason,

tsJl Vosikovsky, O., Journal of Testins and Evaluarion, Vol. 8, 1980, pp. ó8 73. [5ó] Barsom, J. M., in Corosion Fatigue, Chemisrra, Mechanics and Microstmcture, O. De!€rctt)i, A. J. McEvity, and R. W. Sraehle, Eds., NACE, Housron, TX, 1972, pp.424-436. [57] Duquetre, D- J., in Environmenl Sensiti]e F¡acture ot Metals, R. P. Wei, D. J. Duquette, T. W. Crooker, and A. J. Sedriks, Eds., ofñce of Nal,al Research, Arling(on, VA, 1987, pp. l-1ót58l Corrt)sjon Fati$e of Metals in Ma ne Envi¡onments, C. E. Jaske, J. H. Payer, and V. S. Balint, Metals and Ceram¡cs Infomation Center, MCIC-8 l-42, 1981[59] Endo, K., Komai, K., and Kinoshi(a, S., in Pnxeedings of the 22nd tapañ Congrcss on Materials Research, Kto¡o, Japan. 1978, pp. 193-198. [óo] Tumbull, A. and Saenz de Santa Maria, M., Metallursical Transactio¡s A, Vol. 19A, 1988, pp. 1795-180ó. tó./l Gangloft, R. P., in Embri{tlemenr by the Localized C¡ack Envi-

ronment, R. P. Gangloft, Ed., TMS-AIME, Wariendale, PA, 1984, pp.265-290. [ó2] Piascik, R. S. and Canglotf, R. P., Metallurgical Transactions A, v 01. 22A, 199 l, pp. 2415-2428. [óJ] Stoltz, R. E. and Peuou\, R. M., Metallürgical Transactions, Yol. i,1972, pp.2433 2441. [óa] Gan8lofl,R. P. and rú'lly, R. G., Cormsion,vol. s0. 1994, pp. 34s-3s4. [ó5] Thomas, C. J., Ed]'!,ean, R. G. J., and Brook, R., Bioloüling, Vol. 1, 1988, pp. 65 77. E., and Druce, S. G., Corrosion, Vol. [ó6] Scotl, P. M., Trus$ell, 40, 1984, pp.350-357. ^. [67] Lin, F. S. and Starke, E. A. Jr., Materials Science and Engineer'ing, Vol. 45, 1980, pp. 153-165. [ó8] Hippsley, C. A., "Hydrogcn a¡d Tempcr Emb rlement Inter actions in Fatigue of 2 1/4cr-lMo Steel,' Haru'ell Laborator! Report AERE R 12322, Oxon, U.K., 198ó. [ó9] Co{ris, R. A., Malktield, A., Boukerou, A., and Haritopoulos, P., in Enviro¡menr Itrduced Cracking ot Metals. R. P. Gangloff and M. B. Ives, Eds., NACE, Houslon, TX, 1990, pp.223-227. L70l Lin, F. S. and Starke, E. A. Jr., Materials Science and Engineer ing, Vol. 43, 1980, pp. ó5-76. [7.¡] Wei, R. P. and cao, M., Scripla Metallurgica, Vol. I7, 1983, pp.

959-962_

tT2lTurnbull, A., and Ferriss, D. H., Coflosion sciencc, Vo¡.

27,

1987, pp. 1323-1350.

l73l Magnin, T. and Coudreuse, L., Malenak Vol. 72, 1985, pp. I25-134.

Srcience and Engineering,

2ó-

ENVIRONMENTAL CRACK]NG 19

lTal Yan, B., Faringto¡, G. C., and Laird, C., Fatigüe and Fr¿cture of

Engineering Materials and S(n¡ctüt€s, Vol. 8, 1985, pp.259-213of Stress Coúosion and Corosion Fatigue Curves, A J. McEv¡ly, Ed., AsM lnlematioDal, Metals Park, oH, 1990. [7ó] Corrosion Fatigre Technology, ASTM STP ó42, H. L. Craig, Jr., T. w, Cr.ooker, and D. w. Hoeppner, Eds., American Society for Tesring and Materials, Philadelphia, 1978. [77]Corrosion Fatiguer Mechanics, Metallurg]-, Electrochemistry and E.sineerins, ASTM STP 801, T. W. C¡ooker and B. N. L€is, Eds., American Society for Tes(ing and Mateñals, Ph¡ladelpbia, 1984. [78]'Proceedinss ot the Conference on Lo\l Frequency C]clic Loading Eftects in Enr.'ronment Sensitive Fracture." Corrosion Science, Vol. 23, No. ó, Pergamon Press, Oxford, U.K-, 1983. [79] Derrick Jones, W. J. and Blackie, A. P., in Environmentauy Assisted CEcking: Science and Engineenng, ASTM STP 1049, W. B. Lisagor, T. W. Crooker, and B. N. Leis, Eds., American Society for Testing and Materials, Phi¡adelphia, 1990, pp. 447 462. [8r] Lee, E. U., in Conosion Cracking, V. S. Goel, Ed., ASM Inter' na(ional, Mctais Park, OH, 198ó, pp. 123-128. [8.I] Czyryca, E. J., in Metals Handbook: Mechanical Testing, 9ú edition, Vol. 8, ASM l.ternarional, Melals Park, OH, 1985, pp. 366-375. [82] Tanaka, T. Sakai, T., and I§,aya, T., in Fatigüe Livcs and Fa(igue Strengths of Ferrous Metáls, I 986, pp. I 25- l 57. [83] Bemstein, H. and Loeby, C., in Oftshore and Arcric Operations, I. Konuk, Ed.. PD-Vol. 10, ASME, Ne\v York, 1987, pp. 253-2s9. [8¿] Fatigue Crack Gro$'th Measu?ment and Data Analysis, ASTM STP 738, S. J. Hudak and R. J. Bucci, Eds., American Socie[_\, for Testine and Matelials, Philadelphia, 1981. [85] Utah, D. A., Cullen, w. H., Májno, L. C., lrt al., in Me(als Handbook: Mechanical Testing,9th edit¡oñ, Vol. 8, ASM Intema donal, Melals PaIk, OH, 1985, pp. 376-402. [8ó] Crooke¡, T. W., Bogar, F. D., and Yoder, G. R., "Standard Method ol Test tor Constant-Load"Amplitude Fatigue Cmck Groi!1h Rates in Marin€ Environments," Naval Research Laboratory R€port 4594, Washington, DC, 1981. [87] Smilh, H. R. and Piper, D. E., in Stress Corosion Cmckins in High Sire¡gth Steels and in Titanirm and Aluminum Alloys, B. F. Br'o$'n, Ed., Naval Research Laboratory, Washington, DC, 1972. pp. 17-78. [88] Youne, L. M., "Environmenr-Assisted Crackins in Bela Tila' nium AIIoys," M.S. th€sis, University of Virgini¡, Chadottesville, VA, 1991. [89] Ceschini, L. J., Liau, P. K., Rudd, c. E., and Lossdon, w.4., in Environmental-Sensitive Fraclure: Evaluation and Compadson ofTest Melhods, ASTM STP 821, S. W. Dean, E. N. Push, and G. M. Ugiansky, Eds., American Society for Testing and Marerial.. Philadelnhi¡. 1a84, pp.42t -442. t90l Advanc€s in Crack Length Measurement, C. J- BeeveÉ, Ed., EMAS, Cradley Heath, Warle)', Wes( Midlands, U.K., 1982. [91] Fatigue Crack Measurement: Techniques and App¡ications, K. J. Marsh, R. A. Smiih, and R. O. Ritchie, Eds., EMAS, Cradley Heath, Warley, West Midlands, U.K., 1991. t92l Small-Crack Test Melhods, ASTM STP 1149, J- M. t al§en and J. E. Eds., American Society [o. Testins and Malerials, Philadelphia, ^llison, 1992, pp. 116-¡ó8. [9J] James, L. A. and Ceschini, L. J., Journal ot Tesrins and Evaluation, Vol. 13, 1985, pp. 409-415. lgal Donald, J. K. and Ruschau, J., in Fatigue Crack Measuremeni: Techniques and Applications, K. J. Mal.sh, er al., Eds., EMAS, West Midlands. U.K., l99l.pp. ll .3;. [95] Cla¡k, W. O. Jr. and Hudak, S. J. J.., Journal ot Testing and Evaluation, Vol. 3, l9?s, pp 454-470. [9ó] Phillips, E. P., "Resull§ of the Second Round Robin on openinsLoad Measu¡ement," NASA TM 109032, Langley Research Cenler, Hampton, VA, 1993. L75l Atlas

20

CORROSION TESTS AND STANDARDS MANUAL

[97] Van der Velden, R., Eualds, H. L., Schultze, W. A., and

Pun(er, A., in Corrosion Fátigue: Mechanics, Metallurgy, Electmchemislry, and Engineering, ASTM STP 801. T. W. Crcoker ánd B. N. Leis, Eds., American Society tor Testing and Mater¡a¡s, Philadelphia, 1984, pp. ó4-80-

[98] Shoji, T., Takahashi, H., Suzuki, M., and Kondo, T, Journal otEngineering Materials Technolog)-, Transactions of ASME, Vol. 103, 1981, pp.298 304. [99] James, L. A., "Effect of Temperaturc and Cyclic Frequencv Upon Fatigue Crack Gro$th Behavior of Se\eral Steels ¡n an Elevated Tempemtur'e Aqueous Environment," Journal of May 1994. Pressure Vessel Technologl, Vol. I ló, Jaccard, R., JourDal of t,I¿01 Herman, W. 4., HeÍzberE, R. w-, and^SME, Fatigue and Fracture ot tnginecring Materials and Struc'

[/0/]

[¡02]

tures, Vol. I l, 1988, pp. 303-320. Amzallag, C., Mayonobe, B., and Rabbe, P., in Electrochemical Conosion Tesring, ASTM STP 727, F. Mansfeld and U. Berocci, Eds., American Socictl for Tesling and Materials, Philadetphia, 1981, pp. ó9 83. Ganstoff, R. P. and Wei, R. P., in Smáll Fatisue Cracks, R. o. Rilchie and J. Lankford, Eds., TMS AIME, Warrendale, PA, 1986,

pp.239 264.

S. J. and Foü, F. P., in Small Farisue Cracks, R. O. Rilchic and J. Lankford, Eds., TMS AIME, Wanendale, PA,

[/ú3] Hudak,

1986, pp. 289 310. R. P., Metallurgjcal Transactions A,

[/04] Gangloff,

vol. 16A.,

1985,

pp.953 969.

[.¡05] Gansloff, R. P., Slafik, D. C., Piascik, R. S., ánd Van Stone, R. H., in Small-Crack Tesl Melhods, ASTM STP 1149, J. M. Sociely tor Testing t-arscn and J. E. AIlison, Eds-,

^¡¡erican l1ó-1ó8. and Matcrials, Philadelphia, 1992,pp. [./0ó] Piascik, R. S. and Willal.(i, S. A., Fatigue and Fracture ot Engi neering Materials and Stmctures,

vol. 17,1994, pp 1247 1259.

[./07] Gabelta, G. and Rizzi, R., Corosion Sciencc, Vol. 23, 1983, pp. 613-ó20.

[/08] Turnbull, A., Dolphin, A. S., and Rackley, F. A., Corrosion, Vol. 44, 1988, pp. 55 ó1. t.¡091 Andrcsen, P. L. and Youns, L. M., "Crack Tip Chemistrl and Growtlr Rate Measuremcnls in Low Alloy Stecl in High Tem peratúre Water," Co¡Tosion, March 1995. U.¡01 Fo¡man, R. G., Shivakumar, V., Ne$,maD, J, C. Jr.. €t al., ¡n Fmctürc Mechanics (Eighlecnth Symposium), ASTM STP

945, American Society lb¡ Tesling and Materials, Philadel' phia, 1988, pp. 78¡-803. [./1]l Andresen, P. L., Ganslotl, R. P., Coffin, L. F., and Ford, F. P., in Fatiguc 87, vol. III-A, R. O. Ritchie and E. A. sta*e, Jr., Eds., EMAS,wcst Midlands, U.K., 1987, pp. 1723-1751. [1.¡2] Scott, P. M., Memoires er Etudes Scicntitiques Revue de Metallurgie, November 1983, pp. 651-660. [./13] Dover, w. D., lntemational Joumal of Fatigue, Vol. 3, 1981, pp. 52-ó0. [/r4] Vosikovsk], O- and Cooke. R. J., Internationál Journal of P¡essure Vessels & Piping, Vol. 6, l9?8, pp. I 13 129. [,]/5] Canslotf, R. P., Piascik, R. S., Dicus, D. L., and Newman, J C., Jou¡nal ot Aircratl, vol.3l, 1994, pp.720-729[1.Íó] DuÉbility ot Melal Aircraft Structurcs, S. N. Atalüri, C É. Harris, A. Hosgard, et al., Eds., Atlanra Techno¡og] Publishel§, Allanta, G^, 1992. [1/7j EPRf Database for Environmenlall], Assisled Cracking, P¡bi ecl RP2006-2, J. Gilman, Manager, Electric Power Research Inslitute. P¡lo Allo, CA.

[.¡18] "Threshold Colrosion Fati$re ot we¡ded Shipbuilding Steels," Ship SLnrctures Commitiee Repon SSC_36ó, washinslon, DC, 1992. t.//91 Slavik, D. C., wen, J. A., and Ganslofi, R. P., Journal of Ma tcrials Research, Vol. 8 , 1993, pp.2482-2491.

U20l Kobayashi, T. and Shockey, D- A., "Computalional Recon-

struc(ion of Environmentá¡ly Acceleraled Cyclic Cra.k Grc\\'th in Reactor Stecls," Paper 5ó3, Corrosion 89, NACE, Houston. TX, 1989.

MIL-H DBK-5H 1 December 1998

lgo

5LÉr

F E+r -

+*i#

I

g:E 3

or1 o¡r 051 i ! ll

F!rtQ.§.§f r"

,

No

¡t

l,lo. 0!f

|,r* S&Éirt rla &om Foi¡ir _ ro-zl ft ! lo-§ *! E : lo-?t { tBó ; I I t¡ I

.Q

tc *

B. ¡J

k ll

-e

&

q

a c

t-)

I

§

Í ¡

I I t I I

Sirer§ §ñt§{y§ii! F$riür Flsngé. A,{, kr-int¿ Figure 3.7.4.2.9(o). Fotigue-crock-propogotion dofo for 0.250-inch-rhick, 7075f7351 oluminum olloy plote with buckling restro¡trt lReferences 3.2.S.f.9(d) ond 3.7.4.2.9(d)1. Specinen Thickness: 0-250-inch E t,íro fient: 509,0 R.H. Specimen Specitnen

Width: 8, 16,36-inches T)'pe: M(T)

3-399

Tenperatu-e: RT

Or¡ent¿Íioh: L-T

MIL.HDBK-5H

I

December 1998

§úrs

f.ai*rar¿y. llb. a{

R

r{: l{r-¡O É--E o.15i&3X 1&¡0 -+ró"to-q!! !..d¡-0.{4-*.ñ0 tt-gú ñ¿!io.

t{é.

olt

§p,§e¡ñrtra liq§lS1g

-r.

§ rc

l5l

i;

516 :!§r

ul

r

c

7:§

e

w tr--l

it

ü É 5

J

l:f

é

llI,

a

E

E

=* 'ar

6 ¿J

a,

a 6 tr

r'j

J l¡ i¡

ffi

d

h,

§F t_

---r §r-

"-i

t

{

I

St.rrr lntfilr¡|, Fcnor Boógr, Al(. Ir¡-¡!r.l't Figwe 3.7.4.2.9(b). Fotigue-crock-propogolion dolo for O.50O.inch-rhick, 7O75T7351 olum¡num olloy plqle with buckling reslroinl [References 3.1.2.1.ó(i) ond 3.7.4.2.9(d) rhrough (c)1.

Speci en Thicknest: 0.475 to 0.500-inch Ent ironuent: 50-95% R.ll. Specimen llidth: 6,8, 16, 36-inches Temperature: RT Orie tat¡or|. L-T Specimen Type: MG)

3-400

MIL-HDBK-5H

I ,l

t

ti

December 1998

ll t¡

k¡s¡lias .¡ttit¡ür

li ?,

i

**

ar

cr !l

:i

{ t3 €

E

.

§

riB, - l.,i ti¡n.$

ü n

rrrti'l {r*ak LÉr$h, ae". ft{rt{4 Figure 3.7.4.2.1O(o). Residuol slrenglh behovior of 0.ó00-inch+hick 7075-T7351 oluminum olloy plole ql room lemperoture. Crqck orientolion is l-T [Reference 3.1.2.1 .6(e)1.

tiiir

\iar&o,l

cf

¡ül-r$t¡¡ §sct¡df ,i¡rr.rár

ta

¿¡¡"+ * ts'l¡

t

¡{. i¡t$

n¡ st§

j

&e

ñ §

, hq'*Y*o

ü E

E 'ñ

*

'i;

(^ §a!

;\

.d*.$-

lilrdlr C¡a.i L*ttq§, l%, .{r!t§

Figure 3.7.4.2.1O(b). Residuol slrenglh behovior of 1.00-inch-lhick 7O75-T735r oluminum alloy plote ol room lemperqture. Crqck or¡enlqlion is L-T [Reference 3.r.2.1.ó(DI.