S / ~·:b8liC~~'''O~r:~:ht.O,th~~~

THE RELATIVE SEVERITY

~OF

NATURAL AND SYNTHETIC SEAWATERS ON

FATIGUE CRACK GROWTH

IN

CATHODICALLY PROTEctED STEEL BY

<0Cra.ig C. Monahan, B.En,g.

.~ ~

Athesis' submitted to the School of Graduate

I Studies.inpartla1 fulfillment'ofthe requirements f~.-degree of

,Mas~Engineering"

.

,

Newf~oundland. . Faculty of Engineering and Applied Science

Memorial University of Newfou~dland .'Fe1?ruary 1988

St~JOhn's'

.~.'

Per.iuion ~a ^bee,n granted

to ~he Nll"\:.ional Library of Canada' to' .icro·fi"l_ 'thia t.heaia aod to lend or' J8.1l copies of the filII.

The author (copyright'ow~er)

S / ~·:b8liC~~'''O~r:~:ht.O,th~~~

neither the' ·the •.ia nor'

,extend·Ve extract• .fr6~

J

t ,aay be prInted or ot't1erwille '. reprodl.l·ced ,without' bill/her

. wi-i tten perdll8ion.

• L'al.ltori.at.ion a 't.e accord'e . A la Bibl1othlql.le nationale

~~·tt.;a~:::e

^.::.

d:i.~~a:~~·"~

de ven4re dee exellpla1rell dl.l

filII.. .

t,'al.lt.eur,~(titulaire.du droit d'auteur)" ae r~aerv~ lea autre. droit. de publication',

- oi ~a th~h ni ~de'. long.

extraita de celie-ci fte

doivent. ttre' \ iiaprill'. ·01,1' auhtnlientr"~pr9dU1t.Ilai"lll 80n auto~i.a·tionlicrite. . .

ISBN 0-315-43346-9

;y

,~.~ .• n

~~~·&~·;~:~·2~~;.'#¥

"'-1,

, \ ,

THE RELATIVE SEVERITY .OF

NATURAL AND SYNTHETIC SEAWATERS

ON· '

FATIGUE CRACK GROWTH

. . IN .

CAtHODIC~LYPROTECTEDSTEEt

r

-,:

.•..

.

' , '

..

.f','

'To

Jilt.

,'.'~'

'-:':-,..- '

~---CC\---':~---~--tl:~-. _~_~.~-_-_~_-'c._."

,

,

..

,

,

:'i.

.

:"~,:

~':'.

;~':' j

IL;J';i;;ia";'~"'"'.~;,"';=)"j,;",,~,~.J

.

~ ABSTRAcr

Two soi.utions. AS!M-substilUlC ocean water and natural seawater, flTC used more or less interChangeably)o evaluiitethe~fatigue crackgrowth 'properties of steels. There is

:;

.

^.

evidence "in-the literature lo'suggest thallhe mechanismresponsibleforaceeleroled'rQligue

c~acl~wth

incathodically

.p~tected

steel

(i:e.~ter

^reduction)canbesignificaolly , , influencedbybulk solution chemistry. E'X.periments wered~signedto investigatethe

relative seventy o(theso tWO

sol~tions

^whenused

wit~odic"nlly

protected steel - sampl.es.The

inveS~.i~atiOn ~~

^{conductedusing}

~ ~

^o.r

S~imens: w~lded :-Pl~les,

" compacttyPe(~specime~.Sand artifioialere~ices:W.elded.T-p\atcsprovided dataonthe

~nitia;ion

^and

propagati.~n bCh~~ior'of semi~elliplicnt cra~k"s: <ir ~mens, ~ere tes\e~

^to

obtain:gack. growth

~te

^v;rsu.sc.ra«k..tip

stre~s

intensity tinge dam,·since it is difficu.\! tQ

d'riv' S,n\ilil,,"';~sfrom' ^w,ld'" T.;'''' da... Th, artific;;i crevi" ","'rim,;IS. w're .\

combined-

wit~ ~larograPliiC

^and

sc'ann~ng ele~tron mic~probe

^{(SEM)analyses10}

s(u~;­

the chemistrY of calcareous depositsand,their influence on water~~clion^kinetics.

.. ;.,"

~

^{\ I}

ag!!i~~t

^corrbsion.

..

,,

. .The e.xlent to which calcareousdeP'1Sits.affec~~the rate for waterreducti~n^{wit.hjJ1 .} theartifi~ial^crevi,:csd~pended.?n the bulk: soli!tkln chernisfry andelectroe~emical^potential under which Ihe.y formi?rl. Magnesium·rich deposits were lesseff~c~Ye in,reduci~g^{the rate} for water reduction thanwe~calcium rich dc)'Osits.Region~of localiZed~rroslonwere evident.on several of thespeei~ns.A·theory is presented which~xplnin~·how·lot:bli:zed corrosion is possible within crevices andcra~kscathodically pro.tected topotenlia~s^{more •}-~^. negative

fha~

^.•780\mvsCe:ttie generaily accepted·pot.entfal

ror.~~plete

p.n:iec.tion

. ~-

III

... : ,.

.<:;li.lca!eous

d~posits:·cathodic pr'pteetiQ~',

corrosion, fatlgue,.hy<hogen ,embrittl~~nt, s~\Vat~r c~emistty,steel"water!'eduction

~elWords:

For the intenncjiate~evel.~ofcrac~s!!'ess intensity range examined. using

cr

'specimens, lhe"two~1L:ttionsproduced·similar crack growth rate data. Crackprop~gatio.n . rates WC!"e generally .between. 2 and 25 times fasterun~ercathodic protectionin seawater than

i~

^air.

Significaru1~rent

^flitigue

bChaviors)V.~

obServed for

the,)V~lded T-p~.

Paligue.cr!lc~sinitialed early within tJ:Ie specimens. tested "in natural~eaw~ter,but propagated at rates close totho~eobserved in air. Synthetic seawater prolonged the initiation phase but gaveriS~tocra~~propagation nl.les that~ei-eabout2

tim1

^higherthan

.' fhose obs'erved inair, 'these results have been explained in >terms ,ouhe- propenies and .'

Ip~ipitation

Jdneticsof cai"careous

de~si~. ~.

^{: .}

Thisthesis~wascompleted at the Faculty of 'Engineering and Applied Science;.

. Memorial Univ$rsity of

Ne~foundland.

Fundin!g for

th~

^{reseo.r.chWIlS}

provlde~

^by

CANM?T.Spe~ialthanks go'oU[~oDr. O. Vosikovsky ofC~NMETfor the enthusiasm' he has demenscratedto~ardsthe project.

The author isifld~btedto Dr. R.M. Hopkins; who has acted as supervisor Dnd provided advice and "guidance

throughoUl~the

sthdy. Thanks are also due tohr.A.S.!.

Swami~sfor his helpfulsuggestion~an4 SlJpplen;entary financial support.

. . . I ,_ .

Manyth~s,to~s.Carolyn Emerson fo;r Iier assistancewit~ usi~g,the scanning .

:~:::~=probe ·,00

to Teeboiea! siNice,

fO~ th,.i'bricatio•.

of ,pccime", ",d

celi'

'The author would also lik,e to thank NSERC for personal financial_SUp~1tthrough a post-graduate scholarship..

ABSTRACT

TABLEOFCOl'<I'/lNl'S

... 0

:b

":ENVIRONMENTAL YARJABI,.ES

,\ ~ti:·.~~::ZZ~;~~·~~~;;~~~~·~~~·~~i~~: .:..,:... .~ :.:: n

2.3.4. Velocitr of Ihe~nvironment. . 13

2.'?S"Cal~odlc Protection... . 14

\

. .

\.0\INTRODUCTION " ..: , : 1 .',

2.0,yARIABLES WHICH INFLUENCE FATIGUE GRACK fROPAGATION 'RATES IN STEEL.. ..j••...•..•.• ,...

..·.1

2.1 MECHANICAL VARIABLES

2.1.1 Crack Tip Stress Iritensity FaclOf Range. . 4

2.. 1.2 Cyclic: Fre:qucncy 5

2.1.3 Stress Ratio :... 6

i:::~' E~~~~ 5~~~h.~.~~~.~~.~~: ....:..·..· .::::::::::::::: ~

2.2 MATERIAL

~

^AluABLES

"_ ~t1 ~\~~~~~~i~~~·~i'.-i~p~riti~·s :.:::~::: ~O

~_ ACKNO\yLEDGEMENTS .

LIST OF TABLES ...

LIST OF FIGURES ..

NOTATIONS ....

. ... iv

. vili

. ix

... •.•• . xiv

.,'.:

:r:~:':" ".;.<~: ;, .";

3.0. ELECTROCHEMlSTRY..WITHIJ:l FATIGUE CRACKS 16

3.1 -ELECfROCHEMISTRY.UNDERFREE'CORROSION

3.1.1 Potential and 'pH Profiles , : 16 _ • \

3.1.2 MechanismQfAnodic Dissolution : 18

3.1.3 Hydrogen .Embriltlemenl al the Fret Corrosion POlcminl.... 20 3.1.4 Effect of Mechanical Variables on Crack Solution

Modification ,... . . 20

3.2 ELECfROCHEMISTRY UNDER CATHODIC'POLARIZATION 3.2.1 Potential and pH Profiles... .. 23"

~t~ ~~~l~;~h~~!a1~-S~~bl~m:nri~:e;~i~;;~~

^{.... ·.. .. 25 \ "}

.1'Modification... . 29

4.0 CALCAREOUS D/;POSITS· , . .. 32

"

^5.0

....4.1... WHAT CALCAREOUS DEPOSmAR"EAND WHY

THEY FORM , ;... :... ..32

4.2- CALCAREOUS..D:EPQSITS

ANO:~ONMENT.

ASSISTED F'ATIdUE !- 33

4.l

F1,~p~M;~~.I~~~!6E~~~~

^..

~~~s.

³⁴

4.4 THE PROTEcrI\lE~A~YREOFCALCAREOUS DEPOSITS.. :.. 38

• . I

STATEMENT OF OBJECTIVES 42

!

... 46 . 46 47

:.. : 48

6.0 EXPJ;RIMENTAL PROCEDURES 45

/

. 6.1

A:.l~rI~l~;l\~~u~:.~~~.·.~~:.~~IMENTS

^..

6.1.2 Speaimen Preparation ..

, 6.1..3 Daia .Collcction ..

6.2 COMPA ·.TYPE(CT)EXPERIMENTS.. 50

6.2.. 1 Apparatlls... . 50

6.2. Specimen Preparation .. : 51

~.1. Dala. Colleclion 52

-6.3 WLOBO T·PLATE EXPERIMENTS

.3.1 Apparatus : .

I6.3.2 Specimen Preparation ..

6.3,3 Dala Collection, ...

/ _·vl

""

... 63 ..·65

7.0 RESULTS. ^{. : :.67} 7.1 ARTIFICIAL CREVICE EXJ::lERIMENTS ...

;:::~ ~~~~jn~p~~~~:...~~~.~~.i.~~~~:~:

7.1.3 Deposit Analysis ...

7.2 cOMPAcr TYPE(CT)EXPERlMENTS ...

67 . ... 67

.. 68 71 74

80

7.3 W,El--DED T·PLATE EXPERIMENTS 76

7.3.1 Crack Initiation Behavior.,,,.. .. 76

7.3.2 CracJc-Growth Behavior :.; 76

7.3,3 Crack ProfileB~havior... . 77 7.3.4 Beha~orof EnvirOnmental Parameters _, 78

8.0 D1SCUSSIONS ...

~:~

^....

8.1 CORROSION UNDER eArnOBlePROTE<l;rI~ 80

8.1.1 ThePrinc:iple of Cathodic Prote.::tion .

. (Mixed:PotcO:tial. Theor)')... . 8r 8.l.Z

~e~:r~so;~~~~~.;t~:'~.~.~~~.~ ... '

⁸²

8.1.3 'Ollter Evidence

of

Corrosion Under Cathodic

Protection ,... . 84

8.1.4 Possible Ex"planalions.,... . 85 8.2 INFLUENCE OF CALCAREOUS DEPoSITS ONCURRENT

DENSITYAN.D~FATIGUE' BEHAVIOR.... ..91

~..

9.0 CON!=LUSIONS 10.0 SUMMARY ..

REFERENC~~

^...:..

~

^..

APPENDIX .

. /

~J

I

/ (

vii

.. : 95

. · 97

. 1'56

. 163

..

usrOFTABLFS

Tal;>le Title ~ase.

1 . Chemic8.J. 8Jlalysis Wld meehanidal properues of 040.2tGrtld~

3~~'W'f 5I'ee1 ..; .. ; : ~~ 99

2. Waterre?~ction d~ta .o~!aiiledfromartificialcreviceexperiments : 99·

3. ~mpartso.n^ofca1c~eousdepOsits found.withinartificial

crevices polarized to -830 mV (S'CE) at5~C... ....tOO ... 100 101 101 ..., .. 102 4. ComparisOQ.~fcalc~usdeposits found withinartifi~ia1

crevices pol,arized to -900 mY.(~CE) a~ S(lC.:.~:...

5', Comparison of calcareousdeposits folind within ani.fili'ial crevices ,polarized fo -§30 mY_(SO~)lit22°C...

6. L!near regression analysis of CT data.:...~.

"J~ Chemical analysis ofseaw~ters...~.. ( .-,'

viii

',-::(,:",

Figure

. LlSI' OF flGURES

ThIo Page

# ~J' Typical fatigue- cf3ck growth behavior for steel in air [88) 103

·~~lii' Effect of cyclic fTequency(j) on fatigue crackgrowthbehavior

':,'{P

^{for steel.}under_~athodjc p~eclioDin NaCl solution (50)... 104 3. Effect ofSl!esSratio~)on fatigue crack growth behavior

for steelin.seawa,ter .(12)... . , 105

4. Differenllypes- of cyclic10114 wavefonns used in fatigue testing... . 106 5. Effect'ofcathodic potential'onthe

fa~~ue.crack

^growthrIlte

. for sn:uclural steel tested at 0.1 HzIn20°C seawater [65) ...,. 107 . 6. Simplified illustration of the electrochemistry within a fatigue

crack~n.d,er -c,ath~icprotection inseaw~ter.; 108 7; Effect ,ofoxygenon current density~uringcllthodic protection

.of steel in sodiu'!U chloride solution [181. .

. .~.'

...

^.

8. . Effec.lof'Q,iygelfon cup-eilt density duringcat~odicllrotection

. of s}eel in seaw~ter .[18.~ : , \ .

.. 109.

. 109·

9. ·S<:hemaiic·represenul.tionor'pHprofiles near a'film-free cathodically p'olarized st¢'ace fort~water velOCities(~)[17J... . 110 10. 'Plot of'calcareous deposit film'thiCknessasa function ofriominal

seawater· veloclly [21].·.... .

11. CUrrentdensity vs.

time

inpotentibstatic conlrol mode forthree

~yl ~~~Agag~f~e~fc~~~:~~rr~J~~:~~~.~~.~~.~~.

'12. ~:;/I~~i~':o" ~;;r~,m~ ft9t~.::':.d:~g.~~.~~.".~.

~

^{3. :}

~Jtr~~;a:~~~~:'~~.~~~~~~.?~0~~~~~~;·.~.~~~~~~.~.~~ ...:...

~~

^(rp)

requ~

^toJolanze steel

specime~s

·to.830 DiV (vs.

Ag/A:gq reference e1ectrode).,for variouscum:~tdensities- [19J.

15) .

Ph~toinicrograph

^of

~It;el

^showing

~n

^StnlctYrt

~d

^manganese

s~fide incl.usions (X500).... J ,

110

..Ill

III

... 112 .... 112 .. :. 113

ix . -

,.

\

16.' Apparatus used in artificiaf crevite experiments.

Figure 'I1d, Page

113 17. Ice-chest with cover removed, showing eleclfOChemicnl cells 114

18.Artificial crevice' assem·bly.... 114

19. Compact type(CT)specimen geometry liS

20. Seawater enclosure used with

cr

specimells liS

21. Relationship between maximum back face strain and crackd~pth... 116

.. ... 123

. 1

7

⁴

22. Effect of crack plugging on minimum back face strain [40]... 116

23. WeldedT.~lateg·eometry... . 117

24. Apparatlls usedinweldej T-plate ,experiments 118

25. Frame used fot mounting welded T-plates.. .. 118

•

26. Block diagram of data.acquisition systeffi_showing hardware

organization.... . t19

27.' Location of active and refea:nccprobe pairs relative to the weld toe 1~0 28. Seawater circuhltion and refrigeration systems : 120 29. Seawaterencl~sureused with welded T-plates.. : 121 30. Top view 'of T-plate illustrating~CPDprobe positions... .. 121

31. Preparation of welded T-plates .. , 122

3 2. ~ntdecay transientsforartificial crevices polarized to

·830 mV (SeE)@ Soc... :..,...

33. Currentdeca¥ transientsfor artificial cre'(1ces'polarized to

·900 mV(SCE)@ SoC ; :· .

34. Cwrent~ytransientsforanificial crevicell polarized to

-830 mV(SeE)@ 22°C... ... 125

35. .SEM

an~ysis

^of

ca1~us

deposits ne.. tip ofartificialcrevice pol~zCdto -830 mV (SCE)InSoC synthetic seawaler. : 126 36. SEM analysis of.calcareoUs d.eposits nearltip'ofartificial crevice

polarized 10_·~30mY(SeE)in 5°S'nhtural seawater...·.'... 127

Figure TIlle Page

I ^.\

135 134 37. Crevice surfaces.aft,erpolarizationto-900 mV(seE)for

5(}O hours in

sop

,nalural seawater (X6,3) ,.... . U8 38. Crevice surfaces afterpolarization to -900 mV <SCE) for

500 hours in SOCs1filhetic

.

seawater(X6.3)

,

128 39. SEM analysis oJ calcareousdepositsnear mouth of-artifiCial crevice

polarized10-900 mY(SeE)in SoC synthetic seawater... . 129 40. SEM analysis of calcareous deposits near tip of artifigal crevice

polarized 19 -900. mV(SeE)inSoCsynthetic seawater... ... _. 130 4r. SEM analysis of calcareous deposits near mouth of artificial crevice

polllnzedl~-900 mV (SeE)inSoC naturalseawa~er... I3l 42. SEM analysis of calcareous deposits near tip ofartificialcrevice

polarized. 10 -900 mY(SeE)in SoC natu-':3.1'seaw8ter... 132 43. Crevice s'urfaces near mouth of specimenpolarized to

-900

mV .

(SCE)forSOQhoursinSOCsynthetic seawater, be.fore removal

ofd·e.po~its.(X6:3)... . 133

=--ot44r.

-'CrCvice

suifaces:near mouth ofspecimen polarizedto-900-mV .(SCE)forSOO~ounin 5fCsynthetic seawater, afttr Tef!loval

of deposiu'(X6.3) 133

4s. CrevicesulfactS after polarizationto-830 mY.~SCE)for

500hoursin22°C'syn~helicseawater(X(i.3) 134

46. Botlom crevice surface~earmouth~fspecimen polarized to -830 mV(SeE) for S(Xlhoursin22°C synthetic seawater,

aft.er removill of deposits(~6~), .

.47. SEManalysis of calcarw\lS depositsnearmputh (botiom surface) ofanificiaicrevice.poJa.rizedto-83Q-mY (SCE)in22°C~ynthetic seawater...•

48." ..S~ManalySis of calcareous deposits near mouth (top sw:t'ace) ofartificialcrevice polarized to ;.830mY(S~E)in22°C synthclic

.scawat~.r ~ _ .

4~. SEManalysls'ofcal~usdeposits nearlipofartificialcrevice polariied ,to-830 mY(SCE)in'~2°Csynthetic seawater...:...

136 137 50.

~1~ri=~~s_8~O~~~)~Si~2~~n~~~~~::r:~~~.c~~ce

^:

^1.;8)

51. SEMariaIysi;

~f caleeXeoo~

^deposits

near tip

of artificialcrevice polarlzed to-'~30mV(SCE} in 22°C nalUm seawater... . 139

Figure

· "

"."

TIde \ Page

52.

~~~~ft~a~~~~~~~:~~~~_~W~vWcSt~mens

SoC natural seawater.. , •. ,

53. Comparison of fatiguecrilck propagationd31ll for cr specimens testedinairwilh CTspecimens tested at-illamV(SeE)in .

-SoC~yntheticseawater ..

54. Comparison of data in Figures 52 andS3 , .

141 142 55. Comparison of fatiguecrack propagation data for.cr specimens

testedinairwithcr~pecimenstested at-830 mV(SeE)in

22°C natural seawater 143

Calcareous deposits on fracture surface of cr specimen tested

at-830 mV(S~E)in 22°Cnatu~alseawater(X~OO):... 144' Crack depth~a function of elapsed cycles for the welded '.

T-,lalc specimens J ;... 145

Fatigue crack propagationra~as a'function o.r .crack depth

for the welded T-plate specimens.... 146

Comparison of ACPDreadingSwithbeach marksfor the welded T-pJale 'specimen rested in air...-

.

...1 147

Crack

profil~for a we1dedT-plate speciment~tedinSoC'

synthetic seawater... :.. 148

~kprofiles for awelded T-plale specimen testedin SoC

natural seawater ,.... . 149

FracturesUrf~

of

a welded T-plate specimen testedinSoC

natural sea)Vater (NATURAL#2~....... ...150

Fraelluesurface ofa weldelT-plalespecimen'te~tedin SoC synthe!ic seawater(SYNTHETiC#l)...

Currentdecaytransien.ts for welded T-plates tested in5°~

seawaler · .

65.. Principle of cathodic protection[76,77] ...

"66. Potential·pH combinations n'ear external steel surl'aces in •

seawater · ..

. : ISO

152

153 67. PotentIal-pH combinations- within fatigue cracks in seawaler..'.::.... 154

"'68. Proposed mechanismfor corrosion undercathodicprotection.... .... 152 xii

\.

Figure

"

~ ^..

Pago 69. Distributionof particle sizein 2m1of solution fora 100

micron diameter aperture

70. Distributionof particle size in 2 ml;of solution for a 280 micron di~meteraperture ....,

\

"

155 155

./

/

xiii

. .-

~---~-~---.-

'crnckdepth BFS backface strain dafdN fatigyecrackgrowthrate

R sness ratio or.load iatio (rriinimumload/maximum 10lld)

T temperature

/

\

·xiv appliedloadrange electroeheini.cal potential cyclic frequency c~ntde'1sity

crack tiPitress~lensityfactor range

N number'of9'cles

lIK f E·

1.0 INTRODUCTION "

:'

Bnvironment~assisled

fatigue is generallyreferrrd10 as .fatigue dllII1l1.geoccurring underthe combined .acti,on of cyclic IOad,ing-an.d anil8~ssiveenvironment. 9'c1ic loading . 1n an aggressiveenvironmeri( suchas seawaterorlen resultsin a significant reduction in fatigueperfo~ancecompared withthat obtained un<!ercyclic loadingin an inert environment such as air.

/

I~recent years, energyshortages have generilledII.drive t!?wards th.e exploratio.n, : driBing 'and: production of oil and gas

in

offsbore environmerits. Manyofthe,st:n,ietures uSed

i'rim:~ne

~ eriVironmerilS are

, , ' made .

^'

~m

carbon and low alloy steels. Even^{~ .}

though

^{. .}

s~h

steels are susteptible to corrosion(atigu~j the~.are widelyus~ ~causeof their relatively low CO.SI

~d

^{ease of}

.fabri~~tion.

S(nce steiels

~·.subject tci~siY~ degra~ation

ⁱⁿ

mari~

environl.!ients,thelossinfatigue.resistance due to corrosion'ml,lstbetaken into account in engi,neering de,sigo. :Designers. engineers and operators mustbeaware of the possible deleterious effects of environment-assisted fatigue on material perfonnanceandmust assess the potential impact of these effects on the design. safety and' reliability of offshore

. "

lstructures.~

: .

,

Be~ot'CIthe severity of fatigued~gecan' be predicted for agiven'.a11~in a marine envii-on"ment.bo(hthe metalandtheeiectrolyte' mus!beunderstoOd.Un~ilrecently;more""

.. was ,known 'aboutthesieel' structures than abo.u(the,seaw,ater)tsclf.Cons~ue~t1J'^-.

·sea~ater.hasoftenbeentre~ted,45aninyanabic="flui~whosepropenitt arewells~muI.attd by3_S~NaCl,soluri?n.~ so~erespeCts,seawat~.is s~risingly ,~onsist~nt w~tldwide.

Inopen'oceanV/litersQfYar}'ing·saiinity.thc.a~soluieconceritriii.ionsofthe majorjnorgaJ)~

i~ns'change~nS:iderably~butthe ratios0f.tbcircQriceniiations.t.o one-another are.constant. •

\

· However. in tenns of temperature. dissolved'oxyg~nconcenlrallon, salinity and pH, seawater isvaria~lein time,geograp~ica1loCationand deplh.Itis the intent

of

thisst~yto focus on theelectroChe~calprocesses occurring within fatigue cracks in an auemplto

~nde!Stand ~~wthese. processes mightbeinfluenced by variations in bulk senwater chemistry. SeyeritI important mechanicalilrtdmaterial vll!iables which are known to internet with the environmental variables have~beenintroduced near the beginning in order tnltlll c~mpleteunderstanding of the overall corrosionfalig~ecnlck.ing p\ss canbegrasped.

·"'

.

...;,

..

: .... (

^.

)

2.0 VARIABLESWHiCH~UENCEFATIGUECRACK

PROPAGATION RATEs IN STEEL

In recent years, thereba~·beena major effort to quantify the effecls of seawater cOITOsio'n and cathodic protection on thef~tiguebehavior of offshore structural steel - -weldmenlS. The prinwyconcem Is with the initiat;ion and propagation of cracks in welded lubular

s~el

inlersectil;lnS under cyciic loading in marine environments. Joints of this tyPe usually contain weld defects which initiate fatigue cracks.Th.epresence of such flaws

'. " . '

elimln~tesmuch of the fatiguecrack initiation 'period an9Jilligue life isgove~edby crack

pJ;opa~ation

fl'9m a wellftoe,into parenl

m~tal

^{[49], For}

~his

^rea,son.

·t~e

^following - discussion, 'bas beenIimi~d.to inchJde on'ly those variables thought 10 influence 'crack .'

p~pagatio~inthe base metal.

Variables ide":tifiedasaffecting corrosion fatigue crack propagation behavior may

,

begrouped

in~l?

three categories:

~ec~aniCa1.

matfrial.\nd 5nvironmental. Mechanical . -:. variables Include crack tip stress intensityfaclor range, cyclic frequency; stress ratioand cyclic load waveform. Malerial variables of significance are slttnk\handthe distribution of impurities such as manganese sulfide: inclusions."f!'eimpcirtant envirol'lmentaJ variablesare tc:.mperature•. pH, dissolved oxygen conceiltration, velocity of the ·environment and electtoeherriicaJ potential

/

""::..^{". .'~''."}

';"".:'

2.1 MECHANI<;AL VARIABLES

.

2.1.1 CrackTipStrESS Inten;sity Factor

^---

Rnng~

When crack.~"ratedata from a fatigue- experiment are ploned on n log-log scale, the curve depictsasomewhat sigmoidalre~ationshipwiih three distinct regions (Figure1. p.103).In region1,the crack growthra~e'Igoes asymptotically to zero as the stress intensity factor

~nge ~K

^approachesa

threshO~d

^{valueliKl!t.This means}

t~at

^fo;

stress

inte~sitie~'

^below·.6.Kth

th~~

^{is no}

crack'gro~t., I:.~. ·the~\s. a.flltigUi'!i~i!,:3he:

thte,!ihold effect is believed to·be causedbyanumber.o( different processes which. lead to crack.bloclarigfor smail

s~ess inten~iti~s. I~ regi~n 2~\

^{the log of}

cr~c~

^growth.

rntem~y

^..

varyIl~earlywithres~tto"thelog of6.K. Infracture~echanicsterms, the~ck gi'owt~

'mte(daldN)d;w,gthispom(mof the propagation stage. has been sucCessfully related to the

. '

' . .

, linearelas~ccracktipstressintensity factOr range(.:iK''''K",.",~Kmin)according to the Paris·

equation [64}

daldN=C(.:iK)m where' a=cracklength

N"number of cycles

../) e,m

~stants"depending on material variables, . environment,temPe~ture,frequencr, etc.

.

. '

For marine structures design; the Hnem: Portion of the curverepres~,jtingregioni

. ·1 crack'growthis most important since thisisthe portion offatigu~where:a crack undorgoes steadygrowth,As~etransition flQIn region 2 to region 3 occurs, the cratkgro~[h~..:) accelerates drnmatic::aIly as ductile tearing comes into play, Crack growth in reg\on ;3 is of . };

.~

'Ii ':.' '-1· ':'.:.;,

·"~;~.*.Lj/

minor imponance

fo~

^marine

s~clU~n

since lhe cyclic

frequ~ncy

and load spectra are usually such that fU1a1 fractw"C isimmin~nt:

The analytical'exp~ssionrelating crack growth rate to the stress inknsity faclor is especially usefulipevaluating the fatigu!=behalli~of,steels under laboratory conditions.

lhe.shVlCof.th~10g.6.K vs, daldN curve usually~oesnot changesjg~ificantly.ror ~stsin- air. However, tests carried out in ag&ressive enviroJ!ments often

giv~

^{log1U(vs. daldN ,: .. ;}

curvq with unusual shapes that vary significandyWi.~stress mtio. cyclic fr-M,uency and

r" '. . . . _

lem~ralure.In'certain cases ofenviron~entenhancedcraCk.groWthi~dNmay increase by sever8.1~rde'"S:~fn:tagnitude -for only a small change in.6.K M;d thenremainessentiall}- COrBtant overso~~range o.f .6.K value.s~I~e'plateau' regiine), At i.ntermCdiate lev.eIs pf . AK, the .crack ,growth rate- becomes depenaant 'uponmechani~al/environmentalvarj.ab,le interac:ti~ns.At the higher leVels ofA~.the crack growth tate is fast enough to be

. .

insensitive to environmenta1.variables and becomes primarily dependant upon mechanical variables alone.

2.1,2 Cyclic Frequency

.

. For tests in aqueous media, the general trend has beCn that lower

~yclic f~u~ncies

~s.ult

ⁱⁿ

in~reased ~ck

^growtll

rat~s

for iow and intennooiale levelsOfAK. Scon e1.

~l'

[65] 'found that the·c~ck.growth rate was not significantly influenced· by frequency'

·variations in the range of I to10H~,byt that de:erl:asing the frequency from 1'100.1Hz 1"CSuited in crack growth rates about four tofive. times thoseobse,ry~dfor air testS under

comP.~b.lc

^loaethtg

con~tion~.

In a later study, Scott et. al'[1]

~b~~rvCd

^{!hllt on}

ch~ging

^.

~.c cYCllc~~uencr.byadecad~'ara

time,

th~crack growth rate changed.relatively slowly

"

to

that~tcrisliC

^ofthe new cyclic frequency. Nibtw;ring

{~l

^has

~ned

^{a similar}

effect.~

f..a.tigue tests

.~~~ed

^onmildsteel freely.

c~ing

ⁱⁿ

sea~aler.

^a

~Iep

increase

in

cyclicfre;quen~SedIsubstantial reduction in the cracl:growthrtte: The former CIlIcltp~gationflte was re-establishedafterthe crack extended" an additional 5 mmDr'So.VosikoyskYet..a}[SO]haV~.o~edan effect of frequency

on

crack growth ratein theplateau.regime (ie. where crackgro'N\brate isinde~ndentof AK) of t1ic log AK vida/dN curve for steels under cathodic protection in sodium chloridesolution (Figure 2, P:I04>/ Slow'er cyclic

~ucncies .reS~lted i.n-s~bSlantia11~

higher piatcau

c~ck

^growth

~tes.Maahn [321 "has established a similar trend for plateilu crack growth ratesliSII

funCti~~ l;I~ CYCI~C ·frequeri'~y.

Several of these frc<jucl)cy dependant

~rfcct~ ca~

^{be .}

explainedintermsof a hydrogcll eplbrinlcment mechanismandha.vebeen discussed funher under section 3.0.

1.1,) StressR.tio

c.:aanging the sads ratio(R~minimumload/maximum load) has the~st pronouncederfectoncraCkgrowth ntefor lowandintennediateIevelsOf"AK.Thegeneral

.,

. -

trendis an upward shiftinthe log AK vs.daldNcurve withincreases inR (Figure 3,

. ' ~ .

. p.105). This istrUe~ortc,s{Jinairas well':Sin seawater, however, therelativemagnitude ofth~se'shifts can vary for testsperfo~edin seawater where certain e':lvironmenlal v~abfescan have' aninfl~nceon the m;asuredcra~kgrowth ute. The locuion aM extem of the plateau region. appears tobeindependent.p( the .stress'ra'Jo,however, the

• l..extrapol,atedthreshold v,alue ofAI<increaseswithdecreasing Rand.

e.

8.1'-5ult. !,he Jow,er . panofIjtepbS,teaubecomeseliminated

ri2l..

i : f :

.,.,,,;;;, ,'.-;'U;·:.,,:';.,·.· _,.7", :.,.•.

_~".,

^...

,<,..,.:"..,,'; •.',.,'_"'. ,._ '."".:.:. ",,_ ...",,,,<;.,,',. ;.~Ji

2,1.4 Cyclic LoadW.ll~eform

Scott et, al [1) invesligated the 'relative effects of sinu!>oidal, oiangular, positive

&awtooth and negalive sawtooth waveforms on the cOlTOsioh fatiglle behavior of BS 4360 Grade-500

St~l

in seawater at

th~e

^different

potenf~ls

(including free-corrosion). The resullS for the triangular and positive sawtoQth waJleforms agreed closely with.those

.

. obtained from Ihesin~soidal-wa:vefonntests, However, the crack growth rates established for the nega.tive sawtooth wavefonn \v,ereconsi~erablylow,cr. It,was ,concluded Ihat !p_e .environment only contributes tocrackel(tensi~n dl¥i~g th~ris,ing~rtensile p0t't:i:on ofeach cycle, One observation that remained unexplained was the absence of a I?late.aui"n

any

of the crack)l"l?wth curveS other thantho~ genera~using'.a sinusoidal wavefoll1l.

Simple programmed (variable low alnplilude)te~ts h~vebeen-carned·oul by Nibbering[2&]to investigate the 4amaging effect of high loadswh~npUxed with lower loads, on thefati.u~crack propagation rate for mild~leel.The wa..ef;;~.consistcdof repeated sequences of a single high load followed by a number of. low loads. In air, Ihe . crack growlh rate undc/programmed loading

was

an order or'magnitude lower than !hat for cot:lstant amplitude loading, Unde}" free cOrTOSion in

seawat~r: rh~

crack growth rate was not significantly affected by the prescACe of Peaks in Ih;'loading waveform.

The

large beneficial effect of overloads, on fatigue performance in aIr was attrlbu'ted to Ihe

. .

compressive residuals~ssesleft~hin3itt the ctack tip. It was proposed that Ihe absence of slgtU(icantcrackretardationf~the; tests perfonned in~watermay have been due tothe altrllCl:ion of

rabsor~hydrogen to theenlar~edplasde zonegen~ted.t'th~'Crack^tip.

'Maahn [30] has carned O\!t a more

extensi~e

inv,estigation into .the

eff~t

^{of -\}

overloads .onfa,~ig.uecrack8ro~~tate retardation in offshore steeL Crack growth -ret~ationeffects were found:foI- leslScarried out in·air as)'Veilas for tesls conducted

\^.

under cathodic protection in scaw:ater.Ac~nsfto.ntamplhude triangulnr waNotrn with single or multiple overloads was used in all lests. The overloads wUe applied. with 6K

t ,

equal to the baselins .6.K. but with a higher.K.n.... Thc·extentofretardation

'1I~sobsefYcd

^to

beinler-dependant on the overload mtio (ratio of overioadpeakload to baselinepeakload), the number of overloads and the .elc;.ctroehemical potential. ,....single overlondlestscairi~dout in ait and in seawater with cathodic protection did notde~nstrateany significant

~ . ,

.

'retardation effects for. small' overload ratios. Cruc\( retardation w.s obserVed t9 increase VII\th increasing overload ratio and/or increasing number of overloads.M~imumbenefit for a_..civen

o~erl~ad'

ratio wall realiztd:·;.v;ith, about 20'

overl~'ads. ~owe~er.

for thc ·tests corid~ct~in'seawater, thebe~efic.iale,tfcct dt<:aycd \lijthdccreas~~cat,hodic polcntial until a potential of about -950 m¥ (SeE). was reached,. at which point uncltpc;ctedcrack anes!

occurred. A's apo~~ibleOltplana9on for this peculiar occumnce, It was 'suggested thai the

~bsorption

of hydrogen into'-the plast;1.c

Wl1~

at the crack· lip might have a strengthening cffcc.t on the st:el 'bY blocking differeri; "slip'

~ecl:lanisms

^and

i~creasinll

the criticl11 threshold 6K(6~lh)for crack growth',Ifthi.s were' cor:ect, then one would ei'Cpcct the re.tardationeff~Ct, ^graduallyitlcrea~e,rattler than decrease,.wit~.decreasing potential..

Alte~ati~ely{30;2],e~hanced p~ipiUl.tionofc~eousdep?s~tsat ·95p mV (SCB), in combinatio~'t'ilh th-C? residual compressiveslr~ssesat I.he crack tip, ml1Y have~en sufficier:lI to lower thc effective stress intensity level(A.Kott) below. the~hr.c:~h?ldvalue. ' This point is funher disCussed in

sect~on

^2.3.5_SeVeral of the cycli,c load'

wavefo~s

mentioned

he~

^{are alsq}

mus~at~

schematically in Figure. 4 (p.106).

!

/ .

- - 9

2.1.5 Cmck Depth

Ctilck lip.~Iectroehemistryc be significanJ.l.y innueI1t:ed by Cl'lIck dORth.especially .'

wh~re

cathodic; protection

~s

being em

t.~2l.

Jones [24] has shown' that th,e 'fatigue crack growlh rate foras4360 Grade 500 steel underfree corrosion isf&~testfonhallow cracks in the depl'b range 0.5-2.0mn1for low values of stress intensity range (AK< 30 MPa.mln);For

d~rcrnck~

^under

lo~

^s[les$

lev~s

and for all cracks under high' toads, .cra~kdepth did not exen asignific~1influence on crackgrowthrarc. Hant CI.aI [41] have - suggested that researchers specify the crack lenglh associated with eachgrowthrate dam , point such thaian~^{effect of}cr3f~^lengtl!-?ponAK vs:d~dN-curves can be reai'ized. The . effect' of crack~eplhupon crnck gTowth rate has been given f\lrther allention,insection

3.1A.

2.2 , MA.TEJUA1..Y^AlUABtES

1.1.1 StTengt~_~sthe·tensile strength of the steel increases, so does its sensitivity

.

:~hydrogen embrittlemcn.tdamage~ ~us,for lowOJ:mediu.m strength steels, the ,mechanism of h)"lilrogen e·mbr!ttlement mao?be less effective than for highStre~gthsteels due to. the . .ln~ased.ductility^ofthematerial. Although"stnlC~steels arc apparently iU1J11une to static stress corrosion crn.cking(bythe: mechanismof.hydrog<;:ne~briulenHmt)[I,6J. there is no •

p~f~at h~~gen~mb.titti~~nt~'y ~es

placc.above

a.c.e~ain

hardne,s level {6]. The ..

fIl.te..ol~ydrciBen

.

entry into'Bleel is thought tobe enhanced by cyclic strain [51] and

.

cleavage'facetsan:~inter:-il'lI;hular failure,which~haracteriStic'~fhydrOgen embrittle<! .'

;

.

10.

steels, have been found o'\.fraclographic examination of cathodicallY polarized S·N' samples ofas4360 Grade 500 Steellcstedir;anificial seawater at -1000 mV(SeE)[66]'.

2.2.2 Distributionoflmpurities

A slu(iy [27J

int,

tHe influence of manganese sulfide inclusions on the corrosion fatigue ofmildsteel has shown that a sulphur:-enriched band of feITile exists around these inc~usionsand enhanced corrosionOCC~T$inthis contaminated band.~ttinllwas observed tooccu~at the inclusions,inc~~s~on-ma~xinterfaces and in the pearliticreglon~duetothe.

nucleationan~~oalescenceof micropits. In the case of inclusions closet!Jge~her.n, to~bj,n.ationof cyclic, stress~ddissolution appeared to 'breach theliga~entstietwcc:n ' sulfide particles,leading to thefonnation'~fmicrocracks'.It"~asdeduced that the formalian and linking up of these microc.racks eventually lead,to fracture: •

Innatural seawater. where sulfate reducing bacte'ria_are known to exist,m.lInga~~se sulfide inclusions may pose a more serious problem. These bacteria may indirectly influence Ihe cOlJOsion fatigue cracking process through the produclion of .H1.S, An accumulation of Ihis corrosive gas~ilhindte crack might lead to acidic conditions ana accderate n:on ·dissolution. ,Reaction

~oducts

derived fro'",?'HzScan also poison !he 'hydrogen recombination reaction. thereby allowing a. greater pcrceiuage of adsorbed hydrogen at6ms to enter the melal(5.53J~However since the sulphur co"ntent inst~1 Js~

minute and it is

dou~~1

^{that its}

interactiQ~ucing b~teria w'oU'~

^{beof any}

II

2.J ENVIRONMENTAL VAlUABLES

TlJeeffect of seawater temperaluU on the, cOITOsion fatigue crnclr.,growt~RIC in structural siteis was ipvesrigltedby VosikovskyeL al (12J. Underfree cOrrosion, crack growthraleswen: reducedbya faclorof two on changing ·the seawaler temperaturef~

2S"Cto O"C.. Under cathodic prOiCClion,o~lythe plateau growth rates showed a temperature.aependanc~.Thep1b~eau^growthralesW7~aboutfour times lower for a

""ttmpCfit~tc.6f0 c \rnrrrc:'T 2S"C. These trcJ'ldsShOUld~\COmea's as~riscsince anodic dissolution [52) and hydrogen. absorption [4,25J. the tWO mechanisms thought tobe"

.. ' . ^.

responsible forenvironmen,tenhanced"fatiguecrack gr,?wlh inthesesteels, arc bc?th accelerated by temperature~.s.

, .

2.3.2 P~

For low carbon.steels undergoingfTee ccrrosion in sodium chloridesoiutions.~

.

. ...

^{- ,"::,}

is a rnpid increaseinthe rate of crackgrowth

as

^{the pH}ralls'~ow4 and a significant

...

. .

decrease tu:iIrises;above 10. At~ufficlently.highpH values. afatigue~I rea~^on

!heSoNcurve (plot or stress vs. number of C?ycles 10 failUre) for mild Steel in~,5~NaCl .soluti~(15J. changing the pH in the range 5-10 has little« no d"fectontheCQmnlon :

·f4tiJue;Cnlck~wth.ratc[l5,26].·

^{• . '\ . _ .}

:! . . • - _ . . ' . .

I Thereap~10 be little pqbUSh¢ informationCo~e~inglhe influence of..butkilH..

on Ihe faligue crack

grnwlh:

rulet'Ofstttl_U~dercathodic pro(ectl.on.~thodic prolecq~n

.'.

'".", ,':.~.;

has. however. been shown 10.·producev~al,lcaline conditions wilhin faligue cracks [93-2,47]. The solution_~Hwithinfatigue cracks_~discussc;d under section~.O.

Thep~nceof dissolvedoxyg~is one of the key e1emen15 responsible (or Ihe (astO: fatigue crackgrowl;hrales observed(orc~ingSleet in seawater as compared with steel in air. Tests on mild st:ee1 in3% NaCl solution havede~onsttatedthat de·aeradonCM

· reSlore the fatiguelif¢ 10 lhatobserved for comparable air !elts [14,15J, The corrosion mte in the

~ted solutio~~ dJculat,~as

^being

1~

^times

greate~ th~

^.that

~n ~he.de,aeraied

soluJion {14]. SCott et.'all6SJ

te~ted'ccimpact

tension specimens,

m~~ined

(rom OS 4360

<hadeSOD~~'in'anatural'seawater environment at oxygen levels~rImgll,and 7-8 mgll (;ili-saturatioo). Itwasfound thatthereduced ox-yge:n lever

gave

substantially slowercrack, growthrates.·underfree

co~siop

cdnditions,

bu~ad

^nosignificant effect at potentials' tIIOrinegativeilian

~mV(Ag/AkCl).1b~resul151e~supporttothebefieft8atanod;c.

dissolution zpaybe

i!te

predominant cause of enhancedCrackgrowthat

th~

^{freecolTOsion} potential.

"0:..:'

(,:,.:

t~\~;~;"~~:·~~ ^{',(' '} ::: •.""

K':.. ; <;;.

,1

The lack of dependenceofcrack growth ,rates on dissolved oxygen ·content at .

·

~ientiaJ.s ^mole ~egative ~an .800~V (Ag/~gCl)~s q~e.Sti~n~b~)1

^iswelJ

~wn ~at

ⁱⁿ

~waler•.cathodic.polariza~oncau.ses'the.p~~PhatiOnol'anInso~uble calc~usseal,e.on

· the stee;l surface. Higherc~ncen~8tions.qf dln.oIved:oxyge~result In'~Igh.ercurrent.

" den$ities w.hich, intUm.enhance precipitation of,the scale. Thisscale thenliini~stherate at :;

whietr

~ditioha1 9xy~.ri anti/o~ 'wa;~ ~Iec~ies ~ay di'f~U~}~.~_~~J ~~_: ~~ ~:

. "reductionand'the

ca~c,cumnt de~~ty

is substan.tially red.uced.

P~a~s.:.

^for

s~1

^{uRder • '. _}

'~

....

~

,13

cathodic proteCtion in~wate[.lhechemicalamrp~ysica1propenies of .these deposilS may ,_J .establishIhtrate forwaterr~uclionon the f,:"cture surfaces and. in lum, influence the extent of localizedhydrogenembriUle~ent d~ge .a~eadof. the crack tip. Therol~thai calclll:t9usd~positsplay in"the fatigue process istreaiedwith greater det;yl under section 4.0.

.

\ ~:3.4 VdocityoftheEnvironment .

. The relative velocity between

th~

I'tccl'and the envkonmenfhas

~en ~hown

^tohave

~

aSigitifi~antinfluence onthe'fa~guc:crack pr.bpagationrate(9r,steel under.c~thodic . prOlectionin"seawater:

In

g'~neiaI. ~ckgrO'y,rthrates tend~obef~instagnant-~waler'

tJian"~nder flo~ing

condi"ti()ns,

;s~ially

fordeep cracks [671/0ne'c()uld reason that the

'\ " . " . " .

{mer~ra~kgrowth rates -observe.d understagnantconditio'n~couldbedue to the preferential~ucti"onof water within the fatigue crack once dissolved oxygen becomes . scarce.

.

Afresh'supply of so1ution to the crack-tip region. on the other hand. would-favour -

^.

theC?}iyg.en.~uctionreactionandhydrogen~tionontOttl'efraCtu~surfaces wouldbe so~whatless.Th.re~lts,?f hydrogenpe~eationexperi,:"ents onBS 4360~rade500 steel in3.5% NaCl solution have shown that de-aeration promotes hydrogen. absorption into theometal

[~7J.,APp~n~ly.

the stronglr

~.dsorbei.oxygen.ruolecules

^{cause some} hydrogen tobediSj;:liarged'on the surface of the adsorbed oxygen layer rather than~nthe sUrfa.ce

~f th~

^metal

~~8J, Conseque~y;

^there'·isa" reductionin.the

~~U1il of'~ydrogen

entering tile

nietat,

2.3.5 Cathodic Protection

\

It would appear that there is little concurrence in the corrosion fatigue daJ'a publ!shed for offshorestructu~i'steels "subjected to CathOdic protection in mllrine environments [6,7,12,13,14,15],Ithasbe~nreponed that pohuization to'just below the

".f[eeco~sionpotential (ie,polari~ationto~nomV vs,SeE)~ducesthe fntiguecrnck growth rate, but doeS!lotrestore~efatigue life to that found in air tests [6.7] (Figure 5.

~.107),This sIlght increase~n fati~elife has been attributed toItreduced.conlributlC!rt of anodic diss?lutiontocrac~propagation, However, pc;lariz.ation to more negative'potentials has been shown to'cause the fatigiJe.c:rack &rowth

rat~

^toriSe, especially

~nder

^{relatively :"':,} hith ranges of cXclic~tres~intens,it)' [15,25]. Inaddi~on.tests perfonned at veryrieg~tive potentials and low cyclic stress ranges have produced fatigue

~iack

^gi-owth

~tes

^{thnl are} , l!-ctua1)', slower-than those derived,from correspondin'g

air.tests.(.l£L6.26]~

^{.._. _}

This apparent discJ:epanc)' can,h~we¥~,beclarified.- Several researchers '[1,10,12,16,31] have explained, the slower crack growth rates measure'd at low stress

~eve1sunder cathodic protection'in 'seawater (in comparison with those measured in air)to .. bedaused

b~

^Itreduction in the 'effective' stress intens;t)' range,'broughi

a~~1 '·'It

'wedging open' effect due~othe'precipi~tion'ofcalcareous deposits insidethemouth. of the crac~.The calcareous deposits were observed to cause themi~imumvalue of measured , crack openlng t? increase w,hile thema~imum'(alue remained virtually unchanged(1],It . hasal~obeen suggested' that j,alcareoos deposits might exen a strengthening influence in the cr;ck tip vl.cinity

(I~],

At the higher stress levels. the rate of calcareous

d.e~siti~n i~

^I

.n6t fast"enough to'keep up with

crack~exte?sion

^and

~K i~ unaffected.~I~,

the plastic zonc 'generateda~eadof the crack tip willbelarger and-more sensitive'to hyl;l.rogen

15

".' ^..

^{,, , ' .}

:i ....

l-

.embrittlemcnL The nee effect.is aS~tantial[Y faste~crack growth f3.te at the~i~hc:rstte:ss levels when.cathOOicovc:r·pro!ection is being employed.

p'7"··'i"

^;';;~"

^. ..

^n,.;;",

^.

'.", ,

" J::"',

16

3.0 ELECTROCHEMISTRYWITlIINFATIGUECRACKS

A "nowkdge

of

the electrochemistry within fatiguecracksunder different combinations 'of bulk solutionchemistryand environmental parameters is crucial 10 defining the mechanism(s) responsible "for Ihe enhanced fatigue crackg~wlh^rates observed for structural steels'in .aggressive~olutio_ns.The electrolyte within fatigue crncks can become modi,fled relative to the bulk solution when -there is restricted diffusi!," Dnd soluti<;ln pumping, as a consequence of c¥dic lqading, is insufficient forcompl~terenewal oft~ecrack~leCu:olyte.Recentln"estigil.lions [6,9,32.45,47] into the relationship betw.een c.:adk.interiorelectrochemistry andcrackgrowth ratehave

l~ad

^teithe

developm~ni'6f micro.elec~es

.for the direct

measu~meht.o~

^pfland

ele~troeh~icQl

potential within- fatigue cracks. The following discussions _willfocu~on the solUlionco~posilion^and electrode potential within(1)fatigue cracks undergoing free corrosion and(2)fatigue cracks.U~d.ercathodic polarization in sodium chloride solUlions, ;mifida! seawater or naruralseawater.

3.1 ELECfROCHEMlS1'RY UNDER FREE CORROSION

\

..

i ..

J. F'tig~e

e",ck tip

p<it~nti'l

^..d

P.!'

me,mcment' fnc strucm."1 steel, ""deigoing fre~ ~rrosionin marineenviron~nts ~ s:arce~. Based. on available experimental' measurements', and considering the hydrolysis offerrou~ io~sonly, Turnbull [4.5] has estimated the o:tinimum pH possible· in. cavities of struCtural steels at the frc:e corrosion potential tobenot. less

th~n

^6.6.

How~er,

f?r: steels'

'~ntaining ~h~rhiUm,

IC?::;,:Yl1lues

17

.;"';

··.. i

ofpH.-in the rangeD-3.aR:to beexpected since the equilibrium constam for the hydrolysis of

c~liumIon~

isjIlUchiarger thanthatfor (errous'

ion~ (25,41,45~4g1.

.. Potentialmeasurements within artificial crevices[48]simulated cracks[II] and static fatigue craclcs[9J indicate thatthe

crac~

^tippotential is usually

~;hin

a few millivolts of the extsmal surface potential. Scotteta1 Jl] have observed the polentiallJ.t the cracktip of a freely corroding specimen, under cyclic loading in natural seawater,tobeas muchis 60..90 mV mOte negative than the external potential. The crack tippotential was also observed-to. vary periodically with" the load cycle, movrng in the P9sitive d&eotion as the

~cra~k2pe~ed~Ho<Jgkiess et.at[9] found.that during loadcyc1i~gⁱⁿartifiCial~eaw:ter,^the external P:9leni.iai was's!gnificantly (>100 mY) more positive Ihall that near the cr,i£ck.tip at

t~

^beginning

of'~e ex~~nt

but.gradually'decreased. After

;-3 d~y'S, thetwo'~tenti8.ls

usually,remarn.ed very close, but the surface potential was always slightlymorenegative than the crack tip potential. Thisbeha;io~v.:as attributed to the gradual build-up of -oorrosipn products o.n the external surface, resulting in_a.bigher degree of p4?laflzation of

the oxygen reduction rea!<ti9n.

.. Asa panof the same'study [9],

H~gJciess:el:

at.-monitored the

crack

tip 'pH for static and load cycltng conditions. For the caseofzero-load, theera.cICtip pHwasusually between6.3 and 7.0 (asc~mipared "wiih_~.2fonhebulk solution) at.the start ofth,e test but eventu~ly'rose^10valuess~i~htly ~o~ a1ka1i~ethan that forth~bulk seawater.Asiiiiilar trend in the crack tip pH was observed under load cycling. Values of crack-tip pH'in the

~ , ,

.

range6.8-7.2 were.recOrded~u~ng ~ef}l'st fewdays and values were usuallyinthe range

8-9 th~~ler'-'

^HC?wever,

signifi.~1

fluctuations sometimes. occ.med a.nd

~n

increase in crack-tie.potenti8J-waS always

lI;C~p4nitd

^{bya dropin}cr3ck"clppH.

...

18 3.1.2 Mechanism of Anodic DissoJution

For fatigue crack growih under dissolution control (i,e. corrosion fatigue). the mechanistic requirement is enhanced corrosion at theg~ngcrack tip cQ.mpared" to

~elativel)'passive conditions on the crack walls (II]. In corrosion fatigue. the crack lip undergoes plastic cycl!c strain [69]andplastic tensue deformation has been shown to enhance dissolution at, high strain r;ates (35,361.

.Increasing th_e salinity of the corrosi:e environmentwill,.

~I'!der

certain conditions, increaslf the corrosion fatigue crtlck growth rate (I5].It"has been suggested that.a local decrease of the electrolyte pH, as a consequence ofh)'droly~i5within the crack, .will give ris~to enbanced dissolution [32,41,49]. However. Turnbull et. al [46J have shown that changing pH in the-range5-19has aninsignifican~effect on the rate of anodic dissolution for BS 4360 Grade 50D steel in sodium chloride solutions. This observation is consistem withth~fact thatthecorrosi~nfatigue crack growth rate is not significantly affected by pH , changes in the range5-10[15,26).It'You1d appear that, for the case of free corrosion.

~ome

other factor is influencing al\odic

dis50~urio~and

the rate of crack p;opagation. The stability of the protective oxide film(F~OjorFC)OJfonned at the crack tip couldbethe controlling factor. A necesSll'f condition for continuous crack propagation is the repeate<t--- '., fracton: of thisfilm..Yielding ofthe·;;t~at the craCk tip is considemlto assist disSolution

..

^'\ . ' . : ' .

-- .

, by increasing the numQer of active si!es on the dissolving surface [35]. When a metal undergoes plasticdef~ation,slip steps generated in the surfaet grains' providp a clean new sl,lrface upon which electrode~tctionsmay readily ".Ialee place. If the rate of

mech~~

^ruplurt

or

the oiode mQfis

adequi~

^{10 allow}

IO~g

tenn.local dissolution of the bare metal generated at the crack tip, an enhanced crack propl!ogation rale willre~lUlt ..

[25,35:36,,49].~If, h~)Yever,

^the

plasti~ Slrainin~

rate is very high,

mechanicU;fa~·~·:,,::

""""", ,

^{..,.: ,'.' ,,,.,}

,\ 19

damagewill maskany contribution of dissolution 10 the measuredcrackgrowthnlc (49J.

AI_very~oxide breakingJ1ltcs,twodifrerml occum:nces areWSSib1e, depending onthe crack solution chemiStry,butbothwill lead to crackgiowthre~tionor even stoppage .(35,36,491. EOI'

deL

^neutral

c~ditions

^0('

pH

and low concentrations of aggressive.

anions~passiva~on

^may

s~ppress

any significant anodic

~ac.tivity.

^Whereacidic conditions'"

exist and/or an abundance of aggressive anions arc present, conditions will favour pitting' and itismore likely that craclctipbl.unting will occur [49].

Patel [36] hasinvestfgated the influence of chloride ions on the cyclic strain- . enhanced dijlsolutiof!~havi!-lrorqUidslee~in the pH range1-1~.T1tc results suggest,that, .'

11\ pH levels

'll:h~:fiIrris are.~~lY to

^{fonn (pH}8.14),theextent.of

di~~luti~n'viill

^depend

upon.the abilityofthese anions.to affect film stability ialherthantheirability10facWtate the I'elOOval of metal ions from the sleel suiface..For pH in the range 1-3, dissolution was obserl'ed 10 increase with

progrelsiv~ ~~cling,

indicative of &Ctiye

~Ie

generation.

V'" .

However, chapging the; chlOrideconcentr'lllioodidnOt have any appreciable influence on dissolution far this range of

pH.

Asthesolution pH :-vas raised above 8, considerably lower dissolution ralcs were observed for chloride concentrations lessthiP!'.Clr"equalto O.'!,% NaC than for chloride~trations_greaterthan or equalto3.5%~aCl.Evidently, for "pH in

the

range 8-14, the surface film was not significantly weakened allow chloride ,

~eve1s

whereas the

protecliv~

quality of the film

~a.s

deleriorated considerably .by. the

adso~tionofa large number of chlOljde ions.

Tesls {3Sj'canicd out in 0 ..5 M a'queaus solutions of various, aggressive anion!, in ..

Ihe

pH

range 3·14, have demonstrl\cd that sulf.ate ions cause

even,greaie~ s~~-cnh~etd

diSSOlution.err~.~.thanchlorideion~,whereas .. nitrJlte solution exens ale~disruptive . innuenceonthe·ox.ide·fdm'thandoesachloridc solutiOn... ~ ^{•... :}

20

3.1.3 Hydrogen Embrittlement at the Free Corrosion Potential

Itis difficul\tqassessth~relative importance of metal oxidation and"hydrogen ion

. .

reduction on crack propagation enhancementinsteels held

neat

the free-corrosion poteminl.

Although both mechanisms-maybethermodynamically pOssible. under ;hc,conditions of crack-tip pH -and potential outlined in section

3.1.1,

^itis unlikely that hydrogen embrittlementwillplay 11 significant role {n the crack propagation mechanism.

Measurem~n.ts[32] of hyd,rogen uptake inSODst~lnndergoingfree-corrosioninsynthe~­

seawater support thisCOnienti~n.Eyen under an¢ic p?larization with,high corrosionmtes, an insignificantam~untofhy~genentered the metal (32J. Also, discontinuous crac;k propagation behavior,:which is indicative of a hydrogenemblittlemcnt mechanism [2.5J: is seldomobserved for structural steelinthe absence of cathodic protection {261.

3.1A Effect ofMecharucal Variables on.Crack Solution Modification

The altcrnatc opelliilg and closing of a fatigue crack results in Ihc pumping of solution into and out of the crack: p:arkins et.aJ [11Jhavc er:nphasi.zed the imponance of mlck solution renewal In

suslaini~g

crack growth, Crack growlh by a

mech~'m

^{of anodic} dissolution or

hYdrog~

assist will

O~IY

continue if

prod~cts

are' removed

'~d

^rea9tant

depletion does notocc~r.Otherwise, local changes in the solution at the'crack tip will Jtb. suppress further reaction. In the other eXlreIite, excessivepump'~ng ~il1 r~tardCfll.ck . so.lution modification and this will also suppress c<in'osion fatigue crack propagation. The

. .

^.

mean stress-level,cycli~frequency apd crack len.gth all influence the pumping action and

Can

have significant effects ontheex.tenttowhich the crick electrolyte becomes modified.

21

A model developedby-Hanl Ct. al [41]topredict the influence of each of these variables on crack solutionmodiCicalion,pr~jec(edmixing to increase linearly with increasing lrequC'ncy and to increase with the cube of crack length. Mean stress levels that .allo~.crac~closure during a portion of the cycle (i.e. low R ratios)~realso projected 10 increase mixing. This is to be expected. since at low R rRtios:.the crack is praclicall.y emptied for eac:hcyc~.

Turnbull [43.44) has perfomled a theoretical e\la.luarlon oft,he oxygen concentration

',i:

.

.

^.

withinfll.ti~uecnlcks,al the free corrosion potential in a'marine environment. Fot:a given crack iength,inc~asl!1gdK (cons',;nt R) ord~easingR (conslarll dK) was'prcdiclof to increase the concentration ofdissolv~oxygen within tile crnck. The oxygpn concentration4!

at the same distance from lhe crack~pening;was predicted tobegreater for 10nRer crack$..

However,~eoxygen concentration at a given distance from the crack tip was Pte:dictedto decrease.as crack length increased. For values of the mechanical andenvironm~,!laI p.aramelers of practicalrele~cetooff~hon:structUres Oow .6.K, high R and f<!J.lHz)the . crack was calculated to be almOSt completely de-aerated

. The above predic;ions are in relatively ·.close· agreement with experimental

ObSUJllitio~s, Ga~gloff

^[42]

h~~

noticed a decline. in the

environ~en!al·c~l)lponent

^of

fatigue wi.th increasing .6.K and has suggested that the crack electrolyte becomes less ag~ssivewiih in·creasing

crack

opening. For constant 11K; corrosionfaligu~cracks were 'obsetved to~wfaslest for a~rack.depthof I nun'and dec')rease by up 10 an order of magnitudo fOf bolh shorter and longer CTacks:This effect was explained in tenns of the oxygen~o~CCl1trationwithin the crack. Oxygen reduction at the c,rack tip was reasoned 10 inhibit corrosion fati,suebypreventingcra~kacidification and~ucingadsorbe(i-hydrogen covera~e ~ndembrlttlemenl.~~SOIVedoxygen wasth~ughltobesupplied from the bulk

'.

712

. ,. ../J

solution 10 theCf1lC~^tipby diffusionfor very shon...crac1(:$ «I rom)andbyc~>nveclion

ro/ -

long

~cks

(>5 mm), Intermediate le.ngm

Cr~ck~

(l-:..dnmj were believedtobedepicted of ol(ygen.

Exuemely low cyclic frequencies will often I.cad tocrack arrest. either by passivation {35,36,JOTcrack tip blunting (49]. The~ffeclofinc:reasing c;c1ic frcqucl\Cyon masking the environmental contribution to corrosion fatigue crack growth is also~ell ....rccogni~.HighcYclic frequencies genelJllly give

Cfac'k

growthrntc$ that arcc1~^{to those} 0oo.erved for tests in

ar

[26] (Figure 2, p.l04). Onegroup'or'researchers (40] have

~hown

tl).at corroSion fatigue cracks in iil:ructural steelcould'~e ~tardedotev~n'Stopped when fatigued in aeratedseawater at~O H~The~sponslbiemechanism was corrosion product wedging whichcause~crapk closure, Ihen;obyreduti~Jhc~tressint'enSity runge, .6.K. The extent o:f corrosion product,build-up depends on the rale of oX,idep~ipitlllionwhich, in tum, is contl'911ed by the 'concentration of

ox~~en

^"in

the,~U:i~ity

^{of !he}

c~ck

opening, The supply of dissolved oxygen depends on the frequency of the applied load" Craek growth retardatipn due 10 oxide wedging does not occur in structural steels in seawater "for frequencies less "than I H;t; [401.

., ^\

²³

\.

3.2 ELECfR.OCHEMlSTRVUNDERCAmODlCPOLARIZATlON

3.2.1 Potential andpHProfiles

Th~seems10 be,unanimous agreement among,researchers that the electrolyte within fatiguecracksunder cathodicprotectionbecomes alkaline relative to the bulk solution [9.22.32,42.45.47J. Hodgkiess ct. alr9)hav:e measured the pH in staticfatigU~

cracks and in cracks lInder zerocompressionloading, low-AK.loadcyclil}g and high-IlK load-cycling for a range of external potentials (-800 10 -900 mY vs.SeE).'In ,allcases,the

~

.

^~.

track interior pH values rose to between.It and 13 within a 2-4 dayperiod,Turnbull et.al [471 hnvedeveloped a~atherpatica1^modelt?predictthepH and'potcnti;Uprofil~with"in fatigue cilIcks in Sb'Uctu,ra1 sleelc~thodica!JY p~~ectedfn seawater ofdiff~rentbulksolution chemistries 'and have compmd their

p~dic1iOt1"s

with.e;perimental

measurem~nts:

^'Por

ASTMseawater'atSoC. the mOdel predicted pHvll1u~s.atthe crack tip to be Within the rnn~e 1O.6<p~<la.9for potentials Q.etween .-900 and -lbol? mY (SC,:), butsignificantl~ ^{, .} different pH profiles were expected. The pH~asfOreseen to decrease more rapidly ...olh dislance

F:oin mi ~ck

^{tip as}

th~

^external

pote~liarw;s~~de

more positive. The

crac~.

^tip

pH measured at ·800 mY{p~9.8) was slightly, lower than that predicted by the'model (pR.._

10.2) for the same loading conditions.Itwas thought that a'small amo.!1ti.t

of

dissolution andsUb~uent h~lysis ~ighlberespQnsiblef~rJh'eslighlly'lower pH aClUally observed at this Potential. The crack

tip

pH measurementS made at -900 and -1-000 mV(~CE)were .. cons!derably greater(~much as one and~e-half^pointshig~er^{for a}

6.K

arlOMP:,m1fl .. ·1000 mY) than the predictedv~es~'fl:tis d!scrpancy was attributed to the difficulties

~sociated..;ntheffective modelling or-the buffering proc;sses that occur in Se.awater.

',.

24

Turnbull et. al [47J foundgoodagreement between predicted and.experimentally measured,,~uesof cracktippotential for external Potentials.~pqsitive thon

or

^{equal to}

-1000 mV(SeE).For a crackismm' in

le~gth

underloadcycling.with' AK,sCI nt 20 MPa.m1n•

R':'O.5,

f=O.1 Hz and T=5°CiiiASTM seawater, themeanpOlcntilll"drop ranged from about17mV at-&OO-mV{SCE)

t~'

^{about68 mV8t'-)000}

^mV

^{(SCEi The}

mean potentialdropmeastmd at -1100 mV~.s.SCE(ca ..l~OroV) was-slighlly largerth~n the model anticipated for the same potential(ca.lOP mY). The pqten,tinl drop '!Vas predicted 'to decrease as a result of increasing eilher AK orR andto increase withincn:asingcrack depth or

cyclicJfrequ~rfcy, ~

^enhanced_,

po~entiai '.

^{dropsobservedat v:ry}

neg~tive

potentials «-1000 mV'IS.SeE)by Turribull cl. 81 [45;47] and others [11,34Ja~ thou~ht tohe'll

conseq~ence

^Ofhydrogen gas

a~umulatiOn

within the crack..

~e

crack tip potential

meas~~menls

^{made by}

Othe~s

^[9:22,321

~

ⁱⁿ

~asonable

agreement with the aoove model predictions. H?dgkiess el. al {9]recorde~potenHll1 droPs , of up to_50 mV in unloaded fatigue cracks polarized to potentials as negative as -900 m,V~

(SeE),but the potential ,differences wereconsi~raQly^loweru~der IO~-6Kcycling and wereneg1igible.f~rhigh AK values. Maahn [32] has, reported poten.tial drops of the same magnitude as thqse 9bservedbyTu~bullet. a1 [47] forsim~l¥ ex'perime~Uuconditions and • has

~hown

the P?tential drop 10

de~se

with increasing

/iK

or R. However, Maahn (32]

has

~~ed

^the

pote~ti~tl;lrop~deaiease

slightly wi;h'increasing frequency. Hantet,

~l

[22]

ha~e

potential drop

me~urements

within simulated fatigue

crack~

in,3.5% NaCI solution and,\a,t..ural seaw,ater. The potentialp~filewithin the simUfa,led crack y,'IIS obsetved toslee~nandthepote~tia1

drop

to increase~jlhmt?fCnegativ~potentials or:Mth increasing cyclic frequency.

":.:~>': .~,1 . " '...., ..., :

";". -,'

25

,Th~ctack~ip pote~tialhas been noredfQvary sinusoidally wiih the fatigue cycle:- [22,32~41J,^{itbeing more}poslti~eand giving rise to the largestpot~tltia1^dropWit~^the crack,

in

ils most open position (22J.Itis pOssible that, as the CT8ckopens, the_ing~stionof , dissolved .oxygen aClstodepolarize me cril.ck surfaces [22J. The_rise in'cathodic reduction . cilrrent flowing Into Ihe crack is fhus able 10 in.:rease the IR

ru:o

p even though Ihe ohmic

resistanceoftheelectrolyte is.reduced.

--

. ,

3.2.2 ~echan~mofHydrogenEmb..ittlell'!ent

When steel is~raifzedcath.odically in aerated seawaler,IWOreduction reactions

~

^{.e place}on

t~e

^metal

Su1f~~:'oxygel);

^reductionand,hydrogenreduction. Elbeikct.at (38)lilie shown tlJat the cathoqic' rWl:Iclion of dissolved oxygen

O,+2H10+4e-==>40H-

occurs

,~Iow

^-630,nV

{SCE).,~whereas

^{.hydrogen{eduction'docs}'not begin umil the 'polem(aI of the steel is made mom negative than ·720 mV(SeE).

",0

^+e'

is the qharge [J"aDsfer process responsible for hydrogen adsorption 0010 the steel surface.

.

In

neu~'

^or

~irie

soIutio1fs;as is .lhe-cue.iA-seawater (bulk pH 8.2), the

red~cticn.

^{of HzO '}

speciesrterreduction) .

:;:',.', ..:c;.·

Hydrogen.reduction. is possiblevia~ diff~nlmechanisms•.depending on~e P~' of the solution[3,57J. Under jicidic conditions, the~uctionof hydmtedhy~genprotOns

'0'