Astm stp 801 1983

KEY WORDS: corrosion fatigue, fatigue crack growth, fracture mechanics, metals, elec-trochemistry, surface chemistry Nomenclature a Crack length / Cyclic load frequency AK Cyclic ra

St Louis, Missouri, 21-22 Oct 1981

ASTM SPECIAL TECHNICAL PUBLICATION 801

T W Crooker, Naval Research Laboratory, and B N Leis, Battelle Columbus Laboratories, editors

Copyright © by AMERICAN SOCIETY FOR TESTING AND MATERIALS 1983

Library of Congress Cataiog Card Number: 82-83519

NOTE The Society is not responsible, as a body, for the statements and opinions advanced in this publication

Printed in Baltimore Md (b) May 1983

1920-1981

Dedication

Dr Floyd Brown was involved in planning and nizing the 1981 Symposium on Corrosion Fatigue from its earliest inception He died on 16 August 1981 and those concerned with the symposium felt the loss of his wisdom and guidance

orga-Dr Brown received his education at the University of Kentucky, the Carnegie Institute of Technology, and Cambridge University Following an early academic ca- reer at the Massachusetts Institute of Technology and North Carolina State University, Dr Brown joined the Naval Research Laboratory in 1954 as head of the Physi- cal Metallurgy Branch, a position he held until his retire- ment from federal service in 1972 From 1972 until his death, he was a senior research scientist at American University in Washington, D C

Dr Brown was probably best known in ASTM circles for his personal research in stress-corrosion cracking He made some of the earliest and most important contribu- tions to the marriage of fracture mechanics and corro-

sion science Early development of the stress-corrosion cracking threshold parameter, Ki^cc ^ds achieved in large measure by Dr Brown In association with co- workers, he pioneered knowledge of localized electro- chemistry at crack tips in stress corrosion Although less well recognized for his contributions to corrosion fatigue,

he played a guiding role in numerous early studies of rosion-fatigue crack growth His final paper on corrosion fatigue appears in this volume

cor-Dr Brown published and lectured widely during his career, which brought him international recognition and numerous professional awards He was a member of ASTM Committee G-1 on Corrosion of Metals and the Committee on Publications He will be sadly missed by those who benefited from his insight and encouragement when venturing into puzzling fields of investigation in- volving mechanical failure complicated by corrosion

This publication contains papers presented at the Symposium on Corrosion

Fatigue: Mechanics, Metallurgy, Electrochemistry, and Engineering, held in

St Louis, Missouri, on 21-22 October 1981 Sponsors of the event were ASTM

Committees E-9 on Fatigue, E-24 on Fracture Testing, and G-1 on Corrosion

of Metals, and the Metal Properties Council T W Crooker, Naval Research

Laboratory, and B N Leis, Battelle Columbus Laboratories, served as

sym-posium chairmen and have edited this publication

Related ASTM Publications Residual Stress Effects in Fatigue, STP 776 (1982), 04-776000-30

Low-Cycle Fatigue and Life Prediction, STP 770 (1982), 04-770000-30

Atmospheric Corrosion of Metals, STP 767 (1982), 04-767000-27

Design of Fatigue and Fracture Resistant Structures, STP 761 (1932),

04-761000-30

Stress Corrosion Cracking—The Slow Strain-Rate Technique, STP 665

(1979), 04-665000-27

Intergranular Corrosion of Stainless Alloys, STP 656 (1978), 04-656000-27

Fracture Mechanics (13th Conference), STP 743 (1981), 04-743000-30

Fractography and Materials Science, STP 733 (1981), 04-733000-30

to Reviewers The quality of the papers that appear in this publication reflects not only

the obvious efforts of the authors but also the unheralded, though essential,

work of the reviewers On behalf of ASTM we acknowledge with appreciation

their dedication to high professional standards and their sacrifice of time and

effort

ASTM Committee on Publications

ASTM Editorial Staff

Janet R Schroeder Kathleen A Greene Rosemary Horstman Helen M Hoersch Helen P Mahy Allan S Kleinberg Virginia M Barishek

Introdnction

MECHANICS, METALLURGY, AND ELECTROCHEMISTRY

Fracture Mechanics and Corrosion Fatigue—R P WEI AND

G SHIM 5

Discussion 19

Corrosion-Fatigue Cracli Initiation Behavior of Four Structural

Steels—s R NOVAK 26

Anomalous Fatigue Crack Growth Retardation in Steels for Offshore

Applications—R VAN DER VELDEN, H L EWALDS,

w A SCHULTZE, AND A PUNTER 64

Crack Growth by Stress-Assisted Dissolution and Threshold

Characteristics in Corrosion Fatigue of a Steel—K ENDO,

K KOMAI, AND T SHIKIDA 81

Experimental Observations of Environmental Contributions to Fatigue

Discussion 114

Influence of Environment and Specimen Thickness on Fatigue Crack

Growth Data Correlatktn by Means of Elber-Type Equations—

H L EWALDS, F C VAN DOORN, AND W G SLOOF 115

Corrosion-Fatigue Behavior of Ti-6AI-4V fai a Sodhim Chloride

Aqueous Solution—R EBARA, Y YAMADA, AND A GOTO 135

An Analysis of Random Pits in Corrosion Fatigue: A Statistical

Three-Dimensional Evaluation of an Irregularly Corroded Surface—

H KITAGAWA, K TSUJI, T HISADA, AND Y HASHIMOTO 147

Effects of Microstnicture and Frequency on Corrosion-Fatigue Crack

Growtli in Ti-8Al-lMo-lV and Ti-6A1-4V—G R YODER,

L A COOLEY, AND T W CROOKER 159

Corrosion-Fatigue Craclt Growtli Cliaracteristics of Several HY-100

Steel Weldments with Cathodic Protection—D A DAVI S AND

E I CZYRYCA 175

Corrosion-Fatigue Crack Initiation in an Iron-Caustic System—

B N LEIS, R RUNGTA, M E MAYFIELD, AND J A BEAVERS 197

Corrosion-Fatigue Crack Propagation Rates in Commercial 7075 and

P/M X7091 Aluminum Alloys—j s SANTNER AND

M KUMAR 229

Effect of Microstructure and Strength of Low-Alloy Steels on Cyclic

Crack Growth in High-Temperature Water—T SHOJI,

H NAKAJIMA, H TSUJI, H TAKAHASHI, AND T KONDO 256

Fractography and Mechanisms of Environmentally Enhanced Fatigue

Crack Propagation of a Reactor Pressure Vessel Steel—

K TORRONEN AND M KEMPPAINEN 287

Discussion 345

A Theoretical Evaluation of the Oxygen Concentration in a

Corrosion-Fatigue Crack—A TURNBULL 351

Discussion 365

Some Electrochemical and Microstructural Aspects of Corrosion

Fatigue—j CONGLETON, I H CRAIG, R A OLIEH,

AND R N PARKINS 3 6 7

Environmental Influences on the Aqueous Fatigue Crack Growth

ENGINEERING ASPECTS

Implementing Corrosion-Fat^e Crack Growth Rate Data for

Discussion 421

T A PRATER AND L F COFFIN 4 2 3

Fatigue Design Stresses for Weathering Steel Structures—

p ALBRECHT 4 4 5

Discussion 463

Corrosion Fatigue of Welded Steel Joints under Narrow-Band Random

Loading—G S BOOTH 472

Influence of Weld Profile on Fatigue of Welded Structural Steel in

Seawater—s M NEROLICH, P E MARTIN, AND

STP801-EB/May 1983

Introduction

The 1981 Symposium on Corrosion Fatigue: Mechanics, Metallurgy,

Elec-trochemistry, and Engineering was planned with several factors in mind

First and foremost was the realization that the amount of recent and ongoing

corrosion-fatigue research worldwide had quite possibly reached an historic

highpoint This fact alone provided sufficient impetus to proceed with the

event However, the calendar also offered a sense of timeliness; it had been

five years since the most recent ASTM symposium on the subject (Denver,

1976)' and a decade since the last major National Association of Corrosion

Engineers (NACE) general conference on corrosion fatigue (Storrs, 1971)

The broad interdisciplinary nature of corrosion-fatigue phenomena is

reflected both in the extended title of the event and in the co-sponsorship

shared by ASTM Committees E-9 on Fatigue, E-24 on Fracture Testing, G-1

on Corrosion of Metals, and the Metal Properties Council Attempts were

made to present a diversity of views, both at the overview level and at the

topical research level It was intended by the organizing committee to solicit

papers which would accurately reflect the state of the art in the various

aspects of corrosion fatigue In that regard, the organizing committee wishes

to express its appreciation to each of the authors represented in this volume

Finally, special gratitude is expressed to the members of the organizing

committee and session chairmen: Ernest Czyryca, William Hartt, and

Mar-tin Prager Floyd Brown, who was perhaps more instrumental than any of us

in catalyzing efforts to get the symposium underway, died on 16 August

1981 A dedication of this volume in his memory has been made in recognition

of his contributions to the field of mechanical/environmental interactions in

high-strength alloys, of his work on behalf of ASTM, and of his foresight and

efforts leading to this symposium

^Corrosion-Fatigue Technology, ASTMSTP642, H L Craig, Jr., T W Crooker, and D W

Hoeppner, Eds., American Society for Testing and Materials, 1978

Electrochemlstiy

Robert P Wei^ and Gunchoo Shirn^

Fracture Mechanics and

Corrosion Fatigue

REFERENCE: Wei, R P and Shim, G., "Fracture Mechanics and Corrosion Fatigue,"

Corrosion Fatigue: Mechanics, Metallurgy, Electrochemistry, and Engineering, ASTM

STP 801, T W Crooker and B N Leis, Eds., American Society for Testing and

Materials, 1983, pp 5-25

ABSTRACT; The role of linear fracture mechanics is considered in relation to the

impor-tance of integrating chemistry, mechanics, and materials science in the development of a

quantitative mechanistic understanding of corrosion fatigue The value of and need for an

integrated multidisciplinary approach are illustrated by results of studies of

environmen-tally assisted fatigue crack growth in gaseous and aqueous environments Corrosion

fa-tigue of steels in aqueous environments is considered to provide new perspectives for this

integrated approach The need for treating cyclic load frequency as an important variable

and for electrochemical measurements at short times (< 10 s) is discussed

KEY WORDS: corrosion fatigue, fatigue crack growth, fracture mechanics, metals,

elec-trochemistry, surface chemistry

Nomenclature

a Crack length

/ Cyclic load frequency

AK Cyclic range of stress intensity factor

N Number of cycles elapsed

R Load ratio (da/dN) Crack growth rate per cycle

(jda/dN)^ Cycle-dependent component of fatigue crack growth rate in a

deleterious environment

ida/dN)cfj ida/dN)^f associated with ith step of surface reaction

ida/dN)t.fi^ "Saturation" level of ida/dN)ai

ida/dN)^, "Saturation" level of {da/dN)^

' Professor of Mechanics, Department of Mechanical Engineering and Mechanics, Lehigh

Uni-versity, Bethlehem, Pa 18015

^ Graduate Student, Department of Metallurgy and Metallurgical Engineering, Lehigh

Univer-sity, Bethlehem, Pa 18015

{da/dN)f Fatigue crack growth rate in a deleterious environment

(da/dN)^^ "Saturation" fatigue crack growth rate in a deleterious

envi-ronment (maximum enhancement)

{da/dN\ Fatigue crack growth rate in a reference environment (or for

pure mechanical fatigue)

{da/dN\* {da/dN\ + {da/dN),(i^,

{da/dN\^^ Contribution by sustained-load (or stress corrosion) crack

growth

PQ Gas pressure in the external environment

(po/2/) Equivalent exposure

{po/2f\ "Saturation" exposure

kc Reaction rate constant

R Universal gas constant

T Temperature AH; Apparent activation energy of rth step of surface reaction

Tj Reaction time constant associated with rth step of surface

reac-tion

c, Characteristic frequency of reaction; inverse of T,

Corrosion fatigue is a term used to describe the phenomenon' of cracking

(including both initiation and growth) in materials under the combined

actions of a fluctuating (or cyclic) stress and a corrosive (deleterious)

environ-ment Its importance in determining the durability and reliability of

engineer-ing structures is well recognized Unfortunately, the term corrosion fatigue

conjures up the notion of severe disintegration of the material through

chemi-cal attack, accompanied by fatigue cracking In reality, however, relatively

innocuous environments (such as atmospheric moisture) can greatly enhance

fatigue cracking without producing visible corrosion in the commonly

ac-cepted sense (see Refs / and 2 and the references cited therein).-' To avoid this

misconception, the term environmentally assisted fatigue cracking is now

pre-ferred, and the use of the term corrosion fatigue is to be understood within this

rontext Furthermore, with the development of fracture mechanics technology

since the mid-1950s and the increased concern with fatigue crack growth in

many applications, considerations of this problem have been subdivided

natu-rally into two groups: initiation and growth Only the aspects that deal with

environmentally assisted fatigue crack growfth are considered here

Just over ten years ago, a review of the then-current state of the art in

frac-ture mechanics technology as it applied to environmentally assisted fatigue

crack growth (or corrosion fatigue) was given by A J McEvily and R P Wei

at an international conference on corrosion fatigue [/] The assumptions and

limitations of this approach, and its engineering utility and usefulness in

de-3 The italic numbers in brackets refer to the list of references appended to this paper

WEI AND SHIM ON FRACTURE MECHANICS 7

veloping understanding of corrosion-fatigue mechanisms and

phenomenol-ogy, were considered and discussed In the intervening years, fracture

me-chanics technology has become more firmly established and has contributed

significantly to the understanding of environmentally assisted fatigue crack

growth Understanding has come, however, from the recognition that

me-chanics (more specifically, fracture meme-chanics) is but one aspect of the

multi-faceted problem of corrosion fatigue, which involves also chemistry and

metal-lurgy (or, more broadly, materials science)

In this paper, the role of linear fracture mechanics in providing a

quantita-tive framework for corrosion-fatigue research and application is

re-empha-sized The main purpose, however, is to show the importance of interfacing

mechanics, chemistry, and metallurgy in developing a quantitative

under-standing of corrosion fatigue, and the relevance of this underunder-standing to

engi-neering Illustrations are drawn principally from the work of the authors and

their immediate colleagues Environmentally assisted fatigue crack growth in

steels exposed to aqueous environments is considered in the context of these

discussions The readers are encouraged to refer to the published literature

and to proceedings of a number of symposia (such as those cited in Refs / to 5)

to obtain a more complete perspective on developments in this field

Fracture Mechanics Methodology Revisited

One of the significant developments in the understanding of fatigue crack

grovrth and the utilization of crack growth data in design is associated with the

introduction of fracture mechanics technology [1,3,4,6] Through linear

frac-ture mechanics, an appropriate crack-driving force has been defined as a

con-jugate to the rate of fatigue crack growth, which is a measure of the material's

response The driving force is defined in terms of the crack-tip stress-intensity

factor (K) [7,8], or strain energy release rate (G) [7,8], or more generally in terms of the strain energy density factor (5) [9,10] for mixed-mode loading conditions The range of these parameters (AK, AG, or AS), representing the

difference between the maximum and minimum values in one cycle of fatigue

loading, is more commonly used

The use of these linear fracture mechanics parameters to characterize the

mechanical driving force for crack grovrth is based on the recognition that

crack growth is most likely to proceed from the highly stressed region at the

crack tip It is also predicated on the assumption that linear elasticity analysis

results can be applied to an acceptable degree of approximation, and hence

imposes the condition of limited plasticity in their use The assumptions,

util-ity, and restrictions of this approach have been discussed in detail elsewhere

[1,3,4,6] Specific guidelines have been incorporated in ASTM Test for

Con-stant-Load Amplitude Fatigue Crack Growth Rates Above 10~* m/Cycle

(E 647)

Interfacing Chemistry, Meclianics, and Metallnrgy

With respect to environmentally assisted fatigue crack growth, or corrosion

fatigue, fracture mechanics technology contributes in two separable but

re-lated ways Firstly, it provides a formalized framework in which the

crack-driving force is quantitatively defined, and the response to changes in loading,

environmental, and metallurgical variables can be measured, modeled, and

systematically examined Through this formalism, measured crack growth

re-sponse can be analyzed and used in estimating service performance Secondly,

because crack growth is the result of deleterious interactions of the

environ-ment with the microstructure in the highly strained (stressed) region at the

crack tip, some form of fracture mechanics methodology must be incorporated

into the quantitative analyses of these interactions

To further illustrate these two aspects of corrosion fatigue, a schematic

dia-gram and a flow diadia-gram of the various processes that might be involved in

environmentally assisted crack growth by hydrogenous gases are shown in

Figs 1 and 2 respectively [2] Hydrogen embrittlement is assumed to be the

mechanism for the enhancement of crack growth here It is inferred that

envi-ronmentally assisted crack growth is the result of a number of different

pro-cesses operating hi sequence The rate of crack growth is controlled by the

slowest process in this sequence Modeling the influences of gas phase

trans-port in terms of the effective crack opening and of stress-enhanced diffusion in

the crack tip region can be made in terms of linear fracture mechanics [11-13],

and constitutes the interfacing of the three disciplines at one level The second

level involves a quantitative description of the embrittlement process that can

lead to a prediction of the actual growth rates Because this embrittlement

takes place in the highly strained region immediately ahead of the crack tip

FIG 1—Schematic illustration of various sequential processes involved in environmentally

assisted crack growth in alloys exposed to external gaseous environments Embrittlement by

hy-drogen is assumed and is schematically depicted by the iron-hyhy-drogen-iron bond (After Ref 2.)

WEI AND SHIM ON FRACTURE MECHANICS 9

Crack Geom & Dimen

Surface Area &

P T, M

' ,

A D S O R B E D GAS MOLECULES

FIG 2—Flow chart illustrating the processes and parameters that affect environmentally

as-sisted crack growth

where the infinitesimal (small) strain assumption of linear elasticity no longer

holds, linear fracture mechanics analysis is not expected to be directly

applica-ble Additional efforts are needed to link the continuum parameters with the

processes (for example, rupture of the metal-hydrogen-metal bonds) that

oc-cur at the microstructural level Because of these difficulties, the application

of linear fracture mechanics to corrosion fatigue is considered here only at the

first level, that is, only in relation to the coupling between loading and

environ-mental variables

Modeling of Fatigue Crack Growth in Gaseous Environments

Modeling of environmentally assisted fatigue crack growth in pure gases

and in binary gas mixtures, where one of the components acts as an inhibitor,

has been made and verified [//, 12,14,75] Modeling was based on the

proposi-tion that the rate of crack growth in a deleterious environment [{da/dN)^ is

composed of the sum of three components [2,11,12]:

The term {da/dN\ is the rate of fatigue cracic growth in an inert, or reference,

environment, and therefore represents the contribution of "pure"

(mechani-cal) fatigue This component is essentially independent of frequency at

tem-peratures where creep is not important The term (da/dN)^^^ is the

contribu-tion by sustained-load crack growth (that is, by "stress corrosion cracking") at

/^-levels above ^iscc and was first considered by Wei and Landes [16] The

term (da/dN)^^ represents the cycle-dependent contribution, which requires

the synergistic interaction of fatigue and environmental attack, and was

con-sidered by Weir et al [72] and by Wei and Simmons [12]

In the model [11,12], environmental enhancement of fatigue crack growth

is assumed to result from embrittlement by hydrogen that is produced by the

reactions of hydrogenous gases (for instance, water vapor) with the freshly

produced crack surfaces More specifically, {da/dN)c{ is assumed to be

pro-portional to the amount of hydrogen produced by the surface reactions during

each cycle, which is proportional in turn to the "effective" crack area

pro-duced by fatigue during the prior loading cycles and to the extent of surface

reactions The time available for reaction is assumed to be equal to one half of

the fatigue cycle (or to 1/2/, where/is the cyclic load frequency) Based on the

assumptions of Knudsen (or molecular) flow and simple first-order reaction

kinetics, the following relationships were obtained for transport-controlled

and surface-reaction-controlled fatigue crack growth [11,12]:

reaction rate constant respectively The term {da/dN)^., represents the

maxi-mum enhancement in the rate of cycle-dependent fatigue crack growth, which

recognizes that the extent of surface reaction is limited [11,12] These models

provide a quantitative procedure for assessing the influences of loading and

environmental variables, and require the use of (da/dN\ and (da/dN)^(^^ as

experimentally measured limits for the material's response

Two hypothetical cases from Ref 12 are illustrated in Fig 3, one

represent-ing transport control and the other surface reaction control Crack growth

response curves are shown in terms of the ratio {da/dN\/{da/dN\3iS

func-tions of po/2/, for the case where {da/dN\J(da/dN\ = 50 and the ratio of

reaction rate constants is 10' For more complex reactions, such as those

be-WEI AND SHIM ON FRACTURE MECHANICS 11

) @ l k P a

1 I0-'

FREQ ( H 2 ) @ l k P o , 1

FIG 3—Schematic illustration and comparison of gas transport and surface

reaction-controlled fatigue crack growth /12/

tween hydrogen sulfide and steel, the form of the response would differ from

the simple cases shown in Fig 3, and would reflect the different steps in the

reactions [14] For binary gas mixtures containing one inhibitor component,

the response is modified to reflect the competition between the two gases for

surface adsorption sites, and is dependent on the ratio of partial pressures and

reaction rate constants for the two gases [12]

Application to Corrosion Fatigue of Steels in Aqueous Environments

It has been recognized that crack growth in steels exposed to water and

water vapor is controlled by the rate of the water-steel surface reactions

[2,17,18], The reactions with water vapor have been shown to occur in at least

two steps [5,19], The initial, rapid step corresponds to the formation of a patchy

c(2X2) adsorbed oxygen or hydroxyl layer on Fe(OOl) single crystal, and the

second, slower step corresponds to the nucleation and growth of a

two-dimensional FeO layer on Fe(OOl) It appears now that a further reaction takes

place and leads to the formation and growth of Fe304 or 7-Fe203 [20],

Accord-ingly, the environmentally assisted fatigue crack grovrth response is expected

to reflect these different reaction steps

Recent work on high-strength steels showed that the concepts developed for

gaseous environments can be extended to fatigue crack growth in aqueous

environments [21] Data on HY130 and modified HY130 steels, tested in water

vapor and in distilled water at room temperature, support the linkage of

results between the vapor and liquid phases and the expected correspondence

between surface reactions and fatigue crack growth response (Fig 4) The

data in water vapor at low exposures (that is, Po/2f) correspond to the first

FIG 4—Influence offi'equency and exposure (PD/2{) on fatigue crack growth for a modified

HYI30 steel in water vapor and in distilled water at room temperature

step of the reactions with water, whereas those in distilled water correspond to

the slower second step

Based on these data, it is reasonable to suggest the following modifications

to Eq 1 to reflect the existence of at least two steps in the reactions of steel with

water:

High Frequencies:

(da/dN), = {da/dN\ + {da/dN\ix Low Frequencies:

{da/dN\ = (da/dN), + (da/dN)a,x,, + (da/dN),(^2 + {da/dN)„

= (da/dN\* + (.da/dN)a,2 + (da/dN)^^

(4)

(5)

At high frequencies, the time available for reaction is short enough that the

first step of the water-steel reaction is incomplete and the contribution of

sus-tained-load crack growth is negligible At sufficiently low frequencies, the first

step of the reaction is complete and the contributions of the second step and of

sustained-load growth become significant It is now convenient to define a new

"reference" rate [{da/dN)*] which is the sum of the rate in an inert

environ-ment [{da/dN)j\ and the maximum contribution of the first step reactions

{(.da/dN)a,u]

WEI AND SHIM ON FRACTURE MECHANICS 13

Furthermore, it is useful to reinterpret Eq 3 and rewrite it in the form

{da/dN)a,i = (da/dNU^i,Al - exp(-;',/2/)]

(o)

= (da/dNhj,,n - e x p ( - l / 2 / r , ) ] Here, first-order reaction kinetics is again assumed, and the quantity k^po in

the gaseous case is replaced by a characteristic frequency (v;) or by the inverse

of the reaction time constant (T,), where T, = l/c, The subscript / denotes

parameters associated with the ith step in the surface reactions, ands denotes

the maximum or "saturation" value The quantities v,- and T, provide measures

of the reaction rate constants in the aqueous environments, and are related to

the activation energies for the reactions

fi = 1 /T,- OC e x p ( - Aff,/RT) (7)

Although the specific form of Eq 6 must reflect the actual mechanisms of the

reactions, it is useful as a first-order approximation for considering corrosion

fatigue of high-strength steels at the lower frequencies, that is, those that

cor-respond to the slow step of the reactions.'' From considerations of the reactions

of water vapor with iron and steel [5,17,19] and from experimental

observa-tions of fatigue crack growth response, the following assumpobserva-tions and

obser-vations can be made with respect to crack growth in high-strength steels The

term {da/dN\ is observed to be independent of frequency and to be only

mildly dependent on temperature (at least in the range of 10"** to 10~^ m/

cycle) Because of the limited extent of reactions [5,19,20], the "saturation"

values (,da/dN)a,\,s and ida/dN)c(,2,% are assumed to be independent of

tem-perature and frequency The temtem-perature-dependent and

frequency-depen-dent terms, therefore, are {da/dN)^f_2 and (da/dN)^cc- For (da/dN)c!_2' the

temperature dependence is reflected through the temperature dependence of

V2 or T2, and the frequency dependence is given explicitly by Eq 6 The rate

ida/dN)^cc is inversely proportional to frequency [16], and is directly related to

the temperature dependence for sustained-load crack growth Because the

same reactions control both sustained-load and fatigue crack growth, the

same activation energy will be associated with V2 or 1/T2 (or simply f or 1/T)

and (da/dN)scc- Clearly, elucidation of the connection between chemical

reac-tion kinetics and corrosion fatigue cannot be obtained from examining the

temperature dependence alone, and must now include cyclic load frequency as

a significant parameter for fatigue crack growth

To illustrate this point and the overall approach, fatigue crack growth data,

similar to those shown in Fig 4, were obtained on an HY130 and a modified

''it is uncertain at this time that crack growth associated with the first reaction step is

surface-reaction-controlled In water vapor, at pressures below that of capillary condensation, this portion

is expected to be transport-controlled [11,12]

HY130 steel as a function of frequency at different temperatures in distilled

water and in a buffered acetate solution (pH = 4.2) The data were analyzed in

accordance with Eqs 5 and 6, and are shown in Figs 5 and 6 in terms of the

difference [{da/clN)^ — {da/dN)*] or the environmental contribution [{da/

dN)^f2\ versus the inverse of cyclic load frequency (1/2/).^ It is seen that the

results are in good agreement with Eq 6 The fact that environmental effects

were beginning to be observed at higher frequencies at the higher

tempera-tures is consistent with the increased rates of reactions with temperature The

characteristic frequency {v) or apparent reaction time constant (r) at each

temperature was determined from the fatigue data; these data are shown in an

Arrhenius plot in Fig 7 Based on a preliminary analysis, the apparent

activa-tion energy was found to be equal to 50 ± 6 kJ/mol for distilled water and 39

+ 15 kJ/moi for the buffered acetate solution at the 95% confidence level It is

to be recognized that the values of c or T are sensitive to the choice of the value

of {da/dN)^_^ Hence these values and the associated activation energy may

change somewhat with additional data and with refinements in analysis

Nev-ertheless, these values of apparent activation energy are quite consistent with a

value of 36 ± 28 kJ/mol (at the 95% confidence level) for the reaction of water

vapor with AISI 4340 steel [17] The results therefore tend to support the

concept of surface-reaction-controlled crack growth

/ /

1 1 1 1 1 1 1 1 1 1 1

(Hz) 10"

FIG 5—Influence of frequency and temperature on fatigue crack growth for HY130 (solid

symbols) and modified HY130 (open symbols) steels in distilled water

^Because no sustained-load crack growth was observed at the /f-level used in these tests, the

(da/dN)^^ component is not included Possible contribution by additional surface reaction may

have to be included; this is considered in the next section

—

- - - -

-

FIG 6—Influence of frequency and temperature on fatigue crack growth for HY130 (solid

symbols) and modified HY130 (open symbols) steels in buffered acetate solution (pH = 4.2)

FIG 7—Effect of temperature on the characteristic frequency (v) for fatigue crack growth in

aqueous environments (95% confidence intervals are given for the activation energies.)

Discussion

The influence of frequency on fatigue crack growth in high-strength steels

exposed to aqueous environments has been examined by several investigators

[18,22,23] Data from these investigations are shown in conjunction with the

data at room temperature for the HY130 steels in Fig 8 [21] It is apparent

that these data generally conform with Eq 6, with apparent reaction time

con-stants T that differ for the different combinations of environments and steels

Although the fatigue crack growth data strongly support the concept of

surface-reaction-controlled crack growth for steels in aqueous environments,

definitive support must await the development of supporting chemical data

similar to those for the gaseous environments [17,24-26] It is clear from the

fatigue data that the relevant reactions would be those that occur during the

very early stages of the reaction of the environment with the clean steel

sur-faces The time frame for these reactions is expected to be on the order of

milliseconds to tens of seconds, and is extremely short compared with that for

traditional corrosion measurements Chemical or electrochemical techniques,

therefore, must be developed and used for making measurements at these

short time intervals These measurements must be made in environments that

are representative of the solution chemistry at the crack tip Analysis

tech-niques must be developed also to provide unambiguous resolution of the

vari-ous reaction steps from the experimental data These efforts are in progress

According to Eqs 5 and 6, the deviation in data from the "saturation" or

plateau level at the lower frequencies (Figs 5 and 6) would be attributed

1 • 1 ' 1

- ^ 1 ^ ^

• H Y 8 0 Steel in 3.5pct NaC{

* HY130 Steel in Distilled Woter

T H Y I 3 0 Steel in Buffered Acetate

FIG 8—Room-temperature fatigue crack growth response for high-strength steels in water

vapor and in aqueous environments [\S,2l-23J

WEI AND SHIM ON FRACTURE MECHANICS 17

mally to the contribution by sustained-load crack growth, that is, by the {da/

dN)^ component Unfortunately, however, it is not the case here, because the

sustained-load crack growth rates that would be required to account for these

contributions are higher than the actual rates by at least one order of

magni-tude It is also unlikely that this discrepancy can be attributed to the fact that

effective sustained-load crack growth rates during fatigue may exceed the

steady-state rates (because of transient crack growth) More likely, the

devia-tion is to be attributed to the contribudevia-tions from an addidevia-tional step in the

water-steel reactions, that is, the formation of a 7-Fe203 or Fe304 layer

Fur-ther work is needed to clarify this issue

There exists a question of whether the "saturation" phenomenon or plateau

itself may be caused by corrosion product wedging The increase in crack

growth rate above the saturation level at very low frequency levels rules out this

possibility, because the wedging effect would continue to increase with

de-creasing test frequency as more time becomes available for corrosion products

to build up at the crack tip

Sammary

The role of linear fracture mechanics in the understanding of

environmen-tally assisted fatigue crack growth (corrosion fatigue) and in the development

and utilization of data for design is reconsidered It provides the essential

quantitative framework for corrosion fatigue research and application It

must be recognized, however, that linear fracture mechanics (or, more

broadly, mechanics) is but one aspect of the multifaceted problem of corrosion

fatigue, which involves also chemistry and metallurgy (or materials science)

Quantitative mechanistic understanding of corrosion fatigue can be expected

only from investigations that integrate all these disciplines

As an illustration, modeling of fatigue crack growth in gaseous

environ-ments is reviewed Its extension to crack growth in aqueous environenviron-ments is

considered For high-strength steels, fatigue crack growth in aqueous

environ-ments appears to be controlled by the rate of reactions of the environment with

the newly created crack surfaces, and the crack growth response tends to

re-flect the different steps in these reactions The relevant reactions appear to be

those that occur in ten seconds or less To better understand corrosion fatigue

behavior, therefore, it is essential to recognize cyclic load frequency as a

signif-icant variable and to examine the frequency dependence for fatigue crack

growth as a function of temperature These data must be correlated with and

supported by measurements of chemical reaction kinetics at very short times

(< 10 s) in environments that properly reflect the conditions at the crack tip.*

^ References 27 to 35 are cited in the Discussion following this paper

Acknowledgments

Support of this work by the Office of Naval Research under Contract

N00014-75-C-0543, NR036-097 is gratefully acknowledged The authors

ex-press their appreciation to the Research Laboratory of U.S Steel Corporation

for providing the modified HY130 steel used in this investigation

References

[/| McEvily, A ] and Wei, R P in Corrosion Fatigue: Chemistry, Mechanics

andMicrostruc-ture, NACE-2, National Association of Corrosion Engineers, 1972, pp 381-395

(2) Wei, R P in Fatigue Mechanisms, ASTM STP 675, American Society for Testing and

Materials, 1979, pp 816-831

[3] Paris, P C , in Fatigue—An Interdisciplinary Approach, Syracuse University Press,

Syra-cuse, N.Y., 1964, pp 107-132

[4] Wei, R P in Proceedings, Conference on the Fundamental Aspects of Stress Corrosion

Cracking, NACE-1, R W Staehle, Ed., National A.ssociation of Corrosion Engineers,

Houston, 1966, pp 104-111

[5| Wei, R P and Simmons, G W in Stress Corrosion Cracking and Hydrogen Embrittlement

of Iron Base Alloys NACE-5, R W Staehle etal., Eds., National Association of Corrosion

Engineers, Houston, 1977, pp 751-765

(6) Johnson, H H and Paris, P C , Engineering Fracture Mechanics, Vol 1, No 1, June 1968,

p 3

[7] Irwin, G R \n Structural Mechanics, Pergamon Press, Elmsford, N.Y., 1960, p 557

[8\ Paris, P C and Sih, G C in Fracture Toughness Testing and Its Applications, ASTM STP

381, American Society for Testing and Materials, 1965, p 30

[9] Sih.G C.'mMethodsof Analysis and Solution of Crack Problems, G C Sih, Ed., Noordhoff

International Publishing, Leyden, The Netherlands, 1973, pp XXl-XLV

[10] Badaliance, R., Engineering Fracture Mechanics, Vol 13, No 3, 1980, pp 657-666

1/;] Weir, T W., Simmons, G W., Hart, R G., and Wei, R P., Scripta Metallurgica, Vol 14,

No 3, March 1980, pp 357-364

\12] Wei, R P and Simmons, G W., "Surface Reactions and Fatigue Crack Growth," in

Pro-ceedings, 27th Army Materials Research Conference, Bolton Landing, N.Y., July 1980 (to be

published)

[13] Kim, Y H and Manning, S D., "A Superposition Model for Corrosion Fatigue Crack

Propagation in Aluminum Alloys," presented at the 14th National Symposium on Fracture

Mechanics, Los Angeles, 30 June-2 July 1981

\14] Brazill, R L., Simmons, G W., and Wei, R P., Journal of Engineering Materials and

Technology Transactions ofASME, Vol 101, No 3, July 1979, p 199

[/.5] Wei, R P., Pao, P S., Hart, R G., Weir, T W., and Simmons, G W., Metallurgical

Transactions A, Vol 11 A, No 1, Jan 1980, p 151

[/61 Wei, R P andLandes, J D., Materials Research and Standards, Vol 9, No 7, July 1969,

p 9

[17] Simmons, G W., Pao, P S., and Wei, R P., Metallurgical Transactions A, Vol 9A, No 8,

Aug 1978, p 1147

[18] Pao, P S., Wei, W., and Wei, R P m Proceedings, Symposium on Environment Sensitive

Fracture of Engineering Materials, Z A Foroulis, Ed., The Metallurgical Society of AIME,

[21] Wei, R P in Environmental Degradation of Engineering Materials in Aggressive

Environ-ments, M R Louthan, Jr., R P McNitt, and R D Sisson, Jr., Eds., Virginia Polytechnic

Institute and State University, Blacksburg, Va., 1981, p 73

DISCUSSION ON FRACTURE MECHANICS 19

[22] Gallagher, J P and Wei, R P in Corrosion Fatigue, NACE-2, O Devereux, A.J McEvily,

and R W Staehle, Eds., National Association of Corrosion Engineers, Houston, 1972, p

409

[23] Barsom, J M in Corrosion Fatigue, NACE-2, O Devereux, A J McEvily, and R W

Staehle, Eds., National Association of Corrosion Engineers, Houston, 1972, p 424

[24] Lu, M., Pao, P S., Chan, N H., Klicr, K., and Wei, R P., Hydrogen in Metals, supplement

to Transactions of the Japanese Institute of Mete's, Vol 20, 1980, p 449

[25] Chan, N H., Klier, K., and Wei, R V., Hydrogen in Metals, supplement to Transactions of

the Japanese Institute of Metals, Vol 20, 1980, p 305

[26] Lu, M., Pao, P S., Weir, T W., Simmons, G W., and Wei, R V., Metallurgical

Transac-tions A, Vol 12A, No 5, May 1981, p 805

[27] Kim, C E and Loginow, A W., Corrosion, Vol 24, 1968, p 313

[28] Beck, W., Bockris, J C M , McBrecn, J., and Nanis, h Proceedings of the Royal Society of

London, Vol 290, 1966, p 220

[291 Fricke, E., Stiiwe, H.-P., and Vibrans, G., Metallurgical Transactions, Vol 2, 1971, p

2697

[30] Oriani, R A., private communication, University of Minnesota, Minneapolis, Oct 1981

[31] Suresh, S., Moss, C M., and Ritchie, R O in Proceedings, 2nd International Symposium

on Hydrogen, Japan Institute of Metals, Minakami Spa, Japan, 1979, to be published

[32] Fuquen-Molano, R and Ritchie, R O., private communication, Massachusetts Institute of

Technology, Cambridge, Mass., 1980

[33] Gerberich, W W and Yu, W in Proceedings, 5th National Fracture Conference of Canada,

Winnipeg, 3 Sept 1981, to be published

[34] Gerberich, W W and Moody, N R in Fatigue Mechanisms, ASTMSTP675, J T Fong,

Ed., American Society for Testing and Materials, 1979, p 292

[35] Gerberich, W W and Peterson, K A in Symposium on Micro- and Macro-Mechanics of

Crack Growth, ASM, TMS-AIME, LouisviUe, Ky., Oct 1981, to be published

DISCUSSION

W W Gerberich' (written discussion)—I would like to compliment

Profes-sor Wei, his colleagues, and his students at Lehigh University, who, over the

last decade, have made a very substantial and original contribution to the field

of corrosion fatigue There are several areas in the field that almost everyone

agrees upon, and these are in no small way indebted to the pioneering work

accomplished by Professor Wei These areas, along with several areas that

remain in a state of flux, are highlighted below:

• There is no question that a fracture mechanics framework is a formal

framework within which corrosion fatigue crack propagation may be modeled

This is particularly well formulated on a macroscopic scale but is a new

fron-tier on the microscopic scale The micromechanics associated with second

phases, grain size, microcrack branching, crack closure, and microcrack

dis-tributions remains to be clarified with regards to corrosion fatigue

interac-tions

' Department of Chemical Engineering and Materials Science, University of Minnesota,

Min-neapolis, Minn 55455

• With regards to chemical contributions, the formalism of Figs 1 and 2

are particularly apt Professor Wei and his colleagues have been leaders in

demonstrating that surface science is a necessary tool to separate the

trans-port, adsorption, dissociation, diffusion, and reaction steps associated with

time-dependent cracking phenomena

• That the rate of crack growth is controlled by the slowest process is

ac-cepted by workers in the field This leads to a linear superposition of the steps

(for example, Eq 1) if the events are happening concurrently without any

wait-ing time between fast events It is not clear, however, that this is the correct

way to analyze corrosion fatigue in general; see the last point below

• Activation energies of 50 ± 6 kj/mol and 39 ± 15 kJ/mol for the

corro-sion of fatigue of HY-130 steel are compared to the reaction of water vapor

with AISI4340 steel (36 ± 28 kJ/mol) Although these are similar and support

the concept of surface-reaction-controlling crack growth, this concept is not

sufficient to preclude other mechanisms from being rate controlling This is

true even in the case of gaseous environments, as will be discussed

subse-quently

• The synergism between fatigue and environment has been discussed

mostly in terms of Region II (or Paris Law) growth kinetics Firstly, it should

be pointed out that an additional regime of interaction may exist near

thresh-old and that this may be of opposite character (retardation versus

enhance-ment) and differing mechanism Secondly, there may be additional

interpreta-tions of {da/dN)^^ in Region II

• Corrosion-fatigue contributions may be summed by a reciprocal

superpo-sition process if they are sequential events This would be the case if a striation

event were followed by an electrochemical dissolution event which was then

followed by an intergranular separation event However, if the events occur

concurrently, with no lag time, linear superposition may be the best model for

corrosion fatigue

The last three points require additional comment With regard to the

activa-tion energy of the process, there are any number of investigaactiva-tions [27-30] on

trapped diff usivity of hydrogen in steels to show that an activation energy of 36

kJ/mol is very representative.^ At room temperature, this represents a bulk

hydrogen diffusivity of about 10~" m^/s (10~^ cmVs) Given a 20-Hz cycling

frequency, the bulk diffusion distance in 0.05 s is about 1.4 X 10"^ m, which is

just about the cyclic rate of grovrth observed (for example, 10~^ to 10~* m/

cycle) Thus bulk diffusivity as slowed by trapping could be the rate-limiting

step

With regard to fatigue/environment interactions Fig 9 [31] demonstrates

two regimes of interaction For 2V4Cr-lMo steel in 138 kPa H2, there is a

lowering of fatigue threshold to 5.3 MPa • m'''^, which is below the value in air,

^The italic numbers in brackets refer to the list of references given previously

DISCUSSION ON FRACTURE MECHANICS 2 1

• 2 H z , 5 0 H z Air

O 50Hz 1 , D 5H2 138 kPa

FIG 9—Typical corrosion-fatigue crack propagation diagram, showing two regions of

envi-ronmental effects [31]

and there is a deviation in the Paris law regime at 22 MPa • m'''^ The first effect

is frequency independent and has been shown to be an oxide closure effect [32],

since in inert gas or dry air the fatigue threshold is even lower than the

thresh-old in hydrogen The second effect is a true hydrogen interaction, is frequency

dependent, and does have a "threshold" near 22 MPa • m'^^, which is well

below the static threshold of 90 MPa • m'''^ In fact, a collection of data [33]

from ten separate sources indicates that there is a threshold under fatigue

conditions, [/iLiscc(f)l that is always less than its corresponding static

counter-part [/Ciscc] • Thus it may not be necessary to include separate combinations of

(.da/dN)cf and {da/dN\cc as in Eqs 1,4, and 5 if the contribution above K^^^ is

no different, mechanistically, than the contribution above Kj^^fy This is not

meant to eliminate dissolution as a contributing mechanism, but rather to

suggest that "stress-corrosion-cracking" contributions may be occurring well

below the static value of Ki^^ In this context, it may be a matter of

terminol-ogy or formalism as to how one wants to assess fatigue/environmental

interac-tions It is this discusser's opinion that static Ki^^ values may not be the best

indicators of when there is a "stress corrosion" contribution

Finally, we have recently determined in several investigations [34,35] that

mked microfracture modes may occur simultaneously and that in some cases

a sequential process is warranted For example, assume that a crack is growing

along one region by a slip decohesion process (or by striations) and in an

adja-cent region by a cyclic cleavage (or intergranular) process If the cyclic

cleav-age fraction (f^) is known, then it may be shown that the overall growth rate is

given by either

da concurrent: = (da/dNlL + {da/dN)n{l - / c )

dN

or (8)

sequential: — - — = h |- ^ ~ Jc da/dn ida/dN)^ {da/dN)^

If the ductile process {{da/dN)^^] is the slower of the two, then ductility

con-trols the process The sequential treatment is best where the microfracture

processes-' control the effective stress intensity which drives the faster process

forward It is quite probable that in some instances corrosion-fatigue

interac-tions may drive the crack sufficiently beyond the microstructural influence so

that this does not occur The point here is that linear superposition may not

always be the best way to model corrosion fatigue

This discussion prompts three questions:

1 Is it not just as likely that, rather than surface reactions, the rate-limiting

step for corrosion fatigue of certain steels is bulk-trapped diffusivity of

hydro-gen to the embrittlement site?

2 Is ATjscc by itself the best indicator as to when there may be a

sustained-load cracking contribution? Are such contributions completely separable

from those occurring below A'iscc? If that is the case in 2'/4Cr-lMo steel, then

should intergranular fracture by corrosion fatigue (22 < K^^ix < 90

MPa-m^''^ be treated as a different mechanism from intergranular fracture

occur-ring at A"„ax ^ 90 MPa • m^'^?

3 Although it is appropriate to treat chemical sequences in a single

mecha-nism by superposition, where dual processes occur at vastly different rates,

would it not be just as appropriate or even more so to treat simultaneous

micro-mechanisms as sequential processes, as indicated by Eqs 8?

R P Wei and Gunchoo Shim {authors' closure)—The authors appreciate

the thoughtful discussion by Professor Gerberich and his general support for

the approach that has been taken by the authors and their colleagues over the

past decade His comments suggest no substantive disagreements in

philoso-phy They do indicate, however, the existence of substantial areas of

derstanding which require further clarification To help clarify these

misun-derstandings, it is perhaps most efficient to deal directly with the three

questions posed by Professor Gerberich at the end of his discussion

•'This pertains to grains of local orientation or susceptible phases which may fail by one fracture

mode while adjacent grains or other phases fail by a different microscopic fracture mode

DISCUSSION ON FRACTURE MECHANICS 2 3

The first question, although specific in nature, reflects a broader

misunder-standing and is based on incomplete analysis To address this question, it is

useful to refer to Figs 1 and 2, which schematically set forth the conceptual

framework and the various processes that might be involved in environment

enhancement of crack growth, in general, and corrosion fatigue, in

particu-lar Embrittlement by hydrogen (vis-a-vis active path or electrochemical

disso-lution) is explicitly assumed The processes may be grouped in terms of those

that lead to hydrogen production at the crack tip (or processes that are

exter-nal to the material) and those that occur subsequent to hydrogen entry (or

internal processes) Overall, these processes are considered to proceed

sequen-tially, although the possibility for concurrent processes to occur within any

given process shown in Figs 1 and 2 is not excluded The concept of a

rate-controlling process, in this context, simply means that the crack growth rate is

determined by the slowest process in this sequence, nothing more and nothing

less There is no presumption that a single process controls crack growth for all

environments or that a single process remains in control in a given

environ-ment over a broad range of environenviron-mental conditions

The authors' identification of "surface reaction" as the rate-controlling

process for fatigue crack growth in HY130 steels exposed to aqueous

environ-ments in no way precludes hydrogen diffusion or some other process from

being in control in other environments and under other conditions The

possi-bility that hydrogen diffusion (trapped or otherwise) could be in control for the

case at hand was clearly ruled out A more critical analysis of the argument

offered by Professor Gerberich (taken to its logical conclusion) would have led

to the same conclusion To wit, if hydrogen diffusion were in control and

fol-lowed the V D / / relation proposed by Professor Gerberich, the corrosion

fa-tigue crack growth rate would be inversely proportional to the square root of

cyclic-load frequency / a t all frequencies and not only at 20 Hz The rate also

would be proportional to the square root of diffusivity D at each frequency,

which would give rise to an activation energy equal to one half that for

diffu-sion, that is, 18 kJ/mol versus 36 kJ/mol Clearly, these predictions are not in

agreement with experimental data (Figs 5 to 7) The argument for diffusion

control in this case is therefore without merit

The second question is more involved, and incorporates certain

presump-tions regarding the micromechanisms for fracture The answer to this question

has to be a qualified_ves, simply because much is yet to be learned, even though

ayes answer enjoys a fair amount of current experimental support Let us set

the record straight Modeling of corrosion fatigue as a linear superposition of

rates did not follow from the concept of rate-controlling process as Professor

Gerberich believes The model given by Eq 1 was empirically based, and

as-sumed that the contributions by fatigue, cycle-dependent corrosion fatigue,

and sustained-load growth can be treated independently and sequentially

[2,16] The separation of the environmentally assisted contributions into a

cycle-dependent term l(da/dN\f] and a sustained-load term [(da/dN)^cc] was

made in recognition of experimental fact and of physical reality The {da/

dN)^f term gives recognition to the existence of a cyclically deformed region

that is already embrittled and damaged by prior fatigue and environmental

interactions This term is present irrespective of the A'-level in relation to the

static load parameter ATis^j The {da/dN\^^ term, on the other hand, reflects

environmental interactions and the consequent crack growth during the

cur-rent cycle, and becomes significant vi^hen K exceeds /^iscc- The

micromecha-nism for fracture in these two instances may well be the same, but there

ap-pears to be a need to consider the two terms separately

The model is by no means perfect and its specific form is certainly open to

question It does, however, reflect reality reasonably well and is adequately

supported by experimental data The individual terms represent the overall

growth rates for the particular cracking modes, and this representation has

been useful in analyzing crack growth response to changes in environmental

conditions The oxide closure effect alluded to by Professor Gerberich has

been interpreted in terms of a reduction in the effective crack-driving force or

\K through the wedging action of oxides formed on the crack surfaces by

fretting induced reactions at low A7f-levels [32\ This reduction in effective

A A' can overshadow the embrittlement by hydrogen and result in crack growth

rates even lower than those observed in vacuum at the same applied A A'-levels

Significant crack branching in the environment can produce the same results

These effects must be considered in arriving at a more complete understanding

of corrosion fatigue

The issue of threshold, be it for corrosion fatigue or for stress corrosion

cracking, has philosophical and practical implications In a practical sense,

threshold can be defined only in terms of some "consensus minimum" rate

The authors would simply acknowledge that some will find this concept useful

and would elect not to engage in a discussion on this issue at this time

Professor Gerberich raises an important point in his third question,

al-though it is not germane to the authors' use of linear superposition in

for-mulating Eq 1, as indicated previously The essence of his question should be

whether it would be useful to further subdivide the {da/dN)^ and (da/dN\cc

terms to reflect the contributions and rates of individual microfracture

pro-cesses The answer isyes The overall rate would be obtained by using Eqs 8,

depending on whether the microfracture processes occurred concurrently

or in sequence The individual rates in Eqs 8 reflect the particular

hydrogen-microstructure interaction and would depend on the partitioning and rate of

supply of hydrogen to the particular microstructural site Given a ready supply

of hydrogen, the microfracture processes may become rate controlling

(al-though the concept of rate control becomes less well defined when the

proc-esses are not truly independent, for example, for concurrent fracture events)

Which of the microfracture processes predominates (vis-a-vis rate controlling)

depends on how hydrogen is partitioned among the various processes This

subdivision in crack growth rate suggested by Professor Gerberich is

impor-DISCUSSION ON FRACTURE MECHANICS 25

tant for furthering the understanding of the influence of microstructure on

environmentally assisted crack growth The subdivision, however, is not

essen-tial to the understanding of overall crack growth response when one of the

preceding processes in the embrittlement sequence is in control It is hoped

that this important point does not get lost in discussion

It is hoped that both the discussion of Professor Gerberich and this closure

have given the reader a clearer appreciation that corrosion fatigue is indeed a

multifaceted problem Its understanding and solution requires cooperative

efforts by chemists, physicists, mechanicians, and materials scientists to

ad-dress the issues of chemical reactions, hydrogen-metal interactions,

microme-chanics of fracture, and a host of other questions

Corrosion-Fatigue Cracl< Initiation

Behavior of Four Structural Steels

REFERENCE: Novak, S R., "Corrosion-Fatigue Craclt Initiation Behavior of Four

Structural Steek," Corrosion Fatigue: Mechanics, Metallurgy, Electrochemistry, and

Engineering, ASTMSTP 801, T W Croolcer and B N Leis, Eds., American Society for

Testing and Materials, 1983, pp 26-63

ABSTRACT: This study was undertaken to investigate the corrosion-fatigue crack

initia-tion (CFCI) behavior of steels with nominal yield strengths from 248 to 1034 MPa (36 to 150

ksi) Notched specimens were exposed to a 3.5% NaCI solution under

constant-load-am-plitude conditions at a stress ratio R of 0.10 and at a cyclic frequency of 12 cycles per

minute (0.2 Hz) Results of crack-initiation life {N{) were characterized in terms of cyclic

stress-intensity range normalized relative to notch-tip root radius [(A/T/Vp)] and in terms

of cyclic-stress range at the notch tip (Aa^^)

Results showed that the CFCI behaviors for the A36, A588-A, A517-F, and V-150 steels

investigated were virtually identical Each steel exhibited a linear relationship of (AX'/Vp)

versus log A/| for cyclic lives from A'l s 10''to3 X 10* cycles, where the latter value reflects a

continuous testing period of about 180 days (6 months) All four steels exhibited about the

same cyclic-stress range of (AAT/v'p) s 207 ± 21 MPa (30 r; 3.0 ksi) at W, = 3 X lO''

cycles No evidence of a CFCI threshold behavior was determined or apparent up to 3 X

10'' cycles Compared with the respective estimated values of fatigue-crack-initiation (FCI)

threshold in air 1(AA'/Vp),h £ 448, 552, 758, and 1138 MPa (65, 80, 110, and 165 ksi)],

such values of (AA'/v'p) for CFCI behavior correspond to degradations of about 54, 62, 72,

and 82% for the A36, A588-A, A517-F, and V-150 steels, respectively The cited values of

(A/f/Vp) at 3 X 10' cycles were also equivalent to a cyclic-stress range at the notch tip of

Aa„3, s 234 ± 24 MPa (34 ± 3.4 ksi)

The results show that the CFCI behavior of A36, A588-A, A517-F, and V-150 steels is

finite for all current test conditions and is determined by the absolute level of the

cyclic-stress range [(A/f/Vp) or Aa^^^] Such results are in direct contrast with well-e.stablished

fatigue behaviors in air, where the FCI threshold level [(AK/Vp)fj, or (Aa^^jffJ varies

directly with strength level (a^^ or a,,) Thus, the present results show that, whereas FCI

behavior varies directly with strength level, the CFCI behaviors for all four steels studied

were virtually identical and occurred independently of strength level

KEY WORDS: corrosion fatigue, corrosion-fatigue crack initiation, corrosion-fatigue

en-durance, corrosion-fatigue strength, corrosion-fatigue threshold, crack initiation, cracks

from notches, cyclic loading, environmental behavior, environmental cracking,

environ-mental evaluation, environenviron-mental fatigue, fatigue testing methods, ferrite-pearlite steel,

fracture mechanics, high-strength steels, linear clastic fracture mechanics, long-life

be-havior, long-term testing, material bebe-havior, martensitic steel, mechanical bebe-havior,

notched-specimen behavior, notched-specimen fatigue, notched-specimen cracking,

'Senior Research Engineer, U.S Steel Corporation Research Laboratory, Monroeville, Pa

15146

26

NOVAK ON CRACK INITIATION BEHAVIOR 2 7

Stress range, stress-intensity (factor) range, structural behavior, structural integrity,

struc-tural steels, salt-water cracking, sodium chloride solution, stress-concentration effects,

strength-level effects

Corrosion fatigue is generally recognized as an important phenomenon that

can lead to unexpected cracking behavior and failure of structures under

cer-tain conditions Such conditions depend on the specific combination of

mate-rial, cyclic loading, and environment of concern, which in turn represent the

metallurgical, mechanical, and electrochemical components of the

corrosion-fatigue problem, respectively Because it is a synergistic effect of corrosion-fatigue or

cyclic loading (in air) and stress-corrosion cracking (SCC) acting together,

corrosion fatigue can lead to far greater degradation in material load-carrying

capacity compared with either effect acting alone or with expectations based

on linear superposition of the individual effects

Despite increasing study in recent years [1,2], basic understanding of the

complex problem of corrosion-fatigue behavior is still quite limited, with

rela-tively little or no a priori predictive capability available in quantitative terms at

the present time.-^ Recent studies in the area of corrosion fatigue have

empha-sized the linear elastic fracture mechanics (LEFM) approach and crack

growth behavior, a trend that has occurred at the expense of systematic studies

on crack initiation behavior Recent studies have also shown a general lack of

reliable corrosion-fatigue results, and virtually a complete absence of

corro-sion-fatigue crack initiation (CFCI) behavior results, particularly those

ob-tained under the slow-frequency and long-life conditions essential for

predic-tions of long-term structural performance

Because corrosion fatigue can reduce the effective fatigue limit significantly

(as much as a factor of ten) and because crack initiation can account for a large

fraction of total cyclic life (as much as 90% or more), a study was undertaken

to examine the CFCI behavior of four steels that span the general strength

range [207 to 1034 MPa (30 to 150 ksi)] of greatest interest in most engineering

applications The purpose of the study was to establish the baseline CFCI

behavior of typical constructional steels in a salt water environment These

studies were conducted by using a newly developed test technique, and were

performed under slow-frequency and long-life conditions that are of interest in

a number of engineering applications that involve cyclic loading and

environ-mental exposure (bridges, offshore oil platforms, submarines, surface ships)

Materials, Experimental Techniqne, and Stress Analysis

Materials

The four steels used in the current study were ASTM A36, A588-A, and

A517-F plate, and a seamless casing steel designated as V-150 The chemical

The italic numbers in brackets refer to the list of references appended to this paper

compositions and mechanical properties of each steel are presented in Tables 1

and 2, respectively The three plate steels were cut from the same 25.4-mm

(1-in.)-thick plates used in earlier studies [3-5] of fatigue, fracture, SCC, and

cor-rosion-fatigue crack-growth-rate behaviors of constructural steels The V-150

steel was cut from the wall of a 248-mm (9.75-in.)-diameter pipe with a nominal

15 mm (0.600 in.) wall thickness The latter segments were flattened and

subse-quently heat-treated to reestablish original pipe properties (Table 2)

Experimental Technique

The experimental technique employed consisted of testing

single-edge-notched specimens subjected to cyclic loading under conditions of cantilever

bending by using a six-stand test facility (Fig 1) All tests were conducted in a

3.5% solution of NaCl in distilled water, and an optical system was used for

crack detection

Specimen Preparation—Specimens of the design shown in Fig 2 were

pre-pared for each steel in the L-T crack orientation [per ASTM Test for

Plane-Strain Fracture Toughness of Metallic Materials (E 399)].^ The three

25.4-mm (1 -in.)-thick ASTM steel plates were split lengthwise at midthickness and

specimens were machined from each half The specimens from the V-150 steel

were machined from the flattened and heat-treated pipe-wall plates and were

oriented with the specimen length coinciding with the original pipe length All

specimens had a nominal thickness (B) of 10 mm (0.40 in.) The choices used

for the specimen notch depth [a„ = 25.4 mm (1.00 in.)] and notch-root radius

[p = 3.07 mm (0.125 in.)] were somewhat arbitrary, but were still based on

several important considerations involving the ease and extent to which

exami-nation of the notch tip could be achieved both optically and in a reliable

man-ner (positive or unambiguous crack detection)

During specimen preparation, the notch depth was first machined to a^ =

25.1 mm (0.990 in.), with an additional 0.25-mm (0.010-in.) increment

subse-quently removed during the final polishing process at the notch tip Precise

measurements of the notch depth were made after the machining and after the

final polishing steps to ensure adequate material removal The final polish at

the notch tip was obtained by using a 3- to 6-/tm diamond paste; the resulting

surface finish provided an excellent uniform background (at least initially) for

the detection of small cracks This standard notch-preparation procedure

pro-vided a means of minimizing both the presence and influence of residual stress

due to notch-tip machining on CFCI behavior The final step in specimen

preparation prior to actual testing was the cleaning of the notch-tip region with

a degreasing agent

Test Conditions—All specimens were tested under total-immersion

condi-^ L-T = full-thickness crack stressed in the longitudinal (L) direction and oriented to propagate

transverse (T) to the primary rolling direction for plate materials

NOVAK ON CRACK INITIATION BEHAVIOR 29