STP 463-1970

REVIEW OF DEVELOPMENTS
IN PLANE STRAIN FRACTURE

TOUGHNESS TESTING
W. F. Brown, Jr., editor
ASTM SPECIAL TECHNICAL PUBLICATION 463
List price $18.25
AMERICAN SOCIETY FOR TESTING AND MATERIALS

1916 Race Street, Philadelphia, Pa. 19103
NATIONAL AERONAUTICS AND

SPACE ADMINISTRATION
Copyright by ASTM Int'l (all rights reserved); Mon Dec 7 13:19:19 EST 2015
Downloaded/printed by
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized.
(~) BY AMERICANSOCIETYFOR TESTING AND MATERIALS1970
Library of Congress Catalog Card Number: 72-97728
ISBN 0-8031-0058-2
NOTE
The Society is not responsible, as a body,
for the statements and opinions
advanced in this publication.
Printed in LutherviUe-Timonium, Md.

September 1970
Foreword
The object of this book is to bring together a series of papers which de-
scribe recent experience obtained in the United States and in England with
the application of the ASTM E-24 Proposed Method of Test for Plane Strain
Fracture Toughness of Metallic Materials. This information supplements that
which has appeared in ASTM STP 381 on Fracture Toughness Testing and
its Application and in STP 410 on Plane Strain Crack Toughness Testing of
High Strength Metallic Materials. This publication is a cooperative effort of
ASTM and NASA. NASA research personnel have participated in ASTM
Fracture Committee activities since their inception in 1959. This participation
reflects the strong interest that NASA has maintained in the development
of test methods for evaluation of the fracture resistance of engineering
materials. This interest arises from the necessity for the use of high-strength
alloys in critical parts of aerospace structures. By cooperation with ASTM
in publication of this book NASA is helping to fulfill its obligation to
provide the widest practicable and appropriate dissemination of results from
its activities.
George C. Deutsch
Assistant Director for Materials Sciences and Engineering,
NASA Headquarters, Washington, D. C.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions autho
Related
ASTM Publications
Fracture Toughness Testing and Its Applications, STP
381 (T965), $19.50
Plane Strain Crack Toughness Testing of High Strength
Metallic Materials, STP 410 (1967), $5.50
Electron Fractography, STP 436 (1968), $11.00
Contents
PAGE
Introduction to STP 463--w. F. BROWN, JR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
P r o g r e s s in F r a c t u r e T e s t i n g o f M e t a l l i c M a t e r i a l s - - J . G. KAUEMAN . . . . . . . . . 3
Theory ......................................................... 4
Initial Recommendations ......................................... 4
Thickness as a Problem .......................................... 7
Fatigue Cracking ................................................ 8
Emphasis on Thick Sections ...................................... 9
Evolution of Notched Bend Specimens ............................. 12
C o m p a c t K~c S p e c i m e n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Screening Tests .................................................. 15
Thin-Section Problem ............................................ 18
Medium-Strength Materials ....................................... 18
Cautionary Notes ................................................ 18
Summary ....................................................... 20
Evaluation of a Method of Test for Plane Strain Fracture Toughness Testing
U s i n g a B e n d S p e c i m e n - - R . H . H E Y E R A N D D . E. MC CABE . . . . . . . . . . . 22
Pilot Program ................................................... 23
Interlaboratory Tests ............................................. 23
Fatigue Cracking ................................................ 24
Bend Testing .................................................... 30
A n a l y s i s o f K~c D a t a . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Conclusions ..................................................... 40
B r i t i s h E x p e r i e n c e w i t h P l a n e S t r a i n F r a c t u r e T o u g h n e s s (KIt) T e s t i n g - -
M. J. MAY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Collaborative Test Program ....................................... 43
First Stage .................................................... 43
Second Stage .................................................. 44
S t a n d a r d i z a t i o n o f KIc T e s t i n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Influence of Fatigue Precracking Conditions ...................... 47
Crack Length-Specimen Thickness Requirements .................. 51
Draft British Standard ........................................... 61
Summary ....................................................... 61
Discussion ...................................................... 62
T h e I n f l u e n c e o f C r a c k L e n g t h a n d T h i c k n e s s in P l a n e S t r a i n F r a c t u r e T o u g h -
n e s s T e s t s - - M. H. JONES AND W. F. BROWN, JR . . . . . . . . . . . . . . . . . . . . . . 63
Material and Specimen Preparation ................................ 66
Test Procedure .................................................. 67
Analysis of Data ................................................ 68
Bend Specimens ............................................... 68
Smooth Specimens ............................................. 71
Results for Effects of Thickness and Crack Length ................... 72
S c r e e n i n g T e s t s f o r K~c . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
C o r r e l a t i o n o f K~c w i t h T e n s i l e P r o p e r t i e s . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Practical Significance of Results ................................... 85
Appendixes ..................................................... 88
Discussion ...................................................... 92
University of Washington (University of Washington) pursuant to License Agreement. No further reproduction
vi CONTENTS
PAGE
C r a c k T o u g h n e s s Testing o f High-Strength Steels--E, A. STEIGERWALD. . . . . . 102

N o t c h Bend Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Km D a t a for High-Strength Steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
S u m m a r y a n d Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Slow Bend K~c Testing o f M e d i u m - S t r e n g t h H i g h - T o u g h n e s s Steels--s. T.
R O L F E A N D S. R, N O V A K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Materials a n d Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . t25
Specimen G e o m e t r y a n d Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Results a n d Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Kio Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Pop-in Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Specimen Size R e q u i r e m e n t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Effects o f Stress Level D u r i n g Fatigue Cracking . . . . . . . . . . . . . . . . . . . 144
Effect o f Face N o t c h i n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Correlation Between Krc a n d C V N Values . . . . . . . . . . . . . . . . . . . . . . . . 145
S u m m a r y a n d Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Application o f Fracture Mechanics T e c h n o l o g y to M e d i u m - S t r e n g t h S t e e l s - -
W . G . C L A R K , J R . , A N D E. T . W E S S E L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
I n f o r m a t i o n R e q u i r e d in Fracture Mechanics T e c h n o l o g y . . . . . . . . . . . . . 161
Material Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Discussion o f Materials Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Example P r o b l e m . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
General Stress Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Calculation o f Critical Flaw Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Calculation o f Cyclic Life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Material Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
D e v e l o p m e n t o f Inspection Criteria a n d Safety F a c t o r s . . . . . . . . . . . . . 185
Summary ....................................................... 186
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
F r a c t o g r a p h i c Analysis o f the L o w Energy F r a c t u r e o f an A l u m i n u m A l l o y - -
J. P. T A N A K A , C . A. P A M P I L L O , A N D J. R. L O W , J R . . . . . . . . . . . . . . . . . . . . 191
Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
F r a c t o g r a p h i c Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Failure o f Large Inclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Identification o f Void-Nucleating Particles . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
F r a c t o g r a p h i c Observation o f Boundary Between Fatigue C r a c k a n d
Dimpled Rupture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
C o m m e n t a r y on Present P r a c t i c e - - w . F. B R O W N , J R . , A N D J. E. S R A W L E Y . . . . . 216
Specimen Size R e q u i r e m e n t s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Fundamental Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Effects o f Plastic D e f o r m a t i o n in K~c, Tests . . . . . . . . . . . . . . . . . . . . . . 221
Face G r o o v i n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Specimen Preparation and Test Procedure . . . . . . . . . . . . . . . . . . . . . . . . . 227
Crack Starter Configuration and Displacement Gage Length . . . . . . . 227
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions aut
CONTENTS vii
PAGE
Fatigue Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

Nonuniformities in Fatigue Cracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Fatigue Crack Sharpness Requirement . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Testing of Brittle Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
C o m p a r i s o n of Plane Strain Fracture Toughness of Various Alloys . . . . . 232
Correlation of KIc with Other Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Correlation with Tensile Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Tensile Ductility in Material Specifications . . . . . . . . . . . . . . . . . . . . . . . . 240
Correlation with Impact Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Fracture Tests with Subsized Specimens (Screening Tests) . . . . . . . . . . . . . 243
Material Selection for Particular Applications . . . . . . . . . . . . . . . . . . . . . 244
Acceptance Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Alloy Development Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Tentative M e t h o d o f Test for Plane Strain Fracture Toughness of Metallic
Materials ( A S T M Designation: E 399-70 T) . . . . . . . . . . . . . . . . . . . . . . . 249
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions author
STP463-EB/Sep. 1970
Introduction
This Special Technical Publication brings together a set of papers that

were presented at a Panel Session on Plane Strain Crack Toughness sponsored
by the ASTM E-24 Committee on Fracture Testing of Metals during the
1968 ASTM Annual Meeting in San Francisco. The main purpose of this
Panel Session was to review the practical experience in fracture toughness
testing which had developed since ASTM Committee E-24 placed its major
efforts in standardization on methods of test for plane strain fracture tough-
ness. These efforts were started early in 1965, and the initial developments
were described in A S T M S T P 410.1 The paper by Kaufman reviews these
developments and gives a brief history of ASTM Committee E-24 activity.
In the latter part of 1966 Subcommittee I of ASTM Committee E-24 on High
Strength Metallic Materials formulated a Proposed Recommended Practice
for Plane Strain Fracture Toughness Testing of High Strength Materials
Using a Fatigue-Cracked Bend Specimen. Draft copies of this document were
made available early in 1967, and during the next year and a half considerable
experience was gained in applying the Proposed Practice to a wide variety of
metallic materials.
During this trial period various practical problems were encountered,
some due to tke inherent limitations of elastic crack mechanics and others
associated with certain details of specimen preparation and testing. The
members of the Panel, who all had well-established fracture testing programs,
were asked to emphasize these problem areas. As a result of this Panel meet-
ing and subsequent discussions among the ASTM E-24 Comimttee members
several changes were made in the Recommended Practice, and a revised
document was published in the 1969 A S T M B o o k o f Standards. This revision
incorporated a compact tension specimen as well as the bend specimen, and
was designated as a Proposed Method of Test for Plane Strain Fracture
Toughness of Metallic Materials. This Proposed Method was subsequently
further revised as described in the concluding contribution to this STP and
was advanced to a Tentative Method (E 399-70T), a copy of which has been
bound at the back of this volume.
Since the papers which appear in this STP were presented, ASTM Com-
mittee E-24 Method of Test has been incorporated into various specifica-
tions issued by both the U. S. Government and by private industry. It forms
t Brown, W. F., Jr., and Srawley, J. E., Plane Strain Crack Toughness Testing of High
Strength Metallic Materials, ASTM STP 410, American Society for Testing and Materials,
1966.
1
Copyright9 1970by ASTMlntcrnational www.astm.org
2 PLANE STRAIN FRACTURE TOUGHNESS TESTING
a part of the Aerospace Material Document series issued by the Society of

Automotive Engineers and is also the basis far qualifying KIe data for in-
corporation into MIL-Hdk-5. These applications have revealed certain
aspects of the Test Method which could be improved, but, what is more
important, they have shown that there is a strong desire on the part of many
investigators to compromise what they consider to be the overly strict speci-
men size requirements. Suggestions that the required crack length or speci-
men thickness or both could be reduced appear in several of the papers o f
this volume. This concern about specimen size arises naturally out of attempts
to apply the Test Method to materials having lower strength and higher
toughness than those originally intended by the ASTM E-24 Committee.
This volume contains two contributions that were not part of the Panel
Session on Plane Strain Crack Toughness. The first is a paper by J. R. Low,
Jr., and his associates describing some recently completed work on the fracto-
graphic analysis of the aluminum alloy 2014-T6. This paper is quite pertinent
since it provides a plausible explanation for the relatively low plane strain
fracture toughness characteristic of this and possibly other high-strength
aluminum alloy plate. The second is a contribution by J. E. Srawley and
myself entitled "Commentary on Present Practice." This not only summarizes
the salient features of the various papers, but also attempts to clarify certain
aspects of plane strain fracture toughness testing where there appears to be
some confusion.
W. F. Brown, Jr.
Chief, Strength of Materials
Branch, National Aeronautics
and Space Administration--
Lewis Research Center,
Cleveland, Ohio 44135.
J. G. Kaufman 1
Progress in Fracture Testing of

Metallic Materials
REFERENCE: Kaufman, J. G., "Progress in Fracture Testing of Metallic Mate-

rials," Review of Developments in Plane Strain Fracture Toughness Testing,
. 4 S T M STP 463, American Society for Testing and Materials, 1970, pp. 3-21.
ABSTRACT: Progress in the development of fracture-toughness testing techniques

by ASTM Committee E-24 is reviewed briefly. The evolution of the test method
for plane strain stress-intensity factor, K~o, is described, including treatment of
some of the problems dealing with fatigue cracking, thickness limitations and
crack growth detection, and measurement. Attention is given also to the prepara-
tion of the screening test method for thin sections. Some cautions are presented
concerning the necessity for meticulous standardization, despite the time required
to do an acceptable job, and the timing problems in incorporating methods
into specification requirements.
KEY WORDS: fractures (materials), toughness, test techniques, fatigue pre-

cracking, thickness, crack growth measurement, screening, standardization,
evaluation, tests
In 1958, A Special ASTM Committee on Fracture Testing, now known

as ASTM Committee E-24, was established for the purpose of developing
and writing test methods for the determination of the fracture characteristics
of materials as they involve the susceptibility of the material to what has
been commonly called "brittle fracture" or "catastrophic failure" [1]. 2
Failures without warning in certain critical missile applications at stresses
far below those normally expected based upon conventional design tech-
niques provided the primary stimuli behind the formation of the committee.
Despite the prolificacy of the committee (or perhaps because of it), some
confusion exists with regard to the path which has been followed in develop-
ing fracture-test methods in ASTM Committee E-24. It is the purpose of this
paper to review the progress of the committee effort, presenting the reasons
behind the various actions taken and the shifts in emphasis over the years.
1 Section head, Fracture Mechanics and Product Evaluation Section, Mechanical Testing
Division, Alcoa Research Laboratories, Aluminum Company of America, New Kensing-
ton, Pa. Personal member ASTM.
The italic numbers in brackets refer to the list of references appended to this paper.
3
Copyright9 1970 by ASTM International www.astm.org
To preserve the continuity, it is necessary to present the review in rather

general terms, oversimplifying in a number of instances and avoiding deep
involvement in technical details. Similarly, it is impossible to acknowledge
and comment on every individual contribution along the route; rather, the
progress is reported as a committee action.
Theory
Under the guiding hands of J. R. Low and G. R. Irwin, it soon was estab-
lished that elastic fracture mechanics was the vehicle by which the resistance
of materials to unstable crack growth could be best described. Though not
without considerable complication by the fact that most real materials do
not behave in a purely elastic manner on fracturing, the limitations neverthe-
less did not appear insurmountable. It was hoped that, with appropriate
modifications to take into account the finite sizes of structural members and
test specimens, and the crack-tip plasticity, small specimens could be used to
recreate and describe the situation when unstable crack growth develops in
a large structure. Developments over the years have sustained this opinion.
Initial Procedural Recommendations

The first significant step taken by the committee was the establishment of
some recommendations for the fracture testing of materials using the basic
and convenient model of an axially loaded center-notched or double-edge-
notched panel [1 ]. The stress analysis for this stiuation was reasonably well in
hand, and it seemed desirable to proceed with symmetrically loaded specimens
to reproduce the condition of crack growth to fracture. The load and crack
length at the onset of unstable crack growth to fracture were determined,
and the critical stress-intensity factor, K,, and critical strain-energy release
rate, G~, were calculated. Two modifications of the basic equation were made,
one for the width of the specimen and the other for the expected size of the
zone or plastic deformation at the crack tip. Guidelines for making these
tests were laid out in the First Report of the Special Committee on Fracture
Testing [1 ], which we must observe in retrospect has served as one of the
principal guides for fracture testing throughout the intervening years.
Broader use of this procedure, particularly the center-notched specimen
soon unearthed some troublesome problems and resulted in some important
refinements. One concerned the important part of the test involving the
measurement of the "critical" crack length, that is, the crack length when
unstable crack growth began. Originally it was suggested that crack growth
be measured by an ink-staining technique, presuming that the ink would
reliably follow the slow crack growth by capillary action but would not follow
the rapid unstable growth of a crack. Although useful measurements were
made when care was taken with choice and quantity of ink (Fig. la), the
KAUFMAN ON FRACTURE TESTING OF METALLIC MATERIALS 5
FIG. 1--Fracture surfaces o f center-notched specimens with which the ink-stain procedure
was used Jot following slow crack growth (a) with relatively good success and (b) without
success because o f spattering and seepage.
potential variability in fluidity and spattering was intolerable (Fig. lb). The
displacement gage [2] and electrical potential [3,4] means of following crack
growth were developed, followed later by the acoustic [5] and ultrasonic
techniques [6]; in very thin sections, visual observations on the surfaces were
used also [7]. Of these, the displacement gage, which relies on the change in
elastic compliance of the specimen with change in crack length to indicate
the extent of crack growth [8], is probably the most widely used. The type
used at Alcoa Research Laboratories (ARL) is shown in Fig. 2 with some
representative curves [9]; the difference in slope of a line through the origin
to the point of instability from the original slope is the measure of crack
growth, provided it has taken place under essentially elastic conditions. Some
investigators prefer to use a combination of techniques, including the dis-
placement gage and the electrical potential or ultrasonic techniques as auxili-
ary sensors.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorized
z
z
T,
>
C
c'-
o
7,
"2
o
FIG. 2--Load-deformation curve from tension tests o f center-notched fracture toughness specimens in which SR-4 clip
gages were used to measure deformation (and by compliance calibration, the crack growth).
Thickness Emerges Clearly as a Problem

The broad use of more sensitive devices for following crack growth played
an important role in further refinements of fracture testing, as it made it
easier to recognize two things:
(a) The fact that Kc is a thickness-dependent property, as exemplified by
the data for 7075-T6 sheet and T651 plate in Fig. 3. Over a certain critical
range of thickness for each material, the toad and the crack length at in-
stability (and, therefore, Kc) decrease with increase in thickness, reaching a
rather constant minimum value when plane strain conditions are approached
closely. In the near fully plane strain situation, the instability leading to
complete fracture almost immediately follows crack initiation. This point
had been observed earlier [1], but not fully appreciated.
(b) The fact that the initial crack growth, whether stable or unstable, took
place at about the same level of stress intensity for a given material, regardless
of specimen thickness [2]. Thus, it might be possible to learn something about
a plane strain instability, even though complete fracture of the specimen did
not take place under plane strain conditions.
As a result of these two observations, the concepts of a lower level of
critical stress-intensity factor associated with plane strain conditions (Kic)
and of the capability to approximate the plane strain stress-intensity factor
from the conditions under which stable crack growth initiated were solidified
[8]. Both of these points were vital and remain with us today, although both
have changed complexion somewhat.
7O
L~ 60
R
o
-- 5O
z
o
40
r/I~~~, /11~'/// ////~///

/ " Ronge of Meosured KIr Volues
c
~ 2o
,3
0
0 0,2 0.4 0,6 0,8 1,0 1.2
Thickness, in,
F I G . 3--Fracture toughness of 7075-T6, T651 sheet and plate from tests of fatigue-cracked
center-notched specimens (transverse).
8 PLANESTRAIN FRACTURE TOUGHNESS TESTING
As a result of the close attention to crack initiation a "pop-in" concept

developed [10], pop-in being defined optimistically as a small initial burst of
unstable crack growth under plane strain conditions which is arrested because
the crack growth itself caused a departure from plane strain conditions.
An important aspect of plane strain fracture testing became the search for
pop-in, a clear example of which is presented in the second curve in Fig. 2.
More will be said about this later.
Fatigue Cracking
One of the next refinements to develop was a logical outgrowth of greater
concern about crack initiation and how it is affected by the stress condition
and stress concentration at the tip of the notch. Originally, the only require-
ment on the notch-tip radii was that they be <0.001 in. [1], a requirement
readily met by machining many test materials, especially aluminum alloys.
Although data for the aluminum alloys suggested that values of Ko for thin
sheet did not depend on sharpening the notch by fatigue cracking, so long as
other requirements (on net section stress, a / W , etc.) were met, the picture
was different with regard to values of K1 c. The initial crack growth was found
to take place at l0 to 15 percent lower values of stress-intensity factor in
fatigue-cracked specimens than in even the most sharply machine-notched
specimens, and with a less-clear indication of pop-in, as indicated by the
representative curves in Fig. 4 [9]. Since fracture-toughness testing was aimed
FIG. 4--Representative load-deformation curves from fracture-toughness tests, showing the

clear "'pop-in" with machine-notched specimens contrasted with the gradual development of
crack growth in the fatigue cracked specimens. Note also that crack growth starts at signi-
ficantly lower load levels in the fatigue cracked specimens, though the final fracture instability
takes place at nearly the same conditions in the two types of specimens.
at the behavior of metals in the presence of cracks or metallurgical flaws,

these findings solidified the requirement that specimens must have been
fatigue cracked. This brought on one of the most perplexing aspects of
fracture-toughness testing which exists even today--that of providing an
adequate fatigue crack by stressing at a high enough value to obtain reason-
able growth rates, but one which is low enough to avoid stress-history effects
on the fracture test itself.
Emphasis Shifts to Thick Sections

Much of the early fracture-toughness testing was directed at relatively
thin sections, and the procedures prescribed in the First and Second Com-
mittee Reports [11] dealt almost entirely with thin-section problems. As
additional experience was gained, it became clear that not only was thickness
an important variable, as mentioned previously, but also some other con-
ditions were not quite as simple as was believed originally. Of particular
importance, it became clear that a large amount of variation was obtained in
tests of a given thickness of material where different sizes of specimen were
used and there were indications that the elastic response of the testing machine
in which a test was made might be influencing the results significantly. In
short, the possibility that Kc, as defined at that time, might not be a single-
valued property even for a given thickness was indicated.
In view of the above complications, plus the implication that the ap-
parently thickness-independent plane strain fracture toughness might be a
more fundamental property of materials, effort shifted to concentration
upon methods to determine KI c. It is to be stressed that the attention shifted,
not because of less interest in the thin-section problem, but because of the
greater likelihood that this perhaps more fundamental property might
provide a firmer basis for the development and handling of the overall
fracture problem. Knowledge that K~ c provided a "conservative" approach
to the fracture problems and that fatigue and stress-corrosion failures even of
relatively thin members seemed controlled by plane strain fracture conditions
added impetus to the shift. It was anticipated that once the plane strain
problem was solved, attention could return to the thin-section (plane-stress
or mixed mode) problem.
Thus over the next five years, from about 1963 to 1968, attention focused
rather narrowly on the thick section or plane strain problem, and even in this
region we have seen several transitions. A variety of specimens have been
considered for determining the plane strain toughness [12], the first perhaps
being the notched round specimen; however, the large amount of material
and large testing machines required for this type of test discouraged its wide
use. Center-notched specimens also were used [2], but again material and
testing-system problems governed, and attention was soon focused on what
commonly was called the single-edge-notched (SEN) tension specimen [12];
]0 PLANE STRAIN FRACTURE TOUGHNESS TESTING
i / I i I I
36 O B X I~'2 • 6 IN.
A BX3XI21N.
0 I X 4 1 / 2 X 18 IN.
32
0 24
0 I
g a
b
LQ
O' I I I I
0.2 0.4 01.6 0.8 1.0 1.2
Specimen Thickness, in.
F I G . 5--Values of KQ (candidate value of K i t ) ,from tension tests of single-edge-notched

specimens from 1-in. 7178-T7651 plate.
it appeared for a time that the first recommended practice for K~c testing
would incorporate this type of specimen.
During the time in which the SEN tension specimen was used widely, con-
siderable attention was given to thickness requirements. A synthesis of data
led soon to a qualification which still stands that in order to be certain that
plane strain conditions prevailed in the test, the thickness of the specimen
must be of the order of 50 times the radius of plane strain plastic zone, which
is equivalent to 2 ~ times the square of the ratio of KI c to the yield strength
[12]. This is a rather severe requirement; while data for some materials sug-
gest that this factor may be reduced, 3 there are sufficient data suggesting that
it is none too conservative [13] 4 (Fig. 5) and that changes should be made
only after careful study.
During this same period, considerable attention was focused on the use of
face-grooved specimens to improve the likelihood of achieving plane strain
conditions [14]. However, further study showed that, although the instability
point under plane strain conditions appeared to be sharpened a bit by side
grooving, the practice does not consistently increase the ability to develop
plane strain conditions. In fact, it sometimes has the misleading trait of
suggesting by the relatively flat fracture appearance that plane strain condi-
tions have been achieved, when in fact they have not, as indicated by the
extent of local plasticity (Fig. 6) [13]. As a result, the practice of side grooving
remains outside the firm recommendations of the committee, although it still
is being used as a research tool by some investigators.
3 See p. 124.
4 See p. 160.
c
z
0
z
z
0
0
-4
FIG. 6--Fracture surfaces and load-deJormation curves (reversed direction Jor deformation) from tests oJ" regular and face-grooved single-
edge-notched fracture toughness specimens (1 X 4~-in. cross section) o f 2219-T851 plate.
Evolution of Notched Bend Specimens

Greater testing experience soon led to the observation that even the use of
single-edge-notched tension specimens for relatively tough materials re-
sulted in problems in loading capacity and gripping. The immediate solution
to the problems was simply to rotate the plane of the specimen by 90 deg, and
conduct a bend test [12]. Events at the crack tip were identical, and the
differing stress gradients were taken care of in the equations for calculating
K~,. With loading requirements at a practical level and specimen size re-
quirements now reasonably developed, the preparation of the first recom-
mended practice for pIane strain fracture-toughness testing was begun.
Some problems with fatigue cracking had to be dealt with, and one of the
most perplexing was how to provide the best chance of achieving a straight
crack front. With the straight machined notch used over the years with
FIG. 7--Fracture surJ&ees to illustrate fatigue crack fronts: (a) satisfactory from straight
notches, (b) unsatiafactory from straight notches, and (c) satisfactory from chevron notches.
SEN specimens, in some cases the fatigue-crack fronts were straight (Fig.
7a), but in others they were curved broadly as a result of the crack initiating
at a corner instead of at the middle or across the entire front (Fig. 7b). The
introduction of the "chevron" notch improved the situation (Fig. 7c), as the
crack invariably started at the center. Thus, the crack front was usually
rather straight coming out of the chevron, and, if curvature developed, it was
symmetrical and more readily controlled.
Concurrent with the development of the recommended practice, the
problems in determining the load with which to calculate K~c were being
faced. Despite early hopes that it would be possible to rely on the occurrence
of a crack pop-in to assure the presence of a plane strain instability, broader
testing experience with fatigue-cracked specimens indicated that this would
not be possible. It was clear that if pop-in was required, the usefulness of the
test method would be limited appreciably. The large pop-in was usually the
FIG. 7--Continued.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authorize
FIG. 7--Continued.
result of a blunt crack, and, as mentioned previously (Fig. 4), fatigue-

cracked specimens of many materials exhibited only a gradual departure from
linearity.
Study was made of other potential criteria for determining the load with
which to calculate a meaningful value of KI c. Some investigators expressed a
preference for the use of the initial deviation from linearity, but this had the
obvious shortcoming that the resultant value would vary widely with the
sensitivity of the displacement gages and chart recorders, and so the practice
was discouraged. The potential usefulness of a secant offset associated with a
small but specific amount of crack growth evolved. Two percent of effective
KAUFMAN ON FRACTURE TESTING OF METALLIC MATERIALS ]5
crack extension was selected, which is indicated by a 5 percent secant offset for
the notch-bend specimens.
The use of a secant offset instead of an obvious indication of an instability
carries with it the problem of ensuring that crack growth really has taken
place; overtly one cannot be sure whether the departure from linearity on the
load-displacement curve is associated with crack growth or plastic deforma-
tion. Therefore, it is necessary to impose some additional requirements on
the test in terms of criteria for validity of the calculated value. It can be
shown that the shape of the knee in the curve itself is a clear indication of
whether the offset is largely attributable to plastic deformation (very gradual
curvature) or to cracking (relatively sharp curvature) [15]; examples are
presented in Fig. 8. As a result, a requirement was developed that the curva-
ture must be relatively sharp, as indicated by the fact that the offset at 80
percent of the secant-intercept load must be very small compared with (less
than 25 percent of) the offset at the secant intercept load. This requirement,
like the offset itself, has the advantage of not being so much a function of
strain gage sensitivity, and of personal interpretation.
With this problem resolved, and after study and clarification of the fatigue
cracking techniques, the bend-test method was completed, evaluated via a
round robin program [16], and published in Part 31 of the ASTM Standards?
The Compact Kic Specimen

Refinement in the capability for the estimation of K~ c measurement capacity
and some farsighted independent research on the fracture toughness of nuclear
vessel materials [17] led to the conclusion that a compact specimen provided
a more efficient use of material than the notched bend specimen (Fig. 9).
Attention then focused now on the development of the details of the test
procedure with this particular specimen, and the latest version of the test
method for determining K~r incorporates the compact specimen as well as
the notched bend specimen. 5
Screening Tests
Over the same period of time during which the methods for the determina-
tion of K~c have been developing, attention also has been focused on screen-
ing tests, that is, tests useful for merit rating materials with regard to tough-
ness, but which do not provide a number useful for any design consideration.
Some recommendations concerning the use of sharp-notch tension testing as a
useful tool for the relative ranking of materials were made in the initial com-
mittee report [1 ]. These recommendations centered on the 1-in.-wide sheet-
type specimen, but because of: (1) the desirability of enlarging upon the
applicability of the test and (2) experience indicating that the larger size of
5 1968 Book o f A S T M Standards, Part 31.
]6 PLANESTRAIN FRACTURETOUGHNESS TESTING
FIG. 8--Representative load-deformation curves from notch-bend fracture toughness tests o f

X7005-T6351 (top) and 7079-T651 (bottom) plate; the upper curves illustrate the gradual
curvature largely associated with crack-tip plastic" deformation, while the lower curves illus-
trate the sharper curvature indicative o f crack extension.
KAUFMAN ON FRACTURE TESTING OF METALLIC MATERIALS ]7
FIG. 9--Compact tension and notched bend fracture toughness specimen o f equal K t ,
measurement capacity to illustrate greater efficiency o f material use with former.
specimen achieved this greater applicability through a reduction in the frac-

ture strength and therefore in the complication of plastic deformation, a 3-in.-
wide specimen finally was selected, and either center or edge-notching was
permitted (Fig. 4 of footnote 6). This 3-in.-wide specimen was the basis of the
first Recommended Practice for Sharp-Notch Tension Testing of High-
Strength Sheet Materials developed by the committee, 6 a practice which
soon will evolve into an ASTM and USASI Standard Test Method.
This practice provides that the ratio of the sharp-notch strength to the
tensile yield strength, in addition to the sharp-notch strength, should be
calculated and is of significance for screening materials. This is in recognition
that this ratio is a better indication of the amount of plasticity that a material
is capable of developing (instead of cracking) in the presence of a severe
notch (which is a meaningful relative measure of fracture toughness) [18]
than is the commonly used ratio of the sharp-notch strength to the tensile
strength.
It is appropriate at this point to contrast the screening test with the method s
for determining KIe. Contrary to the thoughts of many, an invalid value of
K~ ~(that is, a value determined by a test which does not meet all of the criteria
for validity), is not a reliable and useful relative measure of toughness.
Therefore, while the temptation to use the invalid values "just for screening"
is strong, it may give false information and must be avoided. The reason for
this is that there is no way to gage how far away from the true value of K~ c
the invalid number really is, and whether the invalid KQ is larger or smaller
t h a n KI c. For some materials, an invalid value may be essentially or exactly
equal to K~c, but, for another material, an equally invalid number (that is,
invalid for identical reasons) may be far removed from the real value. An
example of this situation was given in Ref 13 by comparing invalid data for
three aluminum alloys for which the invalid values line up the materials in an
unrealistic manner based on all screening criteria and, more important,
6 1968 Book o f A S T M Standards, Part 31, pp. 963-971.
upon service experience. This point is emphasized solely to illustrate that

invalid K ~ numbers can be safely used only for guidance on how to obtain
a valid test.
Back to the Thin-Section Problem

Since a substantial amount of progress now has been made on the thick-
section problem, it is appropriate to refocus attention on the thin-section
problem. Many of the fracture problems encountered in the aerospace and
missile industry are thin-section problems. Thus, in recent action of ASTM
Committee E-24, a new task group on thin sections has been formed; it
represents and will receive a large part of its guidance from the aerospace
industry.
Medium-Strength Materials
Although this review has been aimed principally at Subcommittee I on
High Strength Materials, it is appropriate to note that a considerable amount
of attention has been given to the use of elastic fracture mechanics for evalu-
ating the fracture toughness of the medium-strength materials, in which the
simplifications on limited plastic zone formation are not so readily applicable.
A review by Subcommittee III of this problem has indicated that, in general,
the same procedures described above can be used, though the specimens have
to be much larger because of the greater toughness; this results in the situa-
tion that: (a) the single-edge notched tension specimen is of little use because
of great load, loading shackle, and material requirements, (b) the notch-bend
specimen is satisfactory from the load and load-application standpoint, but
still requires a lot of material, and (c) the compact K~ c specimen appears to
be the optimum answer. Some hope remains for the double-cantilever beam
specimen [19] and for the incorporation of what commonly is referred to as
elastic-plastic analysis [20] in this area, but much development work re-
mains to be done before we put forth any new methods based on these ap-
proaches.
Cautionary Notes
Despite the progress which has been made, it has not been fast enough or
all inclusive enough to suit many individuals and organizations. To find out
why, it is necessary only to look at several restraints which confront those
involved in the fracture problem.
First, it is essential that fracture data which are to be used in the design
and analysis of structure or cross compared from laboratory to laboratory
be developed from standardized test procedures. Within the past several years,
we have seen the development of methods for making a variety of types of
University of Washington (University of Washington) pursuant to License Agreement. No further reproduc
fracture test. These have ranged from the rather thoroughly evaluated fa-
tigue-cracked notch-bend fracture-toughness test proposed by A S T M Com-
mittee E-24 to tests seemingly picked out of the air to fill a need for "some
kind of test." Some test methods of the latter type have been recognized by
reputable Technical Societies and yet incorporate ambiguous and untried
techniques; there is no testing experience on which to judge the effects of
test variables, such as specimen geometry and loading rate, and there is even
question as to whether the tests are practical. I cite these two extremes to call
attention to the fact that there are right ways and wrong ways to go about
developing test procedures for specifications. It cannot be done haphazardly
and it is up to the material producers as well as responsible people in the
sponsoring agencies to be certain that it is not done in this way.
Development of the test method in the proper way through systematic
investigation by a group such as A S T M Committee E-24 is not an easy
procedure. This development process is often a frustrating process for the
people who are doing it and for the people who are waiting for it. It is not
nearly as fast as most people would like. Nevertheless when a test method is
finally adopted, you can be reasonably certain that: (a) there have been
studies of many of the test variables on the results, (b) the test procedure has
been evaluated in a systematic program involving a minimum of six different
laboratories (in what commonly is called a round-robin test program), and
(c) there is some rather clear indication of the variability of the data and its
significance.
Second, and a logical consequence of the first, we cannot become too
deeply involved or diverted by procedures which, though they serve some
individual purpose, are not amenable to standardization or do not provide
the type of design data to further the needs of industry. As an example, Alcoa
Research Laboratories has used sucessfully the tear test [9] for almost 15
years to measure the relative resistance of aluminum alloys to fracture. The
test has been and still is very useful, because: (a) there are no limits as far as
we know on the applicability of the data because of such things as excessive
plastic deformation, (b)the data has been correlated with other indexes of
toughness such as Kic, and (c) most important, it has correlated on numerous
occasions with service experience. However, the tear test is not suitable for
standardization, because of such things as the dependence of results on test-
ing machine characteristics and the necessity of some judgmental interpreta-
tion of results. Some of the other tests which have been proposed fall into
this category. Others simply are not sensitive enough or do not provide the
type of data which will materially advance the design of structures.
Third, we must not expect that instantly upon the publication of a test
method it will be possible to incorporate a minimum level of K~c (or what-
ever other property is involved) into specifications. It commonly is assumed
University of Washington (University of Washington) pursuant to License Agreement. No further reproducti
that once a recommended fracture test procedure has been adopted it is

appropriate immediately to write minimum values of some fracture param-
eter into specifications. Nothing could be further from the truth. Actually,
after the procedure has been developed, it is necessary to obtain a large
number of data showing the amount of scatter in the results and the influence
of product variation (test direction, product dimension, product heat-treat-
ment) on the values, exactly as has been done with tension test specifications.
Only when this has been done systematically and with sufficient data to
justify statistical analyses, can values for specifications even be considered.
All such tests must be done in a valid manner; it is not satisfactory to con-
sider data developed over the years by procedures perhaps remotely re-
sembling those used in the test practice but differing in many details.
It is up to the material producers to be certain that no minimum values
find their way into specifications before they are justified. In these days of
such great demand for materials it is irresponsible to establish specification
values for any property at levels where vast quantities of acceptable materials
would be rejected needlessly, or where unacceptable materials might be used
with catastrophic end results. For this very reason, we have been extremely
cautious about what values go into design handbooks.
Summary
In summary, a great deal has been accomplished by ASTM Committee
E-24, and it has been accomplished in a reasonable time and by a reasonable
path considering the complexity of the task. Two test methods have been
written, a major addition to one will soon be completed. Guidance has been
given to the use of fracture mechanics in fatigue, stress-corrosion, and high-
rate loading. Much more will be accomplished in the future.
References
[1] "Fracture Testing of High-Strength Sheet Materials," ASTM Bulletin No. 243, Jan.
1960, p. 29; also No. 244, Feb. 1960, p. 18.
[2] Boyle, R. W., Sullivan, A. M., and Krafft, J. M., "Determination of Plane Strain
Fracture Toughness with Sharply Notched Sheets," Welding Journal Research Supple-
ment, Vol. 41, 1962, p. 428s.
[3] Anctil, A. A., Kula, E. B., and DiCesare, E., "Electric-Potential Technique for Deter-
mining Slow Crack Growth," Proceedings, American Societyfor Testing and Materials,
Vol. 63, 1963, pp. 799-808.
[4] Steigerwald, E. A. and Hanna, G. L., "Initiation of Slow Crack Propagation in High-
Strength Materials," Proceedings, American Society for Testing and Materials, Vol.
62, 1962, pp. 885-905.
[5] Jones, M. H. and Brown, W. F., Jr., "Acoustic Detection of Crack Initiation in
Sharply Notched Specimens," Materials Research & Standards, March 1964, pp.
118-128.
[6] Clark, W. G., "Ultrasonic Detection of Crack Extension in the WOL-type Fracture
Toughness Specimen," Materials Evaluation, Aug. 1967, p. 185.
[7] ASTM Special Committee on Fracture Testing of High-Strength Metallic Materials,

"Progress in Measuring Fracture Toughness and Using Fracture Mechanics,"
Materials Research & Standards, Vol. 4, No. 3, March 1964, pp. I07-119.
[8] Irwin, G. R. and Kies, J. A., "Critical Energy Rate Analysis of Fracture Strength,"
Welding Journal Research Supplement, Vol. 33, 1954, p. 193s.
[9] Kaufman, J. G. and Holt, Marshall, "Fracture Characteristics of Aluminum Alloys,"
Aluminum Company of America, ARL Technical Paper No. 18, 1965.
[I0] Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society
for Testing and Materials, 1965.
[11] "The Slow Growth and Rapid Propagation of Cracks," Materials Research & Stand-
ards, Vol. 1, 1961, p. 389.
[12] Brown, W. F., Jr., and Srawley, J. E., Plane Strain Crack Toughness Testing of High
Strength Metallic Materials, ASTM STP 410, American Society for Testing and
Materials, 1967.
[13] Kaufman, J. G., Nelson, F. G., Jr., and Holt, Marshall, "Fracture Toughness of
Aluminum Alloy Plate Determined with Center-Notch Tension, Single-Edge-Notch
Tension and Notch-Bend Tests," presented at the National Symposium on P~acture
Mechanics, Lehigh University, 1967.
[14] Freed, C. N. and Krafft, J. M., "Effect of Side Grooving on Measurements of Plane-
Strain Fracture Toughness," Journal of Materials, Vol. l, No. 4, Dec. 1966, pp. 770-
790.
[15] Srawley, J. E., Jones, M. H., and Brown, W. F., Jr., "Determination of Plane Strain
Fracture Toughness," Materials Research & Standards, Vol. 7, No. 6, June 1967,
pp. 262-266.
[16] Heyer, R. H., "Evaluation of E-24 Proposed Recommended Practice for K~ Testing,"
presented at 1968 ASTM Annual Meeting in San Francisco, California, June 24, 1968.
[17] Manjoine, M. J., "Biaxial Brittle Fracture Tests," Paper 64-Met-3, American Society
of Mechanical Engineers, May 1964.
[18] Kaufman, J. G. and Johnson, E. W., "The Use of Notch-Yield Ratio to Evaluate the
Notch Sensitivity of Aluminum Alloy Sheet," Proceedings, American Society for
Testing and Materials, Vol. 62, 1962, pp. 778-791.
[19] Mostovoy, S., Crosley, P. B., and Ripling, E. J., "Use of Crack Line Loaded Speci-
mens for Measuring Plane Strain Fracture Toughness," Materials Research Labora-
tory, Inc., Jan. 1966.
[20] Hahn, G. T. and Rosenfield, A. R., "Sources of Fracture Toughness: The Relation
Between K~ and the Ordinary Tensile Properties of Metals," Applications Related
Phenomena in Titanium Alloys, ASTM STP 432, American Society for Testing and
Materials, March 1968, pp. 5-31.
R . H . H e y e r I and D. E. M c C a b e 1
Evaluation of a Method of Test for Plane

Strain Fracture Toughness Using a
Bend Specimen
REFERENCE: Heyer, R. H. and McCabe, D. E., "Evaluation of a Method of Test

for Plane Stain Fracture Toughness Using a Bend Specimen," Review o f Develop-
ments in Plane Strain Fracture Toughness Testing, A S T M STP 463, American
Society for Testing and Materials, 1970, pp. 22-41.
ABSTRACT: A task group of ASTM Committee E-24 conducted an interlabora-

tory test program on determination of K~c using the proposed "Recommended
Practice for Plane Strain Fracture Toughness Testing of High-Strength Metallic
Materials Using a Fatigue-Cracked Bend Specimen."
Analysis of K~o data for an aluminum alloy and two high-strength alloy steels
showed that laboratory mean values were within 4-10 percent of grand mean
values. Replication was generally satisfactory, although not as good as for
smooth specimen tension tests.
Information obtained on fatigue cracking and other details of the procedure
will be useful in planned revisions, leading to a standard method of test.
KEY WORDS: interlaboratory tests, fractures (materials), toughness, fatigue

cracked bend specimen, fatigue precracking, plane strain, high strength materials,
aluminum alloys, steels, maraging steel, evaluation, tests
There were several important steps in the development of a linear elastic

fracture mechanics based test for plane strain fracture toughness, and these
are recorded in Ref 1 to 4. 2 Although a number of different specimen types
and loading arrangements could be used, the edge notched bend specimen
was selected at the 30 June 1966 meeting of ASTM Committee E-24 on
Fracture Testing of Metals as the first type to be developed. Subcommittee
I appointed a Task Group to draft a recommended practice using a fatigue
cracked bend specimen, to organize appropriate cooperative test programs
to help identify problem areas, and finally to evaluate the method by inter-
laboratory tests.
1 Supervising research metallurgist and senior research metallurgist, respectively, Armco
Steel Corp., Middletown, Ohio 45042. Mr. Heyer is a personal member ASTM.
2 The italic numbers in brackets refer to the list of references appended to this paper.
22
Copyright* 1970 by ASTM lntcrnational www.astm.org
HEYER AND McCABE ON A BEND SPECIMEN 23
Pilot Program
I n a pilot p r o g r a m six laboratories were supplied five m a c h i n e d specimens
each of a l u m i n u m alloy 7075-3"651, 13/~ by 3 by 13 in. A t the time m a n y of
the participants did n o t have the necessary fixtures a n d displacement gage,
as proposed in N a t i o n a l A e r o n a u t i c s and Space A d m i n i s t r a t i o n ( N A S A )
Technical M e m o r a n d u m X-52209, later published as A S T M S T P 410 [4].
The purpose of the pilot p r o g r a m was to check out the test fixtures, instru-
m e n t a t i o n , a n d analysis of data before c o n d u c t i n g the more extensive inter-
l a b o r a t o r y e v a l u a t i o n program.
The average KIo values obtained by the individual laboratories were in the
range 20.4 to 22.6 ksi in. 1/2, or -4-5 percent. Total variation of single values
was in the range 19.6 to 23.1 ksi in. m, or :k8 percent [5].
Interlaboratory Tests
As experience with the practice broadened, changes were made in the

r e c o m m e n d e d fatigue precracking procedure a n d in interpretation of the
load-displacement records. U s i n g the April 1967 draft of the practice, nine
laboratories tested five specimens each of four materials. These materials a n d
their mechanical properties are listed in T a b l e 1.
TABLE 1--Materials and mechanical properties.
2219-T851 18Ni Maraging 4340--500F 4340-800 F
Yield strength, ksi . . . . . . 51.2 276 238 206

Tensile strength, ksi . . . . . 66.6 285 280 222
Elongation, ~o . . . . . . . . . 10.5 8.0 10.0 11.0
Reduction of area, ~o . . . . . . 56.0 33.9 43.4
KI~, ksi in.~/~. . . . . . . . . . . 32.7 52.0 44.4 79.6
Oblique fracture, ~o . . . . . 10 to 22 12 to 20 2 to 5 l I to 16
C ........................ 0.017 0.42 0.42
Mn . . . . . . . . . . . . . . . . . . . 0.30 0.05 0.71 0.71
P ........................ 0.004 0.010 0.010
S ........................ 0.007 0.012 0.012
Si . . . . . . . . . . . . . . . . . . . . . 0.10 0.09 0.25 0.25
Ni . . . . . . . . . . . . . . . . . . . . . . . 18.8 1.77 1.77
Cr . . . . . . . . . . . . . . . . . . . . . . . . . . 0.80 0.80
Mo . . . . . . . . . . . . . . . . . . . . . . 4.95 0.23 0.23
Co . . . . . . . . . . . . . . . . . . . . . . . 7.2
Ti . . . . . . . . . . . . . . . . . . . . 0.06 0.58
A1 . . . . . . . . . . . . . . . . . . . . balance 0.13
Cu . . . . . . . . . . . . . . . . . . . . 6.3 ...
V..................... 0.10 ...
Zr . . . . . . . . . . . . . . . . . . . . 0.18 ...
24 PLANE STRAIN FRACTURE T O U G H N E S S TESTING
The test specimens were machined to the dimensions shown in Fig. 1 and
heat treated by three of nine participating laboratories. Randomly selected
specimens were supplied to each laboratory for fatigue cracking and testing.
Fatigue Cracking
Information on fatigue cracking equipment and procedures is given in
Table 2. Three types of fatigue loading fixtures were used: three-point bending,
cantilever bending, and pure bending by moments applied at the ends.
Frequencies ranged from 600 to 8000 cpm. Individual laboratories usually
did not vary stress ratio R1 in fatigue cracking the four materials. In most
cases the loading was from near zero tension to maximum tension at the
crack. Laboratories 1 and 3 used reversed bending, with R1 = -- 1.0.
The ratio R2, of maximum K~ in fatigue cracking to K~c, varied with
material as shown in Table 2, with a range of 0.16 to 0.50 for 2219-T851 and
4340-800 F. The 18Ni maraging and 4340-500 F steel specimens were cracked
at higher R2 ratios, in the range 0.25 to 1.06, with most under 0.67. Where a
single average value is given, the range for individual specimens was small.
The number of cycles generally varied for each group of specimens.
Where variation was small, a single average value is shown.
All laboratories encountered problems in meeting fatigue cracking require-
ments. These are numbered in Table 3 and appended to individual K~c values
not meeting the requirement. Values in parentheses were not included in
statistical treatment of the data because they failed to meet certain of the
following requirements.
-q
Ill Section
"( 2W + 0.2 Min. -i- 2W + 0.2 Min. Through
Notch
2219 4340 18-Hi 4340

T85__1 500..__F MAR. 800.____F
B 1.4 0.95 0.5 0.95
W 3.0 2.0 1.0 2.0
a 1.5 1.0 0.5 1.0
FIG. 1--Kzc bend test specimen with nom#lal dimensions o f the materials o f the inter-
laboratory test program.
1. Specimens were lost in fatigue cracking (about four percent of the

number tested), and since no spares were available the number of replicate
tests was reduced to three or four in some cases by failure to meet require-
ments [1].
2. and 3. To minimize plastic deformation at the crack front " . . . the
maximum load of the fatigue cycle shall be no greater than necessary to cause
the final 0.050 in. of crack extension to occur in 50,000 cycles, with a load
range not less than three fourths of the maximum toad. ''~ The load range
limitation is expressed in terms of the stress ratio R~ defined in the footnotes of
Table 2. Only one laboratory failed to meet the load range requirement [3]
and on only one material. However, 27 percent of the specimens were cracked
using less than 50,000 cycles to produce the last 0.05 in. of crack length.
Some laboratories reported only the total number of stress cycles; hence,
there may have been more who did not meet requirements [2].
Fatigue cracking is still under consideration in the committee, and new
information indicates that the 50,000 cycle requirement could be relaxed
somewhat; therefore, it was not used as a basis for rejection of K~c values in
this program.
4. " A fatigue crack is to be extended from the root of the notch for a
length of at least 0.050 in. on each side, as measured at the specimen sur-
faces. ''4 This requirement was not met in 10 percent of the specimens. Two
specimens of Laboratory 4 had cracks which did not extend beyond the
machined notch at the outside surfaces of the specimen. Data for these two
specimens (in parentheses in Table 3) were considered invalid and were not
included in further evaluations. Data for other specimens with cracks which
were slightly short but did extend to the outside surfaces were accepted for
analysis.
5. "Measurements of the crack length shall be made at the most advanced
point of the crack front and at each surface of the specimen: all three values
shall be recorded. For calculating K~ c, the sum of the two surface values plus
twice the value at the most advanced point, all divided by 4, shall be used.
. . . . . if the difference between any two readings is greater than 10 percent of
the thickness, the test is invalid. ''4 This requirement was not met by 7 percent
of the specimens. The problem was principally with the 4340 steel specimens
which tended to have the crack leading near the outside edges rather than at
the midthickness, Fig. 2. Also, some specimens had internal flaws which
resulted in an irregular crack front, as in specimen E2461 of Fig. 2. The
aluminum alloy and the maraging steel had crack fronts leading at the center
and with crack dimensions within limits.
See sections 6.5.2 and 6.5.3 of Ref 8 for revised fatigue precracking requirements.
4 See section 7.2.3 of Ref 8 for revised crack measurement requirements.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductio
Z
m
_r
Z
TABLE 2--Summary of information of fatigue cracking. c

o"1-
z
L a b o r a t o r y Laboratory Laboratory L a b o r a t o r y L a b o r a t o r y L a b o r a t o r y Laboratory L a b o r a t o r y L a b o r a t o r y ~,
1 2 3 4 5 6 7 8 9 o~
Machine . . . . . . . . . . . . . . . BJL-I, Baldwin Tuning MTS Vibraphore Vibraphore Sonntag InHouse Sonntag
IVY SF-1-U Fork SF4 Design SF-10-U O
Frequency, cpm . . . . . . . . . 3500 1800 1400 600 8000 4800 3600 1800 1800
Type loading . . . . . . . . . . . . Cantilever Cantilever Pure 3-Point 3-Point 3-Point Cantilever 3-Point 3-Point
Bending Bending Bending Bending Bending Bending
2219-T851:
Stress ratio, RI" - 1.0 0.4 - 1.0 0.08 N0 ... 0.1 N0 0.05
R~ ~ 0.16 . . . . . . 0.40 0.29 0.23 0.24 0.35 to 0.50 0.35
Cycles • 10 -3 for
last 0.05 in. 12 to 46 75 to 324 . . . . . . 83 to 111 85 to 100 56 48 to 116 38
Cycles • 10 -3 for total . . . . . . 73 to 119 150 ... 160 to 290 . . . . . . 129 to 227
Muraging 18Ni:
Stress ratio, R1 --0.6 .., -- 1.0 0.09 ~0 ... 0.1 N0 0.05
R., 0.25 .., 0.44 0.44 0.63 0.44 1.06 0.40 0.53
Cycles X 10 -a for last
0.05 in. 26 to 50 . . . . . . . . . 58 to 124 19 to 58 20 to 38 30 to 144 47
Cycles X 10-a for total 160 143 to 216 78 to 383 280 to 324 ... 39 to 120 . . . . . . 90 to 215
4340-500 F:
Stress ratio, R~ --1.0 no data -1.0 ~0 ~0 ... 0. l ,~0 0.05
R2 0.38 to 0.58 no data 0.35 ... 0.46 to 0.82 0.35 0,39 to 0.66 0.44 to 0.60 0.35
Cycles >( 10 -a for last
0.05 in. 6 to 34 no data ... 65 to 160 55 30 to 86 9 to 30 84 to 136 45
Cycles X 10 -3 for total . . . . . . 81 to 160 231 ... 115 to 150 . . . . . . 130 to 150
4340-800 F:
Stress ratio, R, --1.0 no data -1.0 0.17 N0 ... 0.1 ,-4) 0.05
R.2 0.19 no data . .. 0.22 0.4l 0.20 0.22 0.24 to 0.50 0.24
Cycles • 10 -a for last
0.05 in. 55 no data . . . . . . 72 to 98 68 to 214 84 to 136 74 to 158 50
175 to 350 . . . . . . 190 to 345 .<
Cycles X 10 -a for total . . . . . . 95 to 179 125 to 356 . ..
>
R1 = ratio of m i n i m u m to m a x i m u m load in fatigue cracking. Z
T o meet the requirement in Section 6.4.1 that the load range be not less than three fourths o f the m a x i m u m load, R~ should be
equal to or less than 0.25 and can be as low as - 1 . 0 for a reversed stress cycle.
b R~ = ratio o f KI m a x i m u m (in fatigue cracking) to K~o.
rn
O
z
>.
t"l
1,O
,,q
z
TABLE 3 - - T e s t results, Kit, ksi in.1/L 8c

O
.-l-
Laboratory Laboratory Laboratory Laboratory Laboratory Laboratory Laboratory Laboratory Laboratory z
Material 1 2 3 4 5 6 7 8 9
2219-T851 . . . . 34.6-2 (28.7-3,4,8) 30.0-4 31.6 33.2 36.8 33.3 31.6-2 35.3-2
33.4-2 34.6-3,4 34.3 32.4 34.l 36.6 32.3 33.8-2 34.9-2
(34.3-2,8) (33.4-3,4,8) 33.3 31.3 33.2 33.9 32.5 29.2 34.7-2
33.7-2 29.7-3 32.6 28.8 32.9 36.4 32.4 34.0 34.6-2
30.4-2 -1 33.6 28.1 29.2 32.5 30.0 30.0 30.9-2
Maraging 18Ni 56.8 51.5-4 48.9-4 49.4 50.0 57. l 56.9-2 43.6-6 52.6-2
52.9-2 5l .0-4 48.8-4 (42.3-4,6) 49.8-4 (53.3-2,5) 59.2-2 47.5-6 55.0-2
54.9-2 52.6-7 49.3-4 47.7-4 48.8 48.9-2,4,6 57.5-2 47.7 53.5-2
51.7 53.9-4 50.3-4 -1 50.7 55.6 54.7-2 46.9-2 51.4
54.0-2 54.0-4,7 50.8-4 -1 49.8-4 56.8-2 54.7-2 49.2-2 55.9-4
4340-500 F . . . 49.0-2 41.0 (40.3-4,5,7) 43.3 43.8 46.2-4,7 46.1-2,7 (42.8-5) 44.6
42.4-2 40.7-4 42.5-7 46.3 44,8 45.1-2,7 42.5-2,7 44.2 43.8-2
46.0-2,6 (45.3-5) (39.8-4,5,6,7) 47.6 45.8 44.4-7 42.4-2,4,7 (42.9-5) 44.l-2
45.6-2 (47.0-5) (44.l-4,5,6,7) 43.3 42.9 41.7-7 44.0-2,7 45.9 42.0-2
48.2-2,7 43.5-4 -1 (52.7-4,6) 46.0 44.0-7 41.9-2,7 -1 46.8-2
4340-800 F . . . 80.7 73.9 78.5 77.4 75.0 84.7-4 78.6-2 78.5 82.5
81.0 71.5 80.3-4 75.4-4 76.0 (79.8-5) 83.8 77.3 (68.4-2,4,5,8)
79.4 76.7-4 79.3-4 (71.4-4,5) 73.5 85.4-4 81.0 80.8 (76.8-2,5)
81.5 78.2-4 78.8-4 72.1-4 78.5 81.5 81.6 -1 85.5
-I 84.1-4 (72.4-4,5,8) 84.6 77.8 84.0-4 80.9 -1 83.3
NOTE--1. Specimen broke in fatigue cracking.

2. Less than 50,000 stress cycles for last 0.05 in. of fatigue crack.
3. Load range in fatigue cracking less than ~ maximum load.
4. Fatigue crack extended less than 0.05 in. from root of notch at surface.
5. Difference in crack front readings greater than 10 percent.
6. Crack front outside limits o f 0.45 to 0.55 times W.
7. Initial slope of load displacement record outside limits of 0.5 to 2.0.
8. Displacement ratio greater than 0.25 (see Section 7.3.2 of Method).
..<
7"
c7
O
7,
7,
c7
FIG. 2--Photographs of crack fronts in 4340 steels.
In several cases where the crack front was dimensionally out of limits the
resulting KIe value was obviously out of line. All specimens which did not
meet the crack front requirement were excluded in the data analysis.
6. "The crack length, a, shall be from 0.45 to 0.55 times depth, W." This
requirement was not met by 4 percent of the specimens. However, the require-
ment was intended primarily to optimize the specimen, and no data were dis-
carded for failure to meet it.
Bend Testing
There were relatively few problems in carrying out the bend test and inter-
preting the load-displacement record. A typical fixture design is shown in
Fig. 3. In Table 3, deviations from the following bend test requirements are
noted:
7. " T h e initial slope of the load-displacement record shall be not greater
than 2 and not less than one half. ''~ This requirement was not met by 8 per-
cent of the records. Variance outside the recommended limits was minimal
See section 7.5 of Ref 8 for revised requirement.
l FIG. 3--Bend test fixture design.
in all cases, and since the requirement was intended as a guideline to aid in
accurate determination of the displacement ratio it was not used as a basis
for rejection of data.
8. This requirement is described in the following sections from the 1967
method. 6 Figure 8 of the 1969 proposed method is reproduced as Fig. 4
in this report. Typical records for the test materials are shown in Fig. 5.
Draw the secant line OP5 through the origin with slope 5 percent less than the
slope of the tangent OA to the initial part of the record. P5 is the load at the inter-
section of the secant with the record. Next, determine the load Po, which will be
used to calculate Ko, as follows: if the load at every point on the record which
precedes P5 is lower than Ps, then Po is equal to P5 (Fig. 4, Type I); if, however,
there is a maximum load preceding P5 which exceeds it, then this maximum load is
PQ (Fig. 4, Types II and III).
Draw a horizontal line representing a constant load of 0.8 Ps. Measure the dis-
tance X1 along the horizontal line from the tangent OA to the record, and measure
the corresponding horizontal distance to Ps. If the ratio of the deviation from line-
arity at 0.8 P5 to that at P5 is greater than 0.25, then the test is not a valid K~c test
(for reasons explained in Ref. 4). Proceed to calculate KQ in order to estimate a
revised specimen size, but do not report Ke as a valid Kx,.
Failure to meet the 0.25 ratio (called displacement ratio in this report)
occurred in three aluminum alloy specimens and in two 4340-800 F speci-
mens, a total of 3 percent. In Table 3, specimens having displacement ratios
out of limits, Requirement 8, are rejected. N o t e that two 4340-800 F speci-
mens are rejected f r o m analysis for out-of-limits crack fronts, Requirement 5,
as well as for Requirement 8.
9. "Specimen size requirements: In order for a result to be considered
valid according to this recommended practice, it is required that both the
6 See sections 8.1.1 and 8.1.2 of Ref 8. The wording has been changed, primarily to
provide for the compact tension specimen.
A A
LoAD /J :
P = 0.8Ps Po
Xl
2ok
vo (o (o
DISPLACEMENT GAGE OUTPUT
NOTE: SLOPE OPs IS EXAGGERATED FOR CLARITY

F I G . 4--Principal types of load-displacement records.
specimen thickness, B, and the crack length, a, shall exceed 2.5 (K~r ~,
and where avs is the 0.2 percent offset yield strength of the material."
The following tabulation shows that the four materials all met the size
limit requirements.
2219 4340 18Ni 4340

T851 500 F Maraging 800 F
2.5 (Ki~/~rvs) ~. . . . . . . . 1.0l 0.09 0.09 0.38

B (thickness) . . . . . . . . . 1.40 0.95 0.50 0.95
a (crack length) . . . . . . 1.50 1.00 0.50 1.00
10. Fracture appearance: The method states that the fracture appearance
shall be noted and the proportion of oblique fracture measured at a location
midway between the crack tip and the unnotched edge of the specimen.
These measurements were quite consistent between laboratories, and the
ranges of all values reported are given in Table 1.
LOAD
- 0.8P5 _/~0.8p5
TSS1 / MAR,
DISPLACEMENT
FIG. 5--Typical load-displacement records.
Analysis of KIe Data

In Fig. 6, the laboratory mean values of Kic are shown as percent deviation
from the grand mean. The largest scatter about the grand mean is for the
maraging steel, approaching 10 percent. Maximum deviations on the basis
of individual test values extended to 16 percent of the grand mean. The
greater scatter between laboratories in tests of maraging steel could not be
attributed to material inhomogeneity because of the random specimen selec-
tion plan. Also relative humidity, although not controlled, did not seem to
be a likely cause. A possible explanation could be the smaller specimen size.
With small specimens it is more difficult to meet dimensional requirements
of the specimen and test setup, and errors in load measurement may be
greater.
In Fig. 7 the replicate variability of Kzo values is shown as -r-two-sigma
limits. In most cases sigma is based on five replicate values, and in no case
less than three. Where sigma was not calculated, a single line represents the
mean value.
Avg,
Percent 2219 4340 18-Ni 4340 All
T851 500 F MAR. 800 F Mats.
IO-
--7
8
--6
6
-6
-9
-b
--6
-9 -I -I
4
-9 --9
--I
2 --7
--4 --7
--8 --I
+ --I --5
-2
-3
O --6,9
--5 --3
--8
2 --7 --4 _=~3

--8 -8
--2 -4
4 --3 -5
L-5
--3
6 --2
--4 LAB. NO.
--4
8-
--8
I0 I
FIG. 6--Kzc mean values f o r nine laboratories as percent deviation from the grand means.
The greatest variability in replication occurred for the maraging steel and
the 4340-800 F steel. In the latter case it can be shown, at the 95 percent
confidence level, that Laboratories 2 and 4 had greater data scatter than the
others. A comparison of tests on the four materials on the basis of percent
of the mean value of KI~ is shown in the following tabulation:
2219 4340 18Ni 4340

T851 500 F Maraging 800 F
+2aKi. + mean Kic • 100

Maximum for one laboratory... 23.6 24.8 30.8 27.2
Average for all laboratories . . . . 22.1 16.1 14.5 11.7
K~.
Average for all laboratories . . . . 32.7 44.4 52.0 79.6
Average replication for all laboratories improves as KIo increases. The

general level of replication in these tests of sharply cracked specimens is in-
ferior to that expected in yield strength or tensile strength determinations
with unnotched specimens. While some improvement may be reasonably
expected as laboratories become more experienced in Kic testing, there are
inherent limitations to good replication.
In Fig. 6, it may be noted that the distribution of laboratory means from
the grand mean for the 18Ni maraging steel shows concentration at the
extremes. Statistical tests (Duncan's test for multiple means) can divide these
data into two significantly different populations. If the deviation of Kic
from the grand mean is averaged over all materials for each laboratory, the
same type of separation is developed. This suggests that it is not the material
but laboratory bias which gives the separation in the case of maraging steel.
Examination of fatigue cracking practices and data did not reveal any
pattern which would explain the grouping of laboratories.
In order to determine if the differences among laboratories is of sufficient
magnitude to warrant closer method control, a linear model procedure for
round robin testing, as described by Lashof [6] and in A S T M STP 335 [7],
was applied to the data. For each laboratory, linear regression lines were
determined for the individual laboratory mean Kic values for the four
materials versus the corresponding all laboratories mean KIo values. The
four materials gave four points for each line, In Fig. 8, the laboratories are
in order of increasing K~c value, average for all materials. The slopes and
goodness of fit standard deviations from the regression lines are plotted
below. Confidence limits for the slopes and means are determined from the
averaged standard deviations.
A slope outside the two-sigma limits indicates that the K~c value of the
laboratory is dependent on the magnitude of the true K~ c of the material.
No laboratory showed this tendency to any statistical significance.
Laboratory 6 is outside the two-sigma limits on the average K~ o plot but is
inside a 95 percent confidence band based on the degrees of freedom (t0.05,16 =
2.12) associated with the averaged standard deviation.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions
KIc
K s i . i n 1/2
90--
80-- 0 0 009 o 0
0 Material - 4340 800 F
70--
0
60-
SO-- Oo 0 -
18-Ni MAR.
0
0 0
0
-0 0
SO 4340 500 F
40
0 oo0o0
30-- 0 mJm
00 00o0o 2219 T851
20--
3 4 5 6
LABORATORY
F I G . 7--Two-sigma ranges of Kit values.
HEYERANDMcCABEON A BENDSPECIMEN 37
56
54 m
AverageKic- KSl In1/2, ~ +-~gp,

52
50
1.10
1.0~
Slopes
1.00 u +20"
.95
99 0 m
3.0
StandardDeviations- Ksi-in1/2
2.0
hO
I I r I I I I I I
4 8 5 2 5 7 I 9 6
LABORATORY
FIG. 8--Linear model analysis of Kz~ data,
Since all laboratories fall within the 95 percent confidence bands, the
method appears to be under control from the standpoint of consistency among
laboratories in the determination of a mean K~,.
Table 4 lists the displacement ratios for individual specimens in the sets.
The aluminum alloy and 4340--800 F, whose test records were Type I
smooth curves (see Figs. 4 and 5) had displacement ratios varyings from zero
to near the rejection limit of 0.25 within a single set of five specimens. Those
few values (aluminum alloy) which were rejected because of failure to meet
this requirement only do not appear to be outside the normal K1 ~ range of
companion specimens. It therefore appears that the 0.25 displacement ratio
requirement risks rejecting good test information. On the other hand, this
requirement, together with the other requirements of the method, will reject
all test records of materials where the KQ determined from the 5 percent secant
is not a valid K~c ,as defined in the method.
The fatigue cracking data compiled in Table 2 are not complete for all
laboratories because some of the information listed was not requested of the
participants. The following observations are made about fatigue cracking
techniques.
There was no detectable effect of the type of fatigue loading on the K1
values. The KI maximum in fatigue loading, as reflected in the R2 values,
does not correlate well with fatigue crack growth rates. The variable R1 ratios
contribute to this lack of correlation. The ratio R2 required to develop a crack
growth rate of 0.05 in. in 50,000 cycles is different for different materials.
The higher R2 ratios used for 4340-500 F and maraging steel evidently did not
adversely affect K~c values. Laboratory 7 with the high R2 ratio of 1.067 in
fatigue cracking maraging steel is the only obvious case of a high R2 ratio
affecting the KI ~ values. The grouping of laboratory means in the Duncan's
multiple means test put this laboratory outside the upper mean population
grouping for tests on maraging steel.
Some laboratories fatigue cracked the last 0.05 in. of crack in less than the
50,000 cycles specified, with no apparent increase in the K~ ~ value. This sug-
gests that the requirement could be relaxed somewhat, to a degree which is
likely to be dependent on the material. Further experimental information is
needed.
Measuring the crack at each surface and at the most advanced point may
not yield the most appropriate crack length for Kit determination. With
crack fronts like those at the bottom of Fig. 2 the important central portion
makes no contribution to the averaged value. Also if small shear lips develop
at the outside surface, the crack length measurements made at the outside
surfaces would not be representative. It is suggested that the recommended
crack length measuring techniques be re-examined.
r R.~ = 0.97 based on the mean K~c for L a b o r a t o r y 7 instead of the grand mean K~c.
HEYER A N D McCABE ON A BEND S P E C I M E N 39
>
O 9 o
>
I >
O
9 ~
< C O O O
c~
Conclusions
l. Individual laboratory mean values of K:o were within • percent of
the grand mean for all laboratories in the case of two materials, and =t=6
percent for the other two materials.
2. Replication of K ~ test results was quite consistent between laboratories
for the two materials with the lowest KI~ values. At higher K ~ levels, there
were larger absolute differences in replication between laboratories. While
the average replication for all laboratories improved on a percentage basis
as Kic increased, the replication was poorer than for tests of smooth tension
specimens, as expected.
3. Individual laboratory mean values of KI c for the maraging steel showed
separation into a high and a low group, and a similar smaller separation in
mean values of K ~ for all materials. This suggested a slight bias among
laboratories. However, application of the linear model method of statistical
analysis showed only one laboratory outside two-sigma limits, and this
laboratory was within a 95 percent confidence band using the appropriate
degrees of freedom.
4. A number of requirements of the method were not always met. The one
which had the most obvious influence on KIo results was the crack front
straightness requirement. Test data for specimens with irregular, out-of-
limits crack fronts were excluded from the statistical analyses.
5. Data from only three specimens were rejected solely for failure to meet
the displacement ratio requirement. However there was some concern over
the range of displacement ratios, from zero to over the rejection limit, within
certain sets of specimens.
6. Failure to meet the requirement of 50,000 fatigue stress cycles for the
last 0.05 in. of crack growth did not result in high or irregular K I e values.
7. In one case, where the maximum KI in fatigue loading was very high,
there was a significant increase in K I c. Need for modification of the fatigue
cracking requirements is indicated.
8. Details of the method of measuring crack length also need further con-
sideration.
9. This program was limited in the materials represented and in the speci-
men sizes selected. It would be of interest to test, in a cooperative program,
materials in specimen sizes close to the rejection limit in thickness.
Acknowledgment
The Task Group acknowledges the cooperation of the following companies
and organizations in carrying out this program: Aluminum Company of
America; Armo Steel Corp.; The Boeing Co., Space Div., Kent Facility;
The British Iron and Steel Research Assn.; Frankford Arsenal; National
University of Washington (University of Washington) pursuant to License Agreement. No further reproduct
A e r o n a u t i c s and Space A d m i n i s t r a t i o n , Lewis R e s e a r c h Center; T h o m p s o n ,

R a m o , W o o l r i d g e ; U n i t e d States Steel C o r p . ; and W e s t i n g h o u s e Electric
Corp.
M a t e r i a l s were supplied and specimens were m a c h i n e d a n d heat treated b y
A l u m i n u m C o m p a n y of A m e r i c a , N A S A - L e w i s Research Center, a n d
U n i t e d States Steel C o r p .
References
[1] Second Report of Special ASTM Committee on Fracture Testing of High-Strength
Materials, "The Slow Growth and Rapid Propagation of Cracks," Materials Research
& Standards, Vol. 1, No. 5, May 1961, p. 389.
[2] Fifth Report of Special ASTM Committee on Fracture Testing of High-Strength
Materials, "Progress in Measuring Fracture Toughness and Using Fracture Mechan-
ics," Materials Research & Standards, Vol. 4, No. 3, March 1964, p. 107.
[3] Srawley, J. E. and Brown, W. F., Jr., "Fracture Toughness Testing Methods," Fracture
Toughness Testing and Its Applications, ASTM STP 381, American Society for Testing
and Materials, 1965, pp. 133-195.
Strength Metallic Materials, A S T M STP 410, American Society for Testing and
Materials, 1967.
[5] ASTM Committee E-24 on Fracture Testing of Metals Subcommittee I, "Task Group
Report on Pilot Program for Determination of K~,," 23 June 1967.
[6] Lashof, T. W., "Ranking Laboratories and Evaluating Methods of Measurement in
Round Robin Tests," Materials Research & Standards, Vol. 4, No. 8, Aug. 1964, p. 397.
[7] Committee E-11 on Quality Control of Materials, A S T M Manual for Conducting an
lnterlaboratory Study of a Test Method, ASTM STP 335, American Society for Testing
and Materials, 1963.
[8] Proposed Method of Test for Plane-Strain Fracture Toughness of Metallic Materials,
1969 Book of ASTM Standards, Part 31, p. 1099.
M. J. May 1
British Experience with Plane Strain Fracture

Toughness (K,c) Testing
REFERENCE: May, M. J., "British Experience with Plane Strain Fracture

Toughness (KIe) Testing," Review of Developments in Plain Strain Fracture
Toughness Testing, ASTM STP 463, American Society for Testing and Materials,
1970, pp. 41-62.
ABSTRACT: This paper reviews the background to the formation of a Working

Group in the United Kingdom concerned with fracture testing of high-strength
metallic materials. The results are given of two collaborative test programs
designed to familiarize participants with testing techniques based on linear elastic
fracture mechanics and interpretation of load-displacement curves.
Results also are presented of specific studies on the influence of fatigue pre-
cracking stress intensity and the specimen dimensions such as thickness and
crack length on Ktc measurement.
The current objectives of the BISRA Fracture Toughness (High Strength
steels) Committee are to produce a British standard for plane strain fracture
toughness (Kro) testing and prove its workability by a further collaborative test
program.
KEY WORDS: fractures (materials), toughness, fracture strength, interlabora-

tory tests, evaluation, tests
In the early 1960's considerable attention was focused, in the U n i t e d

K i n g d o m , on the testing o f ultrahigh-strength steels a n d the variability
between results from different laboratories. A t the request o f the Inter-Services
M e t a l l u r g i c a l Research Council, T h e British I r o n & Steel R e s e a r c h A s s o c i a -
tion, B I S R A , n o w The I n t e r - G r o u p L a b o r a t o r i e s o f the British Steel C o r p o r a -
tion, was asked to set up a W o r k i n g G r o u p under its Engineering P r o p e r t i e s
C o m m i t t e e , to consider the p r o b l e m s o f s t a n d a r d i z a t i o n o f test m e t h o d s for
high-strength steels. The initial aim was to i m p r o v e the u n i f o r m i t y o f testing,
in p a r t i c u l a r with r e g a r d to m e a s u r e m e n t o f transverse p r o p e r t i e s a n d n o t c h
sensitivity, a n d to p r o v i d e d a t a to assist designers in selecting high-strength
steels for critical applications.
1 Head of the Mechanical Testing, Metallurgy Division, The Inter-Group Laboratories

of the British Steel Corp. (B1SRA), Sheffield, England.
42
MAY ON BRITISH EXPERIENCE 43
An informal meeting, to set up the Working Group, was held early in

t964 and the following terms of reference were agreed:
To achieve uniformity in the mechanical testing of ultrahigh-strength steel
(tensile strength greater than 100 tonf/inT) (224 ksi) between room temperature and
--40 C.
The first meeting of the Group was held in March 1964 at which 21 mem-
bers, representing different laboratories, indicated that they wished to partici-
pate in a collaborative test program. A preliminary program of notched
tension tests was proposed, involving tests of both thick and thin section,
high-strength steel in the transverse direction, and the evaluation of in-
instrumented bend tests for thin section material. A Study Group was formed
to consider the test program in more detail and to take into account the
developments in fracture mechanics in the United States. As a result of the
information presented at the ASTM Fracture Toughness Symposium in
Chicago in June 1964, and that gained by the author later in the year while
visiting several laboratories in the United States well experienced in fracture
testing, it was recommended that the present program of the Group, using
machined notches, should be replaced by one using precracked specimens
and based on linear elastic fracture mechanics. A proposal for a collaborative
test program based on these lines was presented to the Working Group in
April 1965, when it was agreed to test two high-strength steels using single
edge, notched, fatigue cracked specimens, loaded in both tension and bend-
ing and following the ASTM test recommendations at that time.
In view of the similar objectives of the BISRA Working Group and Sub-
committee I of the ASTM Committee E-24, Fracture Toughness Testing of
High Strength Metallic Materials, concerning standardization of plane strain
fracture toughness Kx~ testing, it was decided that it would be advantageous
for BISRA to be represented on the ASTM Committee E-24. In this way, it
has been possible for the BISRA Working Group to be fully aware of the
developments in the fracture toughness field and benefit from the considerable
effort in this area in the United States.
Collaborative Test Program
First Stage
The first stage of the collaborative test program was aimed at acquainting
participants, many with no previous experience, with techniques for plane
strain fracture toughness testing and necessitated machining, fatigue crack-
ing, and testing six single edge notch tension specimens from a randomized
sample of a high-strength carbon martensitic Ni-Cr-Mo-V steel. Heat treat-
ment was such as to ensure determination of KIc on specimens of ~6 in.
thick and to produce either a maximum load-failure or pronounced "pop-

in" behavior, thereby minimizing subjective judgment in interpretation of
load-displacement curves.
Specimens were rough machined by participants prior to heat treatment
as a batch and then were ground to size prior to testing. Twenty-eight labora-
tories participated, and a variety of fatigue and tension testing machines were
used together with two basic types of recording instrumentation, X - Y
plotters and ultraviolet recorders. In all cases, during testing, the displacement
was determined using a prestrained beam clip gage, and in addition to this
several laboratories used some form of acoustic pickup also to indicate
crack extension.
The program was monitored at all stages, and the results of the program
were analyzed by BISRA and discussed at a meeting of the Working Group
in July 1966.2 The KIo values quoted by participants ranged from 37 to 116
ksi v/i~.; however, on reanalysis of the test data, 90 percent of the acceptable
results were in the range 43 to 55 ksi x/i-~. Variations in test results arose from
errors in measurement, poor load displacement curves due to inadequate
instrumentation sensitivity, irregular crack fronts, or inadequate fatigue
precracking. In one laboratory or another, almost all the possible problems
associated with plane strain fracture toughness testing using single edge
notch tension specimens were encountered. However, one factor in particular
was found to have a major effect in leading to invalid test data, namely, fatigue
precracking. This arose from irregular or slant fatigue crack fronts due to
multinucleation or nonaxial loading; insufficient crack growth from the
starter notch; or the use of too high loads during fatigue precracking. The
latter was particularly prevalent, where cantilever bending was used to
produce fatigue cracks in less than 10,000 stress cycles and resulted also in the
production of shear lips on the fatigue cracked region. Typical examples of
fatigue cracks produced are shown in Fig. 1. This first stage of the collabora-
tive test program certainly emphasized to participants the necessity for
attention to detail in Kic testing and, in particular, to fatigue precracking.
Second Stage
The second stage of the collaborative test program involved participants
testing single edge notched specimens of the Ni-Cr-Mo-V steel in three- or
four-point bending, and also tension and bend tests of another high-strength
steel (1.5Si-Ni-Cr-Mo-V). However, for this program, nine specimens were
provided for each steel and type of test; three specimens in each case were
heat treated at one of three tempering temperatures, selected so as to produce
different characteristics in the load-displacement records.
2 Walker, E. F. and May, M. J., "Collaborative Fracture Toughness Test Programme
Using Single Edge Notch Tension Specimens," BISRA Report MG/E/401/66, The Inter-
Group Laboratories of the British Steel Corporation.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions auth
FIG. 1--Typical types of fatigue cracks obtained in the first stage of the collaborative
test program.
At this time, as a result of experience gained in the United States in relation

to drafting of a recommended procedure for Kic testing it was recognized that
several changes in the test technique, to that used previously, were necessary.
A status review was produced in 1966 by Brown and Srawley 3 giving current
ideas on specimen dimensions, K-calibrations, instrumentation, etc., which
considerably influenced K~o testing in general and the second stage of the
United Kingdom collaborative test program.
It is perhaps pertinent at this point to highlight several important recom-
mendations that arose from the status review 3 and ASTM E-24 Subcommittee
I discussions.
I. For plane strain fracture, both the specimen thickness and crack length
should be equal to or greater than 2.5 (Kic/o-ys) 2.
2. As "pop-in" behavior was more the exception than the rule it was
necessary to use a linear clip gage to facilitate analysis of the load-displace-
ment record.
3. Rolling contact supports or roller bearings were necessary in bend rigs
to minimize the effect of friction during loading.
4. During fatigue precracking it was necessary to ensure a minimum fatigue
crack extension from the starter notch of 0.050 in. in not less than 50,000
cycles.
5. It was necessary to obtain a symmetrical fatigue crack, and no two crack
front readings should differ by greater than I0 percent of the specimen
thickness.
6. To avoid curved crack fronts and high fatigue loads a chevron notch
starter was recommended to ensure initiation of a single fatigue crack front.
It was not possible for some of the laboratories to incorporate all these
recommendations into their test procedures immediately, and, consequently,
this precluded detailed statistical analysis of variance of the test data. As in
the first stage a large proportion of the results were invalid on the grounds of
unsatisfactory fatigue precracking, for the reasons previously indicated, but
primarily in terms of the recommendations regarding crack length (0.050 in.
minimum from the starter notch), number of cycles (50,000 minimum), and fa-
tigue crack contour (variability to be within 10 percent of specimen thickness).
No laboratory used the chevron notch starter. Analysis of the test data
showed that cam-driven cantilever loaded precracking tended to produce a
far higher proportion of unsatisfactory fatigue cracks than other types of
three- or four-point symmetrically loaded bend fixtures.
The distribution of results for the single edge notch bend specimens for
the Ni-Cr-Mo-V steel and the single edge notch bend and tension specimens
Brown, W. F., Jr., and Srawley, J. E., Plane Strain Crack Toughness Testing of High
Strength Metallic Materials', ASTM STP 410, American Society for Testing and Materials,
1967.
for the 1.5Si-Ni-Cr-Mo-V steel are shown schematically in Figs. 2-4. It was
found that generally the bend test results for the 1.5Si-Ni-Cr-Mo-V steel were
consistently higher than the corresponding tension test results. It was con-
sidered that this could have arisen from several factors, but due to the variety
test procedures used it was not possible to identify a specific reason for the
difference. Certainly, participants who used either bend test fixtures having
loading pins free to rotate or bend test fixtures and antifriction tape obtained
lower K~o values in bending than in tension. The results of this program have
been reported in more detail elsewhere? In conclusion it was considered that
the collaborative tests fulfilled their purpose in familiarizing members of the
Working Group with fracture toughness testing techniques and indicated the
aspects requiring particular care and attention in K~c testing.
Standardization of Kic Testing

Considerable attention has been given, of late, at Working Group meetings,
to discussions regarding the modifications in test procedure arising from
ASTM Committee E-24 recommendations generally and in particular the
various drafts of the recommended practice for K~o testing. Particular atten-
tion has focused on the criteria for fatigue precracking and with regard to
requirements on specimen thickness and crack length, for valid K~c measure-
ments. It was considered that these two main areas were very important
from the point of view of standardization of Kic testing, and further work
was necessary to validate these criteria for all metallic materials. Conse-
quently BISRA and several other members of the Working Group agreed to
put in hand programs to determine:
1. The influence of fatigue precracking conditions on the subsequently
measured fracture toughness.
2. The influence of specimen thickness and crack length on K~ c.
I. Influence of Fatigue Preeracking Stress on Kic Measurement

It was considered that recommendations based on a minimum extension,
as detailed in the early drafts of the ASTM Committee E-24 recommended
test procedure, in a given number of cycles were not particularly satisfactory,
and a more logical approach was to specify the maximum ratio of the stress
intensity during the fatigue precracking, Ke, in relation to the stress intensity
in the subsequent rising load fracture test, that is, to K~o. Consequently, a
program of testing was undertaken involving three martensitic high-strength
steels which were heat treated so as to produce a variety of load-displacement
characteristics. Three-point bend specimens were precracked over a range of
4 Walker, E. F. and May, M. J., "Single Edge Notch Specimens of Two High Strength
Martensitic Steels Tested in Tension and Bending," BISRA Report M G / E B / 3 3 7 / 6 7 , The
Inter-Group Laboratories of the British Steel Corporation.
4~
FREQUENCY
>,,
B-
"r
<si/i,
ZTS
T~MPERING TEMPERRTURE?C
FIG. 2--Collaborative test results for single edge notched bend specimens of the Ni-Cr-Mo-V steel.
FREQUENCY
--9-
B~
lTll 7J
6~
5~
4~
0
Z
S1ondord
m
deviot;on
-r
ksi~n
m
Z
TEMPERING TEMPERATURE.~
FIG. 3--Collaborative test results for single edge notched tension specimens o f the 1.5Si-Ni-Cr-Mo-V Steel.
,o
o
FREQUENCY
z
7-
i
4
#
5
2 C
0','l-
Z
IL ~ , '
,--I
,,,,I
T~
o
//
StQ~Q~d
ksi J-n
F I G . 4--Collaborative test results for single r~tg~oz~d~eetd specimens of the 1.5Si-Ni-Cr-Mo-V steel.
fatigue stress intensity levels, Ks, in an Amsler vibrophore at a frequency of

80 Hz (cps) and subsequently fractured. In order to ensure that any effects
observed were relatable to lower cyclic frequencies a number of specimens
were precracked at frequencies of 10 and 30 Hz. Figures 5a-c show the effect
of final fatigue stress intensity on the subsequent fracture toughness. It was
observed, as can be seen from Fig. 5, that fatigue stress intensities up to the
order of 80 to 90 percent of the subsequent fracture stress intensity did not
appear to influence the KIo value obtained. However, higher fatigue stress
intensities (nominally greater than necessary to fracture the specimen in the
subsequent rising load test) led to apparently high KQ values.
On the basis of the results obtained it was concluded 5 that recommendations
regarding fatigue cracking, namely, that the last 0.05 in. of crack extension
should take place in not less than 50,000 cycles, was conservative, and the
following new criteria were suggested:
1. Where possible, control of the precracking procedure should be by means
of limiting the maximum fatigue stress intensity, Ks, and it is considered that
this can be up to 50 percent of the subsequently measured K~ c value, when
testing and precracking are conducted at the same temperature.
2. In cases where it is impossible to determine the fatigue stress intensity
the last 0.05-in. crack extension should take place in not less than 10,000
cycles.
It was recommended also that in cases where the fatigue fracture surface
exhibited a shear lip then the fatigue crack developed was considered un-
satisfactory, regardless of all other criteria. This information has been taken
into account in subsequent drafts of recommended procedures for plane
strain fracture toughness testing.
2. Crack Length-Specimen Thickness Requirements for Valid Plane Strain

Fracture Toughness Measurements
The requirement that the specimen thickness and crack length should be
greater than 2.5(K~c/~ys)
/ 2
proposed in 1966 arising from the status review 3 was
considerably more stringent than previously adopted criteria and was based
at that time primarily on specimen size studies for maraging steels. Several
members of the Working Group were of the opinion that these criteria were
too conservative and certainly needed justification with regard to low-alloy,
high-strength steels, titanium alloys, and aluminium alloys. A size effect
test program, therefore, was initiated covering high-strength titanium and
aluminium alloys and martensitic and maraging high-strength steels, in
which specimens were successively machined from broken specimens of
Walker, E. F. and May, M. J., "The Effect of Fatigue Pre-Cracking Stress on the
Plane Strain Fracture Toughness of Several Martensitic Steels," BISRA Report MG/E/
306/68, The Inter-Group Laboratories of the British Steel Corporation.
52 PLANE STRAIN FRACTURE T O U G H N E S S TESTING
(A) 4~/4 % NI- Cr-Ha TEHPERED 400~

9O
8O
)
.. ..
~ SO
40
I I I I I I I I I i i i
0 I0 20 30 40 50 60 70 80 90 I00 I10 120
Kr k~4~
120 -(B} 11/2% S i - N I - C r - H e - V TEHPERED Z'/SeC
I10
KEY.
tO0
o FREQUENCY IOr
Standard deviation x
I~mits f~r data fre~ x FREOJJENCY ~Oc.p.s.
"~ 8O - ~laborativn test
programme ( 18 results) x 9 FREQUENCY 80c.ps.
~ 70
o x i~
X X0 9
50 [ I I l I I I I I i , .i
I0 20 30 40 50 60 70 80 90 I00 rio 120
Kf ksl
120 tiC) 1112% $i- Ni-Cr-Mo-V TEMPERED 300'C x
I10
I00
I~ 9o Standard deviation X
limits for data from "o
-~ 80
- r162 test.
~ m e (50 rcsuKs ) x
~ ~o X
6O ~7/~x o~ o.~9 9 "

I I I I I I I I I I I i
I0 ZO 30 40 50 60 70 80 90 I00 I10 120
Kf ksl ~
FIG. 5--Effect o f fatigue stress intensity on Kzc.
larger dimensions. This ensured that the influence of material variability was
minimized.
The results, to date, for a Cu-Mg-Zn aluminium alloy (DTD 5074) are
given in Fig. 6, in which the apparent fracture toughness, KQ, is plotted
against crack length, a, and specimen thickness, B. The data have been
analyzed in terms of whether or not the test and data meet all the necessary
ASTM requirements for valid K1 r measurement, and when deviations occur,
with regard to specimen thickness, crack length, both crack length and
thickness, or deviation at 80 percent PQ, these are indicated. The change in
characteristics of the load-displacement curve for different thicknesses are
shown in Fig. 7. It can be seen from Fig. 6 that there is apparently relatively
little change, if any, in the KQ value arising from deviations with regard to
crack length and specimen-thickness when compared with valid Kic values.
With 18Ni-8Co-5Mo-0.5Ti maraging steel the influence of specimen
thickness and crack length on KQ is shown in Fig. 8 and a selection of load-
displacement curves for various thicknesses in Fig. 9. It is interesting to note,
in this case, that deviation with regard to specimen thickness can lead to KQ
values up to 20 percent lower than K~c values, however, when deviations
occur with regard to either crack length or both crack length and specimen
thickness the difference is greatly reduced.
The results of a similar program with a high-strength titanium alloy Hylite
50 (4A1-4Mo-2Sn) are shown in Figs. 10 and 11, and in this case apparently
valid K~ c results can be obtained in thinner specimens which are considerably
lower than those obtained from thicker specimens. For instance, with a
specimen 0.1 in. thick a valid plane strain fracture toughness K~o value of
20 ksi x/'i~, is obtained, whereas in 0.5 in. thickness the value is between 40
to 50 ksi ~r Behavior of this type is of particular concern for in both cases
the test data met all the necessary requirements for a valid K~ o measurement,
and only on one occasion did the deviation at 80 percent PQ invalidate a test
record.
The influence of inadequate crack length or specimen thickness on KQ in
relation to K~o for a material will be dependent on the influence of crack
length and thickness on the load-displacement record and in relation to this
the influence of crack length in the determination OfPQ for a given percentage
of crack extension. From the preliminary results obtained to date there is no
justification for relaxation of the crack length and specimen thickness re-
quirements. However, it is necessary that further detailed studies be carried
out on the influence of load-displacement characteristics in relation to the
offset procedure, particularly in view of the information obtained to date
with the high-strength titanium alloy.
Considerable discussion also has centered around the mandatory require-
ments in the recommended practice for a chevron notch starter. Very few
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions a
+ VAUD KIc
9 B'<2.5 F K t l Z
LO'~J I'K,'I 2
9 B and a BOTH",=2.5 L~-y~
FK,]'
v DEVIATION AT 80~ Pe>O.2S
50
4O
3o +
at
-.- 20
I0
I 1 I I I I I I L
92 -4 .6 -8 I'0
CRACK LENGTH (,) in
60
511
4O
30 §
+ + ~t § +
er
-.." 20
I0
I [ I 1 I I I I
-2 -4 -6 -8 1.0 I-2 i-4 1-6
THICKNESS (B) in.
FIG. 6--Effect o f specimen thickness and crack length on K s for a high-strength alumi-
nium alloy, DTD 5074.
B -- 0. 375 irL a = .341 (~/= .455 I B', O.ISTin. o=.1361n a/W - , 4 5 2 B = O. IZSIna =.3511n a/W=.469
KIc =, 27.5ksi i ~ Kic = 27-6 ksiVri'n"
3:
0
Z
P,
.-I
._,~f -it"
DISPLACEMENT
FIG. 7--Typical load-displacement curves for high-strength aluminium ahoy, DTD-5074.
t,n
56 PLANESTRAIN FRACTURETOUGHNESS TESTING
1- VALID Ktc
9 B<2"5 [K~'ys~
9 Bond o BOTH< 2.5rl(~q 2
9 9 <25 I-K--'12
L~y~
v DEVIATION AT 80o~ P1>0.25
I00 -
90
+ 4-'1-
80
4-
d
"" 60
5O
40 I I I I I 1 I i
9125 .250 .375 .500 .625 .750 .875 1-00
CRACK LENGTH(o) in
I00
90
4. 4. +
80
it
~7o
il 4-
,.~60
50
4O I I I I I I I I
9125
-250 .375 .500 .625 750 -875 I.O0
THICKNESS(B) in
F I G . 8--Effect of specimen thickness and crack length on KQ for a high-strength maraging
steel, G 110.
B= :>in a = I.O25inn/W = 9 372 I. B= 0.715in a = . 4 7 3 1 n n / W = 4 7 3 B = .lOOin a,, .195ina/W = . 3 9 0 1
Kic= 63,5 ksi I~r~- Klc= 83.3 ksi~" Kic= 79.7 ksi i ~
first acoustic
4~PL 9' / Pa : 12.42 tons

first acoustic
9 Appropriate secant ~=2~3~ns
offset used .eP~
o ,d~..,Po := :328 tens
o
z
c~
-r
OlSPLACEHENT
FIG. 9--Typical load-displacement curves for maraging steel, G 110.
,,q
-I-- VALID Kic
9 B<Z.5 [~-s~ 2
9 B rand a BOTH< 2.5 I~'~,l F_K,I

L-a C..I
9 a<2-5 ~-.~.~l2
v DEVIATION AT BO% Po. :> O- 25
7O
60
9 4-
5O + 4-
++ +
4O
++
30
20 %
I0 I I I I I I I ~ 1 i
Jr
0 O.Z 0.4 0.6 0.8 1.3 I-4
CRACK LENGTH (Q) in
70
60
++§ § +
50 4"
40 O"1-
o:l:
o" 4- 4-
:30 .I-+
.4-
20 4-
V'l"
I0 I I I I I I I I I I ,~ I
0 02 O '4 0-6 0-8 I'0 I-4
THICKNESS (B) in.
F I G . l(~-Effect of specimen thickness and crack length on KQ for high-strength titanium
alloy, Hylite 50.
b-l'OO|n e ='6001. q~cw="457 0 " 2 % Proof" Stress - 147 ksi
Koe: 51"1 ksi ,/'~
Ib=O'495in 9 ='39illn %--.531

Koc= 4 | ' 5 ksi , ( ~
pQ= 3"26 tens
B:O'IOIh a :'2S5in ~ v : "510

Kic:: 22"7 ksi,(~"
PQ
DISPLACEMENT
\
FIG. 1 l--Typical load-displacement curves for titanium alloy, Hylite 50. '<3
laboratories have utilized this type of starter notch in the United Kingdom,
although its benefits have been recognized with extremely brittle materials
such as tool steels. In general, experience has been that it offers no particular
advantage over a flat bottomed notch and is considerably more expensive to
produce. For the majority of metallic materials no difficulty has been ex-
perienced in obtaining fatigue cracks to the necessary requirement that no
two readings along the crack front should differ by greater than 10 percent
of the specimen thickness. Figure 12 shows the fatigue crack contours of a
series of single edge notch bend specimens of a 350 Grade maraging steel
machined, having a variety of crack starter notches. From Fig. 12 it can be
seeri readily that there is relatively little difference in the degree of straight-
ness for the fatigue cracks front produced from the different forms of starter
notch. The linearity of the crack fronts for all the various forms of crack
starter notch were between 4 and 5 ~ percent of the specimen thickness.
FIG. 12--Fracture surfaces Jbr a series o f single edge notch bend specimens showing a
variety o f crack starter notches.
Draft British Standard

Since completing the collaborative test program the majority of members
of the Working Group have carried out specific test programs on a variety
of materials and gained considerable experience in plane strain fracture
toughness testing adhering, in the main, to the ASTM test recommendations.
However, as yet, experience in the United Kingdom in testing of compact
tension specimens is limited, although the results of a recent questionnaire
indicate that twelve members will soon be in a position to carry out such
tests.
Recently the status of the Working Group was changed to the Fracture
Toughness (High Strength Steels) Committee and attention is now to be
focused on the drafting of a British Standard for Plane Strain Fracture Tough-
ness (K~~) Testing. It is aimed to complete the drafting of a British Standard
covering both bend and tension specimens by the end of 1968.
During 1969 a further collaborative test program will be carried out to
prove the workability of the draft Standard for Plane Strain Fracture Tough-
ness of high-strength aluminium, titanium, and st~ee!s. It is anticipated that
any differences between the British Standard and the ASTM Recommended
Practice will be of an editorial nature, and tile aim is to minimize differences
in the operation of the two procedures.
Summary
Over the four years since its conception the BISRA Fracture Toughness
(High Strength Steels) Committee (formerly High Strength Steels Working
Group) has held twelve meetings and provided a forum for discussion of, in
particular, fracture toughness testing, but also with regard to general aspects
of fracture mechanics. Through collaboration it has been possible for a large
number of organizations and laboratories to gain experience in plane strain
fracture toughness, Kro, testing. By close collaboration with the ASTM
Committee E-24 on Fracture Toughness Testing of Metallic Materials, it
has been possible to benefit greatly from the experience in the United States
in this field and more recently to contribute information regarding various
aspects of fracture work in the United Kingdom. The author is particularly
indebted to members of the ASTM Committee E-24, for the help and en-
couragement received which has enabled testing expertise based on linear
elastic fracture mechanics to be developed in the United Kingdom over a
relatively short period.
STP463-EB/Sep. 1970
DISCUSSION
J. M. Krafft ~ (written discussion)--The extraordinary drop in K e values

for the smaller, and thus thinner specimens of high-strength titanium, Fig.
10, is indeed puzzling. Is it possible that this alloy is susceptible to subcritical
crack growth, such as stress corrosion cracking, in room air moisture? If so
it would be possible that a given subcritical growth rate would have a larger
relative effect on the crack opening displacement of the smaller specimens,
and thus depress their indication of KQ. This could be checked with a sus-
tained load in one of the larger specimens, in room air at a K I level of say
20 or 30 ksi %/~., representative of the lower values of Fig. 10.
M. J. May (author's closure) and E. F. WalkerS--We would agree that the

reason for the rapid decrease in apparent valid KIr values for the high-
strength titanium alloy is of concern and difficult to explain. As stated the
influence of a given increment of crack extension due to stress corrosion
would result in a larger displacement in the smaller specimen due to a greater
influence on a~ W. However, in order to account for the difference between
specimens 0.1 and 0.5 in. thick, having Kic values of 20 ksi X / ~ . and 40 to
50 ksi %/~. in these terms would necessitate considerable crack extension.
No evidence of stress corrosion cracking has been observed in this material
under normal testing conditions. Even if one takes the fracture load of the
smaller specimen and corrects for prior crack extension the maximum ap-
plied stress intensity is still less than the critical stress intensity, KIo, deter-
mined from larger specimens.
This type of behavior also has been observed in other high-strength
materials and is thought to arise from test situations where there is an abrupt
change from essentially elastic to largely plastic behavior, that is, a sudden
loss of plane strain conditions due to through thickness yielding. This is
highlighted by a marked difference in the characteristics of the load-dis-
placement records with change in specimen size.
1 Naval Research Laboratory, Washington, D.C. 20390.
2 The Inter-Group Laboratories of the British Steel Corp. (BISRA), Sheffield, England.
62
M . H . Jones 1 and W. F. B r o w n , Jr. ~
The Influence of Crack Length and Thickness

in Plane Strain Fracture Toughness Tests
REFERENCE: Jones, M. H. and Brown, W. F., Jr., "The Influence of Crack

Length and Thickness in Plane Strain Fracture Toughness Tests," Review o f Develop-
ments in Plane Strain Fracture Toughness Testing, A S T M STP 463 American
ABSTRACT: Results are presented from a systematic investigation of the in-

fluence of crack length and specimen thickness on the fracture properties of 4340
steel bend specimens heat treated to yield strength levels between 180 and 213 ksi.
It is shown that Kx~ values determined in accordance with the present ASTM
Tenative Method of Test for Plane Strain Fracture Toughness of Metallic Ma-
terials (E 399-70T) can vary moderately within the specimen size and geometric
limitations imposed by the Test Method. The magnitude of these variations will
depend on the material properties and would be increased by relaxation of the size
requirements.
The possibility of employing subsized specimens for screening materials
regarding their plane strain fracture toughness was explored as well as several
methods for relating Kxc values to uniaxial tensile data. The results indicated
that use of subsized specimens with the ASTM Committee E-24 Test Method
does not constitute a useful screening procedure. For steels subject to single
aging or tempering reactions, it appears that the relation K~c ~ ~r~cs-m may be
useful for estimating K~c values. However, on the basis of present knowledge, it
is not possible to calculate K~ with useful accuracy from uniaxial tensile data alone.
KEY WORDS: fracture touglmess, fractures (materials), plane strain fracture

toughness, steels, evaluation, tests.
Nomenclature
K Stress i n t e n s i t y f a c t o r a s s o c i a t e d w i t h a c r a c k in a l i n e a r
elastic b o d y , c r a c k g r o w t h r e s i s t a n c e K = Y 6 M a l / 2 / B W 2
KI O p e n i n g m o d e ( M o d e I) stress i n t e n s i t y f a c t o r
KI e P l a n e strain f r a c t u r e t o u g h n e s s as d e f i n e d in A S T M E 3 9 9 - 7 0
T, 4.2
x Strength of Materials Branch and chief, Strength of Materials Branch, National Aero-
nautics and Space Administration--Lewis Research Center, Cleveland, Ohio 44135.
Mr. Brown is a personal member ASTM.
63
Copyright 9 1970 by ASTM International www.astm.org
KQ Conditional value of Kic as defined in ASTM E 399-70 T,

8.1 or the K value calculated for any bend specimen on the
basis of the initial crack length and the load corresponding to
an effective crack extension of 2 percent
Ka Stress intensity factor calculated on the basis of the initial
crack length and the load corresponding to detection of con-
tinuous crack sounds
gmax Maximum stress intensity in fatigue cycle
AK Stress intensity range in fatigue cycle
O" True stress in a tension test
ffYS 0.2 percent yield strength in a tension test
E Young's modulus
M Bending moment applied to specimen
P,Po,P1 Applied loads
Ps Secant intercept load as defined in ASTM E 399-70 T, 8.1.1
V,I~o Crack mouth displacement
B Specimen thickness
W Specimen width (beam depth)
r Initial crack length
a,al, a2 Effective crack length
,~a Effective crack extension
do Initial diameter of cylindrical tension specimen
ad Diametral change
l, lo,ll Scale distances
el t Total longitudinal engineering strain in a tension test
ele Elastic component of el t
elp Plastic component of e l t
e3t = e2t Total radial engineering strain in a tension test of a cylindrical
specimen
n Strain hardening exponent or strain hardening index
ezp Plastic component of e~ t
Longitudinal true (natural) strain in a tension test
~lp Plastic component of
~lf True strain at fracture in a tension test
Y Dimensionless linear elastic stress intensity factor K I B W 2 /
6Ma I/~, function of a~ W
F,F(a/W) Dimensionless linear elastic displacement factor EvB/P,
function of a~ W
The recently issued Tentative Method of Test for Plane Strain Fracture
Toughness of Metallic Materials (ASTM Designation: E 399-70 T) 2 in-
JONES AND BROWN ON CRACKLENGTHAND THICKNESS 65
A S
a2
a1
y/ / I-ouch ~,oy
/ ///B,-ittl~ I'"-~
a,,o I'' ;~
t
/ P/2 P/2
0
Displacement, v
FIG. l--Schematic load-displacement curves for two specimens of different crack lengths
with equal thickness and a / W illustrating effect of crack length on Kzc for a tough and a
brittle alloy.
corporates two basic features: specimen size requirements which arise from
the necessity of restricting the application of linear elastic fracture mechanics
to situations of plane strain and small scale yielding [1 ]3 and an arbitrary
procedure for selecting the load [2], which arises from the absence of a
clearly defined abrupt onset of fracture. It is not recognized generally that
these features can result in a variation of valid Kic values due to variations in
specimen size permitted by the ASTM Committee E-24 Test Method.
This geometric influence may be illustrated by Fig. 1 which shows sche-
matic load displacement curves for valid tests on a "tough" and a "brittle"
alloy using two specimens having the same thickness and different crack
lengths but otherwise of standard proportions. For these conditions K~
would be proportional to P/x/a, and this ratio is used as the ordinate of Fig.
1. The secant lines OS represent a 5 percent change in the slope of the linear
portion OA o f the records. This change in slope corresponds approximately
to the 2 percent effective extension of the initial crack specified in the ASTM
Committee E-24 Test Method. For the tough alloy, the intersection of this
line with the test record establishes the Ps/x/a, value used to calculate K~c.
For the brittle alloy, the maximum loads occur between OA and OS and,
therefore, are used to calculate Krc. Note that the tough alloy shows crack
extension under rising load beyond the points of intersection of the secant
line with the load displacement record. In these circumstances the curves for
3The italic numbers in brackets refer to the list of references appended to this paper.
P x / a versus v would have to be identical in the range of small crack exten-

sion i f K ~ is to be independent of the initial crack length. This would require
,5 (a/W) to be independent of the initial crack length. This behavior is most
unlikely and, as will be shown later, the absolute effective crack extension ,sa
is nearly independent of the initial crack length. Consequently, at the same
P/x./a value, the specimen of the tough alloy with the longer initial crack a2
will represent a smaller ,5 (a/W), or conversely the same percentage effective
crack extension in the specimen with the longer initial crack al will require a
higher load. Thus, the measured K~ o values can increase with the initial crack
length of the specimen. The magnitude of this effect will depend on the shape
of the load-displacement record as controlled by both the thickness of the
specimen and the material properties. For the brittle alloy where there is a
maximum load between 0.4 and OS the effect of crack length on K~c will be
negligible.
The geometric effects described can produce under certain conditions an
apparent constancy of K~ o over a range of crack lengths and specimen thick-
nesses outside those established by the size requirements and thus lead to the
erroneous conclusion that the size requirements can be relaxed. Furthermore,
geom,.tic effects on the measured toughness values can be particularly im-
portant ,hen there are large deviations from the size requirements or pro-
portions of the ASTM recommended specimens, such as might be encountered
in use of subsized specimens for "screening" tests.
The purpose of the present investigation was to determine the magnitude
of these geometric effects that could be encountered in plane strain fracture
toughness tests on a commonly used high-strength steel. Systematic tests
were made to determine the influence of crack length and thickness in bend
tests of a SAE 4340 steel. Most of these tests were carried out on specimens
of standard proportions (that is, 0.45 < a/W < 0.55 and 2 < W/B < 4). In
addition, a limited number of tests was made on specimens having a/W and
W/B ratios outside the ranges permitted by the ASTM Test Method. Com-
parison of the 4340 data with that from another investigation [3] of maraging
steel provided some indication of the difficulties which could be encountered
when using subsized specimens for rating materials regarding their plane
strain fracture toughness. The results of this investigation were also useful in
a study of the relation between tensile properties and K~c proposed by Hahn
and Rosenfield [4].
Material and Specimen Preparation

The material A1SI E4340 aircraft quality steel was received in the form of
l-in.-thick hot-rolled and annealed plates (36 by 60 in.) having the composi-
tion given in Table 1.
JONES AND BROWN ON CRACK LENGTH AND THICKNESS 67
T A B L E 1--Compos#ion of AISl E4340 steel'z, composition, weight percent b.
Heat No. C Mo P S Si Ni Cr Mo
Lukens BO 315. 0.42 0.71 0.010 0.012 0.25 1.77 0.80 0.23
" Mill annealed I550 to 1600 F, 1 h furnace cooled to below 800 F.

b Furnished by supplier.
Bend Specimens
Full thickness bend specimen blanks were cut from one plate (R W direc-
tion) with the positions being randomized in respect to the plate width and
the required length? Where possible these blanks were machined to the
standard chevron notch and integral knife edge configuration given in the
ASTM Committee E-24 Test Method. Where the crack length to thickness
ratio was insufficient to permit use of the chevron notch, the slot was ter-
minated in a straight across 60 deg V-notch with a 0.0005 in. maximum root
radius. Specimens with this slot configuration also had integral knife edges.
Specimen blanks with crack starter slots were heat treated in a neutral salt
bath (1550 F, ~ h, OQ + temper, 1 h), and then fatigue cracked in cantilever
bending with Kmax 5 not exceeding 27 ksi v / i n . and AK/Km~ of about 0.95.
The number of cycles necessary to extend the fatigue crack beyond the tip
of the crack starter was in all cases greater than 50,000. Specimens less than
one inch thick were cut from the full thickness fatigue cracked blanks using a
power hacksaw, and then ground on their side surfaces. Any effect of speci-
men location with respect to the center of the plate thickness was within the
scatter of data.
Smooth Specimens
Smooth tension specimens with a 0.3-in.-diameter test section and 2-in.
gage length were cut from the broken halves of ~-in.-thick bend specimens
with the longitudinal axis in the rolling direction.
Test Procedure
The geometrical conditions investigated using bend specimens are sum-
marized in Table 2. The basic test program was confined to specimens
tempered at 750 F. All of these specimens met the size requirements in terms
of crack length and had an a / W = 0.5. An additional series of tests were
run for a 925 F temper in order to investigate conditions in which both
4 An analysis of the fracture toughness data did not reveal any effect of specimen loca-
tion within the area of the plate used.
Estimated from three-point bend specimen K-calibration.
thickness and crack length differed greatly from the requirements for a valid
fracture toughness test. A limited number of tests was made on specimens
tempered at 600 and 850 F in order to establish a relation between K~c and
tempering temperature for this plate of 4340 steel.
T A B L E 2--Geometrical conditions investigated, all tests R W direction.
Temper
Temperature,
deg F ~rys, ksi ao, m. B, in. W, in. a/W W/B
600 . . . . . . . . . 232 1.0 1.0 2.0 0.5 2

750 . . . . . . . . . 213 0.27 0.05 .55 0.5 11 to
to 0.50 1.1
0.50 0.055 1.0 0.5 18 to
to 0.98 1.0
1.1 0.055 2.4 0.5 44 to
to 1.0 2.4
850 . . . . . . . . . 198 1.1 1.1 2.2 0.5 2
925 . . . . . . . . . 182 0.33 1.0 2.0 0.17 2
to 1.5 to 0.75
0.30 0.1 2.0 0.15 20
to 1.4 to 0.7
Bend specimens were tested in accordance with the procedures outlined in

the ASTM Committee E-24 Test Method. Those having width-to-thickness
ratios of ten or larger were provided with lateral support at the span rollers
by clamping flat pieces of metal with sheet Teflon separators to the specimen
at this location. Load-deflection records were produced by an autographic
recorder being fed by the output of the load cell and the clip-in displacement
gage. For some tests the specimen was monitored also with an acoustic
pickup whose output could be correlated with the load [5].
Load-strain curves were obtained for the smooth specimens using a con-
ventional ~-in.-gage length extensometer in conjunction with the load-strain
recorder of the tensile machine. This information was supplemented by
micrometer measurements of the minimum specimen diameter at suitable
load increments up to fracture. These measurements overlapped the extensom-
eter readings.
Analysis of Data
Bend Specimens
It is helpful in understanding the effects of the geometrical variables if
the load displacement records are converted to K versus Aa curves or crack
v = Crack edge displacement
FIG. 2--Schematic autographic load-displacement plot for bend specimen illustrating

quantities measured for calculation o f K versus /xa curves.
growth resistance curves where Aa is the effective crack growth increment 6

from the initial crack length a0. This conversion was made for the specimen
series tempered at 750 F. The procedure may be illustrated by reference to
Fig. 2, which shows an autographic record with the vertical scale converted to
load. The clip gage output is displayed on the horizontal scale and can be
converted to displacement through a calibration factor derived from the
slope of the linear (elastic) portion of the record and a0 (measured after the
test). The relation between displacement, load, and relative crack length
a/IF is given usually in the form of a dimensionless displacement factor [1 ]:
vEB
- f .......................... (1)
P
where:
B = thickness,
E = elastic modulus,
F = function of a/W,
6 The actual crack growth increment will be generally somewhat less than the effective
value determined from an analysis of the load-displacement records. This difference
arises from the contribution of crack tip plastic flow to the measured displacements.
Also the actual crack front is curved slightly outward in the direction of crack propagation.
Differences between effective and actual crack extensions arising from the latter source are
not significant to the type of analysis made in this investigation.
and the other terms as defined in Figs. 1 and 2. An equation for F(a/W) is
given in Appendix I.
The calibration factor for displacement is then
Vo F(ao/W) Po
C- lo- EB Io . . . . . . . . . . . . . . . . . . . . (2)
where v0 is the displacement corresponding to the scale distance 10 at load

P0 on the linear portion of the record. Then at another load PI, vl = liC,
where/1 is the scale distance corresponding to P1. Therefore
vlEB llCEB
P1 P1
Substituting from Eq 2 for C gives
vlEB 11 Po ll
F(ao/W) - a ................. (3)
P1 P1 lo P1
The conversion coefficient a is dependent only on the slope of the linear
portion of the record and F(ao/W). The value of ~a = (a - a0), correspond-
ing to a given point on the load displacement curve may then be found with
the aid of Eq ! for which a polynominal form is given in Appendix I. As
described in Appendix II, the uncertainties of this analysis make the initial
portion of the crack growth resistance curve impossible to determine with
useful accuracy. However, these uncertainties are not troublesome in respect
to establishing the general effects of crack length and thickness on the mea-
sured toughness values.
The K values were computed using the following expression for three-
point bending with a span to width ratio of four:
6Ma~/~
K ~ y - -
BW 2
where:
Y = function of a/W, and
M = applied bending moment.
In standard Kic tests the a value would be the initial crack length. However,
for this analysis the a values were calculated as described above, and there-
fore the computed values of K reflect changes in crack length as well as load.
The autographic plots were reduced to crack growth resistance curves
point by point with the points distributed to best establish the curves in the
region of small crack extensions. For the thickest specimens the load-dis-
placement curves sometimes were characterized by several large steps, Fig. 3,
/ . 0sorTnt 00nts
FIG. 3--Stepped type autographic load-displacement plot showing measurement points

for calculation of K versus Aa curves.
as contrasted with the relatively smooth behavior shown in Fig. 2. These

steps are caused by abrupt advances of the crack accompanied by unloading.
During unloading the rate of change of load and displacement exceeded the
response capability of the autographic recorder. Consequently the unloading
trace has no meaning, its slope being only a function of the relative response
speeds of the X and Y axes pens. Therefore, as indicated in Fig. 3, these types
of records were analyzed by considering only the maximum load points or
those on the rising load portion of the record.
Smooth Specimens
True stress and natural longitudinal plastic strain values were obtained by
reducing the load-strain plots and micrometer measurements of diameters by
the method outlined in Appendix III. Where the extensometer and diameter
change measurements overlapped, the longitudinal strain values derived
from each agreed within about 2 percent. The true stress and natural plastic
strain data were plotted on linear coordinates and on log-log coordinates in
an attempt to define a strain hardening index which would be useful in
checking proposed correlations between KIc and uniaxial tensile properties.
As will be discussed later, the necking strain was obtained by graphical
differentiation of the true stress-natural strain curves.
72 PLANESTRAIN FRACTURE TOI~JGHNESS TESlING
Results for Effects of Thickness and Crack Length

This section will present results from two series of tests: (I) a series with
specimens tempered at 750 F having various thicknesses, and crack lengths
meeting the present size requirements and an a / W ~ 0.5, and (2) a series
with specimens tempered at 925 F having crack lengths and thickness sub-
stantially less than those corresponding to the size requirements.
Crack Growth Resistance Curves for 750 F Temper Series

Typical K versus Aa curves derived from the load-displacement records
are shown in Figs. 4 to 6 for specimens of three different thicknesses. F o r
the crack lengths investigated, the crack growth resistance curves are identical
within the limits of scatter as represented by the bands shown in these figures.
As mentioned in the previous section, the Aa values lack useful precision in
the region near the origin, as indicated by the dashed lines in Figs. 4 to 6.
The inset in each figure characterizes the type of load displacement plot
obtained for the particular specimen thickness tested. The vertical lines
represent Aa values corresponding to 2 percent crack extension for each of
the crack lengths investigated, and the intersection of these lines with the K
versus Aa curves give the corresponding Ko values. 7
It is evident from Figs. 4 to 6 that the KQ value increases with crack length
and that the magnitude of this effect depends on the shape of the crack growth
resistance curve as influenced by specimen thickness. The thickest specimens,
Fig. 4, exhibit the flattest resistance curves and the smallest effects of crack
length. Reducing the specimen thickness results in curves with increasing
slopes and larger effects of crack length on Ko. For the thinnest specimens,
Fig. 6, these effects represent a spread in KQ of nearly 50 percent. This in-
fluence of thickness on the crack length effect might be expected because,
as thickness is reduced, transverse constraints are relieved by crack tip plastic
flow. This results in increased resistance to crack propagation and larger
amounts of stable crack extension. These effects are reflected in the load-
displacement records (see inserts in Figs. 4 to 6) as increased deviations from
linearity before maximum load and in the crack growth resistance curves as
an increase in slope.
Some of the consequences of the interaction between the crack length and
thickness effects on KQ can be seen from Fig. 7, which combines the average
curves for the scatterbands shown in the preceding three figures. If the dif-
ferent thickness specimens had crack lengths of 0.5 in., 2 percent effective
crack extension would represent a Aa = 0.01 in., and essentially no effect of
7 In an actual fracture t o u g h n e s s test the Ko value will be based on the initial crack
length a n d c o n s e q u e n t l y would be lower t h a n that obtained f r o m the K versus Aa curves.
This difference will increase with the absolute crack extension a n d would a m o u n t to a
m a x i m u m o f a b o u t 4 percent for the conditions represented in Figs, 4 to 6.
JONES AND BROWN ON CRACK LENGTHAND THICKNESS 73
g0
~, in. ao, in. B = I in.
4~] P/ /
200 . Ol .02 .~03

A~ in.
F I G . 4--Crack gngwth resistance curve for 4340 steel tempered 750 F, 1 h ( a r s = 213 ksi)
as determined using 1-in.-thick bend specimens.
thickness on KQ would be observed. On the other hand, if the crack length was
1 in., 2 percent effective crack extension would represent a Aa = 0.02 in.,
and Ko would increase as the thickness decreased. The opposite effect of
thickness on KQ would be observed for a crack length of 0.27 in. These effects
of geometry on KQ are quite large for the conditions investigated, where
there are substantial deviations from the specimen size requirements and
proportions recommended in the ASTM Test Method. An estimate of the
magnitude of the geometric effects for valid K~c tests may be made by ex-
amining the KQ values obtained for the entire range of specimen thicknesses
tested.
KQ Valuesfor 750 F Temper Series

The KQ values obtained in accordance with the ASTM Committee E-24
Test Method (based on initial crack lengths) are shown in Figs. 8 to 10. These
data represent a wide range of thicknesses for specimens having initial crack
lengths of 0.27, 0.5, and 1.1 in. Those tests failing to satisfy the ASTM Com-
mittee E-24 Test Method check procedure for at least one percent actual
90
W, in. ao, in. B = O. 27 in.
O. 55 O. 27 I I
80 1.O0 .5
70
.//////.'/ u.)#~= 2 percent ~-~= perce

t
~Xa ~ 2 percent
~" 5O
/
P
40
3O
v
.01 .02 .03

~, in.
FIG. 5--Crack growth resistance curve for 4340 steel tempered 750 F, 1 h (errs = 213 ksi)
as determined using 0.27-in.-thick bend specimens.
crack extension at KQ are designated "excess plasticity." The thickness cor-

responding to the size requirement B --- 2.5 (K~o/avs) 2 = 0.27 in. is marked
on each figure. This thickness value is based on a K~ r = 70 ksi V / ~ .
F o r the shortest crack length, Fig. 8, the trend of KQ is downward below a
thickness which is somewhat smaller than that established by the size require-
ment. At the intermediate crack length, Fig. 9, the KQ values are nearly
independent of thickness over the entire range investigated, and a size re-
quirement appears unnecessary. In contrast, for the longest crack length,
Fig. 10, the KQ values exhibit an upward trend at a thickness somewhat
larger than that established by the size requirement. It is interesting to note,
Fig. 8, that one test which met the size requirements would be rejected by the
check procedure for excess plasticity.
The permissible range in thickness for bend specimens is restricted to
1 < W / B < 4 by the ASTM Committee E-24 Test Method. The effects of
this limitation on the variation in valid K~c values produced by changes in
crack length and thickness within the size requirements is illustrated in Fig.
11. This representation shows the trend curves from Figs. 8 to 10 which are
marked to indicate the W / B limits. The variation in valid K~ c values is about
12 percent of the lowest value.
Another source of variation in valid K~c values arises from the range in
a / W permitted by the A S T M Committee E-24 Test Method. The 5 percent
secant slope of the present method corresponds to an effective crack exten-
sion Aa/ao = 1.8 percent at an a / W of 0.5. For the permitted range of a / W
between 0.45 and 0.55 the corresponding variation in ?xa/ao will be from 2.1
to 1.6 percent. The resulting variation in K~c depends on the slope of the
K versus ~a/ao curve. An example is given in Fig. 12 which shows a plot of
K versus Aa/ao typical for the short crack length specimens having a thickness
of 0.27 in. The vertical lines m a r k the permitted range of a / W and for the
case selected the variation in K~o is about 4.5 percent of the lowest value.
~, in. ao, in.

- 0.55 0.27
5 4 - - B = 0.05 in.
2.2 111
1.1 2percent
60
50 ~ f ~ o . ~ ' 2 P ercent
0 .01 .02 .03

~, in.
FIG. 6--Crack growth resistance curve for 4340 steel tempered 750 F, I h (av~ = 213 ksi)
as determined using O.05-in.-thick bend specimens.
i00
B = 0.05 in. /
0.27 < %_< 1. 1 in.--, /
90 /
B = O.Z7in. /
O. 27_<ao _<1.1 i n . - - x ~
80 ~__...-.-
7O
/ rce, t
--: z percent L B = l O i n
~ 60
9 0.5_< ao_< 1.1 in.
r/ 0~ = rcent
50
/
!
40
300~ .01 .02 .03

Z~, in.
F I G . 7--Composite o f crack growth resistance curves for 4340 steel tempered 750 F, 1 h
(ava = 213 ksi) as determined using bend specimens of three thicknesses.
KQ Values for 925 F Temper Series

While only moderate effects of crack length and thickness were observed
for valid K~c tests, very much larger effects can be encountered when using
undersized specimens to test alloys with high ratios of (K~ c/ays) 2. An example
is the 4340 steel tempered at 925 F (182 ksi yield strength) which has a
(Kt,/ays) 2 of about 0.34 as compared with about 0.11 for the 750 F temper.
The K e values for the 925 F temper are shown in Figs. 13 and 14 as a function
of crack length for thicknesses of ! in. and 0.10 in., respectively. 8 These data
are supplemented with Ka values which have been calculated on the basis of
loads corresponding to detection of continuous crack sounds.
The 1-in.-thick specimens, Fig. 13, met the size requirements at crack
lengths above about 0.85 in., and in this range KQ = K~c was essentially
constant as might be expected9 For crack lengths below about 0.5 the K~
values decrease rapidly and at the shortest crack length K, > Ko, indicating
all of the displacement at the load corresponding to KQ was due to plasticity
8 T h e secant slopes were based on the actual a / W values of these specimens.
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authori
70 ao = 0.27 in.
VV = O. 55 im
65 _ 9 ],,,
60
K 2
9 2 5f :el
55
0 Excess plasticity
5( I I ! ! I I I I
.1 .2 .3 .4 .5 .6 .7 .8
B, in.
FIG. 8--KQ values as a function o f specimen thickness for 4340 steel tempered 750 F, 1 h
Orrs = 213 ksi) as determined using bend specimens with 0.27-in. crack length.
at the crack tip. Those specimens with crack lengths shorter than that given
by the size requirement did not pass the check procedures for excess plasticity.
However, two tests with crack lengths exceeding the size requirement also
did not pass this test. The several vertical lines shown on Fig. 13 represent
the KQ values which would be measured if the size requirements were relaxed.
For example, relaxing the crack length requirement to 1.0 (KIe/O-ys) 2 and
dropping the check procedure for excess plasticity would give a "K:o" value
about 20 percent lower than the presently valid value. Steigerwald 9 has sug-
gested the error in K i e due to relaxing the crack length requirement to a =
1.0(KI ~/avs) ~ would be less than 5 percent.
The data for the 0.1-in.-thick specimens, Fig. 14, involve extreme deviations
from the size requirements and show a strong effect of crack length on the
KQ values over the entire range of crack length investigated. Note that the
KQ value for sufficiently long cracks would exceed the KI r value for this alloy.
For crack lengths below those given by the size requirement Ka > KQ in-
dicating the KQ values are not related to the fracture properties of the speci-
men. Specimens with the longest crack lengths passed the check procedure
for excess plasticity, even though the thickness is only about an eighth of
that required for a valid K: r test.
Screening Tests for KIr

It would be advantageous if specimens considerably smaller than those
required for a valid Kic test could be used to rate materials regarding their
9 See p. 102.
80-
A Excess plasticity a o : 0.5 in.
~ : 1.0 in.
75-
-t ,1.
c2r
65 2
' \ O y s ,]
60 I I I I I /~ I_ I I
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 '.0
B, in.
F I G . 9 - - K Q values as a function o f specimen thickness for 4340 steel tempered 750 F,
1 h (ceys = 213 ksi) as determined using bend specimens with 0.5-in. crack length.
80--
a o : 1.1 in.
W : 2.2in.
75
-Ls
m_
1~ 7o I
25(K,c 2
2 \0ys/
65
[] Excess plasticity
60-- I I r I t I I I I I I
0 .1 .2 .3.6 .7.4 .8 .5
.9 1.0 1.1
B, in.
F I G . I O - - K Q values as a function o f specOnen thickness for 4340 steel tempered 750 F,
1 h (aYs = 213 ksi) as determined using bend specimens with 1.1-in. crack length.
relative plane strain fracture toughness. A reduction in specimen size as-

sumes particular importance in alloy development when the (Kic/~ys)2
ratios are high. A number of possible screening methods have been proposed
including the use of precracked Charpy specimens and those based on cor-
relation of standard Charpy impact energy values with valid KIo results. The
present discussion will be concerned with the possibility of simply using
subsized specimens in otherwise standard Kic tests.
An opportunity to explore this possibility is available through a compari-
son of results of this investigation for 4340 steel with those obtained by
Srawley [3] for a 250 grade maraging steel aged at various temperatures.
This comparison may be made with the aid of Fig. 15, which presents KQ
data for both the 4340 steel and overagedconditions of the maraging steel as a
80
= l-lin.
75
"X,,•
~ w
- - : 4
f 9 .
7O
I
4 i-6_4ksi i~.
65 ra o = 0.27 I
1
60
25 i~/
55
I I I I__ I I
500 .2 .6 .8 .4 1.0 1.2
B, in.
FIG. ll--Composite representation of the influence of specimen thickness on KQ values
for several crack lengths, illustrating the variation in Kzc permitted by A S T M Committee
E-24 Test Method.
90 i - - [ - -
B = ao = 0 . 2 7 i n .
W = O. 55 in.
80
I
/ -
7O
/ r
/,~/
~ 6o "-- O. 45 a/W
/
" I
J\ \
/ O. 55 a/W
/
5O /
I
I
/
401
30
0 I 2 4 5 6 7
Aa/a o x 100
FIG. 12--Example of variation in Kzc with range in a / W permitted by the A S T M Com-
mittee E-24 Test Method. Curve shown is typical for 0.27-in.-thick bend specimens of 4340
steel tempered 750 F, I h (~ra = 213 ksi).
izo
9 Ka Based on acoustic indications B= l i n .
0 Excess plasticity W =2in.
110
100-
9..., 9
6~ 90-- fKa
<. 9 &A
80-
1.0 1.,5 2.0 2.5
\~ 9
7O I I I I I I I I
.2 .4 .6 .8 1.0 1.2 1.4 1.6
a0 - i n .
F I G . 1 3 - - K Q and Ks values as a function o f crack length for 4340 steel tempered at 925 F
(errs = 182 ks 0 as determined using 1-in.-thick bend specimens.
1ZO
-- ~& 9 Ka Based on acoustic indications B = O. 1 in.
tl 0 Excess plasticity ~ = 2 in.
110 --
._ ~ 9
90--
Ka
8oi-
7c I ,u j ] I I l I I
.2 .4 .6 .8 1.0 1.2 1.4 1.6 1.8
ao - in.
F I G , ] 4 - - K Q and Ks values as a function o f crack length for 4340 steel tempered at 925 F
(ars = i82 ksi) as determined using O.l-in.-thick bend specimens.
160 --
Secant, B = O.5 in.
percent
5 10
140 - -
~:\~~ : ~ 42~OMarage
120 - -
a,,,. ~'~ ?50Marage,age

" ~:,
: !: 1000o tollO0o F(3 )
100 --
&
80--
60--
Rangeof valid KIC
40 I I I I I
160 180 200 2ZO 240 zOO
Oys, ksi
F I G . 15--KQ values for subsized specimens and the range o f valid Ktc as a function of
yield strength for 4340 and overaged grade maraging steels.
function of yield strength. The range of valid K~e data for either alloy is
shown by the shaded bands, and it is quite evident that the maraging steel
exhibits a definite superiority in toughness at all yield strength levels in-
vestigated. The KQ values for ~-in.-thick specimens cut from the center of
the thickness of the larger specimens are shown also on Fig. 15. These data
for the 0.5-in. specimens do not represent valid KIr results. Note that the Ko
values for the subsized specimens, based on a 5 percent secant intercept,
would indicate the same toughness level for both steels at 180 ksi yield strength
and no increase in toughness when the yield strength is reduced from 200 to
180 ksi. In contrast the valid Kic values show an increasing superiority of the
maraging steel with reduction in yield strength and continuously increasing
values of K~c for both alloys with decreasing yield strength. At 180 ksi yield
strength level the subsized specimens were unable to distinguish the very
large difference in KI r between the two steels even though B(KQ/~vs) -2 = 2.2.
These results are further evidence against relaxing the size requirements.
An attempt was made to improve the usefulness of the subsized specimens
by basing the Ko values on a l0 percent rather than a 5 percent secant inter-
cept. As seen in Fig. 15, this procedure raised all the values and restored the
trend of increasing toughness with decreasing yield strength but did not
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions au
160 - -
250 Marage ~ Secant, B : 0.5 in.
Age 725~ to ~ percent
850 ~ ,13) ~:/~\ 5 1o
140- ~:~:~ 9 o 4340
9 ~ 250Marage
t20 - -
l~ tOO- o "ii=:::~
80--
60 - - \\ 4340
~ g e of valid KIC \~,
40 l l l l l
160 180 200 220 240 260
Oys , ksi
FIG. 16--KQ values for subsized specimens and the range o f valid Klc as a function o f
yield strength for 4340 and underaged 250 grade maraging steels.
separate the two alloys. It would have been interesting to explore the influence
of further increasing the secant slope values; however, there was an insufficient
number of tests on the maraging steel carried beyond the 10 percent secant
load to permit a further analysis of this type.
A comparison similar to that shown in Fig. 15 may be made between the
results of the underagedconditions of the maraging steel and those for the
4340 steel. The pertinent data are given in Fig. 16, and these substantiate the
observations made from the previous figure. Thus, at the lowest yield strength
level (170 ksi) the KQ values for the subsized 4340 steel specimens are equal
to those for the maraging steel when any reasonable extrapolation of the
trend curves for valid Kic values would indicate the maraging steel is con-
siderably superior in toughness. It also is interesting to note that the sub-
sized specimens would rate the maraging steel to be tougher at 200 ksi than
at 190 ksi yield strength level when again the opposite is true for valid K~c
values.
Correlation of KI~ with Tensile Properties

Relations between smooth tensile properties and KIr are important not
only to a basic understanding of fracture properties but also in a direct
engineering sense as well. Thus, they may offer means for estimating Kt~
values where specimen size requirements make plane strain fracture tough-
ness tests impractical.
It generally is observed for a particular alloy that the strength of cracked
specimens is related inversely to the tensile strength level when this level is
varied by heat treatment, test temperature, or loading rate. Krafft and Sullivan
[6] found their fracture toughness values for mild steel could be correlated
with the upper yield strength to the - 1 . 5 power when the test speed and
temperature were varied. The K~ r data for the 4340 tested in this investigation
is shown in Fig. 17 plotted on a log-log scale against yield strength. These
data are approximated fairly well by the relation K~c = ~vs-3 although the
observed deviations from linearity are systematic and outside the range of
data scatter. Added to Fig. 17 is Srawley's [3] maraging steel data representing
2.20 --
9 725`~ F Age 6 hours
ii00 KIC ~
2. I0 - -
,.oo
1.90
\
\ oo\o
o o\ 950
600~ to 925" F \
9 0 250 Marage (3) \
Average 3 tests \
9 Unlderaged I II\
1.7o I I I
2.20 2.25 2.30 2.35 2.40 2.45 2.50
log Oys
FIG. 17--Relation between plane strain fracture touglmess and yield strength for 4340
and 250 grade maraging steels.
both underaged and overaged conditions. For the underaged conditions

(725 to 850 F) the proportionality between K~ ~ and yield strength is nearly the
same as observed for the 4340 steel. On the other hand, the fully aged and
overaged conditions do not approximate a simple relation to yield strength
possibly because of appearance of a second aging reaction at temperatures
between 850 and 950 F. Data showing the effect of temperature on tensile
strength and Kt~ has been reported by Steigerwald [7] for several types of
steels and a titanium alloy. Analysis of these data indicate an inverse power
relation between Kt, and yield strength is approximated, but that the strength
of the correlation varied among the alloys investigated. There is really no
fundamental reason to expect a simple relation between K~o and a strength
value derived from a tension test. However, the available data indicate an
inverse power relation can be a useful approximation for guidance in esti-
mating Kie values where complex metallurgical effects are not dominant in
determining the changes in tensile strength.
Both Krafft [8] and Hahn and Rosenfield [4] have proposed relations in
which K~c is proportional to a strain hardening index derived from the
tension test. The relation proposed by Krafft includes a process zone size
that would have to be determined by some independent means for the steel
tested in this investigation. However, the relation proposed by Hahn and
Rosenfield involves only quantities of stress and strain which can be deter-
mined directly from a tension test. Thus
Kic = [2/3 E~ysn*-~f] 1/2
where:
E = Young's modulus,
~Ys = conventional yield strength,
n = strain hardening exponent, and
es = true strain at fracture.
The constant 2/~ apparently has the dimension of a length.
One problem with these relations is the difficulty of arriving at a satisfac-
tory definition of the strain hardening index. Frequently this is defined as
the exponent in the parabolic hardening law
O" ~ A~ n
where:
o = true stress, and
e = either the total longitudinal true strain or the plastic component of
this strain.
Unfortunately, as pointed out by Lauta and Steigerwald [9], many materials
do not conform well to this law, and then n must be defined at some arbitrary
strain. Under these circumstances a strain hardening index is unlikely to

have general fundamental significance.
For the present analysis the strain hardening index n was taken as the
strain value where
da
-- O
d~lp
This is the well-known necking condition for uniaxial tension. The value of n
for the various tempers of 4340 steel was obtained by graphical differentiation
of the true stress versus true longitudinal plastic strain curves.
It is possible to examine the usefulness of the relation proposed by H a h n
and Rosenfield by employing it to predict the K~c values of the 4340 steel
from the appropriate tensile data. The necessary information is shown in
Fig. 18 as a function of the tempering temperature. The n values were obtained
as described above. It will be noted that these values are low and nearly con-
stant between tempering temperatures of 500 to 850 F and appear to be
increasing at 925 F. The calculated K~c values differ widely from the valid
data except at 500 F. It might be argued that the disagreement arises from
the fact that 4340 does not follow power law hardening or that the n values
are in error. In order to check these possibilities log a versus log ~lv plots were
made and the slopes of these compared with the n value determined from the
instability condition. Examples of these plots are shown in Figs. 19 and 20
for tempering temperatures of 850 and 500 F, respectively. At 850 F the log-log
plot does exhibit a reasonably straight portion between ~lp = 0.01 and 0.1
with a slope essentially equal to n. This behavior was also observed for the
650 and 925 F tempers. On the other hand, at 500 F, Fig. 20, some imagination
has to be used to find a truly linear portion of the log-log plot. The line
shown has a slope equal to the necking strain (n = 0.027). I f the K~r value
at 850 F temper is used in H a h n and Rosenfield's relation to calculate n, a
value of 0.05 is obtained. This is well outside the error in analysis of the
v e r s u s ~ip curves and, as can be seen from Fig. 19, represents a tangent to the
log a versus log ~lv curve at a strain outside the range of uniaxial tension.
Practical Significance of Results

In what follows an attempt is made to summarize features of the results
which have practical significance to those engaged in fracture toughness
testing.
1. The value of K~ o measured in accordance with the ASTM Committee
E-24 Test Method will vary somewhat with crack length and specimen
thickness. This variation is inherent in the present definition of K~c and the
tolerances placed on the specimen proportions. Its magnitude will depend on
the material properties through their influence on crack growth resistance.
240--
200
18o I I I~'~. I
970
50? l I I
9 06
" i ~ n : Neckingstrain I ~
9 I I I
120--
F2 2 11/2
aoo ----Kzc : L~Eoysn El] ~
60 j//
4( I 1 I
600 700 800 900 1000
Temperingtemperature,~
FIG. 18--Tensile properties and plane strain fracture toughness o f 4340 steel as influenced
by tempering temperature and a comparison between predicted and measured Klc values.
F o r the steel tested in this investigation the total variation was about 15
percent of the lowest value.
2. The variation in measured K~c values could be reduced by either in-
creasing the size requirements or basing Kic on a fixed increment of crack
extension rather than a percentage of the initial crack length. The first means
is in harmony with the nature of elastic fracture mechanics but is otherwise
unpopular. The second means would require a change in instrumentation
JONES AND BROWN O N CRACK LENGTH AND THICKNESS 87
to allow the sensitivity of crack extension measurements to be independent

of specimen size. This might be accomplished by bucking out the elastic
changes in displacement.
3. It is the opinion of the authors that the results of this investigation show
that the present size requirements of a = B => 2.5 (K~o/crys)2 are marginally
sufficient to ensure the small scale yielding required by the basic elastic
crack stress analysis and certainly should not be relaxed for engineering
300 --
- - n = O. 034 0
?80-- ------ n = KIC 0
~"??~
.~
0940- ~ 9 ~
200 ~
i I I I I f
9 001 . Ol .1 1. 0
Elp
FIG. 19--True stress-true plastic strain data for 4340 steel tempered at 850 F (~r~ =
198 ksi).
340
O
4340 Steel 9
3?0 - - 6 0 0 ~ F temper
~ o
300 - -
R~
-~
. ?8O ~ ~ ~' - n=
= O.
9 027J
_~60--
0
0
240--
0 0
2zo l I I I i I
9 001 . Ol .1 I. 0
~ip
FIG. 20--True stress-true plastic strain data for 4340 steel tempered at 600 F OrYs = 213
ksi).
purposes. The results also show that tests to establish specimen size require-
mentw can give misleading indications depending on the particular combina-
tion of specimen geometry and material properties selected.
4. When testing subsized specimens using the ASTM Committee E-24 Test
Method procedures the measured value of K o is likely to be less than K~ o
if both the crack length and thickness are inadequate, and can be greater
than K~c if the specimen is undersized in thickness but oversize in crack
length.
5. If procedures of the ASTM Committee E-24 Test Method are used to
determine KQ values from subsized specimens which are intended for screen-
ing alloys regarding Kic, there is no assurance that the order in terms of Ko
will be that in terms of K~. Furthermore, only rather small deviations from
the size requirements can upset the rating order. On the basis of the pres-
ently reported results there is no clear way of redefining KQ to ensure its
usefulness for screening purposes. However, further research is certainly
desirable and should involve the analysis of complete (to maximum load)
crack growth resistance curves for various sized specimens.
6. The check procedure for determining whether there has been at least one
percent crack extension at KQ will sometimes reject K~ ~ results known to be
valid. This is not surprising since crack growth may sometimes occur below
0.8P,. To the authors' knowledge this difficulty with the check procedure is a
minor problem and is not a reason for discarding it. One possibility of im-
proving the situation would be to reduce the check load point from 0.8P, to
a lower value. This, however, would require increased precision in the auto-
graphic recording system that could be obtained best by bucking out the
elastic displacement change. On the other hand, the check procedure prob-
ably could be eliminated entirely by doubling the size requirements.
7. For steels undergoing single aging or tempering reactions useful guidance
in estimating Kt r values may in some cases be obtained by assuming K~ c
avs -m. However, none of the proposed relations between uniaxial tensile
properties and fracture toughness are sufficiently well developed to permit
useful estimates of Kro from uniaxial tensile data alone.
APPENDIX I
Dimensionless Displacement Factor
Values for the dimensionless displacement factor for three-point bend specimens
were obtained by a boundary value collocation method of elastic analysis [10] for
a/W values between 0.2 and 0.8 in increments of a/W = 0.05. These results can be
represented by the following polynominal:
EvB
log F = log - i f - = 3.48 + 12.7 log a/W + 31.5 (log a/W) 2
+ 40.8 (log a/W) 3 + 20.3 (log a/W) 4
which fits within 2.5 percent for 0.2 < a/W < 0.8. This is sufficient accuracy for
the present purposes. However, if the function of a / W is to be differentiated smooth-
ing procedures must be used on the boundary value collection data. Differentiation
is necessary to obtain the H factors used in the ASTM Committee E-24 Test Method
to obtain the appropriate secant slopes.
APPENDIX II
Errors in Determination of K versus Aa Curves from Load Displacement Records
Errors in the K versus ~a curves arise primarily from uncertainties in the value
of the crack growth increment Aa = (a -- a0) as determined from the load-displace-
ment records. An estimate of this uncertainty may be made on the basis of the
following assumptions: (1) the scale distance coordinates of P and v are about 4
and 2 in., respectively, on the autographic recorder paper; (2) a point on the
autographic curve can be located within a 0.01-in. radius circle, and (3) the value of
a/W = 0.5. For the conditions assumed the error in v/P is
~(v/P) = __AF= 0.006

v/P F
where F is the dimensionless displacement function (see Appendix I). When
A(v/P) < <v/P
AF ~ dF ]-I
(a/W) = F ~- ka(a/rf)J
for an a/W = 0.5, F = 36, and dF/d(a/W) = 200. The differential was evaluated
by an analysis of the boundary value collocation results for the dimensionless dis-
placement factor (see Appendix I).
On the basis of this analysis the uncertainty in the reported values of Aa would
be approximately 0.001 W in. Thus, the K versus zxa curves (see Figs. 4 to 7) are
not established with a useful degree of accuracy near the origin. For the present
purposes it is considered that useful accuracy in ~a is obtained for values in excess
of about 0.0025 in.
APPENDIX III
Calculation of True Stress-Strain Curves in Tension
The autographic load-elongation curves were reduced to true stresses and true
plastic strains as follows. The total longitudinal strain eu was divided into an
University of Washington (University of Washington) pursuant to License Agreement. No further reprod
elastic and plastic part,

elt = e l e -~- e l p
which were determined directly from the load-elongation curves. The true longi-
tudinal plastic strain is then
~lp = log, (1 q- exp)
For values of e , < 0.01 the stresses were computed on the basis of the initial
diameter. Otherwise the stresses were computed on the basis of true areas,
P 4P
- A - rdo 2(1 + e~t)~
where e3t = e2t is the total radial strain. (There was no detectable anisotropy.) The
value of e3t was taken as e~/2 where 0.01 < e~t < 0.1 and as
eat ~ (1 + e l t ) I/2
where e~ > 0.1. This procedure will give the true stresses within one percent.
In the case where the minimum specimen diameter was measured as a function
of load, the true stresses could be determined directly from the diameter readings.
The engineering radical plastic strain e3p was computed as follows:
Ixd 0.3~r
where:
= true stress,
~d = diameter change (negative), and
E = elastic modulus.
For ~3v < 0.05
and for e~ > 0.025

el,, = (1 + e3p)-~ - 1
and
%, = log, (1 + eta)
Acknowledgments
T h e au t h o r s wish to a c k n o w le d g e the assistance in calculations received
f r o m G e o r g e Succop o f the Strength of M a te r ia ls Branch, N a t i o n a l A e r o n a u -
tics and Space A d m i n i s t r a t i o n - - L e w i s , Cleveland, Ohio. T h e y are also
indebted to R. T. Bubsey of the Strength of Materials Branch for his assistance
with the tests and helpful discussions during the analysis of the data.
University of Washington (University of Washington) pursuant to License Agreement. No further repro
References
Strength Metallic Materials, A S T M STP 410, American Society for Testing and Ma-
terials, 1966.
Fracture Toughness," Materials Research & Standards, Vol. 7, No. 6, p. 262.
[3] Srawley, J. E., "Plane Strain Fracture Toughness Tests on Two-Inch Thick Maraging
Steel Plate at Various Strength Levels," Proceedings of the Second International Con-
ference on Fracture, Brighton, Sussex, England, 1969 (to be published).
[4] Hahn, G. and Rosenfield, A., "Source of Fracture Toughness: The Relation Between
Kzc and the Ordinary Tensile Properties of Metals," Applications Related Phenomena
in Titanium Alloys, A S T M STP 432, 1968, p. 5.
[5] Jones, M. H. and Brown, W. F., Jr., "Acoustic Detection of Crack Initiation in
Sharply Notched Sheet Specimens," Material Research & Standards, Vol. 4, No. 3,
March 1964, p. 120.
[6] Krafft, J. and Sullivan, A., "Effects of Speed and Temperature on Crack Toughness
and Yield Strength in Mild Steel," Transactions Quarterly, American Society for
Metals, Vol. 56, 1963, p. 160.
[7] Steigerwald, E. A., "Plane Strain Fracture Toughness for Handbook Presentation,"
AFML TR 67-187, Air Force Materials Laboratory, July 1967.
[8] Krafft, J., "Correlation of Plane Strain Crack Toughness with Strain Hardening
Characteristics of a Low, a Medium, and a High Strength Steel," Applied Materials
Research, Vol. 3, 1964, p. 1964.
[9] Lauta, F. J. and Steigerwald, E. A., "Influence of Work Hardening Coefficient on
Crack Propagation in High Strength Steels," AFML TR 65-31, Air Force Materials
Laboratory, May 1965.
[10] Gross, Bernard, Roberts, Ernest, and Srawley, J. E., "Elastic Displacements for
Various Edge-Cracked Plate Specimens," International Journal of Fracture Mechanics,
Vol. 4, No. 3, Sept. 1968, p. 267.
STP463-EB/Sep. 1970
DISCUSSION
S. T. Rolfe 1 and S. R. N o v a k ~ (written discussion)---In view of the growing

use of fracture mechanics data by engineers, the authors' paper on the in-
fluence of specimen size on K~, values is most timely. They show that K~
values which were determined in accordance with the ASTM Proposed Test
Method for Plane Strain Fracture Toughness of Metallic Materials 3 can
vary moderately (15 percent) within the specimen size and geometrical
limitations imposed by the test method. They further state that these varia-
tions would be increased by a relaxation of the current size requirements.
We agree with this statement and thus agree with the authors that the use of
subsize specimens with the ASTM Committee E-24 Test Method does not
constitute a useful screening procedure.
However, with the development of fracture mechanics to the point where
it is a working tool that can be of significant value to engineers, and the
conc6mitant development of very high-strength steels with good toughness,
the engineer is faced with a serious dilemma. On the one hand, he has a power-
ful tool that can help him in research, material specification, alloy develop-
ment, analysis of subcritical crack growth, and most importantly, in struc-
tural analysis of members that may have cracks in them. Conversely, the
metallurgist continually is developing new materials with crack toughness
that cannot be measured by the currently specified test method because of
the specimen-size requirements. The engineer is then faced with one of
several choices. He can choose not to use fracture mechanics. Under the
rigid requirements that currently exist for Kr~ testing, such a choice would
appear to be proper for very tough materials where valid Kr ~ values cannot be
obtained in reasonably sized specimens. Thus he is forced to use other methods
of determining fracture behavior that may be significantly less accurate.
Furthermore, these less accurate methods cannot relate the critical flaw
size/design stress parameters so important to designers. Because of the
1 Professor of civil engineering, University of Kansas, Lawrence, Kans. 66044; formerly
division chief, Mechanical Behavior of Metals Division, U.S. Steel Applied Research
Laboratory, Monroeville, Pa. 15146.
Research engineer, Mechanical Behavior of Metals Division, U.S. Steel Applied
Research Laboratory, Monroeville, Pa. 15146.
"Proposed Method of Test for Plane-Strain Fracture Toughness of Metallic Materials,"
1969 Book of ASTM Standards, Part 31.
92
Copyright* 1970 by ASTM International www.astm.org
DISCUSSION ON CRACK LENGTH AND THICKNESS 93
tremendous potential of fracture mechanics, this choice should be avoided,

if possible. A second choice (at least until a nonlinear or plastic fracture-
mechanics analysis becomes available) is to use an approximation to K~c
that is useful for engineering purposes.
The authors do not recommend the use of subsize specimens, even for
engineering purposes, and we would agree with this statement. However,
by using subthickness specimens, consistent approximations to K~o have
been obtained within + 15 percent for several high-strength steels. Recent
test results 4 also have shown this same behavior for aluminum. Specifically,
by following the ASTM Test Method in all aspects except the limitation
on thickness, the engineer can obtain an engineering approximation to K~o
(designated as K~E) that can be used in the same manner as K~c values.
Obviously, considerable research is necessary before a KIE test method can be
established in the same manner that the test method for K~ o was established;
however, research in progress indicates that using the current K~, test method
with subthickness (not subsize) specimens, within the limitations described
below, will provide a good approximation to KI c. Because specimen thickness
is usually the limiting requirement (the a and W dimensions of a test speci-
men cut from a plate usually can be made quite large, but the B dimension is
fixed by the original thickness of plate), the use of subthickness specimens is a
realistic solution to the problem of specimen-size limitations.
To substantiate the fact that subthickness slow-bend specimens may give
a good approximation to the K~c values obtained from full-thickness Kro
specimens, slow-bend K~ ~ tests of several high-strength steels were conducted.
In these tests, the a and W dimensions of the specimens were kept large
enough to satisfy the ASTM Test Method. However, the B dimension was
varied so that the results from subthickness specimens could be compared
with results obtained from standard-size specimens.
Preliminary results for an 18Ni maraging steel (192 ksi yield strength)
are presented in Fig. 21. These results show that by keeping the a and W
dimensions sufficiently large, that is, W > 5.0 (K~c/au) 2 and a > 2.5 (Kte/cry)2 ,
test specimens where B is less than the currently specified minimum thickness
(B = 2.5 (K~o/a~) 2) can be used to approximate K ~ within =t=15 percent.
Similar subthickness specimen tests of other high-strength steels with strengths
from 150 to 250 ksi are in progress, and preliminary results substantiate this
behavior.
The data of Fig. 21 also show substantial stable crack propagation above
K~a% (to Kt~,~,) for this alloy. In addition, the data show the dependence of
K~.... on thickness, B, while Kts~ (KtE) appears to be relatively invariant.
4 Kaufman, J. G., "Thickness Effects on Aluminum Alloys," presented at ASTM Com-
mittee E-24 Meeting, 23 Sept. 1969.
9,4 PLANE STRAIN FRACTURE TOUGHNESS TESTING
iiit
I I I I tl I I I I
I
'#''--~'"% KI (MAX LOAD) W= 6.0 ='
/ " ' J -,%- r
ff~""" '~"~'~ % ~ 1 K-r ( MAX LOAD) W=4.0"
,,,, ,,,
160-- N I \
--T .... z)-- - ---~-~z.... --T---- ~Z- ... --"

120
~-~ o o, 'oo----- oo~=e- - - - - -#~ia---~
..... ~ ~ I
80
I I a/w ~ o.5
I I "
I I
Kic= 113 ksi 4 ~ " ~ + 1 5 % I 0 KI 5% (W=4.0")
I II 9 Kz 5% (W= 6.0" )
40 r
I 2 I
iI-~'-B = 1.0( - ~ ) ~ BU,N ACCORDING
, vTO
, Z TEST M
ET
HOD
~-)IN)-i
I
, =(2.
0 J J, I i lit I
0.10 0.20 0.30 0.50 0.70 0,90 1.0 2.0 3.0 4.0
SPECIMEN THICKNESS (B). inches
FIG. 21--Effect o f specimen thickness on K x at 5 percent secant intercept and at maxi-
m u m load f o r an 18Ni (190 grade) marging steel in three-point bending.
The significant feature of this engineering approximation to K t c is that by

making a and W sufficiently large in a bend specimen with a steep stress
gradient, the plastic zone size ahead of the crack tip is sufficiently small so
that an elastic analysis can be used. Although the plastic zone size also is
controlled by the B dimension, the only current technique for determining
the thickness limitation is experimental, and results of subthickness specimens
indicate that specimens having B < 2.5 (KI o/r 2 still yield satisfactory results.
In fact, using subthickness test specimens and making B ~ 1.0 (Kie/r 2
gives a close approximation to KI c.
Analysis of the authors' data also substantiated the fact that subthickness
specimens should give satisfactory results. For the 4340 steel tempered at
750 F (ay = 213 ksi), the average KIc value is approximately 70 ksi ~v/~.
Using this value, the minimum thickness, crack length, and specimen depth
in accordance with the ASTM Recommended Test Method should be 0.27,
0.27, and 0.54 in., respectively. Using the proposed KxE procedure for
engineering approximations, the minimum specimen dimensions should be
approximately 0.11, 0.27, and 0.54, respectively. Note that by using only
subthickness specimens, the a and W dimensions still satisfy the recommended
test method for K~ ~.
Analysis of the authors' data presented in Figs. 8, 9, and 10 of their paper

shows KIc values of 64, 69, and 73 ksi x/in., respectively, at B = 0.27 in. At
B = 0.11 in. the /fiE values are 63, 70, and 77 ksi ~/im., respectively. This
variation between K~c and KIE is well within the 15 percent variation that
the authors observed for the valid K~o test results on specimens tempered at
750 F. Thus, it would appear that subthickness specimens can be used to get
an engineering approximation to K~ ~. For the 4340 steel tempered at 925 F
(~r~ = 182 ksi), the KI~ value is approximately 107 ksi x,/in. On this basis, a
K~E test specimen should have B, a, and W dimensions of 0.34, 0.85, and
1.70 in., respectively. Those test specimens that do satisfy the KxE size
requirement will have a K~E of approximately 105 ksi x/i-re., Fig. 13.
In Figs. 15 and 16, the authors analyze Srawley's test results 5 for a maraging
steel heat treated to various yield strengths to further show that subsize
specimens give misleading results. However, because the subsize specimens
were proportioned in all three dimensions (W, a, and B), only the results on
material tempered from 800 to 1000 F would meet the subthickness specimen-
size requirements. These values obviously agreed with the K~ c values obtained
from the large-size specimens, because even the small-size specimens satisfied
the ASTM Test Method.
In conclusion, the authors are to be complimented on a very timely paper
describing the effects of specimen size on determination of KI ~ values. We
would agree with their conclusion that subsize specimens should not be used
for screening purposes, but we would suggest that subthickness specimens
can be used to obtain engineering approximations to K~c, namely, KIE values.
J. G. Kaufman ~ ( w r f t e n discussion)--This investigation was carried out

with the usual thoroughness of the authors, and we are fully in agreement
with their conclusions that (a) relaxation of present requirements for K~o
testing are unwarranted and (b) the results from undersized specimens are of
little or no value for screening purposes. The data presented in Figs. 8, 9, and
l0 amply demonstrate that values of K o calculated from thin specimens may
be either higher or lower than K~c, dependent upon the interrelation of the
thickness, the toughness, and the absolute crack extension. Data from under-
sized specimens are best considered only as guides for the selection of more
appropriate sizes of specimen.
5Srawley, J. E., "Plane Strain Fracture Toughness Tests on Two-lnch Thick Maraging
Steel Plate at Various-Strength Levels," Proceedings of the Second International Conference
on Fracture, Brighton, Sussex, England, 1969 (to be published).
Section head, Fracture Mechanics and Product Evaluation Section, Mechanical Testing
Division, ALCOA Research Laboratories, New Kensington, Pa. 15068.
A popular misconception regarding data from specimens which are under-

sized with regard to thickness is that, although they may not be valid K~c
values, they are useful as "Ko" values to indicate the "thin-section" toughness
of the material. This is not the case. Disregarding the questions which have
arisen in recent years about the meaning and constancy of Kc values, the
plane strain fracture-toughness test as presently defined tells nothing about
the thin-section instability as it was embodied in the K~ approach. The values
are not quantitative measures of the thin-section toughness in any terms un-
derstood at this time, although there may be some possibility of interpreta-
tion of the load-deformation curves from such tests in terms of resistance
curves, after the fashion proposed by Heyer and McCabe. 7 I would be in-
terested in the authors'comments on this point.
J. M. Krafft s (written discussion)--This excellent paper provides much

needed: (1) systematic studies of the effect of specimen dimensions on the
measure of Kle, and (2) comparisons of plastic flow properties with fracture
behaviors. 1 would like to cast a vote on the alternatives suggested in the
conclusions.
We need not be too concerned that the measure of K~ c is not invariant if
a0 and W depart from required size and ratio. We are usually testing plate or
sheet product with adequate area available to make a0 and W what we would.
And indeed, as the authors note, the whole concept of linear elastic analysis
breaks down if the planar dimensions are inadequate. However the thickness
is usually limited, yet K~o values are of great value. The systematic trends in
KQ with thickness for three specimen widths, Figs. 8 through 11, appear to
be explicable in terms of the effects of these variables on the shape of the crack
growth resistance curve which is sampled at different levels by the 5 percent
secant. Authors' Conclusion 3 would solve this dilemma through control of
the shape of the resistance curve by continuing to require relatively thick
sections. I prefer the alternative, noted in Conclusion 2, of abandoning the
secant intercept in favor of "basing Kx ~ on a fixed (and preferably vanishingly
small) increment of crack extension." Our own experience indicates that if
only the thickness requirement is reduced, this should require no change in
the present instrumentation, down to thicknesses of about 1.0 (Kie/ays) ~.
Such a change would vastly enhance the applicability of the KTc test, and
thus be worth the extra care of crack growth detection.
With respect to the attempted correlations with tensile properties, the
physical event which George Hahn and I attempt to predict is the onset of
7 Heyer, R. H. and McCabe, D. E., "Plane-Strain Fracture Toughness Testing Using a
Crack-Line-Loaded Specimen," presentation at the National Symposium of Fracture
Mechanics, Lehigh University, Bethlehem, Pa., 25 Aug. 1969.
a Head, Fracture Mechanics Consultant Staff, Naval Research Laboratory, Washington,
crack growth, which would be the K I e instability for a very large specimen.
As with decreased thickness noted above, increased toughness, as with
higher tempering temperature, tends to accentuate the value of KQ. This
could account in part for the failure of the n values of Fig. 18 to rise as fast
as K~ ~. It argues further for replacement of the secant intercept method with
a more sensitive measure of the onset of crack extension.
Among materials poorly characterized by parabolic strain hardening and
a constant strain hardening exponent, n, is 4340 quenched and tempered
above 600 F, as authors note from Ref 10 of the paper. In such cases we find
it useful to replot true stress ~ strain ~ curves in an inverted differential form,
da/de/~r versus 1/~. Here parabolic hardening is indicated by a straight line
through the origin of slope n. Whatever the curve, the intercept with the
line da/de/a = 1.0 defines the strain for uniaxial tensile instability. With these
steels, the curve tends to flare out so as to gradually descend below the 1.0
level before dropping toward the origin. This oblique intersection makes
actual intercept sensitive to small variations in curve shape, although it can
be rather accurately defined in this way without exercise of undue "imagina-
tion." Perhaps these refinements in measures of KI~ and of instability strain
would improve prospects of estimating K ~ from plastic flow properties.
M. H. Jones and IV. F. Brown, Jr. (authors' closure)--We are pleased that
Mr. Kaufman agrees it would be unwise to relax the present size require-
ments for KI o testing. It is our opinion that these requirements as they stand
now are barely sufficient to permit the use of an elastic mechanics as the
basis for a quantitative method of fracture testing. In fact, we always have
believed the present requirements should be doubled. As Mr. Kaufman cor-
rectly states, the value of K~c tells us nothing about thin-section fracture
instability as was originally embodied in the Kc approach. It is, of course,
possible that crack growth resistance curves obtained from under-thickness
specimens might be useful in characterizing an alloy's resistance to mixed
mode fracture. This is the approach being followed by Heyer and McCabe, 7
and measurements of this kind should be encouraged. However, it is not yet
clear what feature or features of the K versus a curves should receive attention
for a standardized measure of fracture resistance, nor has the influence of
crack length in these tests been sufficiently well explored.
Both Messrs. Rolfe and Novak and Mr. Krafft call attention to the problem
of evaluation of very tough materials where the thickness needed for practical
applications is not sufficient to meet the ASTM Committee E-24 size re-
quirements. They propose to circumvent this problem by reducing the
thickness requirement. This suggestion has been made before, and we are
quite opposed to it. The size requirements as given in the ASTM Committee
E-24 Proposed Method of Test for Plane Strain Fracture Toughness of Metal-
lic Materials were designed to ensure that conditions of small scale yielding
will be sufficiently approximated in the specimen that a useful engineering

estimate of Kic can be obtained. Any relaxation of these requirements can
reduce the accuracy of this estimate and may lead to serious error in judging
the toughness of a given alloy.
Unfortunately the method suggested by Messrs. Rolfe and Novak for
determining a less precise engineering estimate (K~E) of KI, can result in
K~ E values substantially in excess of the true K~, value. They propose to define
KIE in the same way as K ~ except to relax the thickness requirement to B >
(Kio/au) 2, which really should be stated as B > (KIE/~u) 2 since the true
value of KIo will not be known. Presumably, also they would abolish the
present W / B limits. Difficulties can arise when the crack length is in excess
of that necessary for a valid K~ ~ determination. Under these circumstances,
as pointed out in our paper, the measured KQ value will tend to increase with
decreasing thickness. The magnitude of this effect will depend on the amount
of excess crack length and the change in shape of the crack growth resistance
curve with thickness. This change in turn will depend on the material char-
acteristics. In the case of the 18Ni maraging steel, Fig. 21, tested by Rolfe
and Novak, and for the 750 F tempered 4340 tested by the authors, the
elevation of KQ with decreasing thickness is moderate and where KQ = K~E
overestimates of Kxr would not exceed the 15 percent variation between
K~c and K~E mentioned by the discussers.
However, in some cases we will not be as fortunate in finding so nice a
balance between the effects of crack length and thickness. For example,
cracked bend tests on 6A1-6V-2Sn-Ti, Fig. 22, show steeply rising values of
KQ as the thickness is reduced below that required for a valid K~ o test. These
specimens had crack lengths of about 0.5 in. (approximately six times the
7C I [ I [ I
Ti-6AI-6V-2Sn I in. plate
Age 1100~ F, 4 hr
6C Oys= 174 ksi ~ ~ - -
c~4c
,
~2. 5( i2
\~ys/
v W --w --
3C 9
2~ I I I I I
.i .2 .3 .d .5 .6
THICKNESS, B, IN.
FIG. 22--Influence of thickness on Kto for a high-strength titanium alloy,
University of Washington (University of Washington) pursuant to License Agreement. No further reproductions authoriz
required value) and a/W ratios of about 0.5. These tests were conducted in
accordance with the present ASTM Committee E-24 Test Method and are
valid with the exception of those at the two smallest thicknesses, which
passed the test for excess plasticity but failed to meet the thickness require-
ment. Note that for this alloy specimens meeting the proposed thickness
requirement of B > (KI~/~) 2 can overestimate Kic by 50 percent. Restating
the requirement as B > ( K I E / ~ ) 2 could still result in accepting estimates of
K~c which are over 30 percent too high. It might be argued that this titanium
alloy shows an unusually large effect and perhaps it does; however, there is
no way of knowing in advance of how to adjust the crack length so that a
subthickness specimen will give the "right" estimate of K~ c.
The suggestion by Mr. Krafft that K~ c be based on some very small fixed
increment of crack extension is not in our opinion a workable solution to the
problems introduced by using subthickness specimens. Furthermore, it is
completely incompatible with the present method of test which employs a
clip gage to sense crack mouth displacement. This displacement is sensitive to
both crack extension and plastic flow. It might be possible to develop a new
method of test based on some fixed increment of crack extension, but this
would require a means to sort out plastic flow effects. In any event a "van-
ishingly small" amount of crack extension is not what we would want because
this is essentially impossible to define by measurement. Further, it should be
noted that below a certain unknown limit a given increment of crack extension
may have little significance to the gross fracture characteristic of the sample.
We show poor correlation of K ~ with tensile properties in Fig. 18. Mr.
Krafft suggests that by basing Kt c on a more sensitive measure of crack growth
the correlation with n values might be improved. All the K~ o results shown
in Fig. 18 represent 1-in.-thick specimens, and at 600 and 750 F temper these
are respectively about eight and four times the present thickness requirement.
We are not sure what the "K~ instability" would be in still larger specimens;
however, we assume that it might approach the "KI ~" values calculated on the
basis of acoustic indications from the 1-in.-thick specimens. Acoustic moni-
toring was used on all 4340 steel specimens although the results were not
given in our paper. We designate the K~ value corresponding to the appear-
ance of definite crack sounds as K~. The range of K ~ values obtained at
each tempering temperature for the 4340 steel specimens is given in Table 3.
These values are lower than the K~c values shown in Fig. 18, but still show a
strong dependence on tempering temperature in the range where n is quite
insensitive to tempering temperature. We do not consider these results sup-
port an argument for replacing the secant intercept with a more sensitive
measure of the onset of crack extension. We agree with Mr. Krafft that the
tensile instability strain for 4340 steel tempered above 600 F is difficult to
measure. However, we do not believe any clever constructions will get
TABLE 3--Results from acoustic monitoring ~ o f 1-in.-thick 4340 steel specimens.
Tempering Temperature, Range o f K ~ b Values,

deg F ksix/~.
600 38 to 41
750 53 to 59
850 70 to 80
925 75 to 86
Jones, M. H. and Brown, W. F., Jr., "Acoustic Detection of Crack Initiation in Sharply
Notched Specimens," Materials Research & Standards, Vol. 4, No. 3, 1964, pp. 120-129.
b KIa was calculated on the basis of the load at which the first definite crack sound was
detected.
around the fundamental difficulty which is absence of a well-defined maxi-

m u m in the load-extension curve. It does seem reasonable to search for cor-
relations between flow and fracture properties, and Mr. Krafft has done
pioneering work on this problem. However, we question whether refinements
in measurement of Kjo and tensile instability strains will result in improved
correlations.
In conclusion, we want to emphasize that we are aware of the problems
faced by those who wish to measure the fracture toughness of the new high-
strength steels. However, in a sense they are quite fortunate because when the
thickness of the material required for structural application is insufficient for
a K~c test, plane strain catastrophic failure in service is extremely unlikely.
Under these circumstances the engineer needs new methods of test that will
yield a thickness dependent measure of fracture toughness. Unfortunately,
the development of a mixed mode fracture criterion has turned out to be a
much more difficult task than originally anticipated. However, there is no
reason to believe that a rating of material based o n K I e values will be the
same as a rating based on values derived from a mixed mode fracture test.
Therefore, we believe that attempts to estimate Kx c by some empirical method
are of dubious value when plane strain failure is not a practical consideration
for anticipated structural applications. It is not clear to us that the engineer is
then faced with a true dilemma when he wishes to evaluate the fracture char-
acteristics of these new alloys. In other words, we do not believe that Kxc
constitutes all of fracture mechanics and that no other satisfactory choices are
available.
Where valid K~ c values cannot be measured, we would favor the applica-
tion of another standardized fracture test for alloy development work. This
might be either the center cracked or sharp edge-notch test described in ASTM
Designation E 338-68 or the conventional Charpy V-notch impact test.
Messrs. Rolfe and N o v a k made good use of the standard Charpy V-notch
test in their work on the development of the exceptionally tough 5Ni steels. ~
On the other hand, if plane strain failure is a consideration in anticipated
service applications, then K ~ should be determined as accurately as possible
and any estimates should err on the low side.
9 Manganello, S. J. et al, "Development of a High Toughness Alloy Plate Steel with a
Minimum Yield Strength of 140 ksi," The Welding Journal.
E. A. SteigerwaM ~
Crack Toughness Measurements of

High-Strength Steels
REFERENCE: Steigerwald, E. A., " C r a c k Toughness Measurements of High-

Strength Steels," Review o f Developments in Plane Strain Fracture Toughness
Testing, A S T M STP 463, American Society for Testing and Materials, 1970,
102-123.
ABSTRACT: The influence on fracture toughness of relative specimen thickness in

the center-precracked tension specimen and crack length in the precracked bend
specimen was evaluated. The results indicated that the nature of the crack growth
curve in the precracked sheet tension test could be related directly to the fracture
mode transition temperature. Tests conducted below the fracture mode transition
region provided only catastrophic crack extension, while testing within the transi-
tion region resulted in measurable slow growth which was initiated by a sharp dis-
continuity (pop-in). Above the brittle-to-ductile transition slow crack growth was
observed, but no meaningful fracture toughness could be measured.
The study of the influence of crack length on the fracture toughness obtained in
the notch bend test indicated that valid values of K~ were obtained when the crack
size (a) was 2.5 times the (K/rru) ~ ratio in accordance with ASTM recommenda-
tions. However, it appears that the crack size requirements could be reduced by
a factor of 2.5 without producing variations of greater than 5 percent in the
measured fracture toughness value provided that the thickness requirement was
satisfied.
The available K~o values for a variety of high-strength steel types were also
presented graphically as a function of both strength level and test temperature.
KEY WORDS: fractures (materials), toughness steels, fracture toughness

measurements, evaluation, tests
Crack toughness testing of high-strength materials has been developed to

the point that the American Society for Testing and Materials (ASTM)
recommended procedures are available for determining plane strain fracture
toughness (KIc). Although these procedures are well defined, several practical
problems can arise when they are applied to determining the quality of
actual parts or in making certain engineering decisions. In many cases the
particular part configuration is such that the ASTM recommended practice
for measuring KIo cannot be used to determine specific effects, such as
Materials Technology, T R W Equipment Laboratories, T R W Inc., Cleveland, Ohio
44117. Personal member ASTM.
102
STEIGERWALD ON CRACK TOUGHNESS MEASUREMENTS 103
processing history, on material toughness. Despite these limitations it is often

meaningful to conduct crack toughness tests to provide a relative measure of
material behavior. One purpose of this presentation is to evaluate the in-
fluence of certain geometrical test variables such as specimen thickness and
crack length on the measured fracture toughness parameter to obtain an
indication of the uncertainties that may be introduced when physical con-
straints on the configuration of the part force deviations from ASTM practice.
A second objective is to present available K~c values for a variety of high-
strength steels and provide an indication of the heat-to-heat variations which
can occur in the fracture toughness parameter.
Influence of Geometric Parameters on Precracked Sheet Tensile Results

Although the center-notched sheet specimen is not the optimum configura-
tion for determining fracture toughness measurements there is a relatively
large quantity of data that has been obtained with this specimen type. In
addition, many components are made from sheet material so that the center-
notch specimen represents a convenient method for evaluation. On this basis
there is a need to obtain an indication of the relationship between the speci-
men thickness and the ability to measure fracture toughness values. When
the plastic zone at the tip of a center-precracked sheet specimen is sufficiently
small, slow crack growth occurs initially at the midthickness of the specimen
in a discontinuous manner (pop-in) [1,2]. 2 In cases where the material is
relatively brittle, no slow crack growth occurs, and the initiation of cracking
is coincident with the onset of fast fracture. Tests have been conducted to
determine the conditions that control the occurrence and magnitude of the
discontinuous process [3]. These will be reviewed and related to current
ASTM recommendations for specimen thickness. Results obtained with a
modified H-11 steels are summarized in Figs. 1 through 4, and the smooth
strength properties are presented in Table 1. Details of the test procedure
have been previously described [3]. The crack growth curves were measured
with resistance techniques [4] where the magnitude of the resistance change
can be correlated directly with the degree of crack growth. The notch tensile
properties of the H-11 steel tempered at 1000 F are shown in Fig. 1, while
the slow crack propagation characteristics are shown in Fig. 2. The H-11
steel at this high strength level is relatively brittle, and at test temperatures
below 200 F virtually no slant mode was apparent in the fractures. The crack
propagation curves shown in Fig. 2 indicate that the initiation of crack
3 The H-11 steel was tested as 0.085-in.-thick center-precracked sheet specimens, 1 ~ in.
wide with a crack length-to-width ratio of approximately 0.40. The steel composition in
weight percent was 0.43 carbon, 0.25 manganese, 0.96 silicon, 5.12 chromium, 1.33 molyb-
denum, 0.57 vanadium, 0.01 phosphorus, 0.007 sulfur, balance iron.
I I I I I l I | I
180 -- ,,.~ - 6o ..'2,
160 --
FRACTURE TOUGHNESS~-f~
f~ o
5 0 v-
140
120
2o ~
S~SS,
o 100
-- NOTCH TENSILE a.
STRENGTH O- / NET INITIATION
-- ' N ~ 0"I
8o
- -!- .......
6O
40
2O
100 o
o
75
~ 5o
FRACTURM
E ~ ~ --
25
J, ~ I
0
-100 0 100 200 300
TEST TEMPERATURE ~ OF
FIG. 1--Fracture properties of H-11 steel, tempered at 1000 F.
propagation occurred in a discontinuous manner at all test temperatures

below the transition temperature. 4 In cases where no shear was present the
initiation of cracking was coincident with total failure. In the transition
region slow crack growth occurred, and its onset was accompanied by an
appreciable pop-in. At 300 F the fracture mode ranged from 75 to 100 percent
shear. When the shear was 100 percent the crack propagation curve showed
no discontinuous growth but deviated gradually from linearity.
The transition temperature is defined as the lowest temperature at which specimens
fail with 100 percent shear.
STEIGERWALD ON CRACK TOUGHNESS MEASUREMENTS 105
TABLE 1--Summary of smooth tensile properties of H-11 material.
Test Strength
Temperature, Elongation
deg F Ultimate, ksi 0.2 ~ Yield, ksi in 1 in., ~o
H-11 STEEL: 1000 F TEMPER

- - 100 . . . . . . . . . . . . . 308.4 257.6 9.2
--45 . . . . . . . . . . . . . 300.4 247.7 9.8
40 . . . . . . . . . . . . . 291.6 240.2 10.0
75 . . . . . . . . . . . . . 291.6 242.9 9.5
200 . . . . . . . . . . . . . 281.8 231.6 10.8
300 . . . . . . . . . . . . . 274.4 230.4 11.0
H-11 STEEL: 1050 F TEMPER
- - 100 . . . . . . . . . . . . . 280.4 240.2 10.0
--45 . . . . . . . . . . . . . 270.8 231.4 10.5
40 . . . . . . . . . . . . . 263.9 225.4 10.5
75 . . . . . . . . . . . . . 261.6 223.2 11.5
200 . . . . . . . . . . . . . 255.3 214.6 11.0
300 . . . . . . . . . . . . . 249.6 212.0 11.0
T h e p r o p e r t i e s o f the H - I 1 steel t e m p e r e d at 1050 F are s u m m a r i z e d in

Figs. 3 a n d 4. A l t h o u g h the t r a n s i t i o n t e m p e r a t u r e was shifted to 200 F by
the higher temper, the same general c o r r e l a t i o n s between the m o d e o f slow
c r a c k initiation a n d the t r a n s i t i o n t e m p e r a t u r e were a p p a r e n t . T h e correla-
t i o n between the t e n d e n c y for a p o p - i n to occur a n d the test t e m p e r a t u r e
relative to the m a t e r i a l t r a n s i t i o n t e m p e r a t u r e was f o u n d also in several
o t h e r high-strength m a t e r i a l s [3].
D a t a relating the relative plastic zone size a n d the tendency for the sheet
specimens to show a p o p - i n are presented in Fig. 5. T h e relative plastic zone
size h a s been defined as (K/~ys)2(l/B)
where:
K = c a l c u l a t e d fracture toughness,
ay~ = yield strength, a n d
B = specimen thickness.
T h e c u r r e n t A S T M r e c o m m e n d a t i o n for specimen thickness which specified

t h a t B > 2.5 K2/ay8 ~ is i n d i c a t e d also in Fig. 5. T h e results show t h a t when the
A S T M c o n d i t i o n was satisfied a distinct p o p - i n always occurred in the l o a d -
c r a c k g r o w t h curve. T h e r e l a t i o n s h i p between the presence o f a p o p - i n a n d
the accuracy o f p l a n e strain fracture toughness m e a s u r e m e n t s was e v a l u a t e d
in H-11 steel (250 ksi tensile strength) with r o o m - t e m p e r a t u r e tests utilizing
specimens o f varying thickness (see Fig. 6). T h e fracture toughness of this
1 O0 i I I I 1 I I I I I
9O
80 /
/ .,"
/ .~"
,,:I- -
70
6O
r" /" ~""
/ /.
5O / i. <2o0O~ /
40
k
/ / ()o0oF)
_ 3O
/
,
; /
/
/
o~ 20 i / i
o
I
z 60
==
~.
a= 5o
,/
<E
40 .,.'.
l, FAILURE ,r = FAILURE /
(~_i0
3O
9 O~ /::(-45~ (40~
20
, I I
I I I
10
! I
l , t
I I li i I I I I I I I
10 20 30 40 50 60 70 80 90 100 110 x I0 -7
J
CHANGEIN RESISTANCE( ~ R ) OHMS

FIG. 2--Crack growth curves for H-I1, tempered at 1000 F. Numbers in parentheses
refer to test temperature.
material was relatively constant at the larger thicknesses where a pop-in

was observed. In the thinner specimens where the thickness was less than
approximately 0.040 in. (B > (K/~y)2) the measured value of toughness in-
creased with decreasing thickness. The significant point is that in the center
precracked sheet tension specimens which satisfied the ASTM recommended
thickness of B > 2.5(K/aye)~ a distinct pop-in always occurred. In addition,
the data indicated that in the H-II material evaluated the thickness could
be reduced by a factor of at least two from the ASTM recommended practice
without changing the fracture toughness values. In fact, in the precracked
sheet specimens the fracture toughness usually produced valid K~c values
STEIGERWALD O N CRACK TOUGHNESS MEASUREMENTS 107
whenever a distinct pop-in (discontinuous crack extension of approximately

0.005 in.) existed.
Notch Bend Tests

The currently recommended criteria for obtaining valid plane strain frac-
ture toughness data from the notch bend test indicate that the specimen
thickness divided by the (Kie/O-ys) e ratio and the crack length divided by the
(K~c//~y~)~ ratio should exceed 2.5. A large quantity of results has been pre-
sented in the literature on notch bend specimens with a square cross section
I l P I I I I I I
FRACTURETOOG.NESS /f.~..... ..... 60 ~

5o ~
200 9 40 ~
""~'" ....-- - ......~....-a" ~;

180 30 ~
16o ~-
14o
~. 120 NOTCH TENSILE ~ nf

STRENGTH, Cr N ~. /
o
~- 100
80 o o"" ""
60 "~
STRESS,(7"i
40
20
"t
I O0
75 -X -
~ 50
,.~ FRACTURE MODE
25
0 ~ h r
-100 0 100 200 300
TEST TEMPERATURE ~ ~
FIG. 3--Fracture properties o f H-I1 steel, tempered at 1050 F.
I I I I I ] I [ I I lj
/ f
I00 -- J .z
/.t f //
/
80 --
(300~
/
!
r !
60 -- / l t
/.._~B.r [ 1
= l" (200~ I
./ I
t
40
!
i
(75~ I
#
20
0 --
60' --
/-
..... J
FAI LURE FAI LU RE
40
f .... :
,4o~ __
20
0 I I I I I I I I I I I l
0 20 40 60 80 100 120 x I0 -7
CHANGE IN RESISTANCE (AR) OHMS
F I G . 4 - - C r a c k growth curves for H-11 steel, tempered at 1050 F. Numbers in
parentheses refer to test temperatures.
and with a crack length-to-width (a/W) ratio of approximately 0.2 and 0.3.
On this basis, the thickness could satisfy the recommended 2.5 criterion
while the crack length would not. In an effort to determine the influence of
crack length on the measured K~ ~ value and hence the usefulness of much of
the available KI c values, notch bend tests were conducted on specimens where
both the thickness and the nominal a/W ratio (approximately 0.25) were
maintained constant, but the crack length was systematically changed by
varying the specimen width-to-thickness ratio.
The materials studied were D6AC steel and 2024-T851 aluminum alloy. The
steel was tested with ~-in.-thick specimens, while 1-in.-thick specimens were
STEIGERWALD O N CRACK TOUGHNESS MEASUREMENTS 109
f I r I I I I I I I I I i I I
MATERIAL TEMPER
70 x 10 -7
ASTMRECOMMENDATO
IN H-If IO00~ o
H-1I 1050~ 9
60 2.5 H-I1 llO0~ 9
O"K2
,Ys2 =B 1 Beta TitBnium o
4340 400~ A
97, 50
- ~ 40
~ 30 o N
o N
-- 20
o
-- 10
O I I I I I I 7~'=T'~.=.~.-~..i,.....---.~t~ I a i,, ,o i 0"/16

0.02 0.04 0.I06 0.~08 0.I0 0.12 0.14
K 2 1
RELATIVE PLASTICZONESiZE ~! (7--)
ys F
FIG. 5--Relationship between degree of "pop-in" a n d relative plastic zone size for
several materials.
l I I I I I l l # l
o NO POP-IN
9 OBSERVABLEPOP-IN
80
o o
~ 60 KIC
/
~ z
~o 40
20 I I I I I I I I I t
0.04 O.02 0.06 0.08 O.10
SPECIMEN THICKNESSN INCHES
FIG. 6--Influence of specimen thickness on the stress intensity p a r a m e t e r for slow
crack growth 14-11 steel, 250,000 psi strength level.
used for the aluminum. The compositions, heat treatments, and smooth
strength levels for the test materials are summarized in Table 2.
Notch bend specimens with three- and four-point loading were tested at
various beam heights-to-thickness values (W/B) ranging from ~ to 4.
Major span length-to-beam height (L/W) was varied also between 4 and 8.
o
T A B L E 2--Compositions, heat treatment, and smooth tensile properties for the test materials. z
r
Chemical C o m p o s i t i o n (weight ~ ) ~:
Material C Mn P S Si Ni Cr Mo V A1 Mo Fe Cu
C
O
D6AC ............ 0.45 0.69 0.008 0.006 0.26 0.55 1.08 1.0l 0.08 0.007 ... balance ... z
z
2024-T851 a l u m i n u m ... 0.67 ...... 0.12 0.01 0.01 ...... balance 1.42 ... 4.82 ~,
S u m m a r y of H e a t T r e a t m e n t s a n d S m o o t h Tensile Properties
T~
O
Material T e m p e r a t u r e , deg. F Ftu, ksi Ftu, ksi Elongation, 7o R e d u c t i o n in Area, 7o
D6AC ................ 500 275 231 l 1.7 (1 in. G L ) 42.3

2024-T851 a l u m i n u m . . . . . . 70.3 64.6 7.8 (2 in. G L ) 12.8
N o x E - - D 6 A C austenitized for 20 min at 1550 F in salt, oil q u e n c h e d to r o o m temperature, a n d double tempered, as indicated, for 1 h plus 1 h.
STEIGERWALD ON CRACK TOUGHNESS MEASUREMENTS 1I I
The specimen configurations for four- and three-point loading and formulae
used to compute Kzo values are given in Fig. 7. All specimens were fatigue
precracked at a stress which was less than 20 percent of the 0.2 percent offset
yield strength of the material. Precracking was performed after heat treat-
ment to eliminate uncontrolled crack growth during quenching and to avoid
a possible decarburization effect at the crack tip.
Testing was conducted at r o o m temperature using a crosshead speed of
0.040 in./min. The detection of the crack initiation stress for computing
Kzc was accomplished by the use of a displacement gage consisting of an arm
attached to the loading fixture. The core of a linear variable differential
transformer (LVDT) rested on a contact point on the arm and measured the
relative crosshead motion as a function of the applied load. In all cases the
initiation of crack growth corresponded to maximum load, and this value
was used in the calculation of fracture toughness.
The data obtained from the notch bend tests on all the materials are shown
graphically in Figs. 8 and 9. Variations in the span length-to-width ratio from
four to eight produced no systematic change in the Kzc value. In the steel, no
[- L T L
.,c.= 7"L' [3,. §

NOTCH BEND TEST (THREE-POINT LOADING)
I nro
t'
I
KIC z = ~ [34,7 -~ 55.2 ( +

NOTCH BEND TEST (FOUR-POINT LOADING)
K2
o =o~ + 6TI. O..ysr IN A L L CASES
F I G . 7--Formulae used to compute K i t values.
1 12 PLANE STRAIN FRACTURE TOUGHNESS TESTING
O 4-PT LOADING
9 3-PT LOADING
~-Z 100
a=l.0( K )2 a'2.5 (K )2
80 a"y
z
o 1 o a"yo%
I o o o
o~
2 D6AC STEEL
--~ o-~.- o 9 't ~ 9 ,~ o o 275 Ks~ TENSILE
60 / o 9
o
/ / o STRENGTH
/ o
o o
40 1
o ,2 0.3 0.4
CRACK LENGTH, INCHES
F I G . 8--Variation of measured fracture toughness with crack length, D6AC steel.
26
~ 25
o_
o ~ 2 . 5 ~.,_)y 2 4PT
:~ 24
a=l.0 (+)2 ~ LOADING
~ 23 -o~L~..
~ Ooo ~ "/"
Z
~21: 22
o
21 9 ~ "~1''I~' ~ 3 PT
g 9
"~e LOADING
~ 2o
u..
19 I I I I I
0.1 0.2 0.3 0.4 0.5 0.6
CRACK LENGTH (a), INCHES
F I G . 9--Variation of measured fracture toughness with crack length, 2024-7"851
aluminum alloy.
difference was noted between three- and four-point loading while in the
case of the aluminum, four-point loading provided slightly higher fracture
toughness numbers. The effect of variations in crack length on fracture
toughness, presented in Fig. 8, indicates that the K value for the high-strength
D6AC steel tended to show a slight decrease when the crack length divided
by the (K/~ys) 2 ratio decreased to below approximately unity. In these tests,
the thickness (B) was greater than four times the (K/eye) ~ ratio so that the
change in the observed K value could be associated directly with the crack
length.
STEIGERWALD ON CRACK TOUGHNESS MEASUREMENTS 1 13
The results of the notch bend tests indicate that the recommended ASTM
criterion which specifies that the crack size be 2.5 times the (K/ffys) 2 ratio is
probably a slightly conservative requirement. In cases where sufficient
material is available to allow a W/B ratio of 2, the recommended ASTM
crack size can be easily obtained, and, therefore, there is no real problem in
applying the crack size requirement. When square cross section specimens
must be used however, the results indicate that data obtained on specimens
where the crack size is approximately equal to the (K/~y~)~ ratio will provide
fracture toughness values which probably deviate by no more than 5 percent
from the K~ c obtained from recommended ASTM practice.
Kzc Data for Representative High-Strength Steels

The valid fracture toughness values available in the literature were as-
sembled to provide an indication of the variability of representative data
[4-9]. The format selected for presentation involved plotting the KI c for each
material as a function of either tensile strength or test temperature. In the
latter case, the fracture toughness is shown as a ratio of the K I e at the specific
test temperature to the K~c at room temperature. The fracture toughness
values presented in this manner are given in Figs. l0 through 16 for four
classes of high-strength steels.
The data for the low alloy steels (Fig. 10) consisted mostly of results
obtained with 4340 steel. The curve has been drawn to indicate a slight mini-
mum in K~r in the strength region which corresponds to irreversible "500 F "
temper embrittlement. This representation was selected by giving less con-
sideration to the consistently low set of data reported in Ref 9. Above a
tensile strength of approximately 230 ksi, the K~ o increased relatively rapidly.
The variation in fracture toughness with temperature for the 4340 steel, heat
treated to the 260 ksi tensile strength level, is shown in Fig. 11. Although the
data are limited, this curve characterizes with reasonable accuracy the varia-
tion in K I c which also occurs in the 4140 and 4340 steels heat treated to lower
strength levels [4].
In reviewing the influence of temperature on Kic the question arises as to
whether the decrease of toughness which occurs in a given steel with decreasing
test temperature results simply from the attendant increase in tensile strength.
To examine this question, data were evaluated from 4340 steel heat treated
to room-temperature tensiIe strength levels of either 265 or 210 ksi and then
tested at temperatures between 200 and - 1 0 0 F. The K~c values obtained
from the specific 4340 material tested over a range of temperatures were
plotted as a function of tensile strength level. In this case the variations in
strength resulted from test temperature rather than microstructural differ-
ences due to heat treatment. The results, indicated in Fig. 12 by data points,
are compared with the previous strength level dependence shown in Fig. 10
140 r ] I
120
g
1 O0
~ .~ 80
u
u-
Z
< 6O
a_
40 9 |
20 I I I I I
180 200 220 240 260 280
TENSILE STRENGTH, Ksl
FIG. lO--Variation of KI~ with strength level, low alloy martensitie steels (4340, 4140),
room-temperature tests.
(solid line). At the lower strength levels the temperature dependence of

K: c could be predicted directly from the strength level dependence; however,
at the higher strength levels the correlation was poor. This could be possibly
due to the presence of irreversible (500 F) temper embrittlement in steels
heat treated to that particular strength level range.
The variation in K i t as a function of strength level is shown in Fig. 13 for
the 5Cr-Mo-V steels. The data obtained on sheet, bar, and plate were com-
bined since no definite difference appeared to exist as a function of material
form [5-9]. The variation in fracture toughness as a function of test temper-
ature (shown in Fig. 14) was comparable to that exhibited by the 4340 steel.
The variation of K:~ with strength level and temperature for the precipita-
tion-hardening stainless steels is summarized in Figs. 15 and 16. In the - 5 0
to + 7 5 F range, the decrease in K:c with decreasing temperature was greater
for the precipitation-hardening stainless steels than for the 4340 or 5Cr-Mo-V
steels [4].
The comparison of the fracture toughness as a function of the various
steel classes is shown in Fig. 17. Above 240 ksi, the 4340 steel type had a
toughness which was significantly greater than the 5Cr-Mo-V steels; however,
STEIGERWALD ON CRACK TOUGHNESS MEASUREMENTS 1 15
120 l I I I
110
i--
< TO0
0
0 80
~" 60
U
40
2O I I I I
- 100 0 100 200 300
TESTTEMPERATURE,OF
FIG. 11--Effect o f temperature on the plane strain fracture toughness (Kl~) o f 4340
steel, 260 ksi tensile strength at room temperature.
at the lower strength levels, the two steels possessed comparable properties.
The precipitation-hardening stainless steels had K~c values below those of
the other steels examined.
Although the results summarized in Figs. 10 through 16 provide representa-
tive fracture toughness values which can aid in designing and selecting
material conditions to preclude brittle fracture, the inherent variability in
the K~c value as a function of heat of material must receive adequate con-
sideration. The scatter between heats can be as great as 30 percent, and
the material employed for a specific design should be evaluated prior to use
to ensure that abnormally low fracture toughness does not exist.
Summary and Conclusions

Fracture toughness tests conducted on center-precracked sheet specimens
and precracked bend specimens were evaluated with respect to the influence
that deviations from recommended ASTM practice had on the fracture
toughness parameter. In the case of sheet specimens, the tendency for the
crack growth curve to show discontinuous crack extension at the onset (pop-
in) could be correlated directly to the relationship of the test temperature to
1]6 PLANE STRAIN FRACTURE TOUGHNESS TESTING
140 I I I I t
120
1 ~~RoTREND LINE
~-~ 1O0 OM TEMP. TESTS
t~
8o
\
za:
o
o\o
ta 60 o~
7-
u_
\
4o
o 4340 (-lO0 TO +200~ TESTS)5000F TEMPER
%o
0%0
4340 (-I00 TO +200~ TESTS)800~ TEMPER ~,o
o~
20 I I I I I
180 200 220 240 260 280
TENSILE STRENGTH - KSI
F I G . 12--Variation in Ki~ as influenced by tensile strength changes produced by heat

treatment and test temperature.
lO0
i r A N D PLATEi
9 SHEET
a F O R G I N G (8")
:~ 80
u
Z 60
z
oi---
<
Z
<
~ 20
<
g_
I I I I
200 220 240 260 280 300
TENSILE STRENGTH, Ksi
FIG. 13--Variation o f Klc with strength level, 5Cr-Mo-V steel, room-temperature tests.
STEIGERWALD O N CRACK T O U G H N E S S MEASUREMENTS 1 17
180
i , , r I
160
140
120
F-
o t~
ne
~ 8o
60
40
20 I I I I
- I CO 0 I O0 200 300
TEST TEMPERATURE, OF
F I G . 14---Effect of temperature on the plane strain fracture toughness (K/,) of
5Cr-Mo- V steel.
the brittle-to-ductile transition temperature of the material. When the test

temperature was below the transition temperature no slow crack growth
occurred, and the initiation of cracking was coincident with total specimen
failure. In the transition region (mixed mode failure), slow crack growth
occurred, and the cracking initiated in a discontinuous manner. The magni-
tude of this discontinuity (pop-in) generally decreased as the percent shear in
the fracture surface increased. Above the transition temperature no sharp
discontinuity in the crack growth was present, and meaningful fracture
toughness values were not obtained. When the ASTM recommended criterion
for specimen thickness was satisfied a sharp discontinuity was always ap-
parent in the crack growth curve for high-strength steels, and no ambiguity
existed in determining a load value for calculating KI o.
GO
160 I I ; I
140 I I I I
z
9 17-7 PH
13 PH 15-7 Mo
I ~Q
140
o A M 355 Z
120
).
u
12o C
100
C
o O
O "I-
o ~
o 1oo Z
80
I.-
u
~
u_
8o
J
z
60 O
5
40
60
~
40 I I I I
20 I I I I
-200 - 1oo 0 1oo 200 300
160 180 200 220 240 260
TENSILE STRENGTH~ Ksl TEST TEMPERATURE, OF
F I G . 15--l,'ariation in room-temperature fracture toughness F I G . 16~Effeet of temperature on the plane strain fracture
with strength level for precipitation-hardening steels (17-7PH, P H toughness (Kz,) of 17-7PH, 15-7Mo, and A M 355 precipitation
15-7Mo, and A M 355) in plate form, room-temperature tests. hardening stainless steel.
STEtGERWALD ON CRACK TOUGHNESS MEASUREMENTS 1 19
1,10 I - I I I I
120
._ % j ALLOY STEELS 4340
100
r
i 80 -Mo-V STEELS
~ 60
>..
PRECIPITATION-HARDEN I N G ~ "~
40 STAINLESS STEELS
20 I I I I I
180 200 220 240 260 280 300
TENSILE STRENGTH, Ksi
FIG. 17--Variation of fracture toughness with strength level for various steel types.
A review of tests performed with precracked bend specimens on one high-

strength steel and one aluminum alloy indicated that variations in K~o of
less than 5 percent occurred at crack sizes that were equal to (K/aye) 2 provided
that the ASTM recommendation for thickness was satisfied.
The available KIo data for high-strength steels were presented graphically
to show typical variations as a function of both smooth tensile strength level
and test temperature.
References
[1] Boyle, R. W., Sullivan, A. M., and Krafft, J. M., "Determination of Plane Strain
Fracture Toughness with Sharply Notched Sheets," Welding Journal, Vol. 41, No. 9,
Sept. 1962, p. 428-S.
[2] Steigerwald, E. A. and Hanna, G. L., "Initiation of Slow Crack Propagation in High-
Strength Materials," Proceedings, American Society for Testing and Materials, Vol. 62,
1962, p. 885.
[3] Steigerwald, E. A. and Hanna, G. L., "Influence of Test Conditions on the Determina-
tion of Plane Strain Fracture Toughness from Sheet Specimens," Proceedings, Ameri-
can Society for Testing and Materials, Vol. 64, 1964, p. 1128.
[4] Steigerwald, E. A., "Plane Strain Fracture Toughness of High-Strength Materials,"
to be published in Journal of Engineering Mechanics, 1969.
[5] Wessel, E. T., Clark, W. G., and Wilson, W. K., "Engineering Methods for the Design
and Selection of Materials Against Fracture," ATAC Contract DA-304)69-AMC-602
(T), Final Report, 24 June 1966.
[6] Hanna, G. L. and Steigerwald, E. A., "Fracture Characteristics of Structural Metals,'"
AD 411509, ER-5426, June 1963.
[7] Amateau, M. F. and Steigerwald, E. A., "Fracture Characteristics of Structural
Metals," AD 611873, Jan. 1965.
[8] Jones, M. H., Fisher, D. M., and Brown, W. F., Jr., "Progress Report on NASA-NRL
Cooperative Fracture Testing Program," ASTM E-24, American Society for Testing
and Materials, Jan. 1967.
[9] Backofen, W. A. and Ebner, M. L., "Metallurgical Aspects of Fracture at High-
Strength Levels," AD 406167, May 1963.
STP463-EB/Sep. 1970
DISCUSSION
IV. F. Brown, Jr., 1 and M. H. Jones 1 (written discussion)--Some of the data

reported by Dr. Steigerwald were obtained apparently before ASTM Com-
mittee E-24 formulated the present Proposed Method of Test for Plane Strain
Fracture Toughness of Metallic Materials, and represent tests using different
types of instrumentation than specified in the ASTM Committee E-24
Method. We wonder if these differences might affect the measured toughness
values. Thus, the apparent plane strain fracture toughness results for H-I 1
shown in Figs. I through 5 correspond to the load required for the "initia-
tion" of crack growth as indicated by the appearance of a discontinuity
(pop-in step) in a record of load versus change in electrical resistance of the
specimen or a deviation of this record from linearity. The electrical resistance
should be much less sensitive to plastic flow than the crack mouth displace-
ment measurement specified in the ASTM Committee E-24 Method. Fur-
thermore, this Method specifies a fixed percentage of apparent crack extension
as a basis for selecting the load used to calculate Ko. Strictly speaking, we
would expect agreement between fracture toughness values obtained by the
ASTM Committee E-24 Method and Dr. Steigerwald's electrical resistance
technique only for very brittle behavior (for example, the subzero test records
in Figs. 2 and 4 of the author's paper).
Dr. Steigerwald suggests that the present crack length and thickness re-
quirements of the ASTM Committee E-24 Method could be reduced. A
modification of the thickness requirement is based on the data for the H-11
steel. For the reasons mentioned above we would be very cautious regarding
any implications of these data concerning specimen size requirements as they
relate to the present ASTM Committee E-24 Method. The results obtained
for D6ac and 2024T851 aluminum alloy, Figs. 8 and 9, are cited as evidence
supporting a reduction in the crack length requirement. While the total varia-
tion in fracture toughness values was small for the aluminum alloy, these
data appear to support the present crack length requirement. The D6ac steel
data scatter rather badly with the extreme values being approximately •
percent of an overall mean of about 60 ksi %/i~. This can be compared with
the spread of about • percent for valid K~c values of 4340 steel reported by
1 Chief and research engineer, respectively, Strength of Materials Branch, N A S A Lewis
Research Center, Cleveland, Ohio 44135.
121
Copyright9 1970 by ASTM lntcrnational www.astm.org
Heyer and McCabe 2 from results of the ASTM Committee E-24 round-robin
program. In our opinion the trend line shown by the author in Fig. 8 could
just as well have been drawn with a positive slope and the data used to sup-
port an argument for increasing the crack length requirement rather than
reducing it.
While we would agree that the size requirements will vary with the material
fracture characteristics, it is necessary to formulate them in such a way that
applicability to a wide variety of materials is ensured. With this in mind we
have always believed that both the crack length requirement and the thickness
requirement should be increased, not reduced. This belief appears to be
supported by the data for Hylite 50 reported by May?
We would appreciate the author's comments on his reason for comparing
the various steels in terms of K ~ on the basis of tensile strength rather than
yield strength. The order of rating would in many cases be different depend-
ing on which of these two bases were used. More specifically, it is not clear
to us how the strength potential of a metallic alloy can very much exceed
the yield strength in normal structural applications.
E. A. Steigerwald (author's closure)--The four points raised by Messrs.

Brown and Jones are:
1. The validity of the K~, values obtained from resistance measurements on
sheet specimens.
2. The validity of the ASTM criteria for crack length and specimen thick-
ness.
3. The variation in apparent fracture toughness that occurs by relaxing the
ASTM crack length criterion.
4. The reason for plotting K~ c as a function of tensile strength rather than
a function of yield strength.
I will try to answer each of these points in order. Heat tinting methods
indicated that the resistance technique had the ability to detect crack ex-
tensions of the order of 0.005 in. On this basis most of the fracture toughness
values in the transition range did conform to the ASTM Committee E-24
criterion, but those above did not. Therefore, the K values listed as "Plane
Strain Fracture Toughness" on the right hand ordinate of Figs. 1 and 3
should be more accurately termed "Fracture Toughness" to avoid the indica-
tion that they represent valid plane strain values over the entire test tempera-
ture range. This correction, however, does not alter the basic purpose for
presenting the sheet data which was to show a correlation between the nature
of the crack growth initiation (pop-in) and the fracture mode transition in
sheet materials.
2 S e e p . 22.
3 See p. 42.
DISCUSSION ON CRACK TOUGHNESS MEASUREMENTS 123
I do not recommend that the ASTM criteria for either thickness or crack
length be reduced. The recommended practice should be applicable to a
wide range of materials and not relaxed. I am in complete agreement with
Messrs. Brown and Jones on this point which is really the basic philosophy
presented in the discussion. The purpose of indicating that in some materials
the K value is not significantly altered by crack lengths in the range between
1.0(K/ays)2 and 2.5(K/~ys)2 is to provide a proper perspective to engineers
who often do not have the luxury of test data on very large specimens. In fact
the data shown in Fig. 6, aluminum alloy D T D 5074, and Fig. 8, maraging
steel, of May's paper 3 is in complete agreement with the conclusion obtained
in my paper from the 2024 tests concerning crack size effects. May's data on
Hylite 50 cited by Brown and Jones also is relatively insensitive to crack size
effects. In general, however, as May points out, the results for Hylite 50 are
anomalous and do not give constant K~o values when ASTM recommended
practice is followed.
The various steels are compared on the basis of tensile strength in hopes
of providing a more sensitive and perhaps more meaningful correlation with
fracture toughness. High-strength steels such as 4340 can be tempered over
a range of temperatures to provide a variety of structures. Tensile strength
is very sensitive to tempering temperature while yield strength often is not
(for example, 4340 steel tempered in the 400 to 600 F range). The tensile
strength was then selected because it was believed to provide a parameter
which was more discriminating than yield strength and more meaningful from
a metallurgical standpoint.
S. T. Rolfe 1 and S. R . N o v a k 2
Slow-Bend K,0 Testing of Medium-Strength

High-Toughness Steels
REFERENCE: Rolfe, S. T. and Novak, S. R., "Slow-Bend KIe Testing of Medium-

Strength High-Toughness Steels," Review of Developments in Plane Strain Fracture
Toughness Testing, ASTM STP 463, American Society for Testing and Materials,
1970, pp. 124-159.
ABSTRACT: Linear-elastic fracture mechanics (LEFM) has been used extensively

to analyze the fracture behavior of steels having yield strengths greater than 200
ksi. Although theoretically L E F M can be used to analyze the behavior of all steels,
it has not been used extensively to analyze the behavior of medium-strength (and
lower-strength) high-toughness steels, because large plastic-zone sizes develop
prior to fracture of these steels and test results are therefore difficult to analyze.
Therefore, to establish the limitations of the LEFM approach to the analysis of
medium-strength steels, a series of slow-bend Kt~ tests were conducted on various
steels, with different levels of toughness.
The present paper describes the test procedures and analyses used in slow-bend
K~c tests of steels having yield strengths from 110 to 246 ksi and Charpy V-notch
impact energy absorption ( + 8 0 F) in the range 16 to 89 ft.lb. The range of K~
values for various pop-in criteria is presented as well as a discussion of the various
factors affecting the slow-bend Ktc testing of medium-strength high-toughness
steels.
The results indicate that L E F M can be used to analyze the fracture behavior of
medium-strength high-toughness steels. K~o values ranging from 87 to 246 ksi in.
were obtained for the eleven steels investigated. Although the specimen sizes tested
generally 2 in. thick) did not satisfy completely the recommended ASTM Com-
mittee E-24 size requirements (in some cases, the suggested sepcimen thicknesses
were greater than 6 in.), the K~c values are considered to be representative on the
basis of general strength and Charpy V-notch toughness levels and expected
service performance. Therefore, the measured K~ values are estimated to be
reliable for general engineering use, although some of the specimen sizes did not
satisfy the current ASTM Committee E-24 requirements completely.
When Ktc and Charpy V-notch energy-absorption values were normalized on
the basis of yield strength, an empirical correlation was established over a wide
range of yield strengths. Although such a correlation increases the confidence in
1 Professor of civil engineering, University of Kansas, Lawrence, Kans. 66044; formerly

division chief, Mechanical Behavior of Metals Division, U.S. Steel Applied Research
Laboratory, Monroeville, Pa. 15146. Personal member ASTM.
Research engineer, Mechanical Behavior of Metals Division, U.S. Steel Applied Re-
search Laboratory, Monroeville, Pa. 15146. Personal member ASTM.
124
Copyright* 1970 by ASTM International www.astm.org
ROLFE AND NOVAK ON SLOW-BEND KI~ TESTING 125
both types of approaches to fracture analysis, the use of Charpy tests to predict
KI. values is not recommended until additional test and service experience is
obtained.
KEY WORDS: plane strain, fractures (materials), toughness, fracture strength,
steels, Charpy test evaluation, tests
Linear-elastic fracture mechanics (LEFM) has been used extensively to

analyze the fracture behavior of steels having yield strengths greater than
200 ksi [1,2]. 3 Although theoretically L E F M can be used to analyze the
behavior of all steels, it has not been used extensively to analyze the behavior
of medium- and low-strength (yield strength below 200 ksi) high-toughness
steels, because large plastic-zone sizes develop prior to fracture of these steels
and the test results are therefore difficult to analyze. However, because of the
increasing interest in high-toughness steels having yield strengths less than
200 ksi, and because fracture mechanics can be used to characterize the frac-
ture behavior of steels in terms of stress and flaw size (measured as a critical
stress-intensity factor, K~c), information about the KIo behavior of medium-
strength high-toughness steels is required.
The accuracy of LEFM is good provided that the plastic zone at the tip
of the crack is small compared with the general specimen dimensions [3].
As the ratio of plastic-zone size to specimen size increases, L E F M becomes
less accurate. However, from an engineering viewpoint, this decrease in
accuracy may not be sufficient to prohibit the use of L E F M for medium-
strength high-toughness steels. To establish the limitations of the L E F M
approach to the analysis of medium-strength steels, a series of slow-bend
KI c tests was conducted on various steels with different levels of toughness.
For comparison, tests also were conducted on a 246-ksi yield-strength steel
that was expected to exhibit little plasticity prior to fracture.
The present report describes the test procedures and analyses used in slow-
bend K~c tests of steels having yield strengths from 110 to 246 ksi and Charpy
V-notch impact energy absorptions (-t-80 F) in the range 16 to 89 ft. lb. The
range of K~, values for various pop-in criteria is presented as well as a dis-
cussion of the various factors affecting the slow-bend K~ o testing of medium-
strength high-toughness steels.
Materials and Experimental Procedure
Materials
The chemical composition of the eleven steels studied is shown in Table l,
and the tensile and impact properties are shown in Table 2. The steels had
TABLE 1--Composition of steels investigated, percent (check analysis).
Steel and Heat

Melting Practice No. C Mn P S Si Ni Cr
A517F A M . . . . 73A377a 0.16 0.79 0.01 0.016 0.23 0.87 0.53

4147 A M . . . . . . V1612 0.49 0.98 0.011 0.028 0.29 ND 1.05
HY-130(T) A M . X53588 0.120 0.91 0.008 0.005 0.22 5.26 0.62
4130 A M . . . . . . V1622 0.32 0.72 0.020 0.022 0.32 ND 0.97
12Ni-5Cr-
3Mo A M . . . . . . X14689 0.023 0.088 0.004 0.008 0.094 12.10 5.21
12Ni-5Cr-
3Mo VM . . . . . . 50169 0.002 0.019 0.004 0.005 0.042 12.10 5.03
12Ni-5Cr-
3Mo VM . . . . . . 50250 0.002 0.018 0.004 0.005 0.048 12.20 5.07
18Ni-8Co-3Mo
(180 Grade) A M X53914 0.014 0.032 0.005 0.012 0.022 18.30 ND
18Ni-8Co-3Mo
(180 Grade) A M X53014 0.014 0.032 0.005 0.012 0.022 18.30 ND
18Ni-8Co-3Mo
(180 Grade) VM 50446 0.003 <0.02 0.003 0.007 0.005 17.90 ND
18Ni-8Co-5Mo
(250 Grade) VM 50447 0.003 <0.002 0.001 0.004 0.003 17.10 ND
T A B L E 1--Continued.
Steel and
Melting Practice Mo Co Ti AI b A1c B N O
A517F A M . . . . 0.43 0.26 a ND ~ ND 0.031 0.003 ND ND

4147 A M . . . . . . 0.19 ND ND ND 0.025 ND 0.006 ND
HY-130(T) A M . 0.59 ND ND ND 0.010 0.014 0.011 0.003
4130 A M . . . . . . 0.20 ND ND ND 0.042 ND 0.012 ND
12Ni-5Cr-
3Mo A M . . . . . . 2.86 ND 0.24 0.38 ND 0.0037 0.009 0.001
12Ni-5Cr-
3Mo VM . . . . . . 2.90 ND 0.21 0.22 ND ND 0.008 0.0005
12Ni-5Cr-
3Mo VM . . . . . . 3.12 ND 0.21 0.25 ND ND 0.008 0.13004
18Ni-8Co-3Mo
(180 Grade) A M 2.60 8.09 0.15 0.18 ND 0.0031 0.007 0.002
18Ni-8Co-3Mo
(180 Grade) A M 2.60 8.09 0.15 0.18 ND 0.0031 0.007 0.002
18Ni-8Co-3Mo
(180 Grade) VM 2.96 7.73 0.20 0.007 ND ND 0.004 0.0012
18Ni-8Co-5Mo
(250 Grade) VM 4.65 7.60 0.50 0.052 ND ND 0.003 ND
Ladle analysis, d Copper.

b Acid soluble. ~ N D means not determined.
Total.
TABLE 2--Mechanical properties of steels investigated.
Charpy
Yield Elongation in 1 In., 7o Reduction of Area, ~o V-Notch
Strength Energy
Plate (0.2 ~ Tensile At At Absorption
Thickness, offset), Strength, Maximum At Maximum At at +80 F,
Steel and Melting Practice in. ksi ksi Load Fracture Load Fracture ft. Ib
A517F AM a . . . . . . . . . . . . . . . . . 2 110 121 8.0 20.0 11.3 66.0 62 >

4147 AM . . . . . . . . . . . . . . . . . . . . ~ 137 154 7.0 15.0 8.7 49.0 26 z
o
HY-130(T) AM . . . . . . . . . . . . . . 2 149 159 7.0 20.0 9.6 68.4 89 z
4130 AM . . . . . . . . . . . . . . . . . . . . ~ 158 167 7.0 14.0 7.9 49.2 23 o<
12Ni-5Cr-3Mo AM . . . . . . . . . . . 2 175 181 3.0 14.0 3.0 62.2 32 >
12Ni-5Cr-3Mo VM b. . . . . . . . . . 1 183 191 4.5 15.0 6.9 61.2 60 O
12Ni-5Cr-3Mo VM . . . . . . . . . . . 2 186 192 5.9 17.0 9.3 67.1 65 z
18Ni-SCo-3Mo (180 Grade) AM 2 193 200 3.0 12.5 4.1 48.4 25
0
18Ni-8Co-3Mo (180 Grade) AM 2 190 196 3.0 12.0 3.2 53.7 25
18Ni-8Co-3Mo (180 Grade) VM 2 187 195 3.0 15.0 4.2 65.7 49
18Ni-8Co-5Mo (250 Grade) VM 2 246 257 2.2 11.5 3.4 53.9 16 z
AM signifies an air-melted basic-electric-furnace steel.

b VM signifies a vacuum-induction-melted steel.
,,4
r P/2
ST LI_
. L+A
$8
SECTION A-A
FIG. l--Four-point slow-bend Kic specimen.
yield strengths ranging from 110 to 246 ksi and Charpy V-notch impact-
energy absorption values at + 8 0 F ranging from 16 to 89 ft.lb. All steel
plates were obtained from production heats and were rolled and heat-treated
on production facilities.
Specimen Geometry and Analysis

A four-point slow-bend test specimen, Fig. 1, was used to evaluate the
eleven steels. Where possible, the general size requirements currently being
recommended by ASTM Committee E-24 on Fracture Testing of Metals
were followed in selecting the specimen dimensions [4]. These size require-
ments are as follows:
Crack depth, in. = a > 2.5 ( K I e ) 2

- - . . . . . . . . . . . . . . . . . . (1)
\ ayl
Specimen thickness, in. = B > 2.5 ( K i c y - - . . . . . . . . . . . (2)

\~y /
Specimen depth, in. = W _-> 5.0 (Kaoy - - . . . . . . . . . . . . . . (3)
kay /
Even for specimen thicknesses of 2 in. and specimen depths up to 8 in., these
requirements were not always satisfied for some of the lower strength steels.
The specific sizes of each specimen tested are listed in the Appendix, Tables
6 through 16.
A modified double-cantilever-beam displacement gage identical in principle
to that designed by Srawley [5] was used to measure crack-opening-displace-
ment during testing. The gage consisted of two cantilever beams and a
ROLFE AND NOVAK ON StOW-BEND KIe TESTING 129
FIG. 2--Double-cantilever-beam displacement gage.
spacer block with four strain gages, and was mounted on the specimen, as
shown in Fig. 2. Load data from the testing machine and data from the dis-
placement gage were fed into an X - Y recorder to obtain a continuous load/
crack-opening-displacement (P-A) record. In addition, a load-deflectometer
record was obtained as a "backup" measurement. Figure 3 shows the overall
test setup, including the loading pins that were employed to minimize friction.
The general procedures used in the present study were as follows:
1. Heat-treat material and machine specimens, including face notches if
used.
2. Fatigue-crack at nominal stress less than 25 percent of yield stress.
Propagate fatigue crack approximately 0.05 W. Nominal stress and number
of cycles for each specimen are shown in the Appendix, Tables 6 through 16.
FIG. 3--Four-point slow-bend Klc test setup.
3. Instrument specimen (see Fig. 3) and perform slow-bend test. Obtain

P-A record.
4. Measure specimen geometry (including fatigue-crack depth) and analyze
P-A record by using various criteria to be discussed subsequently. Obtain
load for each criterion used (for example, pop-in, secant intercept, maximum
load, etc.).
5. Calculate Kxc for each criterion by using an equation developed by
Srawley and Brown [2]:
KIo - 6Mal/2[
B W2 1.992 -- 2.468 (W) q- 12.97 (W) 2 -- 23.17 (W) 3
q- 24.80 (;)41 . . . . (4)
where:
M = Maximum moment in four-point bending (M = X L: where LP/2
is the distance between support and nearest load point, L = S~
Sr/2
Appendix, Tables 6 through 18);
a = total crack length, including fatigue crack;
B = specimen thickness; and
W = specimen depth.
ROLFE AND NOVAK ON SLOW-BEND KIe TESTING 1 3"J
To obtain KIo values when face-notched specimens are used, Freed and
Krafft [6] have suggested that the KI ~ value (calculated by using B) should be
multiplied by the factor (B/B,3", where m may range from 1/~ to 1 depending
on the anisotropy of the material and the depth of the face notches. For
isotropic material behavior and the shallow face notches used in this investi-
gation, m approaches ~ and thus B is replaced by (B B,) a/2 in the above
equation. Tests of smooth (not face-notched) and face-notched specimens
have shown [7] that this correction yields results essentially equal to those
obtained from smooth-face specimens as long as the face notches are not too
deep.
6. Calculate indexes of reliability based on specimen geometry. Using the
current ASTM Committee E-24 criterion, the following minimum values
should yield valid results:
a
> 2.5
(KIo~
~--_
\fly I
B
> 2.5
( '_~_, o ~ =
kay /
W
> 5.0
(K
_ _, o ~ --
kay /
Results and Discussion
Kit Values
The detailed results of 46 KI~ tests on the eleven different steels are pre-
sented in the Appendix, Tables 6 through 16. The specimen sizes tested
ranged from about one third of the suggested sizes (based on the current
ASTM Committee E-24 criterion) to about six times the suggested sizes.
The pop-in records ranged from almost ideal LEFM behavior (very small
deviation from linearity prior to complete fracture) to behavior evidencing
considerable plasticity ("roundhouse" load-displacement curves). Nonethe-
less, the K1 c values obtained from the various criteria selected for use in the
present study represent values that generally would be expected for these
steels on the basis of their general strength and Charpy V-notch toughness
levels and their known service performance. Thus, although the "best
estimate" values of K~c that will be described subsequently do not completely
satisfy the ASTM Committee E-24 criterion, they are felt to represent realistic
s%
10%
//
LARGE POP-IN
>> 0.05 OFFSET
TYPE I TYPE 2 TYPE 3
5%
I0%
oo
% SMALL POP-IN
0 0.05 OFFSET
.J
ABRUPT CHANGE IN
SLOPE ~ 3 5 %
TYPE 4 TYPE 5 TYPE 6
DISPLACEMENT {~)
FIG. 4--Types of load-displacement records obtained.
engineering values. The question of whether these values are within ! 2 or

~ 1 0 percent of the true K~c values cannot be answered until larger size
specimens with the same metallurgical structure are tested, or until additional
service experience with these steels is obtained.
The various p - A records were classified into six types, as shown schemat-
ically in Fig. 4. Examples of the five types that exhibited nonlinearity are
shown in Figs. 5 through 9, and fracture surfaces corresponding to these
records are shown in Figs. 10 through 12. Note that all fractures were macro-
scopically fiat. As can be observed from these records, different criteria can
be used to define pop-in. The Krc and nominal stress values selected by using
different criteria are presented in the Appendix, Tables 6 through 16. In
addition, indexes of reliability, based on ASTM Committee E-24 recom-
mended specimen sizes, were calculated for each of the criteria selected.
These indexes of reliability provide a quantitative measurement of the validity
of Kz c values based on current size requirements.
K~c values of 87 to 246 ksi V / ~ . were obtained for the eleven steels in-
vestigated, and the results are summarized in Table 3. The range of Kzc
values selected by using the various criteria is shown, as well as a "best
estimate" of K~c for each steel. The best estimate Kx ~ values were based on
results for specimens that most closely satisfied the ASTM Committee E-24
size requirements or were based on a pop-in value that usually occurred at
about the 10 percent secant offset.
ROLFE AND NOVAK ON SLOW-BEND KIe TESTING 133
TABLE 3--Summary of K1, results.

Charpy
V-Notch
Energy Kxo, ksi ~v/i-m.
Yield Strength Tensile Absorption
Steel and (0.2 ~o offset), Strength, at +80 F, Best
Melting Practice ksi ksi ft .lb Range Estimate
A517F AM . . . . . . . . . . . . 110 121 62 160 to 177 170

4147 AM . . . . . . . . . . . . . . 137 154 26 91 to 121 109
HY-130(T) AM . . . . . . . . 149 159 89 238 to 279 246
4130 AM . . . . . . . . . . . . . . 158 167 23 84 to 109 100
12Ni-5Cr-3Mo AM . . . . . 175 181 32 125 to 137 130
12Ni-5Cr-3Mo VM . . . . . 183 191 60 211 to 230 220
12Ni-5Cr-3Mo VM . . . . . 186 192 65 207 to 233 226
18Ni-8Co-3Mo
(180 Grade) AM . . . . . . . . 193 200 25 100 to 110 105
18Ni-8Co-3Mo
(180 Grade) AM . . . . . . . . 190 196 25 110 to 113 112
18Ni-8Co-3Mo
(180 Grade) VM . . . . . . . . 187 195 49 152 to 161 160
18Ni-8Co-5Mo
(250 Grade) VM . . . . . . . . 246 257 16 86 to 87 87
Pop-In Criteria
E-24 5-Percent Secant Intercept--In a recent d r a f t [4] o f " R e c o m m e n d e d
Practice for P l a n e - S t r a i n F r a c t u r e T o u g h n e s s Testing o f H i g h - S t r e n g t h
M e t a l l i c M a t e r i a l s U s i n g a F a t i g u e - C r a c k e d Bend S p e c i m e n , " by S u b c o m m i t -
tee I o f A S T M C o m m i t t e e E-24, a p r o c e d u r e to calculate K~c on the basis o f
a secant intercept o f 5 percent less t h a n the slope o f the initial p o r t i o n o f the
P - ~ curve is described. Because m a n y m a t e r i a l s d o n o t exhibit pop-in, this
p r o c e d u r e was established to define K~c as the stress intensity at which the
c r a c k reaches an effective length 2 p e r c e n t greater t h a n t h a t at the beginning
o f the test. A s p a r t o f this p r o c e d u r e [4], the l o a d at the intersection o f the
5 percent secant line a n d the original p - A line is determined. T h e h o r i z o n t a l
d e v i a t i o n f r o m the original P-~X line is m e a s u r e d at this l o a d (designated as
PG) a n d called APG. T h e m a j o r p a r t o f the deviation at P~ is s u p p o s e d to result
f r o m a c t u a l c r a c k extension r a t h e r t h a n from plastic d e f o r m a t i o n a r o u n d
the c r a c k tip. Briefly, the p r o c e d u r e consists o f m e a s u r i n g the d e v i a t i o n
(0.80/)5) at some lower load, P = 0.80/'5. F o r valid results this deviation
should be less t h a n one f o u r t h the value at PG.
Values o f PG, 0.80/5, ~PG, A0.80P~, a n d also P10 a n d ~P10 (the l o a d a n d
deflection at a l0 percent secant intercept) are t a b u l a t e d in T a b l e 4 for all
specimens tested. A l l b u t seven ( o u t o f 46) test results satisfied the c o n d i t i o n
t h a t A0.80P~/zX P5 be less t h a n one fourth. A s expected, these seven specimens
were ones which deviated f r o m the r e c o m m e n d e d sizes.
#x
z
T A B L E 4--Analysis o f P-A records using secant intercept method.
A0.80P5
Specimen
Steel and Melting Practice No. Plo, lb P~, lb 0.80P5, lb. APlo, in. APs, in. A0.8OP~, in. AP5
A517F A M . . . . . . . . . . . . . . . . . . T1K-1 56 500 50 000 40 000 0.0042 0.0018 0.0006 0.3333

T1K-2 52 900 47 800 38 200 0.0039 0.0018 0.00048 0.2665
T1K-7 130 000 122 000 100 000 0.0040 0.0020 0.00020 0.100
4147 AM . . . . . . . . . . . . . . . . . . . . 47-1 6 650 6 075 4 860 0.0012 0.0006 0.00012 0.200
47-2 6 650 6 100 4 880 0.0012 0.0004 0 0
47-3 7 320 6 890 5 510 0.0013 0.0006 0 0
47-5 3 160 2 800 2 240 0.0028 0.0012 0.00029 0.242
HY-130(T) A M . . . . . . . . . . . . . . 5NKF-2 199 000 175 500 140 500 0.0046 0.0022 0.00085 0.3860
5NKF-3 171 000 154 500 123 500 0.0040 0.0020 0.00005 0.0250
5NKF-5 122 600 103 300 82 700 0.0025 0.0010 0.00024 0.2400
4130 AM . . . . . . . . . . . . . . . . . . . . 30-1 4 480 4 300 3 440 0.0014 0.0006 0.00014 0.233
30-2 4 330 3 800 3 040 0.0014 0.0006 0.00024 0.400
30-5 4 540 4 320 3 460 0.0016 0.0008 0.00016 0.200
30-6 4 230 4 160 3 320 0.0014 0.0007 0.00002 0.0286
12Ni-5Cr-3Mo A M . . . . . . . . . . . 12NA-1 103 500 98 500 78 800 0.0027 0.0012 0 0
12NA-2 39 200 36 750 29 400 0.0021 0.0009 0 0
12Ni-5Cr-3Mo VM . . . . . . . . . . . 2-A 69 000 64 200 51 400 0.0014 0.0008 0 0
2-B 67 000 55 000 44 000 0.0017 0.0008 0.00026 0.3250
2-C 68 300 58 000 46 400 0.0012 0.0009 0.00032 0.3560
2-D 65 000 54 500 43 600 0.0020 0.0010 0.00030 0.3000
12Ni-5Cr-3Mo VM . . . . . . . . . . . 12NV-1 41 0O0 37 500 30 000 0.0025 0.0011 0.00009 0.0818
12NV-3 48 850 50 000 40 O00 0.0030 0.0014 0 0
12NV-4 72 500 67 000 53 600 0.0039 0.0018 0.00039 0.216
12NV-6 135 O00 129 750 103 900 0.0039 0.0018 0.00018 0.100
12NV-5 127 250 126 0O0 127 250 0.0040 0.0020 0.O0015 0.075
18Ni-8Co-3Mo (180 grade) AM 18NF-1 31 450 31 000 24 800 0.0022 0.0012 O 0
18NF-3 28 200 28 000 22 400 0.0025 0.0012 0 0
18NF-5 33 350 32 700 26 100 0.0019 0.0009 0.00009 0.1000
18NF-7 17 200 16 650 13 300 0.0024 0.0012 0.000135 0.1125
18NF-8 39 750 40 050 32 050 0.0022 0.0010 0 0
18NF-10 30 450 30 450 24 350 0.0020 0.0009 0.000135 0.1500
18NF-11 38 850 39 150 31 300 0.0021 0.0010 0 0
18N-2 36 800 37 0O0 29 600 0.0021 0.0009 0 0 8
18N-3 32 500 3l 500 25 200 0.0020 0.0009 0.00012 0.1335
18NF-12 37 100 37 100 29 600 0.0016 0.0008 0 0 z
t7
18Ni-8Co-3Mo (180 grade) AM DCBS-1 19 800 19 900 15 900 0.0018 0.0009 0 0 z
DCBS-2 22 900 22 950 18 350 0.0016 0.0009 0.00003 0.0333 O
<
DCBS-3 12 790 12 300 9 850 0.0013 0.0006 0.00009 0.1500
DCBS-4 13 700 13 200 10 600 0.0013 0.0006 0.0000375 0.0626 O
DCBS-5 24 200 23 600 18 800 0.0018 0.0009 0.000125 0.1390 Z
o~
DCBS-6 24 400 24 500 19 600 0.0018 0.0008 0 0
18Ni-8Co-3Mo (180 grade) VM 2O0-A 56 400 53 100 42 500 0.0031 0.0014 0.00033 0.236
2O0-B 44 950 44 O00 35 200 0.0083 0.0036 0 0
C7
18Ni-8Co-5Mo (250 grade) VM 250-1 34 0O0 34 000 27 200 0.0006 0.0003 0 0
250-2 19 250 19 050 15 200 0.0011 0.0004 0 0
24
I I I I ]/5*/. I I I I
= 22, 9 5 0 Ibs ~ . / ~ ) ~ . , ~ .
P5 = PMAX MAX LOAD
22
2C
16
.~ 12
I0
TYPICAL TYPE 2 RECORD

I B N i ( 1 9 0 ) MARAGING STEEL
SPECIMEN D C B S - 2
TABLE J
0 I 1 I I I I I I I
0 3 6 9 12 15 18 21 24 27 30
V ( C . 0 . D . ) , i n e h x 10 -3
F I G . 5--Reproduction of actual type 2 load-displacement record.
The purpose of this procedure is to ensure that the major portion of any
deviation from linearity is due to crack extension rather than to plasticity.
However, tests that were interrupted after a 5 or 10 percent secant intercept
was reached showed that the recommended procedure does not necessarily
ensure that crack extension occurs rather than plasticity. In these tests, the
specimens were loaded to a 5 or 10 percent intercept, unloaded, fatigue-
cracked, and retested to failure. In three specimens tested to between 5 and 10
percent deviation, no crack extension was observed even though the ratio of
2x 0.80Ps/A P5 was less than one fourth. A fourth specimen, loaded beyond
pop-in, exhibited an actual crack extension of about 0.05 in. compared with
ROLFE AND NOVAK ON SLOW-BEND KIc TESTING 137
a predicted crack extension (based on slope of the p-A record) of about 0.12
in. These results indicate that the deviation prior to pop-in in these tests was
apparently caused by plasticity, rather than by crack extension.
Other criteria--In addition to the 5 percent secant intercept criterion, a
pop-in step, an abrupt change in slope (greater than 35 percent), and a maxi-
mum-load criterion were used to calculate KIo values. These other criteria
were investigated because it appeared that for medium-strength steels pop-in
steps or abrupt changes in slope generally coincided more closely with a 10
percent secant intercept than with a 5 percent secant intercept. The particular
criterion used for a given series of specimens depended on the type of P-Ax
records obtained.
Analysis of the results of the bend tests, Appendix, Tables 6 through 16,
leads to the following conclusions regarding the other criteria used to select
K I e in medium-strength high-toughness steels:
1. All steels that had Kic/ae ratios of 0.60 or less generally exhibited pop-in
either before, or at, the 5 percent secant intercept. These steels all exhibited
Type 2 or 3 records, Fig. 4, and satisfied the current ASTM Committee E-24
minimum size requirements [4].
I I I I I I Y I
,5%
21-- 10%
PMAX = 19,250 Ibs

18--
15
.~ 12
Jr
9 TYPICAL TYPE 3 RECORD

18Ni (250) MARAGING STEEL
SPECIMEN NO. 2 5 0 - 2
TABLE L
or I I i I I I 1 I I I
o 3 6 9 12 15 18 21 24 27 30
V(C.0.D.) ,inch xl0 -3
F I G . 6--Reproduction of actual type 3 load-displacement record.
I I I I I I I I I
ALLOWABLE DEVIATIO
FROM LINEARITY
(OLD CRITERION )
I//io%
60
p = 5 8 , 2 0 0 Ibs
MAX MAX LOAD
ABRUPT CHANGE
IN SLOPE
54 P=o= 5 6 , 4 ~
P5 =53,100 --
48
42
Q.
36
~ 30
24
18 TYPICAL TYPE 4 RECORD

18Ni(180) MARAGING STEEL
SPECIMEN NO. 2 0 0 - A
12 TABLE K
oV' I I I I I I I I I I
0 6 12 18 24 30 36 42 48 54 60
V (C.O.D.), inch x l O - 3
FIG. 7--Reproduction ofactual type 4 load-displacement record.
2. All steels that had Kic/ay ratios greater than 0.60 exhibited pop-in
either before, or at, the 10 percent secant intercept. The only exception was a
l,in.-thick 12Ni-5Cr-3Mo VM steel, Table l l, which did exhibit an abrupt
change in slope but at a secant intercept greater than 10 percent. With two
exceptions, these steels did not satisfy the ASTM Committee E-24 size re-
quirements.
3. Two steels with Kic/~y ratios greater than 0.60 did satisfy the current
ASTM Committee E-24 mininmm size requirements, Tables 10 and 15, but
exhibited pop-in at the 10 percent secant intercept rather than at the 5 per-
cent secant intercept.
In general, a 10 percent secant intercept corresponded more closely with
pop-in or abrupt change in slope than did the 5 percent secant intercept,
particularly for the lower strength steels. For the highest strength steels, where
ROLFE AND NOVAK ON SLOW-BEND Kzc TESTING 139
pop-in occurred at a 5 percent secant intercept, the 10 percent secant inter-

cept obviously also corresponded to pop-in. Thus, although additional ex-
perimental (and analytical) work is required, the 10 percent secant intercept
may be a possible criterion for determining KIe values for medium-strength
high-toughness steels.
Specimen-Size Requirements
Although the primary purpose of this investigation was not to investigate
minimum specimen sizes required for plane strain (by systematically varying
the specimen geometry), some general observations on specimen size re-
quirements can be made. As shown in Tables 8 and 12, increasing W (in the
range less than the current ASTM Committee E-24 size requirements) in-
creases the apparent KI c. Low values of W suppress the K~ ~values [8] because
yielding occurs sooner than it would in specimens with larger values of W.
Comparison of Kic values obtained with small specimens [9] with values
I I I I I/.5o/. I I I I
90
/ /' / o ,.
PMAX =eo,15OIbs-7 ////// //,5"/o
I .20%
II///~ 25"/.
80
60-
-= 50-
40"
3C TYPICAL TYPE 5 RECORD

12Ni5Cr3MoMARAGING STEEL
20 SPECIMEN NO. 2B
TABLE F
IO
o I I I I I i I I 1
o 4 8 12 16 20 24 28 32 36 4C
V ( C . 0 . D . ) , inch x 10 - 3
FIG18--Reproduction of actual type 5 load-displacement record.
obtained in the present work on large specimens, Table 5, shows that the
K~c values for the smaller size specimens were suppressed.
TABLE 5--Comparison o f Kl~ results obtained with various specimen sizes.
K~cb with Specimen Sizes

K~oo with Specimen Sizes for K~o Results
1 by 1-In. Shown in Shown in
Steel and Yield Strength, Specimens, Next Column, Previous
Melting Practice ksi ksi x/i-~, ksi x / i n _ _ Column, in.
A517F AM . . . . . . . . 110 105 170 2 by 6

HY-130(T) AM . . . . . 145 160 246 2 by 8
12Ni AM . . . . . . . . . . 175 130 130 1 by 3
12Ni VM . . . . . . . . . . 185 169 220 1 by 5
18Ni(180) AM . . . . . . 190 118 112 2 by 4
18Ni(180) V M . . . . . . 190 142 160 2 by 4
Results based on cantilever-beam tests without instrumentation. However, pop-in leads

to failure because of method of loading (Ref 9).
b Results based on four-point slow-bend tests with instrumentation.
Until the K ~ values presented in Table 3 are verified, either by service

experience or by systematic variation of specimen thickness, no conclusions
should be made regarding the minimum thicknesses or crack depths required
for plane strain behavior of medium-strength steels. However, the K~c values
I I I I / I I I I
/
_ ALLOWABLE DEVIATION //~0:~ PM'X= 6 6 ' 0 0 0 'bs
71.5
FROM LINEARITY
,OLD CRITERION)
// /-
~ ~/////~,11~
* J.
, II
57.2
-- PIO CLICK
42.9
E// TYPICAL TYPE 6 RECORD

28.6 - - /'/// T - - I STEEL
//// SPECIMEN T I K - I
14.3
Z,,,,
0 12 24 36 48 60 72 84
'
96
I
108 120
V (C.O.D.), inch x I0 -3
FIG. 9--Reproduction o f actual type 6 load-displacement record.
ROLFE AND NOVAK ON SLOW-BEND Kic TESTING 141
FIG. 1D--Fracture surfaces corresponding to type 2 and type 3 records.
obtained appear to be representative of the K~c values that would be expected

for these steels on the basis of general strength levels, Charpy V-notch
toughness levels, and service experience, even though the specimen thicknesses
for some steels were only one third the recommended sizes. Therefore, it
would seem possible that, for the medium-strength high-toughness steels,
the general size requirements necessary to obtain valid K~ values may be
less than those currently being recommended by ASTM Committee E-24 for
steels having yield strengths greater than 200 ksi. Investigation of the mini-
mum size requirements for KIo testing of medium-strength high-toughness
steels is currently in progress.
F I G . ] 1 - - F r a c t u r e surfaces corresponding to type 4 and type 5 records.
ROLFE AND NOVAK ON SLOW-BEND Kic TESTING 143
FIG. 12--Fracture surface corresponding to type 6 record.
Effects of Stress Level During Fatigue Cracking

Results of a limited study of the effects of stress level during fatigue crack-
ing of 18Ni-8Co-3Mo AM maraging steel are summarized in Table 13 and
Fig. 13. The bend-test results showed that as long as the stress for fatigue
cracking was less than about 20 percent of the yield strength--(K~/ay) 2 of
0.04 or less--the K~. values averaged 104 ksi ~v/in. Stressing at about 30
percent of the yield strength--(K~/~y) 2 of 0.09--increased the K ~ value to
120 ksi ~v/~. This behavior is similar to that observed by Brown and Srawley
[2] and confirms their statement that fatigue cracking of K~c specimens
should be conducted at the lowest practical stress level.
130
125
OI
120
115
I10
o
ff
105
YIELD STRENGTH
I00
= 195 ksi
95 I ,
o ,o 2o 5o 55
YIELD STRENGTH USED IN FATIGUE CRACKING { O - T ) , percent
FIG. 13--Effect offatigue stress level (O-T) on K i t for 18 Ni-SCo-3Mo A M steel (Table 13).
ROLFE AND NOVAK ON SLOW-BEND Kxc TESTING 145
Effects of Face Notching

Most of the specimens tested in this investigation were face notched 5
percent or less on each side. Initially, face notches appeared to reduce the
specimen-thickness requirements for completely plane strain behavior by
adding constraint to plastic flow at the specimen surfaces. Further investiga-
tion, however, indicated that the shallow face notches did not decrease the size
requirements, despite the fact that shear lips essentially were eliminated.
Apparently, elimination of the shear lips merely changes the shape of the
p-A curve after pop-in occurs, and therefore does not significantly increase
the tendency to plane strain behavior. However, the fracture load did ap-
proach the pop-in load, and therefore KI o values could be computed approxi-
mately by using the maximum load, although this calculation is not recom-
mended for general use.
A comparison of the results for smooth and face-notched specimens [7],
a study of Krafft and Freed [6], and limited results presented in this study,
Appendix, Table 14, indicate that for specimens with shallow face notches,
the Kic values are essentially equal to those obtained on smooth specimens.
Correlation Between K~c and CVN Values

An empirical correlation, Fig. 14, was developed between the Kic values
and the Charpy V-notch impact-test results presented in Table 3. When
fully confirmed, such a correlation will increase the usefulness of both Charpy
V-notch impact tests and LEFM tests by demonstrating that both types of
tests predict the same service performance for a given steel. Thus, the engineer
can extend the use of a newer approach such as LEFM (which has great
potential because it predicts behavior in terms the designer can use, namely,
stress and flaw size) while still using the traditional types of fracture tests that
are required by current codes and specifications. Subsequent tests of additional
steels have substantiated the fact that the correlation presented in Fig. 14
does exist for steels having a large range of K1 c, cry,and CVN values.
Although the KIo-CVN correlation presented in Fig. 14 is empirical,
Clausing [10] recently has shown some conceptual basis for such a correla-
tion. Specifically, Clausing showed that the state of stress at fracture initiation
in the CVN impact specimen is plane strain Thus, the state of stress in both
the K~ and CVN specimens is similar. Two differences, strain rate and notch
acuity exist between the two types of specimens, namely, strain rate and
notch; however, changes in either strain rate or notch acuity probably merely
would change the slope of the correlation line, Fig. 3. Also, the strain rate
should not affect the CVN shelf value markedly. Therefore, because the
Charpy V-notch specimen is used widely, the correlation was developed by
using the standard V-notch and loading rate for Charpy tests.
TESTS CONDUCTED AT -I-80 F

VM -- VACUUM- MELTED
AM - AIR-MELTED
5.C
HY-130(T) A 7
•,• 2.0
12Ni VM
1.0 .~118Ni{ 190) VM

//9-4.-25 VM
~'e / 9 18Ni (180) VM
AM
/:~18Ni
0•415•4147
2Ni AM
AM
(180)AM
9 f 18Ni ( 2 5 0 ) VM
0 I I I I I I I
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
CVN/O'y, f t Ib/ksi
FIG. 14--Relation between plane-strain stress-intensity factor, Klc, and Charpy
V-notch energy absorption (CNV).
All tests were conducted at + 8 0 F. At this temperature the CVN values

were "shelf" values for all steels presently evaluated, and the specimens
exhibited 100 percent fibrous fracture. Although the correlation presented is
limited to shelf values, preliminary results indicate that a similar correlation
might prevail for low-temperature K~c and CVN values also [11]. That is,
low temperature Kie test results show that K ~ (as well as CVN values) de-
crease with decreasing temperature. Because most steels do exhibit a ductile-
brittle transition behavior (regardless of the specimen used to measure the
transition, that is, whether CVN, Kic, dynamic tear, or any other type of
specimen), this behavior would not be unexpected. However, in the transition
temperature range, strain rate (as well as notch acuity) may become an
extremely important variable and should be accounted for.
ROLFE AND NOVAK ON SLOW-BEND Kzc TESTING 147
Although the present results indicate that the Charpy V-notch test does
yield a qualitative indication of the K~ c values for various steels, and, in effect,
reinforces the confidence in the Charpy V-notch impact test as a reliable
small-scale laboratory test, it should not be inferred that Charpy V-notch
impact values can be used exclusively to predict K1 o values. Such a conclusion
is not warranted because sufficient test data and service experience have not
been acquired with many of the steels used to establish the present relation.
Summary and Conclusions

Four-point slow-bend K~c tests were conducted on eleven different steels
having yield strengths in the range 110 to 246 ksi. The tests were conducted
to investigate the applicability of linear-elastic fracture mechanics (LEFM)
for analyzing the fracture behavior of medium-strength high-toughness steels.
The results of that investigation may be summarized as follows:
I. L E F M can be used to analyze the fracture behavior of medium-strength
high-toughness steels. KI, values ranging from 87 to 246 ksi x / i n . were
obtained for the eleven steels investigated. Although the specimen sizes tested
did not satisfy completely the recommended ASTM Committee E-24 size
requirements (in some cases, the suggested specimen thicknesses were greater
than 6 in.), the KIo values are considered to be representative on the basis of
general strength and Charpy V-notch toughness and expected service per-
formance. Therefore, the measured K~c values are believed to be reliable for
general engineering use.
2. Although the K~c values were calculated by using several different cri-
teria, pop-in or an abrupt change in the slope of the load-displacement
curve occurred at a maximum secant intercept of 10 percent. Thus a 10
percent secant intercept is believed to be a realistic criterion for determining
K~ c values for medium-strength high-toughness steels.
3. Interrupted slow-bend K~ ~ tests showed that the deviation from linearity
on the P - A records was apparently due to plasticity rather than to crack
extension.
4. When K~ c and Charpy V-notch energy absorption values were normalized
on the basis of yield strength, an empirical correlation was established, even
though the yield strengths of the steels tested ranged from I l0 to 246 ksi and
the Charpy V-notch energy absorptions ranged from 16 to 89 ft.lb. Although
such a correlation increases the confidence in both types of approaches to
fracture analysis, the use of Charpy tests to predict K~ ~ values is not recom-
mended until additional test experience is obtained.
In general, the results of four-point slow-bend tests on medium-strength
high-toughness steels showed that, although the current ASTM Committee
E-24 requirements for plane strain behavior were not satisfied completely,
LEFM can be used to obtain realistic estimates of K~o values for these steels.
APPENDIX
T A B L E 6--Analysis of K i t results for A517F steel (110 ksi yieM strength).
KI~ Values for Various

Criteria, ksi ~ N o m i n a l Stress, ksi
Nominal
Stress for Type 10 7o Abrupt 10 % Abrupt
Fatigue Number of Secant Change Secant Change
Specimen Cracking, of P-A Inter- Pop- in Inter- Pop- in
No. psi Cycles R e c o r d a cept in Slope cept in Slope
TIK-1 .... 36 000 1 2 0 0 0 0 6 177 177 177 147 147 t47

TIK-2 .... 35000 73000 4 160 ... 160 131 ... 131
TIK-7a... 35000 80000 3 175 180 ... 146 150 ...
Specimen G e o m e t r y b a n d Indexes o f R d i a b i l h y c
a B 14z
Specimen
No. a, in. (Kic/~y) 2 B, in. B~, in. (Kxc/~y) 2 W, in. (Kid~cry)2
TIK-1... 2.60 1.01 1.84 1.81 0.71 6.00 2.32

TIK-2... 2.55 1.20 1.85 1.81 0.87 6.00 2.83
TIK-7 a . . 1.78 0.67 1.85 1.81 0.69 6.00 2.24
a T e s t interrupted at 10 percent secant intercept, unloaded, fatigue-cracked, a n d retested

(three-point loading - - SB = 24 in.).
bSB = 48 in., ST = 12 in.
c Based o n Ktc at 10 percent secant intercept.
d T h e classification o f the various p_A records is s h o w n in Fig. 4.
TABLE 7--,4nalysis Of Kle results for 4147 steel (137 ksi yieM strength).
KI. Values for Various

Nominal Criteria, ksi w/i'-~. Nominal Stress, ksi
Stress for
Fatigue Number Type of 5~ 10~o 15~ 5~o 107o 157o
Specimen Cracking, of P-A Secant Secant Secant Pop- Secant Secant Secant Pop-
No. psi Cycles Record c Intercept Intercept Intercept in. Intercept Intercept Intercept in
47-1 . . . . . . . . . . 25 000 23 700 6 94 102 107 107 125 136 143 143
47-2 . . . . . . . . . . 25000 19 300 6 98 107 114 114 130 141 152 152
47-3 . . . . . . . . . . 25 000 19 500 5 106 112 116 121 140 146 154 160
47-5 . . . . . . . . . . 25000 20 500 3 91 103 103 103 147 166 166 166
c7
z
Specimen Geometrya and Indexes of Reliability b O
<
a B W
O
Specimen z
t~
NO. a, in. (Kic/O-y)
2 B, in. B,,, in. (Kic/ay ~) W, in. (KIe/O-y~)
i
47-1 . . . . . . . . . . 0.77 1.26 0.50 0.45 0.82 2.25 3.68
47-2 . . . . . . . . . . 0.80 1.15 0.50 0.45 0.72 2.25 3.25
C7
47-3 . . . . . . . . . . 0.75 1.12 0.49 0.45 0.73 2.25 3.36
47-5 . . . . . . . . . . 1.37 2.42 0.50 0.50 0.89 2.25 4.00
a SB = 1 8 i n . ; S r = 4.5 in.
b Specimens 47-I and 47-2 based on K~. at pop-in and 47-3 and 47-5 based on Kx. at 10 percent secant intercept. O
c The classification of the various P-A records is shown in Fig. 4.
7.
~o
TABLE 8--Analysis o f KI~ results for HY-130(T) steel (149 ksi yield strength).
Ktc Values for Various

Criteria, ksi x/i-re. Nominal Stress, ksi
Nominal
Stress for Type of 10 70 Abrupt 10 70 Abrupt
Speci- Fatigue Number P-A Secant Change Secant Change
men Cracking, of Rec- Inter- in Max Inter- in Max
No. psi Cycles ord" cept Slope Load cept Slope Load
5NKF-5. 16 000 11 800 4 202 211 249 209 218 258

5NKF-3. 22 00(3 6 100 4 238 238 264 203 203 225
5NKF-2. 14 000 36 200 4 246 246 279 180 180 204
TABLE 8--Continued.
Specimen Geometry ~ and Indexes of Reliability b
Speci- a B W
men
No. a, in. (Kic/ay) 2 B, in. Bn, in. (Kic/ay) ~ W, in. (Ki~/ay) 2
5NKF-5. 1.00 0.50 1.95 1.55 0.98 4.0 2.00

5NKF-3 . 1.45 0.57 1.87 1.47 0.73 6.0 2.35
5NKF-2. 2.16 0.80 1.94 1.54 0.71 8.0 2.94
a SB = 24, 36, and 48 in., respectively; ST = 8, 12, and 16 in., respectively.

b Based on Kic at abrupt change in slope.
The classification of the various P-A records is shown in Fig. 4.
ROLFE AND NOVAK ON SLOW-BEND KI= TESTING 151
T A B L E 9--Analysis of K z, results for 4130 steel (158 ksi yieM strength).
Kic Values for Various

Criteria, ksi v/in~. N o m i n a l Stress, ksi
Nominal
Stress for Type of 5% 10% 5~ 10%
Spec~ F a t i g u e N u m b e r P-A Secant Secant Secant Secant
men Cracking, of Rec- Inter- Inter- Pop- Inter- Inter- Pop-
No. psi Cycles ord ~ cept cept in cept cept in
30-1 . . . . . 25000 15 300 3 96 100 100 133 138 138

30-2 . . . . . 25000 12 300 6 84 95 97 116 133 135
30-5 . . . . . 28000 17 900 3 103 109 109 137 144 144
30-6 . . . . . 21 000 22 900 3 92 93 93 127 128 128
Specimen G e o m e t r y ~ a n d Indexes o f Reliability ~
Speci- a B W
men
No. a, in. (Ki,/ay) ~ B, in. B,, in. (Kit/o-y) ~ W, in. (Ki,]~ry)~
30-1 . . . . . 0,89 2.22 0.39 0.34 0.98 2.29 5.73

30-2 . . . . . 0.90 2.50 0.40 0.34 1.11 2.30 6.40
30-5 . . . . . 1.00 2.10 0.39 0.39 0.82 2.28 4.80
30-6 . . . . . 0.88 2.54 0.39 0.34 1.13 2.29 6.62
a SB = 1 8 i n . ; S r = 4.5 in.
b Based o n K~c at 10 percent secant intercept.
T h e classification o f the various P-Zx records is s h o w n in Fig. 4.
TABLE lO---Analysis of Kt~ results for 12Ni-5Cr-3Mo A M steel (175 ksi yield strength).
Kxc Values for Various

Criteria, ksi x/i~. Nominal Stress, ksi
Nominal
Stress for Type of 5~o 107O 57O 107o
Speci- Fatigue Number P-A Secant Secant Secant Secant
No. psi Cycles ord c cept cept in cept cept in
12NA-1 24 000 96 200 6 131 137 137 104 109 109

12NA-2 30 000 30 900 4 120 128 125d 123 136 132d
TABLE lO--Continued.
Specimen Geometry~ and Indexes of Reliability b
Speci- a B W
men
No. a, in. (KIc/~y) ~ B, in. B,~, in. (Kx~/o-y)~ W, in. (K1~/o-y)2
12NA-1 1.61 2.64 2.00 1.81 3.28 5.95 9.75

12NA-2 1.14 2.13 2.00 1.81 3.74 3.49 6.50
a SB = 36 and 30 in.; S r = 12 and 7 in.

b Based on KI~ at I0 percent secant intercept.
c The classification of the various p-A records is shown in Fig. 4.
d Abrupt change in slope.
ROLFE AND NOVAK ON SLOW-BEND K~e TESTING 1.53
T A B L E ll--Analysis o f K i t results for 12Ni-5Cr-3Mo V M steel (183 ksi yield strength).
K~c Values for Various

Criteria, ksi w/i~. N o m i n a l Stress, ksi
Nominal
Stress for T y p e o f 10 7o Abrupt 10 7o Abrupt
Speci- F a t i g u e N u m b e r P-A Secant Change Secant Change
men Cracking, of Rec- Inter- in Max Inter- in Max
No. psi Cyles ord ~ cept Slope Load cept Slope Load
2A . . . . . . 20000 60000 5 175 ... 211 216 ... 260

2B . . . . . . 20000 73000 5 172 ... 206 214 ... 256
2C . . . . . . 20000 65 000 4 186 214 230 200 23l 247
2D ...... 20 000 71 000 4 182 223 233 195 239 250
Specimen Geometry~ a n d Indexes o f Reliability b
Speci- a B W
men
No. a, in. (Kic]~y) z B, in. B,, in. (Klcflry)~ W, in. (Ki,/ay) ~
2A . . . . . . 0.49 0.54 0.99 0.79 1.08 3.0 3.28

2B . . . . . . 0.46 0.52 1.00 0.80 1.13 3.0 3.39
2C . . . . . . 0.63 0.6l 1.00 1 00 0 97 3.0 2.92
2D ...... 0.65 0.66 1.00 1.00 1.01 3.0 3.03
a S B = 16 i n . ; S r = 5in.
b B a s e d o n Kio at 10 percent intercept.
c T h e classification o f the various P-A records is s h o w n in Fig. 4.
T A B L E 12--Analysis of Kzc results for 12-Ni-5Cr-3Mo VM steel (186 ksi yield strength).
K~c Values for Various

Criteria, ksi x/in~. N o m i n a l Stress, ksi
Nominal
Stress for Type o f 5~o 10~o 15~ 5~o 10~o 157o
Speci- Fatigue Number p.A Secant Secant Secant Secant Secant Secant
men Cracking, of Rec- Inter- Inter- Inter- Inter- Inter- Inter-
No. psi Cycles ord c cept cept cept cept cept cept
12NV-1.. 20 000 96 200 4 172 189 198 246 270 283

12NV-3.. 24 000 25 500 3 2052 207 ... 234 237 ...
12NV-4.. 22 000 48 000 4 207 a 220 226 205 218 224
12NV-6.. 23 000 33 200 4 217 226 233 189 197 203
12NV-5 ". 22 000 50 000 3 218 220 . .. 191 193 . ..
12NV-2/. 22 000 45 000 5 158 169 176 182 192 200
Specimen Geometry~ and Indexes o f Reliability b
Speci- a B W
men
No. a, in. (Ki~&y) ~ B, in. B,,, in. (Kicflry)2 W, in. (Kic/ay) ~
12NV-1.. 0.72 0.64 1.87 1.70 1.65 1.99 1.76

12NV-3.. 0.92 0.74 1.89 1.69 1.52 3.00 2.42
12NV-4.. 1.29 0.92 1.88 1.69 1.34 4.00 2.86
12NV-6.. 1.44 0.97 1.89 1.70 1.27 5.00 3.36
12NV-5" . 1.34 0.96 1.90 1.70 1.36 5.00 3.58
12NV-2:. 0.95 1.06 1.89 1.69 2.1l 3.00 3.35
a SB = 16, 28.9, 33, and 33 in., respectively; ST = 4, 6, 8, and 12 in., respectively.

b Based on Kx~ at 10 percent intercept (except Specimen 12NV-1 based on 15 percent).
c The classification o f the various p - h records is shown in Fig. 4.
Small step in p-A record (0.2 offset).
" Test interrupted at 10 percent secant intercept, unloaded, fatigue-cracked, and retested
( S t = 10 in.; S t = 33 in.).
: Test interrupted after pop-in (at 15 percent secant intercept) unloaded, fatigue-cracked,
and retested (KI~ after retest = 189 ksl ~ / ~ . based on pop-in at 5 percent secant offset).
ROLFF AND NOVAK ON SLOW-BEND Kie TESTING 155
]'ABLE 13--Analysis of KI~ results for 18Ni-8Co-3Mo AM steel (193 ksi yieM strength).
/tic Values for Various

Nominal Criteria, ksi w / ~ . Nominal Stress, ksi
Stress for
Fatigue Number Type of 57o 10~o 57o 10~o
Specimen Cracking, of P-A Secant Secant Secant Secant
No. psi Cycles Record" Intercept Intercept Intercept Intercept
18N-2 . . . . . 57 000 8 000 2 119 119 118 118

18NF-8a... 57 000 6 400 2 121 121 115 115
18NF-11... 57 000 6 700 2 121 121 116 116
18NF-1 . . . . 33 000 31 900 3 106 108 106 108
18NF-7 . . . . 31 000 33 600 3 104 107 124 128
18NF-5 . . . . 24 000 67 4130 3 103 106 102 105
18NF-3 . . . . 28 000 70 800 3 99 100 101 102
18NF-10... 11 000 156 000 3 106 107 106 107
18N-3 . . . . . 18 000 123 000 3 107 110 108 111
18NF-12 d. . 45 000 15 300 2 113 113 90 90
TABLE 13--Continued.
Specimen Geometry a and Indexes of Reliability b
a B W
Specimen
No. a, in. (Kit/cry)2 B, in. B~, in. (KI~tar) 2 W, in. (Ki~/~ry)~
18N-2 . . . . . 1.30 3.42 1.81 1.64 4.77 3.99 10.5

18NF-8a.. 1.28 3.26 1.82 1.78 4.65 4.00 10.2
18NF-11.. 1.32 3.36 1.82 1.78 4.65 4.00 10.2
18NF-1... 1.40 4.45 1.81 1.63 5.76 3.99 12.7
18NF-7... 2.25 7.30 1.82 1.65 5.35 4.00 13.0
18NF-5... 1.28 4.20 1.81 1.64 5.95 3.99 13.1
18NF-3... 1.47 5.45 1.81 1.65 6.70 3.99 14.8
18NF-10.. 1.43 4.65 1.81 1.63 5.88 4.00 13.0
18N-3 . . . . . 1.28 3.94 1.81 1.64 5.05 3.87 11.9
18NF-12 ~. . 1.32 3.90 1.78 1.62 5.21 4.00 11.6
a S B = 33 i n . ; S r = 8in.
b Based on Kxo at 10 percent secant intercept.
c The classification of the various P - ~ records is shown in Fig. 4.
a Test interrupted at 3 percent secant intercept, unloaded, fatigue-cracked, and retested.
TABLE 14---Analysis of Kzc results for 18Ni-SCo-3Mo A M steel (190 yield strength.
K~ Values for Various

Criteria, ksi v/i~. Nominal Stress, ksi
Nominal
Stress for Type of 5 ~o 10 7o 5~ 10 ~o
DCBS-5. 24000 40 100 3 110 113 113 109 112 112

DCBS-6. 36 000 13 100 2 118 118 118 117 117 117
DCBS-1. 24 000 37 400 2 110 110 110 116 116 116
DCBS-2. 24000 44400 2 113 113 113 118 118 118
DCBS-3. 24 000 30 000 3 90 93 a 93 a 139 145 145
DCBS-4. 24 000 44 400 3 91 94" 94 e 141 147 147
Specimen Geometry - and Indexes of Reliability b
Speci- a B W
men
No. a, in. (Kio/ay)2 B, in. B,~, in. (Ki,/o'y)2 W, In. (Kiefiry) z
DCBS-5. 0.94 2.66 1.24 1.24 3.50 3.50 9.90

DCBS-6. 0.98 2.54 1.24 1.24 3.22 3.50 9.10
DCBS-1. 1.10 3.28 1.24 1.11 3.70 3.50 10.4
DCBS-2. 0.95 2.69 1.24 1.12 3.50 3.50 9.90
DCBS-3. 0.92 3.82 1.23 0.50 5.10 3.50 14.5
DCBS-4. 0.85 3.52 1.24 0.50 5.14 3.50 14.5
S ~ = 33 i n . ; S r = 8 i n .
b Based on Kxc at 10 percent secant intercept.
c The classification of the various P-A records is shown in Fig. 4.
a Using m = 0.7, Kic = 114.
Using m = 0.7, Kic = 115
ROLFE AND NOVAK ON SLOW-BEND Kxe TESTING 157
TABLE 15--Analysis of Kto results for 18Ni-SCo-3Mo (180) VM steel (187 ksi yield strength).
Kx~ Values for Various

Criteria, ksi x / i n . Nominal Stress, ksi
Nominal
Stress for Type of 5% 10% 5% 10%
200 A . . . 36 000 9 100 4 152 161 161 d 154 163 163 ~

200 B . . . 30 000 19 900 3 155 159 159 166 170 170
Specimen Geometry~ and Indexes of Reliability b
Speci- a B W
men
No. a, in. (Kic/~ry)
2 B, in. B,~, in. (Kxc/ay)~ W, in. (Kxo/~ry)2
200 A . . . 1.45 1.95 1.97 1.78 2.65 3.98 5.35

200 B . . . 1.75 2.42 1.99 1.79 2.75 3.97 5.50
a S h = 30in.; S r = 8in.
b Based on KIo at 10 percent secant intercept.
The classification of the various P-A records is shown in Fig. 4.
d Abrupt change in slope.
TABLE 16--Analysis of Kl~ resultsfor 18Ni-SCo-3Mo (250) VM steel (246 ksi yield strength).
Various Criteria,
ksi Nominal Stress, ksi
Nominal
Stress for Type of 5 ~ 107o 5~ 10~o
No. psi Cycles ord~ cept cept in cept cept in
250-1 . . . . 27 000 29 900 2 87 87 87 86 86 86

250-2 . . . . 27 000 31 100 3 86 87 87 99 99 98
Specimen Geometry ~ and Indexes of Reliability b
Speci- a B W
men
No. a, in. (KIe/O-y) 2 B, in. B,, in. (Kio/cry)2 W, in. (K~o/ay)~
250-1 . . . . 1.25 10.0 1.94 1.74 15.0 4.00 32.0

250-2 . . . . 2.08 16.6 1.92 1.73 15.4 4.00 32.0
$ 8 = 30 i n . ; S r = 8 i n .
b Based on K~o at 10 percent intercept.
c Classification of the various p-A records is shown in Fig. 4.
References
[1] Fracture Toughness Testing and Its Applications, ASTM STP 381, American Society for
Testing and Materials, April 1965.
[2] Brown, W. F., Jr., and Srawley, J. E., Plane Strain Crack Toughness Testing of High-
Strength Metallic Materials, ASTM STP 410, American Society for Testing and Ma-
terials, Feb. 1967.
[3] Irwin, G. R., "Fracture," Encyclopedia of Physics, Flugge, S., ed., VoL 6, Elasticity
and Plastieity, Springer Verlag, Berlin, Germany, 1958, pp. 551-590.
[4] "Recommended Practice for Plane-Strain Fracture Toughness Testing of High-
Strength Metallic Materials Using a Fatigue-Cracked Bend Specimen," TRP prepared
by ASTM Committee E-24, 18 April 1967.
[5] Fisher, D. M., Bubsey, R. T., and Srawley, J. E., "Design and Use of a Displacement
Gage for Crack Extension Measurements," NASA TN-D3724, National Aeronautics
and Space Administration, 1966.
[6] Freed, C. N. and Krafft, J. M., "Effect of Side Grooving on Measurements of Plane-
Strain Fracture Toughness," Journal of Materials, Vol. 1, No. 4, Dec. 1966, p. 770.
[7] Rolfe, S. T. and Novak, S. R., "Effect of Face Notches on Slow-Bend K~ Fracture-
Toughness Testing," Applied Research Laboratory Report No. 37.018-001(1), 1 Jan.
1966 (obtainable through the Defense Documentation Center).
ROLFE AND NOVAK ON SLOW-B~:ND KI~ TESTING 159
[8] Brand, R. A. and Rolfe, S. T., "Procedures for Determining KI~ for 180/210 Ksi
Yield-Strength Steels," Applied Research Laboratory Report No. 40.018-002(23),
1 Dec. 1964 (available through Defense Documentation Center).
[9] Rolfe, S. T., Novak, S. R., and Gross, J. H., "Stress-Corrosion Testing of Ultraservice
Steels Using Fatigue-Cracked Specimens," presented at ASTM annual convention,
Atlantic City, 27 June 1966.
[10] Clausing, D. P., "Effect of Plane-Strain Sensitivity on the Charpy Toughness of
Structural Steels," Applied Research Laboratory Report No. 89.018-020(1), 15 May
1968 (available from Defense Documentation Center).
[11] Barson, J. M. and Rolfe, S. T., "Correlation Between KIc and Charpy V-Notch Test
Results in the Transition-Temperature Range," Impact Testing of Metals, ASTM
STP 466, American Society for Testing and Materials, 1970.
W. G. Clark, Jr.,1 and E. T. WesseP
Application of Fracture Mechanics

Technology to Medium-Strength Steels
REFERENCE: Clark, W. G., Jr., and Wessel, E. T., "Application of Fracture

Mechanics Technology to Medium-Strength Steels," Review of Developments in
Plane Strain Fracture Toughness Testing, A S T M STP 463, American Society for
Testing and Materials, 1970, pp. 160-190.
ABSTRACT: A review of fracture toughness and fatigue crack growth rate data
currently available for medium-strength steels is presented. Data are included
for three commonly used structural steels (AISI 1045, 1144, and 4140); two
pressure vessel steels (ASTM A533 Grade B, Class 1 and A216 WCC grade); and
three grades of rotor forging steels (ASTM A469, Class 4; A470, Class 8; and
A471, Class 4). These data clearly illustrate that under conditions of sufficient
restraint to ensure plane strain loading, existing linear elastic fracture mechanics
technology is applicable to medium-strength steels. The advantages and overall
potential of fracture mechanics technology as a quantitative tool for the prevention
of failure is demonstrated by means of an example problem.
KEY W O R D S : fracture mechanics, fractures (materials), pressure vessels,

toughness, fatigue, cracks, fracture strengths, steels, evaluation, tests
Linear elastic fracture mechanics or "fracture toughness" technology

provides a unique structural design concept which yields a quantitative
relationship between applied stress, flaw size, and material properties.
Specifically, from the proper consideration of fracture mechanics data for
both the slow growth and rapid propagation phases of fracture it is possible
to:
1. Select materials to provide the desired reliability against fracture.
2. Develop a quantitative evaluation of the brittle fracture potential of
components in specific situations.
3. Predict the useful life expectancy of components under sustained or
cyclic loading conditions or both.
4. Establish realistic material selection and nondestructive inspection speci-
fications that will assure the desired degree of immunity from brittle failure
for the required life of the structure.
I Senior engineer and manager, Fracture Mechanics, respectively, Westinghouse Re-
search Laboratories, Pittsburgh, Pa. 15235. Mr. Wessel is a personal member ASTM.
160
Copyright + 1970 by ASTM lntcrnational www.astm.org
CLARK AND WESSEL ON FRACTURE MECHANICS TECHNOLOGY | 61
Until recently, the application of linear elastic fracture mechanics tech-

nology has been limited to relatively high-strength materials which charac-
teristically exhibit plane strain (brittle) fracture behavior in relatively small
section sizes. However, in view of the tremendous potential of fracture
mechanics as a quantitative tool for the prevention of structural failures,
considerable interest has developed with regard to the applicability of the
technology to low-to-intermediate strength materials. Consequently, several
extensive programs have been initiated to assess the applicability of linear
elastic fracture mechanics technology to medium-strength materials and to
establish the appropriate material properties [1-6]. ~ Although none of these
programs are complete, sufficient progress has been made to illustrate clearly
that under the conditions of adequate restraint necessary to develop plane
strain loading, the existing technology is applicable to medium-strength
materials. This conclusion is particularly significant in view of the increasing
use of low-to-intermediate strength materials in larger section sizes (increasing
restraint) where it is realistic to assume that plane strain conditions prevail
over some portion of the operating temperature range.
The purpose of this paper is to review the pertinent fracture mechanics
data currently available for medium-strength steels and to demonstrate,
through the use of a simple example problem, the advantages and overall
potential of the technology. A brief review of the information required to
utilize the technology also is included.
Information Required to Use Fracture Mechanics Technology

The successful use of fracture mechanics technology requires basic informa-
tion concerning material properties, stress analysis, and defect characteriza-
tion. 3 In addition, an appropriate expression (generally referred to as the
stress intensity expression or K-calibration) which describes the relationship
between these factors for the structural configuration of interest must be
available [7].
The basic material property essential to all fracture mechanics considera-
tions is the material's inherent fracture toughness, K~c at the temperature of
interest. From a knowledge of this parameter (as well as the appropriate
stress analysis, defect characterization, and stress intensity expression), it is
possible to calculate the critical defect size necessary to cause brittle failure
as the result of a single application of load. However, few structures are
built to withstand only a single application of load in a nonhostile environ-
ment. Consequently, the useful life of a structure depends upon the rate at
which an existing defect will grow to the critical flaw size as the result of fatigue
Fracture mechanics technology is predicated on the assumption that a crack or crack-
like defect exists in the structure.
or environmental crack growth or both. Therefore, an additional material

property basic to fracture mechanics considerations is the rate at which a
crack will grow under the loading conditions of interest. In addition, due to
the dependence of fracture properties on metallurgical parameters, a fracture
mechanics analysis of a structure fabricated by welding requires K~c and
crack growth rate data for both weld and heat affected zone material as well
as the base plate.
The pertinent stress analysis information (excluding K expressions) re-
quired for use with the fracture mechanics technology essentially involves
general stress analyses based upon the principles of force equilibrium which
yield an evaluation of the applied nominal stresses on the component. Accurate
information concerning the size, shape, orientation, location, and distribution
of defects is also vital to all fracture mechanics considerations. Therefore, it is
imperative that the type of defect likely to be encountered at a given location
in a particular structure be known. Optimum nondestructive inspection
techniques must be employed for this purpose.
The final requirement essential to the proper utilization of fracture mechan-
ics technology is an expression which relates the applied stress and defect
characterization information to the fracture toughness and crack growth rate
data for the geometry and loading conditions of interest. Such expressions
have been developed for a variety of structural configurations and are
available in the literature [8,9].
Of the information required to utilize the fracture mechanics approach to
the prevention of structural failures, the absence of adequate material proper-
ties has been the primary factor preventing the use of the technology for
relatively low-strength, high-toughness materials. In view of the nature of
medium-strength materials, large toughness specimens are required to
establish adequate K~ data, and an initial problem was the development of
an efficient specimen suitable for testing high-toughness materials. This phase
of the application of fracture mechanics to medium-strength materials has led
to the development of the compact tension fracture toughness specimen with
which the majority of data presented here were generated. A detailed dis-
cussion of the development of this specimen has been presented in the litera-
ture [10].
Material Properties
This section presents a summary of the pertinent fracture mechanics data
currently available for a variety of medium-strength steels. Included are
fracture toughness (Kx~) versus temperature data and room-temperature
fatigue crack growth rate data for three commonly used structural steels
(AISI 1045, 1144, and 4140); two pressure vessel steels (ASTM A533 Grade
B, Class 1 and A216, WCC grade) and three grades of rotor forging steels
CLARK AND WESSEL ON FRACTURE MECHANICS TECHNOLOGY 163
(ASTM A469, Class 4; A470, Class 8; and A471, Class 4). The chemical
compositions and heat treatments for these alloys are presented in Table 1. A
summary of the room-temperature mechanical properties is given in Table 2.
Unless noted otherwise, all the fracture toughness data presented in this
paper represent valid plane strain K~ measurements made in accordance with
the latest ASTM recommendations [11,12]. The KIo testing was conducted
with X- and T-type wedge-opening-loading (WOL) and compact tension (CT)
fracture toughness specimens [10 ]. Figure 1 presents the general configuration
and dimensions (relative to the specimen thickness, B) of the various compact
fracture toughness specimens involved. With regard to the subsequent data,
the number preceding the specimen type identification (X WOL, T WOL, or
CT) refers to the test specimen thickness in inches.
All K~c tests were conducted at a stress intensity loading rate,/f, of 100
ksi ~v/in~./min. The low temperature tests were conducted in a nitrogen vapor
environment, and tests at and above room temperature (75 F) were conducted
in laboratory air.
The fatigue crack growth rate tests were all conducted with side notched
(5 percent of the specimen thickness on each side) T-type WOL specimens at
75 F in laboratory air under sinusoidal tension-tension (0 to maximum load)
loading conditions at a frequency of 10 Hz. The extent of crack growth en-
countered during each test was measured and recorded with an ultrasonic
crack growth monitor developed specifically for use with the WOL specimen
[13]. This crack growth monitor exhibits a crack length measurement sensi-
tivity of ~0.010 in. and provides a continuous record of crack length versus
number of elapsed cycles. The crack growth rate data were determined from a
computerized curve fitting analysis of the crack-length versus elapsed cycles
data. 4
Further information concerning additional details of both the KI~ and
fatigue crack growth rate testing techniques used to establish the data re-
ported here is available in the literature [4,5,13].
Kxc Fracture Toughness Data

Figure 2 presents the influence of temperature on the fracture toughness
and yield strength of AISI 1045, 1144, and 4140 steels [4]. 5 These tests were
conducted with 4-in.-thick WOL specimens and with the exception of the
1045 steel test at 75 F, all results represent valid K~c measurements. The
AIS1 1045 and 4140 steel specimens were prepared from large, 4-in.-thick,
normalized and quenched and tempered plates, respectively. The K~c tests as
4 Federowicz, A. J. and Powell, B. A., unpublished Westinghouse Research Laboratories
data.
This work conducted for the U.S. Army Tank-AutomotiveCommand under contract
DAAE 07-67-C-4021.
O~
TABLE 1--Chemical compositions and typical heat treatments for alloys investigated.
Chemical Composition, weight ~ (max unless otherwise noted) Typical Heat Treatment
Stress
Type Steel C Mn P S Si Ni Cr Mo V Cu Normalized Austenitized Tempered Relieved
AISI 1045 . . . . . 0.52 0.83 0.006 0.028 0.25 0.12 <0.10 0.03 <0.01 0.29 1650 F for
6 h, air
cooled o
AISI 1144 . . . . . 0.52 1.46 0.016 0.25 0.20 1550Ffor 900F for6
6 h, oil h, furnace
quenched cooled
A1S1 4140 . . . . . 0.42 0.78 0.007 0.023 0.23 0.22 0.89 0.21 <0.01 0.25 ... 1550 F for 1200 F for
6 h, oil 6 h, furnace
quenched cooled
ASTM Grades:
A533 Grade
B, Class 1.. 0.24 1.42 0.01 0.017 0.22 0.70 ... 0.50 ... ... 1675 F 4- 1600 F -4- 1225 F -4- 1150 F -4-
75F 50 F for 4 25 F for 4 25 F, hold 40
h, water h, furnace h, air cooled
quenched cooled
A216 W C C . . 0.24 1.15 0.008 0.011 0.44 0.37 0.09 0.02 ...... double nor- 1200 F for
realized at 8 hr
1650 and
1750 F for
8h
A469 Class 4. 0.27 0.70 0.015 0.0180.15to 3.0 0.50 0.20to 0.03 ... 1700Ffor 1480Ffor ll60Ffor 1125Ffor
0.30 min 0.60 rain 30 h 30 h, water 30 h, air 30 h, fur-
quenched cooled nace cooled
A470 Class 8. 0.25 1.00 0.015 0.018 0.15 0.75 0.90 1.0 0.20 ... 1850 F for 1750 F for 1225 F for 1175 F for
to to to to to 30 h 30 h, air 30 h, air 30 h, fur- >
z
0.35 0.30 1.5 1.5 0.30 cooled cooled nace cooled o
A471 Class 4. 0.28 0.60 0.015 0.018 0.10 3.25 1 . 2 5 0.30 0.05 1750 F for 1550 F for 1120 F for 1075 F for
to to to to 30 h 30 h, water 30 h, air 30 h, fur- o,
4.0 2.0 0.60 0.15 quenched cooled nace cooled
0
z
-.r
>.
z
R
G~
Z
z
0
Q
-<
o~
t,n
Dwo
W=B
a = 0.SB wl:l.3 B I T a--'i
H = 0. SB T I", ' -f
II ii H
L ~ " x ~ Thcls, T
,It,
~- - W1
"X" Type WOL Specimen
W = 2.55B
a = LOB D i- - W-
H = 1.24B
D = O.70B
2H
WI = 3.2B
T = 0.625B
~IB "J'
tl
iI H
;!
~--ThdsT, l
iI,, 7
W1
"T" Type WOL Specimen
W = 2.0B
a = I.OB
H =1.2B
D = 0. SB
WI= 2.5B
HI = 0.65B
,/.~ 2H
Compact Tension Specimen

FIG. 1--General configuration and dimensional proportions of the various types of com-
pact fracture toughness specimens.
T A B L E 2--Typical room-temperature tensile properties of altoys investigated.
0.20 ~o Yield Tensile Reduction

Type Steel Strength, ksi Strength, ksi Elongation, ~o in Area, 7o
A I S I 1045 . . . . . . . . . . . 37.5 85.5 25 38.8

AISI 1144 . . . . . . . . . . . 78.0 122 5.2 6.9
AISI 4140 . . . . . . . . . . . 65 101 16.8 47.3
ASTM Grades:
A533 G r a d e B,
Class 1 . . . . . . . . . . 50.0 min 80.0 to 100.0 16 63
A216 W C C . . . . . . . . 40.0 min 70.0 22 35
A469 Class 4 . . . . . . . 85.0 to 95.0 100.0 to 110.0 18 to 20 45 to 60
A470 Class 8 . . . . . . . 85.0 to 95.0 105.0 to 120.0 15 to 18 40 to 50
A471 Class 4 . . . . . . . 120.0 to 130.0 135.0 to 145.0 1 6 t o 18 4 0 t o 50
well as the crack growth rate test for these steels were conducted with longi-
tudinal specimens in which the major plane of the precrack was in the trans-
verse direction through the thickness of the plate (RW). The AISI 1144 steel
was quenched and tempered in the form of 10 by 12-in. forgings 4 in. thick.
The direction of crack propagation for the specimens taken from the forgings
was along the 12 in. dimension.
Figure 3 presents the influence of temperature on the K~ c fracture toughness
and yield strength of a 12-in.-thick plate of ASTM A533 Grade B, Class 1
I00 , l i I J , I~
~Z 80 80
o
b~
-~ 60 60
KIc i,, I
c
KIc
S 40 Kic 40 ~
>-
2O 20
AIM 1045STEEL AISI II44STEEL AISI 41ZlOSTEEL
I I I J J J
0 75 0 75 0 75
Test Temperature, ~
F I G . 2--Influence of temperature on the toughness and yield strength of three low-alloy

steels (4T WOL toughness specimens).
180
I t I I I i I 180
A533 Grade B Class 1 Steel
0 IX WOL Plate, 12" Thick, (HSST Plate #2)
160 z~1TCT
132TCT
'~' 4TCT
U 6TCT
(3 IOTCT
| 12TCT J t40
120
120
100
0
L 80
ta_ 2
~D
60
0 v
40
20
2~l"
0 I i
Valid ASTM-E_24
I I I
z|
,~
I
-250 -150 -50 50 100
Temperature, ~
F I G . 3--Temperature dependence of the Klc fracture toughness of a 12-in.-thick A533
Grade B, Class 1 plate.
pressure vessel steel [1,5]. 8 All test specimens were taken from a very large,
quenched and tempered, 12-in.-thick plate and were oriented such that the
fracture plane was normal to the primary rolling direction (R W) of the plate.
In all cases, the center of the specimen thickness corresponded to the center
of the original plate thickness. Note that test specimens of increasing size are
necessary to maintain plane strain conditions as the KT~ toughness increases
and yield strength decreases with increasing temperature. Also note the
rapid increase in K~ c which occurs in the temperature range between 0 F and
room temperature.
Additional fracture toughness data were generated also for another plate
of A 533 Grade B, Class 1 steel, and, in general, the results were quite corn-
6 From programs being conducted and jointly sponsored by the Pressurized Water
Reactor Plant Division ( P W R - P D ) o f the Westinghouse Electric Corp. (T. R. Mager,
cognizant engineer) and the Empire State Atomic Development Associates Inc.; P W R - P D
and USAEC-EURATUM: and PWR-PD and Oak Ridge National Laboratories ( H e a v y
Section Steel Technology Program).
CLARK A N D WESSEL O N FRACTURE M E C H A N I C S T E C H N O L O G Y 169
parable [5]. A limited amount of data are also available for A533 Grade B,
Class 1 steel weld metal and heat affected zone (HAZ) material [5]. These
data indicate that the weld and HAZ material have essentially equivalent
toughness which, in turn, is somewhat superior to the toughness of the plate.
Figure 4 presents the effect of temperature on the K,c fracture toughness
and yield strength of ASTM A216 Grade WCC cast steel [1,3,5]. Note that
although specimens up to 8 in. thick were tested, it was not possible to obtain
valid plane strain fracture toughness data at temperatures in excess of - 2 5 F.
However, the invalid data are useful in indicating the general high level of
toughness and, therefore, are included on the curve. Additional tests involving
larger specimens (up to 12 in. thick) are being conducted currently in an effort
to extend the Kx~ data to higher temperatures representative of the practical
operating temperature range.
Figure 5 presents a summary of the K,o fracture toughness and yield
strength versus temperature data for the three most commonly used turbine-
generator rotor forging steels [2]. The data summarized for the Ni-Mo-V
steel (ASTM A469 Class 4) were obtained from five different heats. The
Cr-Mo-V (ASTM A470 Class 8) and Ni-Cr-Mo-V (ASTM A471 Class 4)
160 l ' [ ' l ' I ' [60

/
/
MO // I~
Speci men Type /

o 2TCT /
120 o 6TCT 9 / 120
~ 8TCT I
40 40
20 Open Points Satisfy ASTM 20

Criteria for Validity
0 I I I I
-300 -25o -2~ -1~o -16o -5o ~ 50 1~o ~o
Temperature, ~
F I G . 4--Temperature dependence of yield strength and Klc fracture toughness for an

A216 Grade WCC steel casting.
"~ NiMoV Steel j

1~
,, ~ . ~ . ~~176 / /
~~ FVI,49
lO
i I i I I I
I I I 1 !
NiCr~V S t e e l
Forging Ident Z60
- - 3178
----- HV9241
----- HD~O ZZO
- --......
180
140
I[I0
4o
I I I i i I i.
160
ISO
140 ~ ~ ---- HV6892

\\ ---- HV6839 /
130
~/%s //
I1o
t~3
i ~ / ///
70
60
5o
3o
Io
Temperature, ~F
FIG. 5--Summary of the fracture toughness behavior of various rotor forging steels.
steel data are from three heats of each alloy. Several types and sizes of com-
pact tension specimens as well as spin burst tests [14] were employed to
obtain these data. All of the rotor forging steel data presented here were
determined with test specimens oriented such that the plane of fracture was
in the radial direction of the original forging.
Fatigue Crack Growth Rate Data

As pointed out earlier, an adequate fracture mechanics analysis of a
structure subjected to cyclic loading or a hostile environment or both re-
quires data regarding the rate at which a subcritical size flaw will grow to the
critical size under the prescribed loading conditions. This section of the
paper presents a summary of the room temperature fatigue crack growth
rate data available for the medium-strength steels under discussion. Although
the overall programs involve determination of the influence of environment on
fracture behavior, no such data are presently available. Consequently, the
following discussion will be limited to the results of the fatigue studies.
The following fatigue crack growth data are expressed in terms of the
log of the rate of crack growth (da/dN) versus the log of the change in crack
tip stress intensity factor per cycle (/XK). In general, a linear relationship was
found between the log da/dN and log /XK for the majority of materials in-
vestigated. Therefore, the data can be expressed in terms of the generalized
fatigue crack growth rate law developed by Paris [15]. This fatigue crack
growth rate law has the form of:
da/dN = CoAK"
where n is the slope of the log da/dN versus log /XK curve and Co is an em-
pirical constant determined from the data. The n and Co parameters for the
alloys studied are noted on the respective crack growth rate curves. The Co
parameters reported here are for the case of the crack growth rate expressed in
inches per cycle and the stress intensity factor expressed in psi ~v/i~.
Figures 6, 7, and 8 present the fatigue crack growth rate data for the AISI
1045, 1144, and 4140 steels, respectively. Note that with the exception of a
few data points at the lowest and highest portions of the /XK ranges investi-
gated, a linear relationship exists between log da/dN and log /XK. Note also
that the 4140 steel exhibits the highest rate of change of crack growth with
increasing /XK as indicated by the slope n = 10 compared to n -- 4 and 5 for
the 1045 and 1144 steels, respectively.
The crack growth rate data generated for the ASTM A533 Grade B, Class
I and A216 Grade WCC pressure vessel steels are presented in Figs. 9 and
10, respectively. The n and Co parameters based on the upper scatter band of
the data also are noted. Comparison of the room temperature fatigue crack
100 I ' I ' I ~ I '

80 AISI 1045 Steel #
60 ~'S = 37500psi
- 75~ Air
40 -- Environment ]
o~
_ 10 cycles/second 7
4T WOL Specimen I 1
t'--
Ap = 40000Ibs I ~
U
t,--
20
E
10 _ lo-S_
-~J-~ -
8 e-
- )
e-- 6
-
r
4 R
2-
I , I I ,I, 160-
10 20 40 60 80 100
Stress Intensity Factor Range, AK,
ksi
FIG. 6--Fatigue crack growth rate versus stress intensity factor range for .4/S/1045 steel.
growth rate properties of both pressure vessel steels studied indicates that
the A216 steel exhibits a somewhat faster rate of growth than the A533
Grade B, Class 1 steel.
Figures 11, 12, and 13 present typical fatigue crack growth rate data for
ASTM A469, Class 4 (Ni-Mo-V), A470, Class 8 (Cr-Mo-V)and A471, Class 4
(Ni-Cr-Mo-V) rotor forging steels, respectively [6]. Comparison of these
data yields a significant variation in the rate of crack growth between the
different grades of steel. Although the Cr-Mo-V steel (forging 124K406)
exhibits the greatest change in growth rate with increasing ~K, the rate of
growth at a given ~ K level is less than that observed for the other rotor steels
over most of the AK range studied.
Discussion of Material Properties

The results presented in this paper clearly illustrate that K~ ~ fracture tough-
ness data can be obtained for medium-strength, high-toughness steels pro-
100 I ' I
' I'1'_ 1~4
80
AISI 1144 Steel
60 _ OyS = 18000 psi
!
40 /5~ Air Environment ! oL
w 10 cycles/second
4T WOL Specimen
AP = 45000 Ibs
e.-
,- 20 r
u
E J t--
~1~ 10 t--
g 8 o
x: 6
C)
-g 4
I I I i I iI 10-6
10 20 40 60 80 100
ksi
F I G . 7--Fatigue crack growth rate versus stress intensity factor range for .41S1 1144 steel
vided sufficient restraint exists to develop plane strain conditions. However,

it is also apparent that relatively large section sizes are required to provide
the necessary restraint at temperatures of practical interest. In addition, it
appears that medium-strength steels characteristically exhibit a rapid rise in
fracture toughness with increasing temperature.
With regard to the room-temperature fatigue crack growth rate data, it is
obvious that the majority of data established for medium-strength steels are
consistent with the generalized fracture mechanics fatigue crack growth rate
law (da/dN = Co AK") initially established for relatively brittle materials. In
addition, there does not appear to be a significant effect of state of stress
(plane stress or nonplane strain) on the fatigue crack growth rate behavior
of these alloys. Specifically, most of the fatigue crack growth rate tests were
conducted with test specimens of insufficient size to maintain plane strain
conditions throughout the entire test. Consequently, the state of stress as
defined by the existing criteria [11,12] changed during the test, and yet there
100 10-4
80
60 AISI 4140 Steel
~- Oys = 65000 psi
40 ~ 75OFAir
M., Environment nt
]0 cyc l es/second
,nd
"~
J~ 4T WOL Specimen P..
20 AP = 45000 IbsiS
P 6
u
E
a;
10-5
e-
N _ 10 -
P
= 2x lo AO ~
dN
11 I I I I ] J I i 10--6
10 20 40 60 80 100
ksi
FIG. 8--Fatigue crack growth rate versus stress intensity factor range for A / S / 4140 steel
is no evidence of a transition effect in the crack growth rate data (the log
d a / d N versus log AK curve remained linear).
Table 3 presents a summary of the pertinent room-temperature material
properties associated with each of the alloys discussed in this paper. For
those alloys which did not exhibit plane strain behavior in the K ~ tests
conducted at 75 F, an apparent toughness value, based upon an extrapolation
of the valid K ~ data from the lower temperatures, is reported. Note that
there is no correlation between yield strength and Kic or relative toughness
(KIe/O-ys) for the alloys studied. Arbitrarily selecting a relative toughness
value of one or greater as a criterion for high-toughness properties (in view of
existing Kr ~ data for a variety of materials, this is probably a realistic crite-
rion), it is apparent that all but three of the alloys discussed in this paper
(AISI 1144 and 4140 steels and ASTM A470 steel) exhibit excellent toughness
properties.
T A B L E 3--Summary o f room-ternperature rnaterial parameters for alloys investigated.
0.2 % Yield Relative

Strength, KI~, Toughness,
Material ksi ksi V/~.. Kio/~y~ n~ Cob
A I S I 1045 steel . . . . . 37.5 50 , 1.33 4 5.6 X 10-.4

A I S I 1144 steel . . . . . 78 60 0.77 5 3.1 X 10 -~9
A I S I 4140 steel . . . . . 65 56 0.83 10 2 X 10 -5~
A S T M A533 G r a d e B,
Class 1 steel . . . . . . 68 190 " 2.8 2.2 1 X 10 -~5
A S T M A216 W C C
grade steel . . . . . . . 48 155" 3.23 3 2.3 X 10 -19
A S T M A469, Class 4
steel . . . . . . . . . . . . 74 90 ~ 1.22 2.7 4.4 X 10 -~8
steel . . . . . . . . . . . . 93 55 0.54 6.7 1 X 10 -~6
steel . . . . . . . . . . . . 113 200~ 1.77 1.4 2.9 X 10 -]2
a n = slope o f log d a / d N versus log AK curve.

b Co = empirical constant f r o m log d a / d N versus AK data.
c A p p a r e n t toughness based on extrapolation of existing KIr data.
100 10--4
6O
- Upper Scatter Band
_~ 40 (slopen=2.2) ~ /
~ ~ ~__ LowerScatter -~
20 /~-- Ban, --
10 -- ~ -]5
Co=, ,o _ o
~ 6_ //t T tTemp =75% _ ~'~
// 7a SpecimenType "-
o 1TWOE ~-
. 2TWOL
2 c~ 3f WOL
4TWOL
1 I I I I I I lrl I J 10--6
10 20 40 60 80 100 200 400
Stress Intensity RangeAK, ksi
F I G . 9--Fatigue crack growth rate as a function o f AK for ,4533, Grade B, Class 1
steel (11~-in. plate).
300 [ ~ I ~ [ I ill ' L
200
I00
Upper Scatter
80 - (slopen = 3) 09
(a>
I
u
60 -Lower Scatter
Band
~ 40
C
03
,.1= 20
r
~ 10 10-5
8 ~ = Co Z~Kn
Co= 2.3x10-I9
/ ~ ~ oT e da Temp= 75~
Test
4 9 st Frequency= 600 cprn
SpecimenType
9 3T WOL
2 o 1T WOL
[] 1T WOL
z, 1T WOL
1 10-6
20
IO 40 60 80 i00 200 300
Stress Intensity Range, AK, ksi
F I G . lO--Fatigue crack growth rate as a function o f AK for A216 Grade WCC casting.
The toughness properties reported here for the AISI 1045, 1144, and 4140
steels are based on a limited number of tests from one heat of each alloy.
In addition, the tests were conducted on 4-in.-thick sections which did not
respond to heat treatment as would smaller sections [4]. Therefore, these
data cannot be arbitrarily considered truly representative of similar tests of
the same alloys. Considerably more data are required to characterize these
materials adequately and ultimately provide information suitable for design
purposes. However, the results obtained for these steels demonstrate that
existing fracture mechanics concepts are applicable to these grades of steel,
and they also provide a heretofore nonexistent basis for the level of fracture
resistance that we can expect for some commonly used structural alloys.
The fracture mechanics data reported for the ASTM A533 Grade B,
Class 1 pressure vessel steel and the rotor forging steels represent a substantial
200 I 1 [ I
I 'I'I
Ni-Mo-V Steel (124J357)
82 ksi Yield Strength
100 75~ Air Environment 10. 4
15 cycles/second
80 2T WOL Specimens
60 A P = 16,000 Ib
a~
40
9 ~- Slope I. 9
E 20
a;
I0-- IF 5
.c (..9
8 da = 1.9 x IO-14AK I"
(.9
6 m
r
Spec Ident
(_)
4 9 9 No. I07
o No. 108
1 I i I , I I [li 110-6
10 20 40 60 80 I00 200
Stress Intensity Factor Range, AK, k s i ' ~ .
FIG. l l - - F a t i g u e crack growth rate as a function o f A K f o r a 82 ksi ~u, N i - M o - V alloy.
number of tests as well as different heats. Consequently, these data can be

used as a realistic basis for design considerations for the respective material.
Example Problem
In order to demonstrate the use of fracture mechanics concepts in design,

a hypothetical example problem is included and the pertinent considerations
and computations described in detail. The problem involves selecting the
most desirable of two materials for an application involving a four-point
loaded beam in pure bending. Material selection will be limited to the AISI
1144 and 4140 quenched and tempered steels for which data have been re-
ported earlier. F o r example purposes, we will assume room-temperature
loading conditions in a nonhostile environment. A summary of the pertinent
I ' I ' I
10- 4
Cr-Mo-V Steel (124K406)

93.5 ksi Yield Strength
IS~ Air Environ ment
40 I0 cycles/second
~D
21"WOL Specimens
AP = 20,000 Ib
11)
r
r
da = 1 x 10-36~ K6"7
r
.m
20 dN
2 e.-
U
.m
r
r
.m
E 0 0 0
z - Spec I d e n t ~0 0 _ lO"-5
10 r~
- z~ NOo I r
8
- oNo. 3
e.- 2
6 - ,No. 4 0
o lJ
C.9 ~- Slope 6. l
4
r A
2-
I I , I , I Ill 10 - 6
10 40 60 8 0 1 0 0
20
Stress Intensity Factor Range, AK, ksi ~/Tn.
F I G . 1 2 - - F a t i g u e c r a c k g r o w t h rate a s a f u n c t i o n o f A K f o r a C r - M o - g steel.
room-temperature properties of the two steels used in the example are given
below:
Yield Kto, Fatigue Crack

Material Strength, psi psi ~ / ~ . . Growth Data
A I S I 1144 steel . . . . . . . . . . . . . . . . . . . . . . . 78 000 62 000 Fig. 7

A I S I 4140 steel . . . . . . . . . . . . . . . . . . . . . . . 65 000 56 000 Fig. 8
It will be assumed that both steels are supplied to a guaranteed minimum

yield strength of 60,000 psi, and this value will be used for all subsequent
CLARK AND WESSEL O N FRACTURE MECHANICS T E C H N O L O G Y 179
design considerations. The solution to the problem involves five general

areas of consideration:
(a) general stress analysis,
(b) computation of critical flaw sizes,
(c) calculation of cyclic life,
(d) material selection, and
(e) development of inspection criteria and safety factors.
Each of these areas of consideration are described in detail in the following
discussion.
Operational Requirements for E x a m p l e Problem

The following requirements are stipulated for the hypothetical problem:
1. Supply a beam of rectangular cross section to be loaded under four-
point bending (pure bending) to a maximum load of 26,800 lb.
2. Dimensional requirements dictate that the beam be 4 in. high with a
major and minor span of 8 and 6 ft, respectively.
100 I ' I ' I ' I 'l 9 1 4u ^ -
80 Ni-CF-Mo-V Steel (3178) []

J
60 - 112. 6 ksi Yield Strength /
- 10 cycles/sec0nd ~"
4O - 21- WOL Specimens ~,g'o
a, A P = 20, 000 Ib ,'~o. **
- /t.! _
-N = 2.9 x 10-12AKI'4 r
20 P..
,t-
e-
t.)
"F= E
~~ 10
lO-5
rv.
,~ 8
e-
6
- . Spec Ident - r
~ 4 9 No. I x_
9 ~ 2
oNo. 3
9 No. 4
1 1 I I I I LI Jl 10--6
10 20 40 60 80 100 200
Stress Intensity Factor Range, ZXK,ksi ~i-fi.
FIG. 13--Fatigue crack growth rate as a function o f A K for a Ni-Cr-Mo-V steel.
3. The beam is to be subjected to zero to maximum load cyclic loading at

75 F in a nonhostile environment.
4. The minimum cyclic life requirement is 100,000 cycles.
5. The maximum applied nominal stress is not to exceed one half the
material yield strength. For this example, the maximum applied stress re-
quirement is equivalent for both materials, that is, 30,000 psi.
General Stress Analysis

The application described above essentially involves a simple beam in
pure bending and can be analyzed accordingly with the bending moment and
flexure equations. The bending moment M is equal to half the applied load
times the moment arm L or M = PL/2. For the proposed problem, P equals
26,800 lb and L equals 12 in. (half the difference between the major and minor
spans) resulting in a bending moment of 161,000 in.-lb. In order to satisfy the
maximum nominal stress requirement of 30,000 psi, we can compute the
beam thickness B from the flexure equation a = M/z where ~ is the maximum
outer fiber tension stress (30,000 psi), M is the bending moment (161,000
in..lb), and z is the rectangular section modulus (z = Bw2/6, w equals the
beam height, 4 in.). Solving for B yields a beam thickness requirement of 2 in.
Figure 14 illustrates the loading conditions and pertinent stress analysis
equations.
Calculation of Critical Flaw Sizes

Since it is generally realistic to assume that all structures contain discon-
tinuities of some type it is necessary to establish the critical defect size which
will result in failure. For the problem at hand, we will limit our consideration
of defect sizes to edge cracks between the minor span where the tensile
stresses are at the maximum. For the case of pure bending, the outer fiber
tensile stresses are essentially constant within the minor span; as a result, our
considerations will be applicable to edge cracks at any location within the
minor span.
In order to compute the critical flaw size in terms of fracture mechanics,
we must first establish if linear elastic fracture mechanics concepts are
applicable to the loading conditions of interest. To do this we must know the
K~, and yield strength for the material at the minimum operating tempera-
ture. We then solve the specimen size criteria formula [B = 2.5 (Kt c/aye)2] for
the minimum thickness for which plane strain conditions prevail [9,10]. The
minimum thickness for plane strain conditions at room temperature for the
1144 and 4140 steels are 1.48 and 1.72 in., respectively. Since our problem
involves a 2-in.-thick beam, fracture mechanics concepts are considered
applicable, and we can proceed.
-P= 13,400 Ib P = 13,400 Ib

2 2
I- 6' -I
2"
;.-t:l'
r
4 w:4,,
t| - 8' 2-I
Bending Moment, M = ~ = 161,00(I in.-Ib
Outer Fiber Tension Slress,o = M-

Z = 30,0(10 psi
BW2
Rectangular Section Modulus, 7 = T
F I G . 14--Loading conditions for example problem.
Once it has been established that fracture mechanics concepts are applicable
to the problem, we must then have available an adequate stress intensity
expression which relates flaw size and applied stress to the stress intensity
factor for the loading configuration of interest. The stress intensity expres-
sion (K-calibration) for a beam in pure bending is presented generally in the
literature [8,9,11] as:
6 M a 1/~
KI- Y
BW ~
where Y is a function of the ratio of crack depth to beam width (a/W). In

addition, a graphical (or tabular) relationship between Y and a/W such as
that shown in Fig. 15 usually is included [11]. This form of the K-calibration
was developed primarily for use in Kxc testing where the crack depth a is
known and K~ is to be determined. However, for the case of an actual appli-
cation problem where a is to be established, the K-calibration must be
modified such that we can solve graphically for a/W and utlimately, a. A more
desirable form of the K-calibration for use in the example problem is:
KIB W 1"5
X-
6M
where X is a function of a/W. The graphical relationship between Y and

a/W for the case of pure bending (four-point loading) in Fig. 15 was con-
verted to the above form by letting X equal Y (a/W) 1/~ and plotting X
against a/W. The resulting modified K-calibration is presented in Fig. 16.
Using the expression:

KIB W 1.~
X-
6M
and the curve in Fig. 1 6 a n d settingB = 2in., W = 4in., M = 161,000
in..lb, and K~ equal to 62 and 56 ksi ~v/in.. (the KIc values for AISI 1144 and
4140 steels, respectively) yields critical flaw depths of 1.2 in. for the AISI 1144
steel and 1.06 in. for the AISI 4140 steel. These flaw sizes represent the critical
defect sizes which will cause failure as the result of a single application of
load under the operational conditions described for the example problem.
Calculation of Cyclic Life

Assuming the initial presence or early development of a defect in a cyclically
loaded structure, it is obvious that knowledge of the critical flaw size alone
does not permit an estimate of the useful life of the structure. The component
life is dependent upon the rate at which an existing defect will grow to the
critical flaw size under the prescribed loading conditions. Consequently, we
3.4
3.2 P
3.0
_P
2.8
2 r
2
6Mal/2. Y(JW)
KI BW2 1 I
2.6
WhereM Is TheBendingMoment
y B = Thickness
2.4 I
2.02"2 ~ ,,,''"3-Point.w
S-=8
1.6
0 0.I 0.2 0.3 0.4 0.5 0.6
a
w
FIG. 15--K-calibrations for bend specimens.
must compute the maximum initial allowable defect size (edge crack depth)
that will not grow to the critical size for failure during the required life of the
structure (100,000 cycles). In order to make these calculations we must have
fatigue crack growth rate data expressed in terms of the stress intensity factor
as well as the other data needed to compute the critical flaw sizes. The fatigue
crack growth rate data for AISI 1144 and 4140 steels are presented in Figs. 7
and 8, respectively. These data clearly illustrate the dependence of the rate of
crack growth on the change in stress intensity per cycle, AK. Since the AK
parameter increases as the crack length increases, it is apparent that to com-
pute the number of cycles to failure for a given initial flaw size one must
integrate the crack growth rate data over AK values representing the/XK level
at the start of life (that corresponding to the initial flaw size) to the/XK level
at failure, AK = K~ ~.
Interpreted in another way, our goal at this point in the problem is to
establish the crack length versus number of elapsed cycles curve for the
component geometry and loading conditions of the example problem from
crack growth rate data established with entirely different loading conditions.
~utve ~87292-A
3.0 I
I '
PR P/Z
2.5
T. -T
KI B W1" 5 W
X- 6M f0r 2< -~<8
2.0
PL
Moment M=
2
X
1.5
1.0
0.5 - /
0 -.--L_
0 0.1 0.2 0.3 0.4 0.5 06
a/W
FIG. 16--K-calibration of four-point loaded notch bend specimen.
The ability to do just this is the primary advantage of the fracture mechanics
approach to fatigue.
Since we are interested in calculating the maximum initial allowable flaw
size for the prescribed loading conditions, we must select a lower limit of
initial flaw size and corresponding AK value with which to begin the integra-
tion. This lower limit of flaw size is selected generally as the minimum flaw
size which can be reliably detected with existing nondestructive inspection
techniques. For this example, let us assume that the minimum detectable
flaw size is an edge crack 0.125 in. deep. In addition, let us incorporate an
inspection safety factor of two at this point yielding a minimum detectable
flaw size of 0.250 in. The stress intensity factor corresponding to this flaw
size then becomes the lower limit for the integration procedure. Solving the
stress intensity expression for a equal to 0.250 in. yields a KI value of 29 ksi
~v/in. Since the loading conditions involve zero to maximum cyclic load,
K~ = AK.
Although the required integration procedure can be readily adapted to a
simple computer program, for example purposes, we will proceed with a
numerical integration technique.
Assuming that the linear relationship between log da/dN versus log 2xK
shown on the respective crack growth rate curves extends from AK = 29 ksi
~v/i~. to /XK = K i c , we can integrate as follows. Select a convenient interval of
elapsed cycles and assume that the growth rate remains constant over this
interval. The increase in crack length during this interval of growth is then
added to the crack length at the beginning of the interval and a new AK value
computed. The crack growth rate associated with the new zXK value is read
off of the appropriate da/dN versus AK curve and the increase in crack
length over the cyclic interval at the higher growth rate computed and again a
new AK value established. This process is repeated until the AK value reaches
Kt~, and at this point failure is eminent. The total number of elapsed cycles
required to go from the crack length at the start of life to the critical flaw size
is then the number of cycles to failure. The results of the numerical in-
tegration are presented in graphical form in Fig. 17. Presented in this manner,
the results clearly demonstrate the fatigue crack growth behavior of the two
steels under the loading conditions outlined in the example problem.
In order to satisfy the minimum cyclic life requirement of 100,000 as
stipulated in the example problem, it is apparent from Fig. 17 that the
maximum initial allowable crack depth must not exceed 0.260 in. for the
AISI 4140 steel or 0.510 in. for the AISI 1144 steel.
Material Selection
From the fracture mechanics analysis of the hypothetical problem it is
apparent that under the given loading conditions the critical flaw size neces-
CLARK A N D WESSEL O N FRACTURE MECHANICS TECHNOLOGY 185
sary to cause failure varies little between the two materials (1.2 in. for the
AISI 1144 steel and 1.06 in. for the AISI 4140 steel). However, in view of the
difference in fatigue crack growth rate properties, there is a substantial dif-
ference in the initial allowable flaw size which will just grow to failure in
100,000 cycles of loading (0.260 in. for the AISI 4140 steel and 0.510 in. for
the AISI 1144 steel). Although either material would adequately satisfy the
prescribed operational requirements and still comply with the nondestructive
inspection detection limit, it is obvious that the AISI 1144 steel is superior.
Consequently, assuming that availability and economic factors are essentially
the same for each material, the AISI 1144 steel would be selected for use in
the application.
Development of Inspection Criteria and Safety Factors

The development of adequate nondestructive inspection criteria and
design safety factors for a structure assumed to contain a defect are closely
related. Specifically, design safety factors can be established on the basis of
applied stress or cyclic life which, in turn, influence the defect size which
must be detectable. In addition, a safety factor can be incorporated in a
design situation on the basis of the inspection level such as was employed
for this example problem.
If we select a safety factor of two on the cyclic life for the AISI 1144 steel
beam, the maximum initial allowable flaw size (at 200,000 cycles from Fig.
17) is 0.370 in., and we must have a nondestructive inspection level capable of
1.2 I I f I I I I I
1.1
1.0
13,400 lbP = 26' 800113b,400 tb
0.9
0.8
\ l" -6 "l A
/ - - AISI 1144 Steel J~'~ ~,"~1' ,~a i L/]I~/
.~- o.7 I. ,,'
y, 0.6 \ M= =l,,O ln ib
0.5
0.4
0.3
~ ISl 4140 Steel
0.2
0.1
0 I I t P I ; ~ I I
0 40 80 120 160 200 240 280 320 360 400
No. of Cycles to Failure, thousands
F I G . 17--1nitial flaw size versus cycles to jailure for example problem.
detecting this flaw size. If we now wish to incorporate a safety factor of two
on the inspection level sensitivity over and above the safety factor on cyclic
life, defects 0.185 in. deep must be readily detected. The cyclic life of the
beam now becomes something in excess of 400,000 cycles (from Fig. 17)
yielding a total safety factor larger than four. If a larger safety factor is
desired on either aspect, the minimum detectable flaw size nmst be decreased
even further or a material with crack growth properties or a KIo or both
superior to AISI 1144 steel must be used or the design stress reduced accord-
ingly.
Discussion of Example Problem

The example problem presented here does not illustrate all of the con-
siderations that can be and are employed in the fracture mechanics approach
to the prevention of failure. Specifically, no attempt was made to evaluate
the case of a component operating in a hostile environment where environ-
ment assisted crack growth may occur. However, the basic aspects of calcu-
lating the critical flaw size and useful life data have been demonstrated, and
these techniques need be only modified slightly to incorporate other variables
into the design considerations. Example problems demonstrating the use of
fracture mechanics technology in other loading conditions are available in
the literature [1,3,4,7].
Summary
A review of fracture toughness versus temperature and room-temperature
fatigue crack growth rate data currently available for medium-strength steels
was presented. These data clearly illustrate that under conditions of sufficient
restraint to ensure plane strain loading, existing linear elastic fracture mechan-
ice technology is applicable to medium-strength steels. It was noted that
these steels tend to exhibit a very rapid increase in K~o with temperature in
the temperature range of practical interest. Hence, for most of the materials
studied, the critical defect sizes necessary to cause failure under normal
service conditions are extremely large.
It was shown also that the room-temperature fatigue crack growth rate
properties of medium-strength steels could be expressed in terms of the
crack tip stress intensity factor range, AK. Specificially, the fatigue crack
growth rate properties of these steels conform to the general behavior ob-
served for higher-strength steels.
The practical usefulness and overall potential of fracture mechanics
technology as a quantitative tool for the prevention of structural failure was
demonstrated with a simple example problem. The details involved in
computing the critical defect size for failure as well as the maximum initial
allowable flaw size for a specified cyclic life were included in the example.
References
[1] Wessel, E. T., Clark, W. G., Jr., and Pryle, W. H., "Fracture Mechanics Technology
Applied to Heavy Section Steel Structures," Fracture 1969, Chapman and Hall,
London, 1969, p. 825.
[2] Greenberg, H. D., Wessel, E. T., and Pryle, W. H., "Fracture Toughness of Turbine-
Generator Rotor Forgings," Engineering Fracture Mechanics, VoL 4, 1969.
[3] Greenberg, H. D. and Clark, W. G., Jr., "A Fracture Mechanics Approach to the
Development of Realistic Acceptance Standards for Heavy Walled Steel Casings,"
presented at the ASM Metal Congress, Detroit, Oct. 1968, to be published by American
Society for Metals, 1969.
[4] Wilson, W. K., Clark, W. G., Jr., and Wessel, E. T., "Fracture Mechanics Technology
for Combined Loading and Low-to-Intermediate Strength Metals," U.S. Army Tank
Automotive Command Final Report No. 10276, 18 Nov. 1968.
[5] Wessel, E. T., "Linear Elastic Fracture Mechanics for Thick-Walled, Welded Steel
Pressure Vessels: Material Property Considerations," presented at Symposium on
Fracture Toughness Concepts for Weldable Structural Steel, United Kingdom Atomic
Energy Authority, Reactor Materials Laboratory, Culcheth, England, 29-30 April
1969.
[6] Clark, W. G., Jr., "Fatigue Crack Growth Characteristics of Rotor Steels," to be
published in Engineering Fracture Mechanics, 1970.
[7] Wessel, E. T., Clark, W. G., Jr., and Wilson, W. K., "Engineering Methods for the
Design and Selection of Materials Against Fracture," DDC Report AD 801001,
Defense Documentation Center, 1966.
[8] Paris, P. C. and Sih, G. C., "Stress Analysis of Cracks," Fracture Toughness Testing
and Its Applications, ASTM STP 381, American Society for Testing and Materials,
April 1965, p. 30.
[9] Hofer, K. E., Jr., "Equations for Fracture Mechanics," Machine Design, 1 Feb. 1968,
p. 109.
[10] Wessel, E. T., "State-of-the-Art of the WOL Specimen for K~c Fracture Toughness
Testing," Engineering Fracture Mechanics, Vol. 1, No. 1, June 1968, p. 77.
[11] Plane Strain Crack Toughness Testing of High Strength Metallic Materials, ASTM
STP 410, American Society for Testing and Materials, Jan. 1967.
[12] "Method of Test for Plane-Strain Fracture Toughness of Metallic Materials," ap-
proved by ASTM Committee E-24 for publication in ASTM Standards, Part 31, 1969.
[13] Clark, W. G., Jr., and Ceschini, L. J., "An Ultrasonic Crack Growth Monitor,"
Materials Evaluation, Aug. 1969.
[14] Sankey, G. O., "Spin Tests to Determine Brittle Fracture Under Plane Strain,"
Journal of Experimental Mechanics, Vol. 8, No. 12, Dec. 1968.
[15] Paris, P. C., "The Fracture Mechanics Approach to Fatigue," Proceedings of the Tenth
Sagamore Army Materials Research Conference, Aug. 1963, Syracuse University Press,
Syracuse, N.Y., 1964.
STP463-EB/Sep. 1970
DISCUSSION
S. T. Rolfe 1 (written discussion)--As the authors well know, I agree com-

pletely with their conclusion that existing linear elastic fracture mechanics
technology is applicable to medium-strength steels. This fact has been well
documented, not only by the work of the authors and their colleagues at
Westinghouse, but also by numerous investigations at the Applied Research
Laboratory of U.S. Steel Corp. 2-7 In fact, research in progress 8 indicates that,
providing certain conditions of restraint are satisfied, the usefulness of the
fracture mechanics approach may be extended even further than the authors
propose. The data shown in Fig. 4 that are invalid according to the criteria
for plane strain behavior substantiate this fact. Thus, the authors' conclusion
regarding the general applicability of fracture mechanics to medium- (and
low-) strength steels is well founded.
The authors correctly note that there is no correlation between yield
strength and K ~ for the alloys studied. However, the K~c data shown in
Fig. 5, along with test results of five structural steels having room-tempera-
ture yield strengths in the range 39 to 246 ksi, have been analyzed by Barsom
and Rolfe ~ to develop an empirical correlation between slow-bend K~o test
1 Professor of civil engineering, University of Kansas, Lawrence, Kans.; formerly division
chief, Mechanical Behavior of Metals Division, U.S. Steel Applied Research Laboratory,
Monroeville, Pa.
2Rolfe, S. T., Barsom, J. M., and Gensamer, M., "Fracture Toughness Requirements
for Steels," presented at the First Annual Offshore Technology Conference, Houston, Tex.,
18-21 May 1969.
Barsom, J. M. and Rolfe, S. T., "Klo Transition-Temperature Behavior of A517F
Steel," presented at the National Symposium on Fracture Mechanics, Lehigh University,
Bethlehem, Pa., 25-27 Aug. 1969.
4 Shoemaker, A. K. and Rolfe, S. T., "The Static and Dynamic Low-Temperature Crack
Toughness Performance of Seven Structural Steels," presented at the National Symposium
on Fracture Mechanics, Lehigh University, Bethlehem, Pa., 25-27 Aug. 1969.
See p. 124.
6Barsom, J. M. and Rolfe, S. T., "Correlations Between Ktc and Charpy V-notch Test
Results in the Transition-Temperature Range," Impact Testing of Metals, ASTM STP 466,
American Society for Testing and Materials, 1970.
7 Barsom, J. M., Imhof, E. J., Jr., and Rolfe, S. T., "Fatigue-Crack Propagation in High-
Yield Strength Steels," presented at the National Symposium on Fracture Mechanics, Lehigh
University Bethlehem, Pa., 25-27 Aug. 1969.
s Novak, S. R. and Rolfe, S. T., "Experimental Analysis of Sub-Thickness K~oSpecimens
to Establish Engineering Crack-Toughness Values (Kie)," Progress Report to ASTM
Committee E-24, Cleveland, Ohio, 23 Sept. 1969.
188
Copyright9 1970by ASTMlntcrnational www.astm.org
DISCUSSION ON FRACTURE MECHANICS TECHNOLOGY 189
results and standard Charpy V-notch (CVN) impact test results in the transi-
tion-temperature region. This empirical correlation is shown in Fig. 18. It
should be emphasized that this correlation, as well a previously published
upper-shelf Kic-Charpy V-notch correlation based on results obtained on
eleven steels having yield strengths in the range of 1 10 to 246 ksi 2,~, are both
empirical.9 Because of differences in loading rate and notch acuity between the
KI, and Charpy tests, no theoretical justification for the correlation currently
exists. However, various investigations 1~ have shown that for the particu-
lar geometry of the Charpy V-notch test specimen, maximum constraint
exists at the tip of the notch. Therefore, it is not unreasonable to expect
that the conditions of constraint ahead of the notch in the Charpy V-notch
and KIo test specimens are similar. Thus, although the authors state that
there is no correlation between yield strength and K1 c ,an empirical correlation
that is based in part on their data does exist between Charpy V-notch impact
energy and KI o.
The authors' example problem is extremely useful in demonstrating the
application of fracture mechanics to problems involving fatigue loading. A
similar analytical procedure for relating subcritical crack growth to inspec-
tion requirements that involves both fatigue and stress corrosion has been
presented by Sinclair and Rolfe. 13 In this procedure, various assumptions
regarding the frequency of loading of structures were made so that crack-
size/time curves could be developed to predict the life of a structural member.
This procedure, as well as the example problem described by the authors,
demonstrates the relative importance of stress level, yield strength, material
properties, and initial crack size with respect to the time required for a crack
to grow to a given size.
In conclusion, the authors' paper, as well as numerous other studies of the
fracture and fatigue behavior of medium-strength steels, conclusively dem-
onstrates the usefulness of the fracture mechanics approach as a quantita-
tive tool that can be used for the prevention of failure of structures fabricated
from medium-strength steels. For this, the authors are to be complimented.
9 See also Fig. 14 of Rolfe and Novak paper on p. 124.

10Clausing, D. P., "Effect of Plane-Strain Sensitivity on the Charpy Toughness of
Structural Steels," AD836314 L, 15 May 1968 (available from Defense Documentation
Center).
11Hollomon, J. H., "The Notched-Bar Impact Test," Transactions, American Institute
of Mining, Metallurgical, and Petroleum Engineers, Vol. 158, 1944, pp. 310-322.
12Gross, J. H., "The Effect of Strength and Thickness on Notch Ductility," Impact
Testing of Metals, A S T M STP 466, American Society for Testing and Materials, 1970.
13Sinclair, G. M. and Rolfe, S. T., "Analytical Procedure for Relating Subcritical Crack
Growth to Inspection Requirements," presented at Conference on Environmental Effects in
Failure of Engineering Materials, Metals Engineering Division, American Society of
Mechanical Engineers, Washington, D.C., 31 March-2 April 1969.
I I I I I I I
0 A 5 1 7 - F STEEL Z~ Ni-Cr-Mo-V STEELS

o A 3 0 2 - B STEEL Cr-Mo-V STEELS
]]~ ABS - C STEEL Ni-Mo-V STEELS
o HY-130 STEEL WDATA FROM REF 12
V 18Ni(,250) STEEL
2 3,'2
(CVN)
IOOC - -
800 --
//~ /,g //
- ,/ / /
~-. _
~4oc - /SX"///
zOO1-- . ~ ~,~"
/ ,,~/o
L, ,,,,~_~-~"
~-4-~
0 I0 20 :30 40 50 60 70
ABSORBED ENERGY IN CVN IMPACT TEST AT VARIOUS TEMPERATURES, f l - l b
FIG. 18--Relation between Kz~ and C V N values in the transition-temperature region.
J. P. Tanaka, ~ C. A. Pampillo, ~ and J. R. L o w , Jr. ~
Fractographic Analysis of the Low Energy

Fracture of an Aluminum Alloy
REFERENCE: Tanaka, J. P., Pampillo, C. A., and Low, J. R., Jr., "Fractographic
Analysis of the Low Energy Fracture of an Aluminum Alloy," Review of Develop-
ments in Plane Strain Fracture Toughness Testing, A S T M STP 463, American
ABSTRACT: The brittle fracture of many high-strength alloys occurs by dimpled

rupture. It is recognized that, as in ductile materials, voids which nucleate at non-
metallic particles, grow and coalesce during deformation leading to final rupture.
The density and distribution of impurity particles, therefore, may affect strongly
the toughness of high-strength alloys without contributing significantly to their
strength. To understand better this problem the fracture process of a high-strength
aluminum alloy (2014-T6) was studied. Two types of large impurity particles (3
to 10 um in diameter), where voids were nucleated, have been identified by micro-
probe analysis. Voids were nucleated at these inclusions by fracture of the inclu-
sions, and this happened at rather low strains. From the examination of the fracture
surface of plane strain fracture specimens, by replicas and directly with the
scanning electron microscope, it was concluded that the joining of the large voids
initiated at the large inclusions occurred through the matrix by a mechanism of
void sheet formation. The void sheet formation seems to be developed through
the formation of voids at small particles (0.1 to 0.2 um in diameter) distributed
uniformly in the matrix. These particles should be regarded as a dispersion which
strengthens the matrix.
KEY WORDS: fracture strength, toughness, fractures (materials), rupture

(materials), inclusions, voids, aluminum alloy, evaluation, tests
The brittle fracture of many so-called high-strength aluminum, titanium,

and iron-base alloys has been shown to occur by dimpled rupture [1]. 2 This
mode of fracture, as contrasted, for example, with the cleavage mode, is
associated normally with ductile materials, and various studies have shown
that the fracture process is, generally, nucleation of voids at impurity particles
and the growth and coalescence of these voids during plastic deformation
1 Metallurgy and Materials Science, Carnegie-Mellon University, Pittsburgh, Pa. 15213.
Mr. Tanaka is now with Technical Research Center, Nippon Kokan Steel Co., Kawasaki,
Japan. Dr. Low is a personal member ASTM.
191
Copyright 9 1970 by ASTM lntcrnational www.astm.org
to produce final rupture. These observations on ductile materials suggest that

the fracture toughness of the more brittle, high-strength materials, may be
affected strongly by the density and distribution of impurity particles which
do not contribute to the strength. For this reason a study of the fracture
process in a high-strength aluminum alloy (2014 T6) was undertaken to
identify the void nucleating particles in this material, to determine their
composition, and to suggest means by which they might be eliminated with-
out loss of strength.
Material
A 2-in.-thick plate of commercial 2014 aluminum alloy, heat treated to
the T6 condition, was used to make specimens for fracture testing. The
chemical composition and tensile data of the alloy are listed in Tables 1 and
2.
TABLE 1--Composition of 2014 aluminum-copper alloy (weight percent).
Upper
Cu Si Mn Mg Fe A1 Limits
Composition 3.9 0.5 0.4 0.2 0.7 Ti-0.15

remanent Zn-0.25
Limits . . . . . . 5.0 ~1.2 ~1.2 ~0.8 Max Cr-0.10
Actual . . . . . 4.64 0.84 0.72 0.51 0.2

Analysis . . . . 4.71 0.81 0.73 0.51 0.2
TABLE 2--Typical mechanical properties of 2014-T6 alloy.
Yield Strength, Tensile Strength, Elongation

psi psi in 2 in., ~o
60,000 . . . . . . . . . . . 68,000 10
Fracture toughness tests were made at the National Aeronautics and Space
Administration Lewis Research Center by M. H. Jones and R. T. Bubsey.
Figure l shows the shape and dimensions of the fatigue-cracked toughness
specimens. Toughness measurements were all made in the " T R " testing
direction, that is, the fracture surface lies perpendicular to the thickness of the
plate and the crack propagation direction is parallel to the rolling direction.
A mean value, of six tests ofK~c = 17.6 ksi N/in. +0.4 ksi x/in~, was obtained.
TANAKA ET A L O N LOW ENERGY FRACTURE 193
I 07
)-V-
g
I-
~j
LLI
F,
60 ~
1_ 1.50"
.J
g .05" I
QI:
I
iJ
f
i
LSO" . . . . . . . ! i '
THICKNESS DIRECTION
FIG. 1--Compactfracture toughnessspecimen.
Fractographic Study
Two techniques were used to study the fracture surfaces of the cracked
specimens. One was the observation, with the electron microscope, of two
stage plastic-carbon replicas of the fracture surface; and the other was direct
observation of these surfaces with the scanning electron microscope.
Replicas
Figures 2 and 3 show typical electron microscope fractographs. A dimpled
topography was found over the entire fracture area observed. The most
salient feature to be noted here, is that two widely different sizes of dimples
are present; moreover, they are not distributed randomly but form groups, or
colonies, of large (10 to 30 pm) dimples (A) separated from similar areas by
bands of small (0.5 to 2 #m) dimples (B). In some areas of some of the large
dimples, features typical of a cleavage pattern (C) are observed. These
cleavage patterns are thought to be due to the fracture of an inclusion which
failed by cleavage, nucleated a void, and remains in the void. The greater
proportion (50 to 60 percent) of the fracture surface is covered by large
dimples, and it is very likely, then, that the most important factor leading to
the low toughness of the alloy is the existence of the particles which nucleate
the large size dimples.
Scanning Electron Microscopy

Although of lower resolution than the transmission electron microscope,
the scanning electron microscope (SEM) allows the direct observation of the
fracture surfaces and also observations at much lower magnifications of
University of Washington (University of Washington) pursuant to License Agreement. No further reprod
194 PLANE STRAIN FRACTURETOUGHNESS TESTING
FIG. 2--Two stage plastic-carbon replica o f fracture surface. Large dimples at A, small
dimples at B, cleavage pattern at C.
FIG. 3--Same as Fig. 2.
areas that can then be magnified extensively for more detailed study. This
allows the observer to relate the submicroscopic features seen at high magni-
fications to the grosser features of the fracture surfaces.
Figure 4 shows a low magnification fractograph taken with the SEM of the
fracture surface of one of the toughness specimens. By careful examination
it is possible to distinguish areas with the same large dimples seen in the
replicas and smoother areas which at higher magnifications, Figs. 5 and 6,
TANAKA ET AL ON LOW ENERGY FRACTURE 195
FIG. 4--Scanning electron microscope micrograph showing large dimples and smoother
areas at A (secondary electrons).
FIG. 5--Same as Fig. 4 at larger magnification (secondary electrons). A is the same spot
as in Fig. 4.
FIG. 6 - - S a m e as Figs. 4 and 5 at higher magnification showing small dimples on smoother

areas o f Fig. 4 (secondary electrons). A is the same spot as in Figs. 4 and 5.
show the topography of the small dimples. A marks the same point in the
three micrographs of Figs. 4, 5, and 6. All these micrographs were taken
with secondary electrons. Figures 7 and 8 compare, at low magnification, the
same area taken with secondary electrons and with back scattered electrons.
When back scattered electrons are used, deep shadows appear where sec-
ondary electrons showed areas of small dimples (smooth at low magnifica-
tions). This means that these areas of small dimples are steps on the fracture
surfaces joining areas covered by large dimples that lie at different levels
(back scattered electrons give an effect similar to highly oblique light in the
optical microscope). The reason why these areas did not appear as steps on
the fractographs obtained by replicas is that these steps are too large and the
replicas probably collapse, showing large size and small size dimples at the
same level (see Figs. 2 and 3).
Failure of Large Inclusions

Figure 9 shows an optical micrograph of a mechanically polished section
of the material. Large inclusions, 3 to 10 um in diameter, separated by a
distance of 5 to 15 ~m from each other, may be seen. In some cases, grouping
of these inclusions into colonies were observed, but the general distribution
was quite uniform except for a certain tendency of the inclusions to be
aligned along the rolling direction (marked as R in Fig. 9).
TANAKA ET AL O N LOW ENERGY FRACTURE 197
FIG. 7--Low magnification scanning electron micrograph taken with secondary electrons,
to compare with back scattered electrons in Fig. 8.
FIG. 8--Same as Fig. 7 taken with back scattered electrons.
F I G . 9--Optical micrograph of a mechanically polished surface of the alloy.
Two types of inclusions may be distinguished by their shape and color in

the optical microscope. One with a more rounded shape appears yellowish
under the microscope (clear gray in the micrographs) and the other of a
more irregular form appears as a darker yellow (darker gray in the micro-
graphs).
A preliminary experiment was made to study the nucleation of cracks in
the alloy by bending prepolished specimens in a three-point bending jig. This
method of preparing and deforming the specimen was changed later to tensile
deformation with final polishing after straining in an Instron machine. It was
found that the preparation by polishing seemed to disturb the particle matrix
interface leading to incorrect conclusions about the mode of failure of the
inclusions. However, one valid result obtained was that after a significant
amount of deformation at the bent surface, short cracks appeared which
were not related to the inclusions. By slightly etching the surface with Kellers
reagent (0.5HF, 1.5HC1, 2.5HNO3, 95H20) it was found that these cracks
were located at grain boundaries, as shown in Fig. 10 by arrows. These cracks
did not grow to produce total fracture, and further deformation resulted in
sudden fracture of the specimen. Its fracture surface presented the typical
dimpled topography, already shown, with no clear evidence of intergranular
failure.
Further experiments, which we believe show clearly the nucleation of the
fracture process, were made by observations on tension specimens deformed
in tension by different amounts in an Instron machine. After deformation,
FIG. lO--Optical micrograph. The polished surface was etched to delineate grain bound-
aries after cracks (arrows) had appeared as a result o f bending deformation.
the specimens were mechanically polished very carefully with metallographic

papers through the 0000 grade and finally lapped with a magnesium oxide
powder as the abrasive. Two polished surfaces were observed, one correspond-
ing to the midsection parallel to the tension axis and the other just below the
free surface of the specimen. The ratio of cracked particles to the total number
of particles greater than 1 um in diameter is shown in Fig. 11 as a function of
deformation. The solid circles correspond to the observations made at the
free surface and the open circles at the midsection. It is clear that the particles
5O
I I I I I
o~
40
30
IJU
d
zo
~o
I 2 3 4 5
DEFORMATION (%)
FIG. 11--Fraction o f the inclusions greater than 1 um jbund cracked as a function o f
tensile plastic deformation. Solid circles correspond to observations made at the free surface
o f the specimen and open circles at the midsection.
FIG. 12--Polished surface of fractured tension specimen showing growth of cracks. S is

the tensile axis and R the rolling direction.
start cracking at the early stages of deformation. The fraction of cracked

inclusions increases steadily up to a point where the maximum load is reached,
after which the rate of cracking increases sharply.
The cracks formed at the inclusions were only seen to propagate into the
matrix for the largest deformations, and this was observed more frequently
at the midsection, Fig. 12 (S shows the stress direction). Blunting of the
cracks formed at the inclusions were observed in some cases, and this was
due to failure at the particle matrix interface, as shown in Fig. 13 by arrows.
Some evidence of grain boundary cracking is found in specimens with the
larger deformations. This is shown in Fig. 14 by arrows. The different ap-
pearance of the crack as compared with that in Fig. 10 is believed to be due
to rounding of the edges by polishing after cracking in the case of Fig. 14.
Transmission Electron Microscopy

The small size dimples seen on the fracture surfaces were nucleated very
likely at particles too small to be observed with the optical microscope. More-
over the observation that some grain boundary cracking occurred without
showing typical intergranular fracture appearance in the fractographs sug-
gested that a precipitation free zone around grain boundaries might exist in
the alloy. This is known to lead, in some cases, to grain boundary failure with
a dimpled fracture topograpy [2]. With this in mind, thin films of the alloy
were examined in the electron microscope. The thin films were prepared by a
jet electropolishing technique [3] using a mixture of 2/~ methanol and
nitric acid. Figure 15 shows a typical transmission electron micrograph.

Small particles about 0.1 to 0.2 um in diameter a distance of about 0.5 to 1 um
apart clearly are seen. Because of their size and distribution, these should be
regarded as a dispersion which strengthens the matrix. The work hardening
characteristics of the alloy should be affected especially by them [4]. This
coarse dispersoid, which contributes significantly to the strength of the alloy
in addition to the fine 0' precipitate, has been identified as AI~2(MN,Fe)aSi [8].
At higher magnifications, Fig. 16, the matrix shows a structure formed
by fine elongated precipitates 0.01 to 0.05 um long which are believed to be
the principal hardening precipitates (0'). No precipitation free zone was
found around grain boundaries, but elongated inclusions 0.3 to 0.5 ~m long
were observed, Fig. 17. This finding may well explain the grain boundary
cracking observed in the bending and tension specimens. The fact that the
fracture topography was of the dimple type and did not show any evidence
of grain boundary interface failures may also be understood. In effect, the
existence of grain boundary failure showing a dimpled topography is possible
even if no precipitation free zone exists. If inclusions lie along grain bounda-
ries and either are at small distances between each other, or nucleate cavities
FIG. 13--Polished surface o f tension specimen showing blunting o f the cracks in the
inclusions (arrows). S is the tensile axis.
FIG. 14--Polished surface of tension specimen showing grain boundary cracking (arrows).
FIG. 15--Transmission electron micrographs showing a fine dispersion of particles.
TANAKA ET AL O N LOW ENERGY FRACTURE 203
FIG. t6--Transmission electron micrograph showing dispersed particles and hardening

precipitates.
FIG. 17--Transmission electron micrograph showing small inclusions at grain boundaries.
before the randomly distributed inclusions, a dimpled topography should

show in the fractographs as a result of void growth and coalescence at these
inclusions. In this case, the dimples should not be shallow (that is, approxi-
mately of the thickness of the precipitation free zone, as those found by
Ryder and Smale [5]), and therefore may be difficult to distinguish from
dimples formed at the randomly distributed particles found in the matrix.
Identification of Void-Nucleating Particles

Since the largest part of the fracture surface was covered by large dimples,
the large inclusions, believed to be responsible for them, probably control
the toughness of the alloy. Therefore, it is necessary to determine the elements
present in these inclusions and, if possible, their specific composition.
Polished specimens, in which inclusions were seen easily in the optical
microscope, were first analyzed to determine the elements present in all of the
large inclusions. Figure 18 shows typical micrographs obtained with the
microprobe analyzer. The top left picture shows the specimen current by
which it is possible to show the location of the particles. The remaining
picture show the distribution of each of the specific elements investigated.
Particles containing iron, manganese, and some silicon are found in general,
together with some particles containing large amounts of copper. In these
polished-section micrographs, it is not possible to decide if the particles
contained aluminum because of interference from the aluminum matrix.
Extraction replicas of the fracture surface were then analyzed to compare
with these results on polished surfaces (a description of the technique used
to extract the particles is found in the Appendix). Figures 19 and 20 show
electron micrographs of typical extraction replicas. Many particles were
extracted and seem to remain in the place corresponding to the dimple they
had originated. However, some were found moved from their original posi-
tion. The microprobe showed that the particles contained copper, silicon,
and aluminum, as shown in Fig. 21, and some only silicon. Comparing
these results with those of the polished specimens, it seems that either the
particles containing iron, silicon, and manganese were not extracted from
the fracture surface or that they do not contribute to the formation of dimples.
However, direct analysis of the fracture surface with the microprobe did
detect the incusions which contained iron. This and the fact that both types
of inclusions were found to crack in the deformed specimens indicate that
the inclusions containing iron were not extracted from the fracture surface.
The composition of the particles containing iron, manganese, and silicon
was estimated by comparing the intensity of the microprobe X-rays of the
specific elements.
TANAKA ET At ON LOW ENERGY FRACTURE 205
FIG. 18--Electron microprobe analysis of a polished specimen of the alloy.
The estimated composition of these particles is shown in Table 3 together

with that of a possible intermetallic phase? The agreement is reasonable.
For the particles containing copper, it was possible to obtain electron
diffraction patterns whenever they appeared thin enough in the extraction
replicas. Figures 22a and b shows diffraction patterns for the particles A and
B shown in Fig. 20 and in the microprobe micrograph in Fig. 21. An analysis
of the diffraction patterns obtained of the particles in the extraction replicas
might correspond to A14Cu9 or Cu15Si4 and pure silicon or Cu~Si or CuA12.
3Anderson, W. A., AluminumCompanyof America, private communication.
FIG. 19--Electron microscope micrograph o f an extraction replica.
FIG. 20--Same as Fig. 19.
Therefore, we may conclude that two varieties of large particles are found
in the alloy. One, which has been estimated to be very likely Al12 (FeMn)a Si,
corresponds to the inclusions which in the optical micrographs appeared
with a more irregular form and a darker gray. The other type, which cor-
responds to the more equiaxed, lighter colored particle in the optical micro-
graphs, does not contain iron or manganese, and contains aluminum and
silicon combined with copper and possibly magnesium since magnesium was
not included in the microprobe analysis of these extracted particles.
FIG. 21--Electron microprobe analysis of particles in extraction replica shown in Fig. 20.
TABLE 3--Microprobe estimate of chemical composition of Mn-Fe-Al-Si type particle

(weight percent).
Mn Fe Si A1
12N13 8~10 3~4 > 50
Chemical Composition of:
(Mn Fe) Si A1
All~(FeMn)~Si 32 5 63
Discussion
F r o m the fractographic study made of the fractured toughness specimens

of the alloy, several conclusions may be drawn.
The geometrical correlation between size and distribution of large dimples
and the size and distribution of the large particles readily seen in the optical
F I G . 22 a and b--Electron diffraction patterns of particles A and B of Fig. 20.
microscope made it apparent from the beginning that these inclusions are
responsible for the nucleation of voids leading to the large dimples. Further
evidence is provided by the extraction replicas in which it was possible to
relate in many cases a large dimple with its nucleating inclusion. Although
it was not possible to extract the inclusions containing iron, manganese and
silicon, direct microprobe analysis of the fracture surface, and the fact that
they were found to fracture in the deformed tension specimens, showed that
these inclusions were responsible also for the large dimples.
These large dimples cover more than 50 percent of the fracture surface, and
hence the inclusions which have acted as nuclei of the voids must be signifi-
cantly reducing the toughness of the material.
In general, voids are nucleated by fracture of the inclusions, and, as already
mentioned, all large inclusions regardless of type were observed to crack
in the tensile deformation experiments, suggesting indirectly that all of the
large inclusions were nucleating cavities. Around 10 percent of the particles
were found cracked at strains of 3 percent and 45 percent at strains of about 5
percent. It is very likely then that almost all of the inclusions within the plastic
zone at the tip of the crack of the toughness specimens were cracked well be-
fore the unstable fracture started. Most of the rest of the fracture process
leading to the critical instability must have been growth of these cavities
through the matrix and coalescence with other cavities.
The fractographs seem to show a situation by which several voids nucleated
at large particles coalesce somewhat at the same level and parallel to the
main fracture surface. Thereafter they join with other similar areas by a
mechanism of void sheet formation [8] leading to the rupture of the ligament
between these areas. The void sheet formation mechanism seems to be
developed through the formation of voids at small particles distributed
uniformly in the matrix, as may be inferred by comparing the distances
between small dimples (about 0.5 to 1 tzm, Fig. 6) and that between small
particles, Fig. 15. In some cases, these could have been nucleated also at the
small inclusions found at the grain boundaries. Of the areas covered by small
dimples it is difficult to distinguish the regions which are related to grain
boundaries and those which are not. However, features such as those marked
by the arrows in Fig. 14, could be well related to grain boundary failure.
We would like to point out here, that although large inclusions are un-
doubtedly detrimental to fracture toughness, because the joining of large
voids nucleated at them is made through the matrix by a void-sheet mecha-
nism, the resistance of the matrix to this aspect of crack propagation should
contribute to the overall toughness. The fact that the large inclusions cracked
at rather low deformations means that their main role during the fracture
process may be only to establish the size of the "ligament" or distance
through which the matrix must fail; in fact, this same conclusion may be
drawn from the experiments of Edelson and Baldwin where the volume
fraction of the second phase was important rather than its character. The
same concept is developed in the process zone theory of Krafft [14], and one
of the properties of the matrix which according to experimental correlation,
plays an important role is the strain-hardening exponent of its stress-strain
curve. The higher this is, the more difficult it should be to produce the local-
ized shear plastic deformation necessary to produce the void-sheet mechanism.
One of the microstructural components we have pointed out already that may
be enhancing the work hardening ability of the material is the small dispersoid
seen in the transmission electron micrographs. However, this beneficial
effect may be well counterbalanced if these particles act as easy sites for
nucleation of voids. This may be the case in the present alloy where the small
dimples found on the fracture surfaces are thought to be nucleated at these
small dispersed particles.
Fractographic Observation of the Boundary Between Fatigue Crack and

Dimpled Rupture Regions of Fracture Surface
Figure 23 shows an electron fractograph taken from a two stage plastic-
carbon replica made at the boundary between the fatigue (F) and the dimpled
ruptures (D) regions of the fracture surface. The fatigue region in this material
does not show the striations typical of fatigue. However, the features of the
dimpled rupture region are quite different from those of the fatigue region
which may be seen more clearly in a low magnification fractograph taken
with the scanning electron microscope, Fig. 24.
At the boundary between the two regions, a step may be seen, Fig. 24,
which does not appear clearly in replicas due to collapse of the carbon film
which wilt not support such a step. In any event, a topography different from
both the dimpled rupture and the fatigue region is seen at the boundary
(shown between arrows), Fig. 23. This transition area has been called the
stretched-zone [13], and its topography of the surface is said to be made up
almost wholly of "serpentine glide" due to slip [9]. It is apparent from Fig.
23 that the extent of this region is quite irregular. Two scanning electron
micrographs at higher magnification of the stretched zone are seen in Fig. 25.
The step seen at lower magnification in Fig. 24 now appears to be rounded.
The sketch shown in Fig. 26 shows, we believe, the situation at the tip of the
crack just before critical instability. We therefore believe that the stretched
zone is produced during blunting of the crack tip and perhaps some stable
crack growth. Several correlations have been made recently between the size
of the stretched zone and fracture mechanics data. The result seems to be that
the size of the stretched zone is of the order of GI o/4ay, where GI c is the critical
crack extension force and ~y the yield stress [10]. If we take Gic/2~y as a
Z
>.
>.
t-
O
Z
5
0..<
FIG. 23--Electron microscope micrograph o f a two stage replica o f the boundary between fatigue (F) and overload region (D). r
1,0
2]2 PLANESTRAIN FRACTURE TOUGHNESS TESTING
FIG. 24--Low magnification electron micrograph taken with the scanning electron micro-
scope o f an area similar to that in Fig. 23.
measure of the "crack opening displacement" (COD) at instability, 4 then,

roughly, the radius at the tip of the crack just before instability is of the order
of one half to one times the crack-opening displacement in size.
The formation of the stretched zone must involve some complex slip
mechanism at the crack tip. We would like to suggest here that this mecha-
nism should be similar to that proposed by Rogers [11] to explain the forma-
tion of the lips of a ductile copper tension specimen that failed in a double
cup fashion. The mechanism has been called "ductile cutting." As in the case
of the formation of those lips, two factors, which we believe also are involved
necessarily in the formation of the stretched zone, are large amounts of plastic
deformation necessary to provide the separation or "cutting" of the material
and low triaxiality which inhibits void growth during its formation.
At some point during the formation of the stretched zone, void growth
ahead of the blunting crack tip should commence. This may begin between
the free surface at the crack tip and the elastic plastic boundary ahead of it.
The exact location should meet the requirements of high triaxiality and
extensive plastic deformation such that the work performed by the hydrostatic
component of the stress at that point is maximum. Unfortunately, as com-
pared to cleavage fracture where the direction of crack propagation may be
identified by a study of the river patterns [13], in dimpled rupture it may
prove difficult to find experimentally the sequence of events leading to un-
stable fracture.
4 It is assumed that the uniaxial yield strength ~u is doubled under plane strain conditions
(G. Irwin, private communication).
TANAKA ET AL ON LOW ENERGY FRACTURE 21 3
FIG. 25--High magnification o f areas in Fig. 24 at the boundary.
STRETCHED
ZONE
FIG. 26---Sketch showing blunting o f the crack tip and the stretched zone.
Conclusions
1. All of the fractographic evidence indicates that the fracture process is
dominated by cavity nucleation, caused by fracture of the large impurity
inclusions tentatively identified as All~(Fe,Mn)~ Si and a combination of
excess copper with either aluminum, silicon, or magnesium
2. From their size and distribution, it is unlikely that these large inclusions
contribute to the strength of the alloy and presumably greatly decrease its
fracture toughness.
3. These first two conclusions suggest that modification of the composition
to reduce the density of the two major types of large inclusions should bring
about an improvement in fracture toughness without significant loss of
strength.
4. These conclusions should be tested by making up a 2014 alloy, with its
composition so modified as to eliminate or minimize the number of these
two types of inclusions and testing its strength and fracture toughness after
a suitable heat treatment.
5. It is recognized that some microstructural constituents which strengthens
the alloy (small dispersoid) may act also as void nuclei and therefore affect
the toughness of the material.
Acknowledgment
This investigation was made possible by a Research Grant from the
National Aeronautics and Space Administration, Grant No. N G R 39-087-
300.
APPENDIX
The technique used to extract particles from the fracture surface while
keeping them related to the dimple they had produced and appeared on
the replicas was the following: The fracture surface was electroetched in a
solution of 1/~ nitric acid 2/~ methanol, by applying a voltage of 0.5 V for 30 s
and left without voltage for an extra 60 s. After carbon deposition the surface
was coated with a special wax (Ladd Laboratories) which was melted and
then poured over the carbon-coated fracture surface. Scoring the surface into
small squares and electropolishing with 10.5 V at - 2 5 C freed the small
squares which then floated on the surface of the solution. After retrieving the
replica fragments from the solution the wax was removed by soaking the
fragments in xylene. By this method it was frequently possible to extract
nonmetallic particles from the fracture surface and, as shown in Figs. 19
TANAKA ET At ON LOW ENERGY FRACTURE 215
and 20, retain them in position with respect to the dimples which they nu-
cleated. The method successfully extracted the inclusions containing copper,
silicon, and aluminum but failed to extract those containing iron, manganese,
and silicon.
We m a y mention that attempts to make g o o d extraction replicas by an
oxide replica technique developed by Keller and Zeigler [7] gave negative
results. This may be because the fracture surface was too r o u g h and the
thickness o f the oxide layer too irregular to obtain any good piece o f oxide
to examine in the electron microscope.
References
[1] Low, J. R., Jr., Engineering Fracture Mechanics, Vol. 1, No. 47, 1968.
[2] Kaufman, J. G. and Holt, M., Alcoa Research Laboratory, Technical Paper No 18,
1965.
[3] Schoone, R. D. and Fiscione, E. A., Review of Scientific Instruments, Vol. 37, No. 1351,
1966.
[4] Ashby, M. F., in Strengthening Methods in Crystals, Kelly, A. and Nicholson, R. B.,
eds., Elsevier, New York, 1969.
[5] Ryder, D. A. and Smale, A. C., in Fracture of Solids, Interscience, New York, 1963,
p. 237.
[6] Phragmen, G., Journal of the Institute of Metals, Vol. 77, No. 489, 1950.
[7] Keller, F. and Zeigler, A. H., Transactions, American Institute of Mining, Metallurgi-
cal, and Petroleum Engineers, Vol. 156, No. 82, 1944.
[8l Rogers, H., Ductility, American Society for Metals Seminar, 1968.
[9] Beachman, C. D., and Meyn, D. A., Memorandum Report 1547, Naval Research
Laboratory, 1964.
[10] Brothers, A., ASTM Task Group of E-24, Philadelphia meeting, March 1969.
[11] Rogers, H., Transactions, American Institute of Mining, Metallurgical, and Petroleum
Engineers, Vol. 218, No. 498, 1960.
[12] Wolff, U. E., Reactor Primary Coolant System Rupture Study, Research and Develop-
ment Report GEAP-5474, Atomic Energy Commission, Jan. 1967.
[13] Low, J. R., Jr., in Fracture, Wiley, New York, 1959, p. 68.
[14] Krafft, J., Applied Materials Research, Vol. 3, No. 88, 1964.
W. F. Brown, Jr.,: and J. E. Srawley I
Commentary on Present Practice

Myself when young did eagerly frequent
Doctor and Saint, and heard great argument
About it and about: but evermore
Came out by the same door where in 1 went.
Omar Khayyam
REFERENCE: Brown, W. F., Jr., and Srawley, J. E., "Commentary on Present

Practice," Review o f Developments in Plane Strain Fracture Toughness Testing,
A S T M STP 463, American Society for Testing and Materials, 1970, pp. 216-248.
ABSTRACT: Emphasis on specimen size requirements and attempts to correlate

K~o with simpler mechanical test results because these are areas of considerable
controversy among those engaged in plane strain fracture toughness testing are
presented. Problem areas in connection with plane strain fracture toughness
testing indicate that the Tentative Method of Test for Plane Strain Fracture
Toughness of Metallic Materials (E 399-70 T) may need revision before being
advanced to a Standard and that additional experimental data are needed in order
to make a judgment concerning the need for revision.
KEY W O R D S : fractures (materials), toughness, fracture strength, plane strain

fracture toughness, tests
Nomenclature
K Stress intensity factor associated with a crack in a linear elastic
body
K: Opening mode (Mode I) stress intensity factor as defined in A S T M
Tentative Method of Test E 399-70 T, 4.1
Kie Plane strain fracture toughness as defined in ASTM Tentative
Method of Test E 399-70 T, 4.2
KQ Conditional result of a K:c test as defined in ASTM Tentative
Method of Test E 399-70 T, 8. I
K: Maximum stress intensity in fatigue cycle
Y Dimensionless linear elastic function of a/W
Applied stress remote from a crack
1 Chief, Strength of Materials Branch, and head, Fracture Section, respectively, National
Aeronautics and Space Administration, Lewis Research Center, Cleveland, Ohio 44135.
Personal members ASTM
216
Copyright 9 1970 by ASTM International www.astm.org
BROWN AND SRAWLEY ON COMMENTARY ON PRESENT PRACTICE 217
aYs 0.2 percent offset yield strength in a tension test as defined in ASTM
Standard E 8-68
P Applied load
P, Secant intercept load as defined in ASTM Tentative Method of
Test E 399-70 T, 8.1.1
E Young's modulus
B Specimen thickness
W Specimen width (depth of a bend specimen)
a Effective crack length
ao Initial crack length
v Crack mouth displacement
to Initial tip radius of a simulated crack
R Ratio of maximum to minimum load in a fatigue cycle
CVN Charpy V-notch energy
4, Elliptical integral function of ratio of the length to depth of a
semielliptical surface crack
.4 Initial crack area of a precracked Charpy specimen
The purpose of this contribution is to discuss problem areas and recent

developments in plane strain fracture toughness testing with particular
reference to the papers contained in this volume. We have placed special
emphasis on specimen size requirements and attempts to correlate Ktc with
simpler mechanical test results because these are areas of considerable con-
troversy among those engaged in plane strain fracture toughness testing.
From what follows it will be evident that we believe the Tentative Method
of Test for Plane Strain Fracture Toughness of Metallic Materials (ASTM
Designation: E 399-70 T) may need revision before being advanced to a
Standard and that additional experimental data are needed in order to make
a judgment concerning the need for revision.
Specimen Size Requirements

The most important issue among those interested in plane strain fracture
toughness testing is the specimen size required for a valid KI~ test. The
larger the value of the material property (K~~/ays)2 in inches, the larger will
be the thickness and other dimensions of the smallest specimen necessary
to obtain a useful approximation to the value of K~c, where KIo is the plane
strain fracture toughness in ksi. in. 1/~ and gYs is the ASTM 0.2 percent offset
tensile yield strength in ksi. Conversely, the greatest value of K~c that can be
determined usefully with a specimen of given dimensions is limited in propor-
tion to the yield strength and the square root of which ever of the specimen
dimensions is most critical (either thickness or crack length in an ASTM

standard specimen). The size requirements as stated in the 1970 E 399
Tentative Method of Tesff were developed empiricaUy [1,2] 3 from a rather
limited amount of data, largely confined to steel and aluminum alloys. It is
evident from some of the papers in this volume, that the present size require-
ments as stated in ASTM Method E 399-70 T, 8.1.5, are considered excessive
particularly by those who are interested in testing low yield strength high
toughness alloys and those who wish to use small specimens for screening
purposes.
Arguments for reducing the size requirements are presented by Rolfe
and Novak in a discussion to the paper by Jones and Brown, and also by
Steigerwald in his paper. We do not believe the data presented support re-
duction of the size requirements, and the reader is referred to authors' reply
to the discussion by Rolfe and Novak and to the discussion of Steigerwald's
paper for reasons supporting this belief.
On the contrary, we have maintained consistently that the present size
requirements are barely sufficient and should be increased, not reduced. Our
opinions in this respect are reinforced by the data presented in the paper by
Jones and Brown and in the paper by May. The paper by Jones and Brown
makes three pertinent points, namely:
1. The value of KI~ measured in accordance with the ASTM Method E
399-70 T can vary significantly with variations in crack length and specimen
thickness. The magnitude of this variation will depend on the material
properties, and for the 4340 steel investigated it would be negligible if the
requirement for thickness were twice that of ASTM Method E 399-70 T.
2. Deviations from the size requirements and specimen proportions speci-
fied in ASTM Method E 399-70 T can result sometimes in significant over-
estimates of KI c.
3. Unless carefully planned, investigations designed to establish the size
requirements can give misleading results due to interactions between crack
length and thickness.
The results presented by May for Hylite 50 are particularly interesting
since they represent the first published data from a systematic study of the
effects of crack length and specimen thickness in plane strain fracture tough-
ness tests on a titanium alloy. Figure 10 of May's paper shows a large varia-
tion in KQ with specimen thickness. An average value of KQ = 51 ksi~v/in.
is established by specimens exceeding a thickness of 0.6 in., and in this
range the KQ values are constant within the limits of scatter (approximately
49 to 56 ksiv~in.. If 51 ksix/in, is regarded as the true K~c value, the KQ
values for the thinnest specimens tested (B = 0.1 in.) are less than one half as
2 T h i s Tentative M e t h o d a p p e a r s at the back o f this volume.
great, and all but one of these results is valid by the present ASTM Committee
E-24 Test Method. In fact, the trend of the valid Kt c values is to decrease
with thickness below about 0.5 in. If the size requirements were based on the
K value calculated from the maximum load, 4 the results from all but two of
the specimens with thicknesses less than 0.2 in. would be invalid. These two
specimens have thicknesses of about 0.100 in. However, four specimens with
thicknesses between 0.252 and 0.257 would still give valid results even though
their Ko values range from 34 and 43 ksi~v/i~. If the thickness requirement
were doubled (that is, 0.2B > KQ2/ffys 2) and KQ selected to correspond to 2
percent effective crack extension (as in E 399-70 T ) t h e n all but one result
from specimens with a thickness less than 0.5 in. is invalid. This one result is
from a 0.100-in.-thick specimen and was marked as invalid on the basis of the
supplementary requirement of 8.1.2 in ASTM Method E 399-70 T.
The results presented in the papers by May and by Jones and Brown
certainly do not support arguments for relaxing the present size requirements
in ASTM Method E 399-70 T, 8.1.5, but show that further review of the size
requirements is necessary before the present tentative method is adopted as
an ASTM standard. With this in mind, it is worthwhile to review the basis of
the specimen size requirements and to indicate the type of additional informa-
tion that is needed to ensure that ASTM Committee E-24 Method of Test
for KI ~ is compatible with its basis in linear elastic fracture mechanics.
Fundamental Concepts
The concept of K~r due to Irwin et al [3] entails two independent size
effects associated with the two essential conditions of: (1) linear elastic
behavior of the material over a field which is large compared with the plastic
enclave that surrounds the crack front and (2) tritensile plane strain con-
straint within this enclave and somewhat beyond it. T o a practical degree of
approximation the specimen thickness must be sufficient to maintain this
constraint up to the test load that corresponds to K~r and the other dimen-
sions must be sufficient to maintain boundary conditions around the plastic
enclave which match the crack tip stress field components of a linear elastic
model.
Any useful Krc test method must provide for these basic limitations by
restriction of the valid range of a test to that for which the plastic enclave
size factor (Ki/crys) 2 is within some specified fraction of the most critical
dimension of the specimen. In ASTM Method E 399-70 T, for example, this
factor is restricted to the smaller of 0.4a0 or 0.4B, where a0 is the initial crack
length and B is the specimen thickness, and if the material property (KI c/ays) 2
is beyond this limit the specimen is not large enough in at least one respect to
determine K1 c for the material concerned. The crack length limit in ASTM
4 Data by private communication from M. J. May.
Method E 399-70 T also entails other specimen dimensions such as the width
W (depth of a bend specimen), which are held proportional to a0 within fairly
narrow limits. The two independent limits for crack length and thickness were
established empirically [1,2], and their coincidence when a0 = B is not of
special significance. These limits will vary somewhat with the properties of
the material and must be chosen sufficiently low that the probability of
obtaining spurious results on a new material is minimized. When the pro-
visional result KQ of ASTM Method E 399-70 T fails the provisions of 8.1.5
the only recourse is to test larger specimens of the material concerned, and,
if this is not possible, then K1 r cannot be determined for that material by the
ASTM Method E 399-70 T Method of Test.
Irrespective of the ASTM Test Method, however, it has to be accepted
that it will frequently not be possible to determine K I , for a given material
in any meaningful sense, usually because the material is not available in
sufficient thickness, but sometimes for another reason. The concept of K1 c is
fundamentally incompatible with a test method that could be applied to all
available forms of materials, and neglect of this fact can only result in mis-
representation of the relative merits of different materials. The issue, there-
fore, is whether or not the ASTM Test Method could be made less restrictive
with regard to specimen dimensions without significant risk that spurious
results would be obtained in some cases. In our opinion there is no con-
vincing evidence that the provisions of ASTM Method E 399-70 T are unduly
restrictive; on the contrary, there is reason to suppose that they are not
restrictive enough for some materials which are of technological importance.
The test method should be applicable to any material that is available in
sufficient bulk, and its provisions ought to be restrictive enough to guard
against spurious results from the least favorable material. Of course, this
means that the same provisions would be more than enough in other cases,
but it is not feasible to discriminate between the more and the less favorable
materials. It follows that the discovery of any number of favorable cases
does not justify any relaxation of the provisions of the test method, though
the discovery of a single unfavorable case may make it necessary to further
restrict the test method. There is a further consideration which should not be
neglected. Although the direct application of the ASTM Method E 399-70 T
is restricted unavoidably to those materials that can be obtained in sufficient
bulk, it does serve as a primary reference method for other fracture toughness
tests that are based more empirically but less restricted in scope of applica-
tion. This is an important reason why any margin of error in the provisions
should be on the side of restrictiveness rather than uncertainty about the
meaning of the results.
We do believe that the 1970 ASTM Tentative Method E 399 needs some
revision before it is offered to the ASTM as a standard method, but the kind
of revision we feel is needed would increase the restrictiveness of the method

to some extent, not decrease it. Our concern here is that there are circum-
stances in which it is possible for the operational definition of KI c in the 1970
ASTM Tentative Method to produce spurious results that are essentially a
consequence of plastic enclave deformation, yet are accepted as plane strain
fracture toughness values by the provisions of the method. Some of these
spurious results are substantially less than the Kr~ values of the materials
concerned, but it is quite possible to obtain a spurious result that happens to
be close to Krc by compensation of effects. When a specimen is too thin the
crack extension resistance in terms of K will be greater the smaller the thick-
ness, but the plastic deformation resistance in terms of K will be less the
smaller the ligament length (W - a0), and also the smaller the thickness.
These effects influence the secant intercept load P, in opposite directions, and
can compensate one another almost exactly. With the present restriction of
the plastic enclave size factor (K~/ays) 2 to be 0.4B or less and where a0 < B
it is unlikely that a result greater than KI ~ could be admitted as valid, but any
substantial relaxation of this restriction would greatly increase this risk.
These dubious results have on thing in common; the secant intercept load
P, is substantially lower than the ultimate failure load for the specimen.
This suggests that the nonlinear component of displacement at Ps (which, by
definition, is 5 percent of the total displacement) might be due largely to
plastic deformation in at least some instances. It is true that this contingency
was anticipated, and that it is controlled supposedly by the supplementary
provision of section 8.1.2 of the 1970 ASTM Tentative Method (explained in
Ref 2). But experience has shown that the procedure of this provision is
impractically delicate in relation to the precision required of the test record.
There may be more practical alternatives, the simplest being to limit the
ratio of ultimate load to P, for a valid test to some value near unity, say
1.1. However, this procedure would not have eliminated the thickness de-
pendence of valid K~c values reported in the paper by May. We feel that it is
necessary to obtain specific information about the complex effects of plastic
deformation in K~. tests before any decision is taken about the revision of the
test method, and we have started a research program at the National Aero-
nautics and Space Administration-Lewis Research Center for this purpose.
While our preliminary results are not sufficiently refined to resolve the ques-
tions about the test method, they serve to illustrate the nature of the data
we expect to obtain and its relevance to the test method.
Effects of Plastic Deformation in Kic Tests

In a properly conducted ASTM Method E 399-70 T KI c test the nonlinear
component of the displacement as a function of load is the sum of two parts,
one due to plastic enclave deformation, and the other to actual crack exten-
sion. To examine the detailed behavior of a specimen during a test it is con-
venient to derive from the direct test record a plot of crack extension resistance
in terms of K against effective (or apparent) crack extension (see, for example,
the paper by Jones and Brown). The effective crack extension is simply a
function of the ratio of load to displacement, P/v, and of the parameters a0
and I4z which are constants of the test, and it represents the sum of the effects
of the actual crack extension and of the plastic enclave deformation. To
really understand what has happened at all stages of the test it would be
necessary to separate and compare the components of the effective crack
extension as functions of K, but the test record of P against v does not contain
sufficient information for this to be done. Our approach is to obtain the
necessary additional information by two methods: supplementary tests on
specimens which have blunt simulated cracks, so that crack extension is
deferred to beyond the range of interest, and digital machine computations on
finite element models to simulate tests in which specimens are deformed with-
out crack extension. These two methods are complementary rather than com-
pletely comparable because it is not yet feasible to employ a three-dimensional
finite element model with sufficiently refined elements. But the two-dimen-
sional model can provide results for the plane states of both strain and stress,
which should constitute upper and lower bounds on experimental results for
specimens of any thickness. The tv, o methods therefore provide a useful
check on one another. The computational program can employ any monotoni-
cally increasing stress-strain relation, and therefore can simulate a wide
range of practical materials rather closely. The program was developed by
J. L. Swedlow, 5 who collaborated with E. Roberts 6 to adapt it to our purpose.
The relation between the load and the plastic component of displacement
depends on a number of independent parameters, which are of two kinds:
specimen dimensions and plastic properties of the material. For specimens
with proportions otherwise fixed, the dimensions concerned are: B, a0, I4z,
and the initial tip radius of the simulated crack, to when this is not negligible.
It should be possible to adjust results to compensate for the effect of to from
information obtained in a preliminary study in which to is varied. The plastic
properties of the material can be expressed in various ways, and it is appro-
priate to select the 0.2 percent offset tensile yield strength, ~ s , as the major
plastic property since it is used in the ASTM Test Method. Residual dif-
ferences between materials then need to be considered in terms of parameters
of the second order of importance (such as the ratio of yield to ultimate
tensile strength, or some other index of strain hardening), and perhaps higher
Professor, Department of Mechanical Engineering, Carnegie Mellon University,
Pittsburgh, Pa.
6 Strength of Materials Branch, NASA-Lewis Research Center, Cleveland, Ohio.
orders. In view of the importance of oys it is convenient to use the plastic

enclave size factor (K~/ays) 2 as the resistance variable instead of K~, as in
Fig. 1. This factor (K1/crys) 2 is equal to (YPa~/2/BWays) 2, where Y is the
dimensionless linear elastic function ( K B W / P a ~/2) of a / W , and a is the ef-
fective crack length obtained from another dimensionless linear elastic
relation between a / W and E B v / P ; E is Young's modulus. These dimensionless
relations are given for a number of cases by Brown and Srawley [1].
Figure 1 is an example of the results obtained by the finite element method
when the internode distance at the crack tip was 1/32 of the crack length.
The element map has since been refined, and we expect to obtain results for an
internode distance at least as small as 1/320 of the crack length. This refine-
ment is necessary to obtain accurate results for deformations within the range
of interest. However, the available information will serve to illustrate several
important points. The material simulated was a 18Ni maraged steel with a
yield strength of 186 ksi and an ultimate tensile strength of 200 ksi, hence
designated Steel 186/200. The engineering stress-strain curve shown in Fig. 2
is fairly typical of such high-strength steels in that there is little strain harden-
ing beyond about 1 percent total strain9 In Fig. 1 the dashed line represents
Irwin's [5] estimate of the plane strain plastic "correction" to the crack
5S / teel 1861200
/J
/ //plane strain J J Steel 1861200
.E .4 tress
e,J>-
%/ 9 --
.p.o
._~
,.4 .2
~u
(J
._~ /
_~ n estimate
/ I I I I
950 .51 .52 .53 .54
Effectivecrack length, a, inch
FIG.l--Effect of Plastic enclave deformation obtained by finite element simulation of
tests of A S T M Kzc bend specimens: W = 1 in., ao = 0.5 in., material stress-strain relation
maraged steel 186/200 (see Fig. 2). Dashed line represents lrwin's 1961 estimate of the
plane strain plastic "correction" to the crack length as proportional to the plastic enclave
size factor.
Engineering stress-strain curves

Steel 1861200 Pseudosteel 981200
200
ksi
150
I00
50
I
.002 .004
I
.006 .008
l' I
.01
Strain
0
I
.05
I
.10
I
.15
I
.20
Strain
FIG. 2--Engineering stress-straincurves used in finite element digital computer program
for simulation deformationofKl~ bendspecimens.
length which is, in round figures, (Ki/avs)2/18. This estimate is evidently not
a good approximation to the plane strain results obtained by finite element
simulation, and it should be appreciated that the estimate was not intended
for this purpose, but rather to adjust the value of KI to account for relaxation
of stress around the crack front. Nevertheless, the Irwin estimate commonly is
employed for other purposes than that for which it was intended originally,
and therefore needs to be considered in the present context.
The fact that the effect of plastic enclave deformation in terms of effective
crack extension may not be directly proportional to (K~/ays) ~ is indicated
by the experimental (dashed) curve in Fig. 3 for the same material. This curve
lies between the (full) curves for plane strain and plane stress, as would be
expected. The specimen from which this curve was obtained had W = 1 in.,
a0 and B -- 0.5 in., and to = 0.03 in. (the finite element results are essentially
dimensionless, but the curves represent a specimen with W = 1 in. and a0 --
0.5 in., so that they can be compared with experimental results). The experi-
mental curve is closer to the plane stress curve than it would be if to had been
small enough to be negligible, but it is nevertheless clear that the two methods
give consistent results, and that these results do not agree with the Irwin
estimate. It would be unwise, however, to draw any firm conclusions with
regard to the ASTM Test Method from these preliminary results. We need
results that are more refined, and also more extensive with regard to the
effects of each of the parameters: ao, B, a o / W , to~B, and at least a second order
plastic property.
We have obtained preliminary results which suggest, fortunately, that
strain hardening may not be a strong factor in the effect of plastic enclave
deformation. These are shown in Fig. 4, where the dashed curves are again
the plane strain and plane stress curves for Steel 186/200, and the full curves
were obtained with a stress-strain relation which was devised to provide a
marked contrast in strain hardening, designated Pseudosteel 98/200 and
shown in Fig. 2. Both stress-strain curves in Fig. 2 have the same Young's
modulus (27,000 ksi), proportion limit (80 ksi), and ultimate tensile strength
(200 ksi), but Pseudosteel 98/200 has an essentially constant strain hardening
"exponent" of 0.2 (in terms of plastic strain, not total strain), which estab-
lishes the yield strength of 98 ksi. This degree of strain hardening is well
beyond the range of practical steels of the same strength level; the "exponents"
for such steels are less than 0.1. The corresponding curves for the two different
materials in Fig. 4 differ by no more than 25 percent, far less than the dif-
ference between plane strain and plane stress for either material. It therefore
seems likely that the more refined and extensive results to be obtained will
confirm that strain hardening is indeed a second order factor, though perhaps
not entirely negligible.
5 9
9F /Plane strain / Experiment /
/ / / / B: 0.5 in. / Pane
,_. 3 L/ // " J
., l,,'J
~_'~ 1~ , ~ Material; Steel1801200

0,,[,, ] I 1 I
.50 .51 .52 .53 .54
Effectivecracklength, a, inch
FIG. 3--Experimental data on plastic enclave deformation of ASTM Kz. bend specimen
with W = 1 in., B = 0.5 in., ao = 0.5 in., to = 0.03 in., material: Steel 186/200 compared
with finite element simulation data.
5 98/200 186/200 98/200 186/200

lane strain
c~
'I-//
/ // /
/ ..-"
t,'/" Planestress
i r162162
Z"
ol I I I I
.50 .51 .52 .53 .54
Effective crack length, a, inch
FIG. 4--Comparison of finite element simulation data for the two markedly contrasted
stress-strain curves shown in Fig. 2, Steel 186/200 and Pseudosteel 98/200.
To summarize our views on the issue of specimen size, we believe that the
available evidence suggests that the provisions of the 1970 ASTM E 399
Tentative Method may not be sufficiently restrictive to ensure that spurious
results could not be obtained. Almost certainly they are not unduly restric-
tive in any respect. In order to clarify this issue we have started a research
program to determine the effects of a number of factors on the plastic enclave
deformation which occurs in K~c test specimens. Hopefully the results from
this program will assist in simplifying the test method and establish K1 ~ as a
reference fracture property. The preliminary results shown here are indicative
only of the nature of the task and its complexity, and should not be used to
derive any firm conclusions.
Face Grooving
Before leaving the subject of specimen size requirements it is advisable to
comment briefly on the use of face grooves in plane strain fracture toughness
test specimens. Sometimes it has been claimed that face grooving will reduce
the thickness necessary for a plane strain fracture test by enhancement of the
transverse constraint. However, the available data do not support this claim
nor do they show any advantage in face grooving insofar as interpretability
of the record is concerned. Face grooves introduce complexities into the
crack tip stress field which are not well understood, and we would not recom-
mend their use. For further details the reader is referred to discussions of this
subject by Brown and Srawley [1] and by Srawley [4].
Specimen Preparation and Test Procedure

The papers by Heyer and McCabe and by May reported results of inter-
laboratory test programs involving altogether some 35 industrial and govern-
ment laboratories. These programs were very helpful in defining problem
areas and indicating where certain changes were desirable in the Proposed
Recommended Practice (Bend Specimen). The following section will sum-
marize the problems revealed by the cooperative programs as well as some
which have been encountered in our laboratory. Consideration of these
problems led to certain changes in the proposed practice, and these are re-
flected in the ASTM E 399-70 T Tentative Method of Test for Plane Strain
Fracture Toughness of Metallic Materials.
Crack Starter Configurationand Displacement GageLength

Objections have been made to the chevron crack starter notch that was
specified in the Proposed Recommended Practice (Bend Specimen), and
May reports that few laboratories in the United Kingdom use this configura-
tion. One advantage of the chevron is the very high K level produced at its
tip which aids in starting the fatigue crack. This eliminates the need for sharp
notches as would be required at the end of straight-through slot configura-
tions. A disadvantage of the chevron is the time required to develop the
fatigue crack between its arms. The ASTM E 399-70 T Method of Test, 6.2.2,
now permits both straight across and chevron types of crack starter configura-
tions.
Recently high precision measurements were made on compact tension
specimens having standard proportions to determine the influence of starter
slot width and displacement gage length on the crack mouth displacements [6].
The results showed an increase in displacement per unit load per unit thick-
ness as the starter slot width was increased from 0.006W to 0.10W. However,
the rate of change of displacement per unit load per unit thickness with re-
spect to a/W was practically independent of slot width, and therefore the
secant offset corresponding to an effective crack extension of two percent
will be independent of slot width within the range investigated. Also, varia-
tions in displacement gage length between 0.17 W and t.2 W were investigated
and shown to have no effect on the measured displacements. Based on these
results the starter slot width limit for both the bend specimen and for the
compact tension specimen has been increased from a maximum of 0.05 W to
W/16. The displacement gage length for the compact tension specimen has
been increased from a maximum of W/4 to any value up to 1.2W. However,

the maximum displacement gage length for the bend specimen remains at
I4"/4.
Fatigue Cracking
No feature of the Proposed Practice (Bend Specimen) caused as much
difficulty as fatigue cracking. The major difficulties were of three general
types: (1) nonuniform crack fronts that did not meet the straightness re-
quirement, (2) deviations from the requirement that the final 0.050 in. of
fatigue crack extension be produced in not less than 50,000 cycles, and (3)
crack fronts that were outside the specified a/I4" limits. The latter difficulty
was associated most likely with inexperienced operators and will not be dis-
cussed here. The former difficulties, however, deserve further comment.
With one exception neither of these could be associated clearly with the type
of fatigue machine or fixturing used by the participants in the interlaboratory
programs. This one exception was reported by May, who stated that cam
driven cantilever bending tended to produce a far higher portion of unsatis-
factory fatigue cracks than did three- or four-point loading.
Nonuniformities in Fatigue Crack Development

It is important to realize that substantial curvature or irregularity in fatigue
crack front development is unavoidable when produced by nonuniformities
in the fracture properties of the metal. Such nonuniformities probably explain
the behavior of the 4340 steel specimens reported by Heyer and McCabe
in their Fig. 2. These specimens nearly all exhibited crack fronts which were
most advanced at the specimen surfaces. In our experience fatigue crack
fronts are most advanced usually at the center of the thickness. Occasionally
a material is so nonuniform in its fracture properties that extremely irregular
crack fronts are produced, and no meaningful value of Kz, can be obtained.
This situation is sometimes encountered in tests on weldments.
On the other hand, nonuniform crack fronts can result from nonuniform
loading of the notch cross section, for example, when the clamping pressure
in cantilever bending is not uniform. As pointed out by May, these effects are
minimized by use of three- or four-point bending.
The crack length specified in the Recommended Practice (Bend Specimen)
is an average of two measurements at the specimen surfaces and one at the
center. This has been modified in ASTM Method E 399-70 T, 7.2.3, which
specifies the crack length shall be taken as the average of three readings, one
at the center of the thickness, and one at each quarter thickness position
either side of center. This new specification recognizes that the central portions
of the crack front are involved primarily in the crack extension occurring in
the Kz o test.
Fatigue Crack Sharpness Requirement

The lower the maximum stress intensity level at which the fatigue cracking
process is finished, the sharper will be the crack. The KQ value measured in
the subsequent toughness test will decrease with increasing fatigue crack
sharpness up to a point beyond which further increases in sharpness will have
no effect on K~,. At the time the Proposed Practice (Bend Specimen) was
formulated there were very little data on which the necessary control of fatigue
cracking conditions could be based, and the requirement that the final 0.050
in. of fatigue crack be produced in 50,000 cycles was a cautious choice. The
requirement was stated in terms of an average rate of fatigue crack propaga-
tion rather than specifying a maximum stress intensity so that satisfactory
control of crack sharpness could be obtained without an accurate knowledge
of the K levels during the fatigue process. Since that time additional in-
formation has been obtained which has led to a modification of these re-
quirements, but more data are still needed.
Details of the fatigue cracking procedure were reported by Heyer and
McCabe for six of the laboratories cooperating in the ASTM sponsored
interlaboratory program. On the basis of this information there did not
appear to be any correlation between the measured K ~ value and deviations
from the requirement of 0.050-in. crack extension in 50,000 cycles. This
observation suggested that this requirement might be relaxed, but the data
from the interlaboratory program was not sufficient to establish a different
requirement.
Information concerning the effects of fatigue cracking conditions on K~o
is given in the paper by May (Fig. 5). These data show the influence of the
maximum stress intensity in fatigue (Ks = AK) on the apparent K~c values
and indicate that KI/K~, ratios as high as unity do not affect the apparent
K~e values. Recent data 7 from our laboratory for a 300 grade maraging steel
(avs = 283 ksi), Fig. 5, show substantially larger effects of Ks on the apparent
K~, values, and indicate fatigue cracking should be carried out in such a way
as to ensure the last portion of the crack extension was produced with a KI/K~c
ratio less than about 0.60.
The effect of Ks on KQ depends on the material and additional data should
be obtained on other materials. In the meantime the KJKro ratio has been
set at a maximum of 0.60 in the 1970 revision of the Method of Test E 399,
6.5.2.
It is of course not possible to determine the KJK~ ratio before fatigue
cracking. For this reason ASTM E 399-70 T Method of Test, 6.5.2, gives
r Data for the same steel was published in A S T M S T P 410 (Fig. 31), and these early
results indicated a Ks/Kto ratio less than about 0.35 was necessary to ensure that varia-
tions in Ks would not influence the measured toughness values. We have determined since
that these data are in error. Specificallythe indicated fatigue loads were considerably below
the actual loads. The new data shown in Fig. 5 were obtained using specimens cut from
those previously tested.
I00
]8 Ni 300 grade maraging steel
0.20 inch sheet; age 900~ F, 8 hr
Oys = 283 ksi
90
- Z~K=O.9 .9/
.~ 70-
o ~ - - - i ~
Kf O ~ ~
60--
50 I ~ I I I I I
0 I0 20 30 40 50 60 70
KI, ksi,
FIG. 5--Influence of maximum stress intensity in the fatigue cracking operation of KQ for
a maraging steel.
another requirement, namely, that the critical final 2.5 percent of the overall
length of notch plus crack shall be produced with the ratio of KJE less than
0.0012~/~. For the data shown in Fig. 5 this is consistent with a KJK~c <
0.60. It should be noted that these new requirements, in general, will cor-
respond to fatigue crack growth rates roughly ten times those originally
specified in the Proposed Recommended Practice (Bend Specimen). These
new requirements focus attention on control of the maximum stress intensity
during fatigue and are preferable to a control based on an averaged growth
rate since the latter would not reflect an "overload" that could occur in the
last few cycles of fatigue.
The E 399-70 T Method of Test, 6.5.1, specifies that the maximum K
level in the final stage of fatigue cracking shall be known within 5 percent.
This requires sufficiently accurate measurement and control of the load as
well as a sufficiently accurate K-calibration for the specimen and loading
arrangement used. If the loading arrangement in fatigue is similar to that
used in the subsequent K~c test then the K-calibrations in ASTM Method
E 399-70 T, 8.1.3 and 8.1.4, will have more than sufficient accuracy. It should
be noted that ASTM Method E 399-70 T requirement for a symmetrical load
distribution about the notch does not permit the use of cantilever bending.
Generally, either displacement or load controlled fatigue may be used. Dis-
placement controlled fatigue has the advantage that the load will decrease
as the crack extends, thereby making it easier to meet the requirements for
the critical final stages of fatigue cracking. However, the relationships neces-
sary to calculate load from the measured displacement and crack length will
depend on specimen thickness and the elastic modulus [6]. For this reason an
independent measurement of load should be made when using displacement
controlled fatigue.
Load measurement and control on some fatigue machines may impose
special problems. In general, we have found that static load calibration on
some types of equipment is not always reliable. For example, when using
machines operating with cam drive, inertial effects due to bearing clearances
and masses attached to the specimen may add a fixed amount to the set
dynamic load. These effects become more important the lower the dynamic
load, and in our experience may be as much as 20 percent of the set load
when fatigue cracking small specimens. Transient overloads may be en-
countered on high-speed machines assumed to be operating under load con-
trol. These transients are associated with resonant effects and can be large
enough to overcome the static load and produce audible rattling or pounding
during periods of machine speed change. Control of load and crack extension
is impractical when using machines that derive their dynamic loads from an
eccentric rotating weight. These machines will provide a constant load only
if the specimen stiffness does not change. When crack growth occurs the
deflections and therefore the loads increase. The increase in K due to crack
extension is then coupled with that due to the increase in load and the crack
growth rate increases rapidly.
Testing of Brittle Materials

An evaluation of the fracture properties of very brittle materials such as
ceramics or intermetallics requires testing techniques and statistical analysis
procedures outside the scope of ASTM E 399-70 T Method of Test for Plane
Strain Fracture Toughness of Metallic Materials. However, there are metals
and metallic alloys which are considerably more brittle than most engineering
materials but tougher than ceramics and intermetallics. Among these brittle
materials are beryllium, tungsten, chromium base high-temperature alloys,
as well as tool, die, and bearing steels.
Investigators working with beryllium have cracked fracture toughness
specimens by two techniques. In one method a fatigue crack is produced at
an elevated temperature (about 700 F) where the metal is considerably tougher
than at room temperature where the Kx o test is conducted. This method will
give accurate room-temperature K~ c values only if the maximum K level in
the elevated temperature fatigue is sufficiently low. How low has yet to be
determined. Harris and Dunegan [7] have cracked beryllium by applying a
compressive force normal to the specimen surface at a point below the crack
starter notch and then driving a wedge into the notch. If the technique is used
properly the crack stops at the compressed zone. This method is attractive
from the standpoint of its apparent simplicity. However, the sharpness of the
crack, the shape of its front, and the direction of the terminal portion of the
crack in the compressed zone are influenced in a complex way by the magni-
tude of the compressive stress, the area of its application, and the thickness
of the specimen. Systematic studies of the effects of these variables are neces-
sary before a comparison can be made between the results of tests using
specimens cracked in this way with results obtained from fatigue cracked
specimens.
Recently M. H. Jones of NASA-Lewis has fatigue cracked specimens
(0.06 to 0.50 in. thick) of both hot pressed and forged beryllium at maximum
stress intensity levels meeting the requirements of ASTM Method E 399-
70 T. These specimens were provided with a slot terminating in a very sharp
(<0.001 in. radius) straight-across V-notch. Completely reversed (R = - 1)
cantilever bending was used with maximum head deflection essentially
constant. Loads were measured by a load cell in series with the specimen.
The use of completely reversed bending in combination with the very sharp
notch permitted crack initiation at an initial value of the maximum stress
intensity of about 0.5 K~o. Satisfactory control of fatigue crack growth was
possible because the applied load decreases with crack extension when the
maximum deflection is held constant. The fatigue cracking of these beryllium
specimens using this method was no more troublesome or time consuming
than the fatigue cracking of a high-strength steel or aluminum alloy.
Comparison of Plane Strain Fracture Toughness of Various Alloys

It is only within the last two years that the ASTM Committee E-24 Method
of Test for plane strain fracture toughness has been generally available,
although the general principles of this method were known and used by a
few investigators for two years longer. The amount of valid K~o data available
at this time is therefore rather limited, but sufficient datahave been pub-
lished to permit a comparison among a wide variety of alloys in the form of
plate and forgings. The pertinent data is given in Table 1 which also identifies
the alloy composition, form of the product, heat treatment where applicable,
and the source of the data. In all but two cases the data were obtained at
room temperature.
For the purpose of comparison, we have chosen to use the ratio K~o2/ays 2,
designated as the plane strain crack size factor, as a measure of the fracture
toughness. This ratio, of course, is essential to determine the specimen
size for a valid K~c test. For the present purpose it provides an index of
crack size tolerance under plane strain conditions. The yield strength is used
as a convenient factor for obtaining the desired dimensions for this ratio,
but it should not be assumed that the critical crack size will be proportioned
TABLE 1.--Alloys in the form of plate and forging for which valid KI~ data is available.
Refer- Direc- Ft~,

Alloy ence c Form Condition b tion c ksi
Steels:
AISI 4140 . . . . . . . . . . . . . 10 4 by 24 by 1550 F, OQ q- 1200 F, RW 65
60-in. plate 6 h, FC
A533-B-I ~. . . . . . . . . . . . . 12 12 by 120 1675 F, AC q- 1600 F, 4 RW 72
by 260-in. h, WQ q- 1225 F, 4 h,
plate FC + 1150 F, 40 h
AISI 1144 . . . . . . . . . . . . . 10 4 b y 10by 1550 F, OQ + 900F, WR 78
12-in. forging 6 h, FC
Cr-Mo-V . . . . . . . . . . . . . . 13 34-in.-dia 1775 F, AC q- 1240 F, axial 93
forged rotor 40 h MR
Ni-Cr-Mo-V . . . . . . . . . . . 13 65-in.-dia 1550 F, AC q- 1090 F, axial 154
forged rotor 40 h MR
17-4PH b. . . . . . . . . . . . . . . . . . 5 by 9 in. 1100 F, 4h, age RT 162
forging
9-4-25 . . . . . . . . . . . . . . . . . 11 3-in. forged 1550 F, 2 h, OQ q- 1000 WR 175
plate F (2 q- 2) h
D6aC-1 . . . . . . . . . . . . . . . . . 14 31/2-inplate 1700 F, ll/z h, SQ ... 207
975 F, 2 h, OQ
h- 1025 F (2 -F 2) h
D6aC-2 . . . . . . . . . . . . . . . 14 31/2-in.plate 1650 F, 11/2 h, SQ ... 207
975 F, 2 h, SQ 450 F
2hr+1025F(2+2) h
H-11M . . . . . . . . . . . . . . . . . . . 1-in. plate 1850 F, 1 h, OQ q- 1080 ... 213
F, l h
4340 . . . . . . . . . . . . . . . . . . . . . 1-in. plate 1550 F, I/2 h, OQ + 500 RW 238
to 900 F, 1 h to 182
300M . . . . . . . . . . . . . . . . . . . . 41/2 by 41/2- 1600 F, SQ, 1000 F, 1 h, RW 243
in. forging OQ, 110 F, 1/2 h, AC q-
575 F (2 + 2) h
18Ni marage 250 grade.. 4 2-in. plate 1 5 0 0 F, AC + 725 to RW 190 to
850 F, 6 h 253
Aluminums:
2219 . . . . . . . . . . . . . . . . . . . . . 1.4-in. plate T851 WR 51
7039 . . . . . . . . . . . . . . . . . . 10 4 by 12 by T651 WR 57
10-in.
forging
2014 . . . . . . . . . . . . . . . . . . . . . 1.8-in. plate T651 RW 66
7075 . . . . . . . . . . . . . . . . . . . . . 13/8-in.plate T651 WR 78
Others:
Be . . . . . . . . . . . . . . . . . . . . . . . 1 by 16-in.- 75 ~o upset at 1400 F center 66
dia forging radial
Be . . . . . . . . . . . . . . . . . . . . . . . Hot pressed . . . ... 39
Ti-6-4r . . . . . . . . . . . . . . . . 15 3-in. forged 1000 F age WR 140
plate
See footnotes on page 234. (Continued)
TABLE 1--Alloys in the form of plate and forging for which valid Ktc data is available.-
Continued
Composition, percent by weight

Ftu~
Alloy ksi C Mn P S Si Ni Cr Mo
Steels:
A I S I 4140 . . . . . . . . . . . . . 101 0.42 0.78 0.007 0.023 0.23 0.22 0.89 0.21
A533-B-1 a . . . . . . . . . . . . . 93 0.24 1.42 0.01 0.017 0.22 0.70 ... 0.50
A I S I 1144 . . . . . . . . . . . . . 122 0.52 1.46 0 . 0 1 6 0.025 0.20 . . . . . . . . .
Cr-Mo-V .............. 117 0.31 0.78 0.009 0.010 0.28 0.07 1.10 1.15
Ni-Cr-Mo-V ........... 174 ...
17-4PH b. . . . . . . . . . . . . . . 167 0.07 1.0 0.040 0.030 1.0 4 17 ...
max max max max max
9-4-25 . . . . . . . . . . . . . . . . . . . . 0.26 0.33 0.008 0.008 0.01 8.41 0.40 0.48
D6aC-1 ............... 233 0.43 0.85 0.65 0.025 0.025 0.75 0.90 0.50
max max
D6aC-2 ............... 232 ...
H-11M ................ 261 0.39 0.31 0.016 0.010 . . . . . . . . . . . .
4340 . . . . . . . . . . . . . . . . . . 280to 0.42 0.71 0.010 0.012 0.25 1.77 0.80 0.23
194
300M ................. 295 0.41 0.81 0.007 0.003 1.77 1.85 0.83 0.40
18Nimarage 250grade.. 192to 0.006 0.06 0.006 0.007 0.01 18.5 ... 4.81
263
Aluminums:
2219 . . . . . . . . . . . . . . . . . . 67 ... 0.3 ..... 0.20 ...
max
7039 . . . . . . . . . . . . . . . . . . 65 ... 0.25 ..... 0.30 ... 0.20

max
2014 . . . . . . . . . . . . . . . . . . 69 ... 1.0 ..... 1.0 .. 0.10
max
7075 . . . . . . . . . . . . . . . . . . 86 ... 0.3 .. 0.5 .. 0.3

max max
Others:
Be . . . . . . . . . . . . . . . . . . . . 84 ......
Be . . . . . . . . . . . . . . . . . . . . 46 ...... ~
Ti-6-4s . . . . . . . . . . . . . . . . 150 0.023 ...
a T e s t e d a t 50 F .
b T e s t e d a t -- 50 F .
c N A S A - L e w i s unless otherwise indicated.
d O Q = oil q u e n c h e d ; F C = f u r n a c e d c o a l e d ; A C = air c o o l e d ; S Q = s l a c k q u e n c h e d .
(Continued)
TABLE l--Alloys in the form of plate and forging for which valid Kt~ data is available.-
Continued
Composition, percent by weight
Alloy V Co A1 Ti Fe Cu Mg Zn Zr
Steels:
AISI 4140 ................ bal.
A533-B-1-. ............... hal. . . ~ ~
AISI 1144 ................ bal.

Cr-Mo-V .............. 0.26 hal.
Ni-Cr-Mo-V .............. hal.
1 7 - 4 P H b. . . . . . . . . . . . . . . . . . hal.
9-4-25 ................. 0.70 3.9 ... bal . . . . . . . . . . . . .

D6aC-1 ............... 0.03 . . . . . . . . . bal . . . . . . . . . . . . .
D6aC-2 .....................
H-11M ...................... bal ....
4340 ........................ bal .... 9 . ~
300M ................. 0.08 ... ..... bal.

18Ni marage 250 grade ..... 7.4 0.11 0.40 bal. o . .
Aluminums:
2219 ........................ bal- 0.06 0.30 6.3 0.02 1.0 0.18
ance max max max
7039 ........................ bal- 1.1 0.40 0.10 2,8 4.0
ance max max
2014 ........................ bal- 0.15 1.0 4.5 0,5 0.25
ance max max max
7075 ........................ bal- 0.2 0.7 1.6 2.5 5.6
ance max max
Others:
Be ..........................
Be . . . . . . . . . . . . . . . . . . . . . . . . . .
Ti-6-4I ................ 4.1 .., 6.3 bal. 0.13 o . .
" F i r s t letter i n d i c a t e s d i r e c t i o n o f n o r m a l t o c r a c k p l a n e a n d s e c o n d l e t t e r i n d i c a t e s
direction of crack propagation: R = finish roiling or forging direction; W = width direc-
tion of plate or of rectangular cross-section forging; T = thickness direction of plate or rec-
tangular cross-section forging; MR = midradius.
z O t h e r e l e m e n t s : 0 , 0 0 4 H~, 0 . 1 7 O~.
to K~o2/ays'~when the applied stress is at the yield strength. However, design

stresses frequently are based on a fraction of the yield strength, and, if this
design stress ratio ~/avs is sufficiently low to ensure small scale yielding, the
critical crack size under plane strain conditions will be proportional to
(Kt~/crys)2 (#/,ys) -2.
The plane strain crack size factor is plotted against the ratio of yield strength
to Young's modulus ays/E in Fig. 6. 8 The crack size factor would be expected
to increase with a decrease in the value of #vs/E, and this appears to be the
general trend for a particular alloy represented in Fig. 6. However, at a given
value of ~vs/E large variations are noted among the materials represented on
this plot. The toughest alloys with ratios of ays/E greater than 0.005 are the
18Ni marage steel and the titanium. The other atloys of comparable strength
fall well below these two toughest materials, with H-11M being the brittlest
material represented. Differences among 4340, 300M, D6aC, and 9-4-25
steels are due, in part, to differences in product form and heat treatment.
At levels of ~ys/E less than 0.005 by far the toughest material is the pres-
sure vessel steel A533B-1. It is quite evident that low strength per se is no
guarantee of high toughness. Thus, the data for the Cr-Mo-V rotor steel,
4140, and 1144 steels lie well below a trend line established by the maraging
steel and the A533B-1 steel The relatively low toughness of the rotor steels is
probably due in part to temper embrittlement during cooling of the heavy
rotors from the tempering (stress relieving) temperature.
The toughness of the aluminum alloys is rather disappointing. Both
2014T651 and 7039T651 are substantially lower in strength than 7075T651
but have essentially the same plane strain crack size factor. We believe high-
strength aluminum alloys have a considerably better potential than provided
by the present compositions. For example, the paper by Tanaka, Pampillo,
and Low shows the fracture process in 2014T6 heavy plate is dominated by
cavity nucleation caused by fracture of large impurity inclusions associated
with iron and manganese as welt as with excess copper. It is tempting to
believe that a reduction in these types of inclusions would improve the tough-
ness without significantly affecting the strength. In this connection, attention
should be directed to an investigation by Piper et al [8] of the mixed mode
fracture strength of the aluminum alloy 7178 as influenced by modifications
to the base composition. These investigators showed that substantial improve-
s Some investigators chose to compare the fracture toughness of materials on the basis
of their ultimate tensile strength rather than yield strength. An argument in favor of
using the tensile strength is given by Steigerwald in his author's reply. The general upward
trend ofK~c~/vys2with decreasing strength would be still observed if the basis of comparison
were the tensile strength. However, the relative position of some of the alloys represented
in Fig. 6 would be altered considerably.
BROWN AND SRAWLEY ON COMMENTARY ON PRESENTPRACTICE 2 3 7
il 4140
O H-11 M
O D6aC-I
Q Ni-C r-Mo-V
C3 Cr-Mo-V
# 1144
10 ,--
0 AS33-B-1(50~ F)
D 17-4 PH(-50 ~ F)
[] 9-4-25
5-- # 300M
9 2014T651
9 2219T851
I~ 70397651
9 \ 9 70757651
18Ni lwarage ~ 9 Be
~>-- 1--
il Ti-6AI-4V
.5 --
<>
C~
N
~ --
9 05 - -
4340k
o
.01 I I I I I I
2 4 6 8 I0 12
Oys/E x 103
FIG96---Plane strain crack size factor for a variety of alloys in form of heavy plate or
forgings (see Table I for further details).
ments in the strength of center crack sheet specimens could be obtained by

"leaning out" of the major alloying elements and reducing the content of
impurity elements9 A recent paper by Kaufman et al [9] presents plane strain
fracture toughness data for forgings of 7075 (T73 and T6) and of a premium
alloy designated as 7175 (T66 and T736). The premium alloy is indicated to
have both superior strength and toughness, but the means by which these
improvements are produced are a proprietary matter.
The results for beryllium indicate crack size factors for the forged disk of
this material are no lower than some of the very high-strength steels. However,
it should be remembered that the toughness of beryllium will depend strongly
on the crystallographic orientation and that very poor properties are obtained
when the applied stress is normal to the basal planes9 In the forged disk
which we tested, the anisotropy was favorable in that the basal planes were
predominately parallel to the disk surface (normal to the crack plane in the
bend specimen). Similar tests on hot pressed beryllium showed this material
to have a slightly lower crack size factor than the forged disk but a yield
strength of only 38 ksi. In this case there is considerably less crystallographic
anisotropy than in the forged disk.
Correlation of KI, with Other Properties

In the past few years there have been numerous attempts to correlate KI o
values with other material properties. These other properties include the
uniaxial tensile characteristics or the results of Charpy V-notch impact, and
drop weight tear tests. In many cases it is difficult to judge the value of these
correlations because of the lack of precision associated with the measurements
of the plane strain fracture toughness coupled with the use of undersized
specimens2 The best documented of the proposed correlations involve the
conventional tensile characteristics or the Charpy V-notch impact energy.
When these correlations are based on valid K ~ data, however, they have
only a limited value in judging the plane strain fracture toughness of engineer-
ing alloys, as is shown by the results discussed next.
Correlations with Tensile Properties

Hahn and Rosenfield [17] have proposed to calculate K ~ from a strain
hardening index and other quantities that could be derived from a conven-
tional tension test. As shown in the paper by Jones and Brown this calculation
did not yield useful results for a common low-alloy steel. A relation proposed
by Krafft [18,19] also involves a strain hardening index as well as a fracture
process zone size that would have to be determined by some independent
means. As discussed in the paper by Jones and Brown, a fundamental diffi-
culty with relations involving a strain hardening index is the variety of def-
initions used for this quantity (see also Srawley [4]). No single parameter can
characterize strain hardening over a wide range of plastic strains, and, as yet,
there is no way of knowing what portion of the stress strain curve is of most
importance in the fracture process of cracked specimens.
It has been recognized for many years that the strength of sharply notched
bars of high-strength steels generally increases as the tensile strength de-
creases. Krafft and Sullivan [20] have proposed that K~c ~ 1/ays for mild
steel where aYs corresponds to the upper yield point and where both testing
speed and temperature were varied. An inverse relation between Kic and
yield strength also was reported in the paper by Jones and Brown for 4340
9For example, in a recent investigation[17] of the correlation of K~cwith drop weight
tear energy, "K~c" values were based on the load at the first distinct instability, and the
majority of the plane strain fracture toughness specimens were undersized with respect
to both thickness and crack length.
steel and underaged 18Ni marage 250 grade steel. For each steel K~c was
approximately proportional to ays -3 although there are systematic deviations
of the data points from a straight line on a plot of log K~ c versus log aYs.
Steigerwald [21] has reported results of plane strain fracture toughness tests
at various temperatures for a variety of steels, but not all of these results were
valid K~o values. Furthermore, in some cases the data for a given alloy in-
clude results from more than one heat and product form. However, it is
possible to examine the relationship between K~c and yield strength for the
same heat and product form for each of four alloys. The results, Fig. 7, show
that while K~c decreases as the yield strength increases with decrease in test
temperature, a simple inverse power relationship of K~ o to yield strength is
not followed by three of the four steels. The results discussed above indicate
1"9I
1.8
1.7--
1.6
1.5
1.4--
1.3-- H-I1M
1.2 I I I I I
2.20 2. 25 2. 30 2. 35 2. 40 2. 45
10g0ys
FIG. 7--Relation between plane strain fracture toughness and yield strength for single
heats of alloy plate stock tested at 200, 75, - 50, and - 100 F (data from SteigerwaM [21]).
that while the KIe value generally increases with decreasing yield strength,
a simple power relationship between these two quantities is likely to apply
only over a limited range of their values.
Tensile Ductility in Material Specifications

The frequent use of minimum values of tensile elongation and reduction
of area in material specifications is a reflection of their acceptance as indica-
tors of "ductility." However, the significance of ductility defined in this way
is obscure. There is now a sufficient amount of KI c data available to demon-
strate that there is no generally useful relation between tensile elongation or
reduction of area and the plane strain fracture toughness. For example,
Fig. 8 shows a plot of the plane strain crack size factor KI c2/ffys~ against the
tensile elongation for many of the alloys listed in Table 1. It is quite evident
that there is no useful correlation between the crack size factors and tensile
elongation. In the case of SAE 4340 steel, a large increase in crack size factor
occurs with essentially no change in elongation. In fact, the elongations and
the crack size factors may suggest opposite conclusions regarding fracture
I --
.7
o
.5 - - 1144
o--
t~ .3 -
4340~ \ \ t
__ Oys:21 , ksit 7075~ II~7039T6

.I
om
07! zJBeH.P. / 2014T6
L
O5
/t6 300M
Be Forged
03
0
H-II M
Oys = 213 ksi
.m I I I I I
0 4 8 12 16 20
Elongation(>4D), percent
FIG. 8--Plane strain crack size factor versus tensile elongation for a variety of alloys
in form of heavy plate or forging (see Table 1 for further details).
1 --
.7--
O 1144
ca>,-
r
d-,
.3 --
Ti -6-4 /
r r 4340
.1 9 I~'
7075T6 7039T6
201#,76
._m 9 07
Be H. P. 0
/
9 05 ?,OOM l OyS = 214 ksi
r
Be Forged
9 03 0
H-IIM
Oys : 213 ksi
.01 I I I I I
0 10 3o 2o 4o 5o
R. A., percent
FIG. 9--Plane strain crack size fitctor versus tensile reduction o f a r e a f o r a variety o f
alloys i n f o r m o f h e a v y p l a t e orforging (see Tablelforfurtherdetails).
toughness when comparing two materials, as seen by comparing H-11M

(~YS = 213 ksi) with 4340 OYs = 214 ksi). The reduction of area, Fig. 9, is
no more helpful in judging the toughness. There are large differences in
reduction of area between the aluminum alloys but very little change in crack
size factor. The material (H-11M) with the lowest value of K~o2/~ys2 has the
highest reduction of area. The information presented in Figs. 8 and 9 shows
that the factors controlling tensile ductility are not related in any simple way
to the resistance to brittle fracture. Thus, the specification of a minimum
value of tensile elongation or reduction of area is likely to be a futile exercise
in attempting to minimize the danger of brittle failure in service.
Correlation with Impact Properties

The standard Charpy V-notch impact test is simple to perform, but its
results are difficult to interpret. It involves a moderate stress concentration
and a strain rate much higher than in a conventional tension test. The small
size of the Charpy specimen severely limits the number of materials that can
be tested with plane strain conditions maintained through the fracture process.
Furthermore, its blunt notch does not reproduce the essential features of the
crack tip stress field in a linear elastic body. Under these circumstances no
general relation would be expected between the Charpy V-notch impact
energies and K~ o values.
Despite these complications, the convenience and economy of the Charpy
V-notch test to alloy producers has led to investigations of correlations
between Charpy V-notch impact results and KIo values. The paper by Rolfe
and Novak presents a plot showing room temperature data for the crack
size factor Kt~2/avs 2 as a function of the ratio between Charpy V-notch
energy and the yield strength (CVN/~ys). These and other room-temperature
data were presented more recently by Barsom and Rolfe [22] in a paper which
discusses correlations between Charpy V-notch energy and Kic values in
greater detail. According to Rolfe and his co-workers, the best correlation
was obtained for metal conditions where the Charpy energies were defined by
Barsom and Rolfe as "upper shelf" values based on the appearance of 100
percent fibrous fracture. Much poorer correlation was obtained for materials
having Charpy energies in the transition region. The plane strain fracture
toughness values used in the upper shelf correlation were derived from tests
on 1-in.-thick plate [22]. According to ASTM Method E 399-70 T, the thick-
ness was sufficient for only four out of the eleven steels used in this correlation.
These four steels covered a range of Ki~2/~ys ~ values from about 0.125 to
0.25 in. which represents the lower 10 percent of the range of crack size
factors over which a correlation was claimed. For this reason, we cannot
agree that a useful correlation has been demonstrated.
A plot of KI o2/~-~s2 against the Charpy V-notch energy for some of those
materials listed in Table 1 is shown in Fig. 10. The Charpy data were obtained
from specimens which were cut from the broken halves of the K~ o tests. All
data shown with the exception of that for the A533B-1 steel was obtained at
room temperature. The tests on A533B-1 were made at 50 F. Generally, the
Charpy energies represent values in the transitition region with respect to
the variation of energy with test temperature. While there is no general pat-
tern to all of these data, above a crack size factor of about 0.1 in. there is an
upward trend of this factor with increasing Charpy impact energy for the
4340 and 18Ni marage steels. These trends, where they exist, certainly would
be helpful in alloy development, and it should be noted that a new series of
steels with very high fracture toughness were developed by U. S. Steel Cor-
poration using the standard Charpy V-notch test [23]. However, it would be
unwise to assume that the Charpy test can be used to compare the relative
plane strain toughness of different alloy types or that an upward trend of
plane strain crack size factor with increasing Charpy energy is always to be
expected. Thus, at equal values of Charpy energy, the Kie2/o'ys 2 values of
4340 and the overaged 18Ni marage steel can be quite different, and below
4 m
A533B-I41,
2--
i--
~
o
G~
r
.8
~_~>" 96,1~44
.4--
4140
_~%4-25
Cr-Mo-V S//O(3Ni-cr-Mo-v
.~_
~'~ .2-- / - , _ , / o DbaC-1
4340~ /
Oys = 214ksi ~ /\\

9 / '-250Marage
,08--
906 - -
l DbaC-2
.04--
I I { I I I I
4 6 8 10 20 40 60
CVNEnergy,-ft-lbs
FIG. lO--Plane strain crack size factor versus Charpy V-notch impact energy for a variety
of steels in form of heavy plate or forging (see Table 1 for further details).
a K~ o2/,Tvs2 value of about 0.1 in. there is no correlation between plane strain
toughness of the 4340 steel and the Charpy energy.
The results in Fig. 10 are confined to steels. Judy et al [24] have discussed
similar data for aluminum alloys and conclude that the correlation between
Kic and Charpy V-notch energy is poor for these materials.
Fracture Tests with Subsized Specimens ("Screening" Tests)
It is often impossible or impractical to make KTo tests in accordance with

ASTM E 399-70 T, yet essential to obtain some estimate of fracture toughness
of a material, preferably one which correlates well with (KI o/ays) 2. Often the
available product form is not bulky enough to provide specimens which will
meet ASTM Method E 399-70 T, but there are other factors which can pre-
vent the use of sufficiently large specimens for KI ~ measurement, for example,
space in a materials test reactor for irradiation damage studies. There are
many different kinds of tests employing subsized specimens (sometimes called
screening tests) that are used under these and similar circumstances, and
many more that could be devised. Most of these tests have considerable
engineering value when properly used, but they can be seriously misleading
if their limitations are not understood clearly. In particular, it must not be
taken for granted that the order of ranking of materials by a given subsized
specimen test will be the same as the order of ranking for K~c, or for (K~ c/
ays) ~. Size effects are of the essence in fracture behavior, and to neglect them
involves the penalty of incomplete information.
It is convenient to divide the various kinds of tests into three classes accord-
ing to purpose: (1) selection of materials for particular applications, (2)
acceptance of materials for particular applications, and (3) development of
new or improved materials. A test which is designed for one of these purposes
may not be appropriate for another.
Material Selection for Particular Applications

In the first place, if it is believed that plane strain fracture toughness is
likely to govern failure in service then no attempt should be made to circum-
vent the requirements of ASTM Method E 399-70 T. As shown in the paper
by Jones and Brown in this symposium, subsized specimens cannot be used
with ASTM Method E 399-70 T test procedures to screen materials for plane
strain fracture toughness. When making material selection for a particular
application the only known satisfactory screening test is the K~o test. If the
largest specimen that can be obtained from the component of concern is
insufficient for a K~ test in accordance with ASTM Method E 399-70 T,
then it may not be possible to make reliable estimates of the relation between
failure stress and critical flaw size on the basis of a K~o value obtained from a
thicker specimen of the nominal material. In that case, the selection of the
material for the application can proceed on the basis of a test which is not
confined to plane strain behavior, such as ASTM Test for Sharp-Notch
Tension Testing of High-Strength Sheet Materials (E 338-68) (up to 0.25 in.
thick). ASTM Method E 338-68 was designed to screen materials for solid
propellant rocket motor cases which needed toughness levels that are asso-
ciated with mixed mode fracture rather than with K~c. Accordingly ASTM
Method E 338-68 is confined to a particular class of materials, a single
specimen size, and a limited range of thickness, and, while it is intended that
its scope shall be somewhat enlarged, it cannot be made into a universally
applicable test method. In general, screening tests for particular applications
have to be designed for the applications, and further discussion of this topic is
beyond the scope of this book.
Acceptance Tests
If material acceptance criteria involved specifications of minimum plane
strain fracture toughness, it would be necessary to measure K~c as a rou-
tine mechanical testing procedure. However, the tendency in most indus-

tries is to dismiss the routine measurement of Kic as impractical and to
search for a simpler test method that will accomplish the same purpose. For a
restricted class of materials it may be possible to find a satisfactory correlation
between some simpler fracture test and K~, over the range of expected Kzc
values. There are no general rules for developing such an acceptance test,
and each situation must be considered individually.
Difficulties are likely to arise when the thickness of the product being
evaluated is substantially greater than the thickness of the acceptance test
specimen. Examples of oversimplified procedures that could yield misleading
results are furnished by the Society of Automotive Engineers' Aerospace
Material Documents AMD23DC and AMD23DE which propose screening
tests for heavy wrought sections. AMD23DC recommends a small (If/ =
0.395 in., a / W ~ 0.40 and 0.25 < B < 0.50 in.) precracked bend specimen
and AMD23DE recommends two surface crack specimens (0.25 < B < 0.375)
both 1 in. wide, one with a crack approximately 0.2 in. long and the other
with a crack approximately 0.1 in. long. In AMD23DE the notch strength
of the surface crack specimen is used as an index of fracture toughness. An
illustration of the misleading results that could be obtained using this index is
given in Fig. 11, which shows the notch strength of surface crack specimens
reported by Sippel and Vonnegut [25] for 18Ni 300 grade maraging steel at
two thicknesses having essentially the same yield strength. The data are
plotted against crack length (the quantity measured during fatigue cracking)
although the same conclusions would be reached by plotting against crack
depth or the Irwin function for the elliptical crack a/4~~. The important point
to note is that, for cracks of essentially equal size, the thinner specimen has
320--
t1,.~ B~ O.Zin.
B = 0.4in.
~Oys : 296 ksi
200
--,
~20! I 1 I l , I
.08 .12 ,16 .20
Surface crack length, Zc, in.
FIG. l l - - N e t ~trength as a function of surface crack length for l-in.-wide specimens of
18Ni 300 grade maraging steel place, a/B is the ratio of crack depth to specimen thickness.
the higher net strength and that this difference increases with crack s~ize. In
fact, at a crack length of approximately 0.14 in. the net strength to yield
strength ratio of the thin specimen is above unity, while that of the thick one
is about 0.8. For the types of behavior illustrated in Fig. 11 it is unlikely that
the 0.2-in.-thick specimen could usefully screen 0.4-in.-thick sections, and
there is no reason to believe that this behavior would be particularly rare.
From this example it is obvious that under some circumstances the use of a
simplified fracture toughness acceptance test could be dangerous in that
material would be accepted that should have been rejected. We would en-
courage the use of ASTM Method E 399-70 T for acceptance testing when-
ever plane strain failure is possible in the service application. The money and
effort expended for equipment and training of personnel are in most cases
slight in comparison to the consequences of catastrophic service failures.
Alloy Development Tests

Only a few years ago high-strength alloy development programs were
carried out largely on the basis of producing the highest tensile strength level
that could be obtained with a satisfactory level of "ductility" as judged from
tensile elongation or reduction of area values. It now is recognized generally
that fracture toughness should be evaluated directly in such programs, but
the selection of a toughness screening test for alloy development is restricted
by the limited amount of material generally available to alloy development
programs. As a result there is a natural tendency to compromise on the size
of the specimen and on the nature of information obtained from the screen-
ing test. While certain compromises may be necessary, it is important to
anticipate their consequences when planning the program. In particular it
should be recognized that the specimen size selected may not permit KIo
values to be determined for the tougher material conditions.
If the goal of the program is to develop alloys having high values of plane
strain crack size factor (Kie/avs) z, the screening test must be sensitive to
changes in K1 e. This can be determined only by a "pilot program." This pilot
program need not involve all of the variables in the alloy development pro-
gram but should be designed to establish whether a useful correlation exists
between the screening test results and the plane strain crack size factor over
the yield strength range of the class of materials considered. For example, as
can be seen from Fig. t0, it would be unwise to use the room-temperature
Charpy V-notch impact test to screen high-strength martenistic low-alloy
steels such as 4340 at yield strength levels above about 200 ksi. At lower
strength levels the Charpy test may be useful in screening these alloys provid-
ing proper attention is paid to the sensitivity of the Charpy test to tempera-
ture and strain rate. However, comparisons of fracture toughness levels
between different classes of materials (for example, steels and aluminum
alloys) on the basis of C h a r p y V-notch impact energy should be approached

with great caution.
A potentially dangerous screening practice on the use of W / A values f r o m
precracked C h a r p y specimens tested in slow bending or impact is to estimate
"K~c" numbers. All t o o frequently one can find "K~c" values over 200 ksi
~r reported f r o m tests on precracked C h a r p y specimens. These values are
likely to be gross overestimates of the valid K~ c. F o r example, a "K~ o" value
of a b o u t 140 ksi~v/~n., was derived from precracked C h a r p y W / A data and
has been reported [26] for a~-in, plate of 9Ni-4Co-0.30C at a yield strength
level of a b o u t 190 ksi, but the valid Kx c value for the material at this strength
level is about 80 ksiv/~n.. [27]. The use of the W / A values themselves as an
index of plane strain fracture toughness is a dubious practice unless a pilot
p r o g r a m is used to establish the existence of a useful correlation between KI o
and the W / A values for the particular class of materials involved. In our
opinion it is preferable to use the standard C h a r p y V-notch test for screening
in alloy development rather than the precracked C h a r p y which has not
been standardized. The precracked C h a r p y offers no demonstrated advantage
over the C h a r p y V-notch as a screening test for K~c, except possibly for very
brittle conditions where the specimen size would be sufficient to permit a KI e
test in accordance with A S T M M e t h o d E 399-70 T, and under these circum-
stances there is no good reason for not making the standard Kic test.
References
[1] Brown, W. F., Jr., and Srawley, J. E., Plane Strength Crack Toughness Testing of
High Strength Metallic Materials, ASTM STP 410, American Society for Testing and
Materials, 1966.
Fracture Toughness," Materials Research and Standards, Vol. 7, No. 6, June 1967,
pp. 262-266.
[3] Irwin, G. R., Kies, J. A., and Smith, H. L., "Fracture Strengths Relative to Onset and
Arrest of Crack Propagation," Proceedings, American Society for Testing and Ma-
terials, Vol. 58, 1958, p. 640.
[4] Srawley, J. E., "Plane Strain Fracture Toughness Tests on Two Inch Thick Maraging
Steel Plate at Various Strength Levels," Fracture, 1969, Proceedings of the 2nd Inter-
national Conference on Fracture, Brighton, England, April 1969.
[5] Irwin, G. R., "Plastic Zone Near a Crack and Fracture Toughness," Sagamore
Ordnance Materials Conference, Syracuse University Research Institute, Aug. 1960.
[6] Jones, M. H. and Brown, W. F., Jr., "Experimentally Determined Displacements for
Compact Tension Specimens," to be published, American Society for Testing and
Materials, 1970.
[7] Harris, D. O. and Dunegan, H. L., "Fracture Toughness of Beryllium," Journal of
Materials, American Society for Testing and Materials, Vol. 3, No. 1, March 1968,
p. 59.
[8] Piper, D. E., Quist, W. E., and Anderson, W. E., "The Effect of Composition on the
Fracture Properties of 7178-T6 Aluminum Sheet," Metallurgical Society Conference,
Vol. 31, Application of Fracture Toughness Parameters to Structural Metals, Gordon
and Breach, New York, 1966.
[9] Kaufman, J. G., Schilling, P. E., and Nelson, F. G., "Fracture Toughness of Aluminum
Alloys," Metals Engineering Quarterly, American Society for Metals, Vol. 9, No. 3,
Aug. 1969, p. 39.
[10] Wilson, W. K., Clark, W. G., Jr., and Wessel, E. T., "Fracture Mechanics Technology
for Combined Loading and Low to Intermediate Strength Metals," Westinghouse
Electric Research Laboratory, Final Technical Report on U.S. Army Tank Command
DAAE 07-67-C-4021.
[1/] Wessel, E. T., "State of the Art of the WOL Specimen for K~c Fracture Toughness
Testing," Engineering Fracture Mechanics, Vol. 1, 1968, pp. 77-103.
[12] Wessel, E. T., "Linear Elastic Fracture Mechanics for Thick-Walled, Welded Steel
Pressure Vessels: Material Property Considerations," Westinghouse Research Scien-
tific Paper 69-1D7-BFPWR-P1, 24 Feb. 1969.
[13] Greenberg, H. D. et al, "Application of Fracture Mechanics Technology to Turbine
Generator Rotors," Westinghouse Research Scientific Paper 69-1D9-MEMTL-P1,
15 Oct. 1969.
[14] Gunderson, A. W., Air Force Materials Laboratory private communication with
Brown, W. F., Jr.
[15] Wessel, E. T., Clark, W. G., and Wilson, W. K., "Engineering Methods for Design
and Selection of Materials Against Fracture," Westinghouse Research Laboratory,
Final Technical Report, U.S. Army Tank Automotive Center, Contract DA-30-069-
AMC-602(T), 24 June 1966.
[16] Freed, C. N. and Goode, R. J., "Correlation of Two Fracture Toughness Tests for
Titanium and Ferrous Alloys," Report 6740, Naval Research Laboratory, 16 Jan. 1969.
[17] Hahn, G. and Rosenfield, A., "Source of Fracture Toughness: The Relation between
Kl~ and Ordinary Tensile Properties of Materials," Applications Related Phenomena in
Titanium alloys ASTM STP 432, American Society for Testing and Materials, 1968,
p. 5.
[18] Krafft, J., "Fracture Toughness of Metals," Report of Progress, U.S. Naval Research
Laboratory, Washington, D.C., Nov. 1963.
[19] Krafft, J., "Correlation of Plane Strain Crack Toughness with Strain Hardening
Characteristics of a Low, Medium, and a High Strength Steel," Applied Materials
Research, Vol. 3, 1964, p. 88.
[20] Krafft, J. M. and Sullivan, A. M., "Effects of Speed and Temperature on Crack
Toughness and Yield Strength in Mild Steel," Transactions Quarterly, American
Society for Metals, Vol. 56, 1963, p. 160.
[21] Steigerwald, E. A., "Plane Strain Fracture Toughness for Handbook Presentation,"
TRW AFML TR 67-187, July 1967.
[22] Barsom, J. M. and Rolfe, S. T., "Correlation between K~e and Charpy V-Notch Test
Results in the Transition-Temperature Range," Impact Testing of Materials, ASTM
STP 466, American Society for Testing and Materials, 1970, pp. 281-302.
[23] ManganeUo, S. J. et al, "Development of High Toughness Alloy Plate Steel with a
Minimum Yield Strength of 140 ksi," Welding Journal Research Supplement, Vol. 43,
Nov. 1964, p. 514.
[24] Judy, R. W., Goode, R. J., and Freed, C. N., "Fracture Toughness Characterizations
Procedures and Interpretation to Fracture-Safe Design for Structural Aluminum
Alloys," Report 6871, Naval Research Laboratory, 31 March 1969.
[25] Sippel, G. R. and Vonnegut, G. L., Third Maraging Steel Project Review, AFML
RTD-TDR-63-4048, Nov. 1963.
[26] Kalish, D. and Kulin, S. A., "Coupling of Ultrahigh Strength and Fracture Tough-
ness in Steels by Means of Thermomechanical Processes," Manlabs, Inc. A F M L - T R -
67-115, May 1967.
[27] Aerospace Structural Metals Handbook, Code 1221, Mechanical Properties Data
Center, Traverse City, Mich., 1969.
STP463-EB/Sep. 1970
Tentative Method of Test for Plane Strain

Fracture Toughness of Metallic Materials
(ASTM Designation: E 399-70 T) ''2
1. Scope
1.1 This m e t h o d covers the d e t e r m i n a t i o n o f the plane strain fracture
toughness of metallic m a t e r i a l s by a b e n d or a c o m p a c t tension test o f a
n o t c h e d a n d fatigue-cracked specimen having a thickness o f 0.25 in. (6.4
ram) or greater.
NOTE 1--Plane strain fracture toughness tests of thinner materials that are sufficiently
brittle (see 6.1) can be made with other types of specimens [1].3 There is no standard test
method for testing such thin materials.
l. 1. ! The b e n d specimen is a single-edge n o t c h e d b e a m l o a d e d by three

p o i n t bending.
1.1.2 T h e c o m p a c t tension specimen is single-edge n o t c h e d and pin l o a d e d
in tension.
1.2 The specimen size required for testing p u r p o s e s increases as the square
of the r a t i o of toughness to yield strength of the m a t e r i a l ; therefore, a range
of p r o p o r t i o n a l specimens is provided.
NOT~ 2--The values stated in U.S. customary units are to be regarded as the standard.
The metric equivalents of U.S. customary units given in the body of the standard and in the
appendix may be approximate.
2. Summary of Method
2.1 The m e t h o d involves tension or t h r e e - p o i n t bend testing o f n o t c h e d
specimens which have been p r e c r a c k e d in fatigue. L o a d versus d i s p l a c e m e n t
across the n o t c h at the specimen edge is r e c o r d e d autographically. T h e l o a d
c o r r e s p o n d i n g to a two percent increment of crack extension is established by
a specified deviation f r o m the linear p o r t i o n o f the record. The K~c value is
calculated f r o m this l o a d by e q u a t i o n s which have been established on the
basis o f elastic stress analysis o f specimens of the types described below. T h e
1This tentative method is under the jurisdiction of ASTM Committee E-24 on Fracture
Testing of Metals. A list of members may be found in the ASTM Year Book.
2 Effective 19 March 1970.
249
validity of the determination of K~c value by this method depends upon the
establishment of a "sharp-crack" condition at the tip of the fatigue crack, in a
specimen of adequate size. To establish a suitable crack-tip condition, the
stress intensity level at which the fatigue precracking of the specimen is con-
ducted is limited to a relatively low value.
3. Significance
3.1 The property K I e determined by this method characterizes the resistance
of a material to fracture in a neutral environment in the presence of a sharp
crack under severe tensile constraint, such that the state of stress near the
crack front approaches tritensile plane strain, and the crack-tip plastic region
is small compared with the crack size and specimen dimensions in the con-
straint direction. A Kic value is believed to represent a lower limiting value of
fracture toughness. This value may be used to estimate the relation between
failure stress and defect size for a material in service wherein the conditions
of high constraint described above would be expected.
Background information concerning the basis for development of this
test method in terms of linear elastic fracture mechanics may be found in
Refs 1 and 2.
3.1.1 The K~ c value of a given material is a function of testing speed and
temperature. Furthermore, cyclic loads can cause crack extension at K~
values less than the K~o value. Crack extension under cyclic or sustained
load will be increased by the presence of an aggressive environment. There-
fore, application of K~o in the design of service components should be made
with awareness to the difference that may exist between the laboratory tests
and field conditions.
3.1.2 Plane strain crack toughness testing is unusual in that there can be no
advance assurance that a valid K~o will be determined in a particular test.
Therefore it is essential that all of the criteria concerning validity of results be
carefully considered as described herein.
3.1.3 It should be recognized that it will not be possible to determine a K~c
value for some materials. In such cases it may be possible to obtain a qualita-
tive indication of the material strength in the presence of cracks by use of
ASTM Method E 338.
3.2 This method can serve the following purposes:
3.2.1 In research and development to establish, in quantitative terms,
significant to service performance, the effects of metallurgical variables such
as composition or heat treatment, or of fabricating operations such as welding
or forming, on the fracture toughness of new or existing materials.
3.2.2 In service evaluation, to establish the suitability of a material for a
specific application for which the stress conditions are prescribed and for
which maximum flaw sizes can be established with confidence.
TENTATIVE METHOD OF TEST 251
3.2.3 For information, and for specifications of acceptance and manufac-

turing quality control, when there is a sound basis for specifying minimum
plane strain fracture values. Detailed discussion of the basis for determining
such minimum values in a particular case is beyond the scope of this method.
4. Definitions
4.1 stress-intensity factor, Kx (FL-3/2)---a measure of the stress-field
intensity near the tip of an ideal crack in a linear elastic medium when de-
formed so that the crack faces are displaced apart, normal to the crack plane
(opening mode or mode I deformation). K~ is directly proportional to applied
load and depends on specimen geometry. See Ref 3, page 30 on "Stress
Analysis of Cracks."
4.2 plane strain fracture toughness, K/c (FL-3/2)--the material-toughness
property measured in terms of the stress-intensity factor K~ by the operational
procedure specified in this test method.
4.2.1 In this method, measurement of K~c is based on the lowest load at
which significant measurable extension of the crack occurs. Significant
measurable extension is defined in terms of a specified deviation from linearity
of the load-displacement curve as described in 8.1. In some instances this
may coincide with the maximum load, but frequently the specimen will sus-
tain a higher load than that at which significant crack extension occurs.
5. Apparatus
5.1 The procedure involves bend or tension testing of notched specimens
which have been precracked in fatigue. Load versus displacement across the
notch is recorded autographically. Testing may be done in various machines
having suitable load sensing devices for instrumenting to an autographic
recorder or recording device.
5.2 Bend-Test Fixture--The bend-test fixture must be designed to minimize
errors which can arise from friction between the specimen and the supports.
The friction effect can be virtually eliminated if the test fixture is designed
to allow the support rolls to rotate and move apart slightly, thus permitting
rolling contact to be maintained. The simple fixture design shown in Fig. 1 is
recommended. The support rolls are allowed limited motion along plane sur-
faces, but are initially positively positioned against stops by low tension
springs (such as rubber bands). The dimensions shown in Fig. 1 apply to
either 1-in. (25-ram) or 2-in. (51-mm) deep-bend specimens. For other speci-
men sizes the roll-stop spacings should conform to the span, and the roll
diameters should be between I4/ and W/2.
5.3 Grips and Fixtures for Tension Test--The general gripping arrangement
for tension testing of compact K~ c specimens is shown in Fig. 2. A clevis and
ALTERNATIVE 4-INCH S P A N 0
FOR I-INCH DEEP SPECIMENS~
SEE ROLLER
L z.o I
0.2 DIAM
BOSSES FOR
RUBBER BANDS
OR SPRINGS
2.0
2.2
2.4
NOTE 1--Dimensions are in inches.

NOTE 2 - - F o r 1-in. and 2-in. deep specimens; proportion accordingly for other specimen
sizes.
METRIC EQUIVALENTS
U.S. Metric U.S. Metric

Customary Equivalents, Customary Equivalents,
Units, in. mm Units, in. mm
0.2 . . . . . . . . 5.1 1.5 . . . . . . . 38

0.3 . . . . . . . . 7.6 1.505 . . . . . 38.23
0.6 . . . . . . . . 15.0. 2.0 . . . . . . . 51
0.75 . . . . . . . 19.1 2.6 . . . . . . . 66
0.998 . . . . . . 25.35 3.495 . . . . . 88.77
1.0 . . . . . . . . 25.0 3.505..... 89.03
1.002 . . . . . . 25.45 3.6 . . . . . . . 91
1.10 . . . . . . . 27.9 4.0 . . . . . . . 100
1.495 . . . . . . 37.97
FIG. 1--Bend testfixture design.
p i n a r r a n g e m e n t is e m p l o y e d a t b o t h t h e t o p a n d b o t t o m o f t h e s p e c i m e n t o
a l l o w r o t a t i o n as t h e s p e c i m e n is l o a d e d [7].
5.3.1 T h e c r i t i c a l t o l e r a n c e s a n d s u g g e s t e d p r o p o r t i o n s o f t h e clevis a n d
p i n s are g i v e n in Fig. 2. T h e s e p r o p o r t i o n s are b a s e d o n s p e c i m e n s h a v i n g
W/B = 2forB > 0.5 in. (12.7 m m ) a n d W/B = 4forB < 0.5 in. ( t 2 . 7 m m ) .
University of Washington (University of Washington) pursuant to License Agreement. No furth
TENTATIVEMETHOD OF TEST 253
If a 280,000-psi (1930-MN/m ~) yield strength maraging steel is used for the

clevis and pins, adequate strength will be obtained for testing the specimen
sizes and aYs/E ratios given in 6.1.3. If lower-strength grip material is used,
or if substantially larger specimens are required at a given ~ys/E ratio than
those shown in 6.1.3, then heavier grips will be required. As indicated in
Fig. 2, the clevis corners may be cut off sufficiently to accommodate seating
of the clip gage in specimens less than 0.375 in. (9.5 ram) thick.
5.3.2 Careful attention should be given to achieving as good alignment as
possible through careful machining of all auxiliary gripping fixtures (see 7.4).
5.4 Displacement Gage--The displacement gage output shall indicate very
accurately the relative displacement of two precisely located gage positions
spanning the crack notch. Exact and positive positioning of the gage of the
specimen is essential, yet the gage must be released without damage when the
"Pin Dia = 0.24W -.005 in.

~,L0~lintj Rod Thd For SpecimensWith Oys > 200 ksi
TheHoles May Be 0.3W + .005 in. Dia
A~ And The Pin 0. 288W- . 005 in. Dia
I~
'1
I
I
I II
II
It
T\ J
1. 25D
I
iI
II
II
It | II II II
I..L._ ~_ _ _lJ + U . . . . LI
.SW
-N ~w~* ~- I.sw ~• 005
5W~.~.O2tS.025W W I [
-- r_ I
\Flat ~-A /~ I "r

'~ x, .6W / ~~
---, .25W .SW+ .015"--~ .25W,---
A - Surfaces Must Be flat, In-Line and
See5.3 " Perpendicular, As Applicable, to
Within0.002in.F.I.R.
NoTE--Dimensionsare in inches(0.005 in. = 0.13 mm).
FIG. 2--Tension testing clevis.
RECORDER
FOIL RESISTANCE
STRAIN GAGE *
20 k
ATTACHABLE / \N T2
KNIFE EDGE \\X
F x, * 5 0 0 OHM GAGES WILL

PROVIDE GREATER
S E N S I T I V I T Y THAN
120 OHM GAGES
NOTElGage details are given in the Appendix.
FIG. 3--Double-cantilever clip-in displacement gage and method of mounting.
specimen breaks. A recommended design for a self-supporting, releasable

gage is shown in Fig. 3 and described in the Appendix. Figure 3 also shows
the manner in which the gage is mounted on the specimen; for this purpose,
the specimen must be provided with a pair of accurately machined knife
edges. These knife edges can either be machined into the specimen itself, or
they can be separate pieces which are fixed to the specimen with screws (see
Fig. 7 and 6.2). It is not the intent of this test method to exclude the use of
other types of gages or gage fixing devices providing the gage used meets the
requirements listed below and providing the gage length does not exceed W/4.
5.4.1 Each gage shall be checked for linearity using an extensometer cali-
brator or other suitable device; the resettability of the calibrator at each
displacement interval should be within +0.000020 in. (0.00050 mm). Read-
ings shall be taken at ten equally spaced intervals over the working range of
the gage. This calibration procedure should be performed three times, re-
moving and reinstalling the gage in the calibration fixture between each run.
The required linearity shall correspond to a maximum deviation of + 0.001 in.
(0.0025 ram) of the individual displacement readings from a least-squares-
best-fit straight line through the data. The absolute accuracy, as such, is not
important in this application, since the test method is concerned with relative
changes in displacement rather than absolute values (see 8.1).
TENTATIVE METHOD OF TFST 255
6. Specimen Configuration, Dimensions, and Preparation

6.1 Specimen Size
6.1.1 In order for a result to be considered valid according to this method it
is required that both the specimen thickness, B, and the crack length, a, ex-
ceed 2.5 (Kic/avs) ~, where ~YS is the 0.2 percent offset yield strength of the
material for the temperature and loading rate of the test [1,4,5].
6.1.2 The initial selection of a size of specimen from which valid values of
K~c will be obtained may be based on an estimated value of K~c for the
material. It is recommended that the value of KI c be overestimated, so that a
conservatively large specimen will be employed for the initial tests. After a
valid KI~ result is obtained with the conservative-size initial specimen, the
specimen size may be reduced to an appropriate size [a and B > 2.5 (Kt~/
~ys) 2] for subsequent testing.
6.1.3 Alternatively the ratio of yield strength to Young's modulus can be
used for selecting a specimen size that will be adequate for all but the toughest
materials:
Minimum
Recommended
Thickness and
ays/E Crack Length
in. mm
0.0050 to 0.0057 . . . . . . . . 3 75
0.0057 to 0.0062 . . . . . . . . 21/2 63
0.0062 to 0.0065 . . . . . . . . 2 50
0.0065 to 0.0068 . . . . . . . . 13`4 44
0.0068 to 0.0071 . . . . . . . . 11/2 38
0.0071 to 0.0075 . . . . . . . . 11/4 32
0.0075 to 0.0080 . . . . . . . . 1 25
0.0080 to 0.0085 . . . . . . . . 3,4 20
0.0085 to 0.0100 . . . . . . . . 1/4 121/2
0.0100 or greater ........ I/2 61/2
When it has been established that 2.5 ( K i r 2 is substantially less than the
minimum recommended thickness given in the preceding table, then a cor-
respondingly smaller specimen can be used. On the other hand, if the form
of the available material is such that it is not possible to obtain a specimen
with both crack length and thickness greater than 2.5 (K~c/~vs) 2, then it is
not possible to make a valid KI r measurement according to this recommended
method.
6.2 Standard Specimens--The geometry of standard specimens is shown in
Fig. 4, bend specimens, and Fig. 5, compact tension specimens, with notch
details in Fig. 6.
256 PLANESTRAIN FRACTURETOUGHNESSTESTING
W
-+ .010
-T-
Crack Starter A--~

Envelope( See Fig. b W_+
and Section6.2 ) ...~=(, ~- .002
L
A.~ t,ill
, : a=k~ 11 3_
2W+ O.lW rain -~ - 2W+ O.lW rain -
NOTE 1--Dimensions and tolerances are in inches unless otherwise indicated.
NOTE 2--"A" surfaces are to be perpendicular to center line of crack-starter envelope
within • TIR.
NOTE 3--Integral or attachable knife edges for clip gage attachment may be used (see
Fig. 7 and 6.2.5).
METRIC EQUIVALENTS
U.S. Customary Units, in. Metric Equivalents, mm
0.002 . . . . . . . . . . . . . . . . . . . . . . 0.05
0.005 . . . . . . . . . . . . . . . . . . . . . . 0.13
0.010 . . . . . . . . . . . . . . . . . . . . . . 0.25
FIG. 4--Bend specimen--standard proportions and tolerances (not a working drawing).
6.2.1 The crack length, a, is nominally equal to thickness, B, and is between

0.45 and 0.55 times the depth, W.
6.2.2 The crack-starter slot configuration must lie within the envelope,
shown in Fig. 6, that has its apex at the end of the fatigue crack.
6.2.3 The length of the fatigue crack shall be not less than 5 percent of the
overall length, L, and not less than 0.05 in. (1.3 ram).
6.2.4 To facilitate fatigue cracking at a low level of stress intensity (see 6.5)
the notch root radius should be 0.003 in. (0.08 mm) or less. However, if a
chevron form of notch is used, as shown in Fig. 6b, the notch root radius may
be 0.01 in. (0.25 ram), or less.
6.2.5 Attachable or integral knife edges for fixing the clip gage to the
specimen shall be provided as shown by the suggested designs in Figs. 7a and
7b, respectively. The displacements will be essentially independent of the
gage length for the bend specimen providing the gage length is equal to or less
than W / 4 . For the compact tension specimen the displacements will be
essentially independent of the gage length for any length up to 1.2W. A
design for attachable knife edges is shown in Fig. 7a. This design is based on
the gage length requirement for the bend specimen and a knife edge spacing of
TENTATIVE METHOD OF TEST 2.57
0.2 in. (5.1 ram). The effective gage length is established by the points o f con-
tact between the screw and hole threads. F o r the design shown the m a j o r
diameter o f the screw has been used in setting this gage length. The No. 8
screw will permit attachable knife edges to be used for specimens having
W < 3 in. (76 mm). A No. 2 screw will permit the use of attachable knife
edges for specimens having W > 2 in. (51 mm). It will be recognized that
more flexibility is possible in the design of attachable knife edges for the
c o m p a c t tension specimen because the gage length may extend to 1.2W.
.25W • 0(]6 Dia

2 Holes, Note 2 - ~
Crack Starter ~,. T ~ J~--~t--~

Envelope ( See Fig. 6 - ~ ~ J~ ~l~[ A-~
andSecli0n6.2) Nx,x I ~ ~'J'~J I
c_-_
- W+.O02 i- _j L B=
-I +.
,- 1.25W+.005
NOTF I--Dimensions and tolerances in inches.
NOTE 2--"A" and "C" surfaces are to be perpendicular and parallel as applicable t o
within 0.002-in. TIR. For specimens of ay, > 200 ksi the holes may be 0.3 W q- 0.005.
NOTE 3--Integral or attachable knife edges for clip gage attachment may be used (see
Fig. 7 and 6.2.5).
METRIC EQUIVALENTS
0.002 . . . . . . . . . . . . . . . . . . . . . . 0.05
0.005 . . . . . . . . . . . . . . . . . . . . . . 0.13
0.010 . . . . . . . . . . . . . . . . . . . . . . 0.25
FIG. 5--Compact tension specimen--standard proportions and tolerances (not a working

drawing).
Specimen
I Edge Recl~iredEnvelo~
(a) N =~ 3
2 L = 0. 45 to 0. 55Wfor Bend Specimens
L = 0.10 to 0.80W for CompactSpecimens
Examples
T Chevron
T Straight Thru
,,,L- FatigueCrack Shall Be N~ LessI~aa~"

5 Percent of Length.L, Nor Less
T 0.05 inch [-
~. Keyhole ... |
NoT~--Dimensions are in inches.

METRIC EQUIVALENTS
0.05 . . . . . . . . . . . . . . . . . . . . . . . 1.3
FIG. 6.--Envelope for crack-starter notches and examples of notches tipped with fatigue
cracks.
6.3 Alternative Specimens--The form of available material may be better

adapted to alternate specimen shapes than to the standard specimen with
B = 0.5W.
6.3.1 Alternative bend specimens may have B = 0.25W to W.
6.3.2 Alternative compact tension specimens may have B = 0.25W to
0.5W.
6.3.3 Crack length, a, shall be 0.45 to 0.55 W, the same as for the standard
specimen.
6.3.4 The support span, S, for the bend specimen shall be 4W, the same as
for the standard specimen.
6.3.5 The size requirements of 6.1 shall be met, as for the standard speci-
mens.
TENTATIVEMETHOD OF TEST 2 5 9
6.4 Other Designs of Compact Tension Specimens--Other designs of com-

pact tension specimens are in use, particularly in nuclear reactors for monitor-
ing long-term irradiation damage. When such specimens are prepared and
tested in accordance with the procedure of this test method the validity of the
results can be judged on the same basis as for specimens of standard geometry,
namely, as specified in 8.1. In reporting results from such specimens the
appropriate expression for calculation of K~ from toad and specimen dimen-
sions should be stated and its origin cited [4].
6.5 Fatigue Cracking--The fatigue cracking shall be conducted with the
specimen fully heat treated to the condition in which it is to be tested (Note 3).
The fatigue crack is to be extended from the notch at least 0.05 in. (1.3 mm)
I
I
J . 07
9 06 2C + Screw Ihd Dia < W
- 4
-- -- -- -IX Knife Edge Width

I\ = Clip Gage Arm Width, rain
- - 6 "
I \
I \
I
f- L _ ..J
L _ _ _
i ]
I /
Screw Hd Dia
I /
2
I /
60o" I/
Specimen . . . . J/_ 1 c
9032 in. rain
/
O.100 in. |
_ _ Notch Centerline ~
NoTE--Dimensions and tolerances are in inches.

METRIC EQUIVALENTS
0.032 . . . . . . . . . . . . . . . . . . . . . . 0.81
0.06 . . . . . . . . . . . . . . . . . . . . . . . 1.5
0.07 . . . . . . . . . . . . . . . . . . . . . . . 1.8
0,10 . . . . . . . . . . . . . . . . . . . . . . . 2,5
FIG. 7a--Example of attachable knife edge design based on the gage length requirements
for the bend specimen.
5
I
I
I
Crack Starter Envelope
I
I ( See Fig, 6 )
I
I
I
I
~ - N -'1
I
I
I
I
~
I
0. 06
o.8
_0.200
0.195
Noa~l--Dimensions are in inches.
< 4 in. but it may be any value up to W/4 (see 6.2.5).
NOtE 2---Gage length shown is for W_-
METRIC EQUIVALENTS
U S Customary Units, in. Metric Equivalents, mm
0.050 . . . . . . . . . . . . . . . . . . . . . . 1.3
0.060 . . . . . . . . . . . . . . . . . . . . . . 1.5
0.195 . . . . . . . . . . . . . . . . . . . . . . 5.0
0.200 . . . . . . . . . . . . . . . . . . . . . . 0.1
FIG. 7b---Integral knife edges.
and sufficiently far to meet the requirements of 6.2, 6.3, or 6.4. T o determine
when this requirement has been met in practice, it is usually sufficient to
observe the traces of the crack on the side surfaces of the specimen. T o
ensure that the fatigue crack will be sufficiently sharp, flat, and n o r m a l to
the specimen edge, the following conditions of fatigue cracking shall be met:
6.5.1 The equipment for fatigue cracking shall be such that the load distri-
bution is symmetrical with relation to the notch, and the m a x i m u m value
o f the stress intensity in the fatigue cycle shall be k n o w n with an error o f not
more than 5 percent [6].
6.5.2 D u r i n g the final stage of fatigue crack extension, for at least the
terminal 2.5 percent of the overall length of notch plus crack, the ratio o f the
m a x i m u m stress intensity of the fatigue cycle to the Y o u n g ' s modulus,
Kj(max)/E, shall not exceed 0.0012 in. 1/2 (0.000192 m~/2). Furthermore, Ks
(max) must not exceed 60 percent of the KQ value determined in the subse-
quent test if KQ is to qualify as a valid K~c result (see 8.1).
6.5.3 The stress-intensity range should be not less than 0.9 Kj(max). Fatigue
cycles that involve reversal of loading are of no particular advantage but may
be expedient with certain types of equipment. The K calibration is very
sensitive to the distribution of clamping forces which are necessary to grip
the specimen for reversed-load cycling. Particular care must be taken that the
K calibration for the fatigue setup properly represents the applied clamping
forces.
6.5.4 When fatigue cracking is conducted at a temperature T1 and testing at
a different temperature 7'2, Kj(max) must not exceed (~ysl/2 avs2)KQ, where
aYsl and avs~ are the yield strengths at the respective temperatures T1 and T2.
NOTE 3--Some materials are too brittle to be fatigue cracked without fracturing. These
materials are outside the scope of the present standard test method.
NOTE 4--The K calibration is the relationship of stress-intensity factor K to load and
specimen dimensions [1]. For example, see 8.1.3 and 8.1.4.
7. Procedure
7.1 Number of Tests--It is recommended that at least three replicate tests
be made.
NOTE 5--Information on variability of test results will be found in Ref 1, pages 26 and
74, and Ref 8.
7.2 Specimen Measurement--All specimen dimensions shall be within the

tolerances shown in Figs. 1, 2, and 3.
7.2.1 Measure the thickness, B, to the nearest 0.001 in. (0.025 ram) or 0. I
percent, whichever is larger, at not less than three positions between the
fatigue-crack tip and the unnotched edge of the specimen, and record the
average value.
7.2.2 F o r a bend specimen, measure the depth, IV, and the crack length, a
from the notched edge of the specimen to the far edge, and to the crack front,
respectively. For a compact specimen, measure these dimensions from the
plane of the center line of the loading holes (the notched edge is a convenient
reference line but the distance from the center line of the holes to the notched
edge must be subtracted to determine W and a). Measure the depth, W, to the
nearest 0.001 in. (0.025 ram) or 0.I percent, whichever is larger, at not less
than three positions near the notch location, and record the average value.
7.2.3 After fracture measure the crack length to the nearest 0.5 percent at
the following three positions: at the center of the crack front, and midway
between the center and the end of the crack front on each side. Use the average
of these three measurements as the crack length to calculate Ko (see 8.1.3
and 8.1.4). If the difference between any two of the crack length measurements
exceeds 5 percent of the average, or if any part of the crack front is closer to
the machined notch root than 5 percent of the average crack length, or 0.05 in.
(1.3 ram) minimum, then the test is invalid. Also if the length of either sur-
face trace of the crack is less than 90 percent of the average crack length, as
defined above, the test is invalid.
7.2.4 If the fatigue crack departs noticeably from the plane of symmetry of
the notch, measure the greatest angle between the fatigue crack surface and
the plane of symmetry. If this angle exceeds 10 deg the test is invalid.
7.3 Bend Testing--Set up the bend test fixture so that the line of action of
the applied load shall pass midway between the support roll centers within
0.5 percent of the distance between these centers (for example, within 0.02 in.
(0.5 mm) for a 4-in. (100-mm) span). Measure the span to within 0.5 percent of
nominal length. Locate the specimen with the notch center line midway be-
tween the rolls to within 0.5 percent of the span, and square to the roll axes
within 2 deg. Seat the displacement gage on the knife edges to maintain
registry between knife edges and gage grooves. In the case of attachable knife
edges, seat the gage before the knife edge positioning screws are tightened.
Load the specimen at a rate such that the rate of increase of stress intensity is
within the range 30,000 to 150,000 psi. in.1/2/min (0.55 to 2.75 MN-m-3/2/s),
corresponding to a loading rate for the 1 in. thick specimen between 4000 and
20,000 lb/min (0.03 to 0.15 kN/s).
7.4 Tension Testing--Eliminate friction effects, and also eccentricity of
loading introduced by the clevis itself, by adherence to the specified tolerances
for the specimen clevis and pins shown in Fig. 2. Eccentricity of loading can
also result from misalignment external to the clevis, or from incorrect position-
ing of the specimen with respect to the center of the clevis opening. Obtain
satisfactory alignment by keeping the center line of the upper and lower
loading rods coincident within 0.03 in. (0.76 mm) during the test and by
centering the specimen with respect to the clevis opening within 0.03 in.
(0.76 mm). Seat the displacement gage in the knife edges to maintain registry
between the knife edges and the gage groove. In the case of attachable knife
edges, seat the gage before the knife edge positioning screws are tightened.
Load specimens at a rate such that the rate of increase of stress intensity is
within the range 30,000 to 150,000 psi.in.1/2/min (0.55 to 2.75 MN.m-3/2/s)
corresponding to a loading rate for the 1-in.-thick specimen between 4500
and 22,500 lb/rnin (0.034 to 0.17 kN/s).
7.5 Test Record--Make a test record consisting of an autographic plot of
the output of the load-sensing transducer versus the output of the displace-
ment gage. The initial slope of the linear portion shall be between 0.7 and 1.5.
It is conventional to plot the load along the vertical axis, as in an ordinary
tension test record. Select a combination of load-sensing transducer and
autographic recorder so that the load, PQ, (see 8.1) can be determined from
the test record with an accuracy of -4-1 percent. With any given equipment,
the accuracy of readout will be greater the larger the scale of the test record.
8. Calculation and Interpretation of Results

8.1 Interpretation of Test Record and Calculation of Kz~--In order to
establish that a valid K~ c has been determined, it is necessary first to calculate
a conditional result, KQ which involves a construction on the test record, and
then to determine whether this result is consistent with the size and yield
strength of the specimen according to 6.1. Referring to Fig. 8, the procedure
is as follows:
8. I. 1 On the test record draw the secant line OPz through the origin with
a slope 5 percent less than the slope of the tangent OA to the initial part of
the record (Note 6). Ps is the load at the intersection of the secant with the
record. Next, determine the load PQ which will be used to calculate KQ as
follows: if the load at every point on the record which precedes Ps is lower
than Ps, then PQ is equal to Ps (Fig. 8, Type 1); if, however, there is a maxi-
m u m load preceding Ps which exceeds it, then this maximum load is PQ
(Fig. 8, Types II and III).
NOTE 6----Slightnonlinearity at the very beginning of a record often occurs during seating
of the specimen and the gage, and should be ignored. However, it is important to establish
the initial slope of the record with high precision, and therefore it is advisable to minimize
this nonlinearity by a preliminary loading and unloading with the maximum load not pro-
ducing a stress intensity level exceeding that used in the final stage of fatigue cracking.
NOTE: SLOPE OP5 IS EXAGGERATED FOR CLARITY
A A
/j A
XI ~.
.,=~ ///. ,
O
fTYPEIII
x'-7]%P=~ x, = o.G.
r , r
DISPLACEMENT GAGE OUTPUT

FIG. 8--Principal types of load-displacement records.
8.1.2 Draw a horizontal line representing a constant load of 0.8 Ps. Measure
the distance X1 along the horizontal line from the tangent OA to the record,
and measure the corresponding horizontal distance to Ps. If the ratio of the
deviation from linearity at 0.8 Ps to that at Ps is greater than 0.25, then the
test is not a valid K~c test [4]. Proceed to calculate KQ in order to estimate a
revised specimen size, but do not report KQ as a valid K~ r
8.1.3 For the bend test calculate KQ from PQ as follows:
PQSI2.9(~)a/z
KQ -- BW 3/~ 4.6 (W)a/2 q- 21.8 ( ~ ) 5n
- - 3 7 . 6 (W)V2 + 38.7 ( ~ ) 9 ' ~ 1
where:
Po = load as determined in 8.1.1, lb,
B = thickness of specimen, in.,
S = span length, in.,
W = depth of specimen, in., and
a = crack length as determined in 7.2.3, in.
8.1.3.1 When using SI units with PQ in newtons and dimensions in milli-
meters the above expression for Ko should be multiplied by 0.0348.
8.1.3.2 To facilitate calculation of KQ, values of the power series given in
brackets in the above expressions are tabulated in the following table for
specific values of a/IV.
a/IV f(a[W) a/W f(a/W)

0.450. . . . 2.28 0.500... 2.66
0.455 . . . . 2.32 0.505... 2.70
0.460 . . . . 2.35 0.510... 2.75
0.465 . . . . 2.39 0.515... 2.79
0.470 . . . . 2.42 0.520... 2.84
0.475 . . . . 2.46 0.525... 2.89
0.480 . . . . 2.50 0.530... 2.94
0.485 . . . . 2.54 0.535... 2.99
0.490 . . . . 2.58 0.540... 3.04
0.495 . . . . 2.62 0.545 .... 3.09
0.550 .... 3.15
8.1.4 For the tension test calculate KQ from Po as follows:
KQ-- BW
'/2P~ I29.6 ( W ) 1/~- 185.5 ( ~ ) at2 q- 655.7 ( ~ ) ~/2
- - 1 0 1 7 . 0 ( ~ ) V ~ + 638.9 ( ~ ) 9 / ~ 1
where:
PQ = load as determined in 8.1.1, lb,

B = thickness of specimen, in.,
W = width of specimen, in., and
a --- crack length as determined in 7.2.3, in.
8.1.4.1 W h e n using SI units with PQ in newtons and dimensions in mil-

limeters the above expression for Ko should be multiplied by 0.0348.
8.1.4.2 T o facilitate calculation o f KQ, values Of the power series given in
brackets in the above expressions are tabulated below for Specific values o f
a/w.
alW f(a[W) alW f(a/W)
0.450 . . . . 8.34 0.500... 9.60
0.455 . . . . 8.45 0.505... 9.75
0.460 . . . . 8.57 0.510... 9.90
0.465 . . . . 8.69 0.515... 10.05
0.470. .. 8.81 0.520... 10.21
0.475. .. 8.93 0.525... 10.37
0.480. .. 9.06 0.530... 10.54
0.485. .. 9.19 0.535... 10.71
0.490. .. 9.33 0.540 . . . . 10.89
0.495... 9.46 0.545 .... 11.07
0.550 .... 11.26
8.1.5 Calculate 2.5 (Kq/ays) 2. I f this quantity is less than both the thickness
and the crack length o f the specimen, then KQ is equal to Ktr Otherwise it is
necessary to use a larger specimen to determine KIr in order to satisfy this
requirement. The increase in dimensions can be estimated on the basis o f
KQ.
8.2 Fracture Appearance--The appearance of the fracture is valuable
supplementary information and shall be noted for each specimen. C o m m o n
types o f fracture appearance are shown in Fig. 9. F o r fractures o f Types a or
b, measure the average width, f, o f the central flat fracture area, and note and
record the proportion of oblique fracture per unit thickness (B -- f ) / B . M a k e
this measurement at a location midway between the crack tip and the un-
notched edge o f the specimen. R e p o r t fractures o f Type c as full oblique
fractures.
9. Report
9.1 The report shall include the following for each specimen tested:
9. !. 1 Thickness, B,
9.1.2 Depth, W,
NOICH
OR OR
a b c
FRACTION PREDOMINANT FULL

OBLIQUE OBLIQUE OBLIQUE
FIG. 9--Common types o f fracture appearance.
9.1.3 Fatigue precracking conditions in terms of:

9.1.3.1 Maximum stress intensity, Ki(max) and number of cycles for
terminal fatigue crack extension over a length at least 2.5 percent of the
overall length of notch plus crack, and
9.1.3.2 The stress intensity range for terminal crack extension,
9.1.4 Crack length measurements,
9.1.4.1 At center of crack front,
9.1.4.2 Midway between the center and the end of the crack front on each
side, and
9.1.4.3 At each surface.
9.1.5 Test temperature,
9.1.6 Relative humidity as determined by ASTM Method E 337, for Deter-
mining Relative Humidity by Wet- and Dry-Bulb Psychrometer,a
9.1.7 Loading rate in terms of Ks (change in stress intensity factor per unit
time),
9.1.8 Load-displacement record and associated calculations,
9.1.9 F r a c t u r e appearance,
9.1.10 Yield strength, a n d
9.1.11 KIe.
APPENDIX
A1. DOUBLE CANTILEVER CLIP-IN DISPLACEMENT GAGE

AI.1 The gage consists of two cantilever beams and a spacer block which are
clamped together with a single nut and bolt, as shown in Fig. 3. Electrical-resist-
ance strain gages are cemented to the tension and compression surfaces of each beam,
and are connected as a Wheatstone bridge incorporating a suitable balancing re-
sistor. The material for the gage beams should have a high ratio of yield strength
to elastic modulus, and titanium alloy 13V-11Cr-3A1 in the solution treated con-
dition has been found very satisfactory for this purpose. If a material of different
modulus is substituted, the spring constant of the assembly will change corre-
spondingly, but the other characteristics will not be affected. Detailed dimensions for
for the beams and spacer block are given in Figs. A1 and A2. For these particular
dimensions the useful operating range of the assembly is from 0.15 to 0.30 in.
(3.8 to 7.6 mm). The range can be altered by substituting a spacer block of appro-
priate height. As discussed in 5.4, the precision of the gage corresponds to a maxi-
mum deviation of • in. (0.0025 mm) of the displacement readings from a
least-squares best-fit straight line through the data. Further details concerning
design, construction and use of these gages are given in Ref 9.
References
Strength Metallic Materials, ASTM STP 410, ASTTA, American Society for Testing
and Materials, 1966.
[2] Srawley, J. E., "Plane Strain Fracture Toughness," Fracture, Vol. 4, Chapter 2, pp.
45-68.
[3] Fracture Toughness Testing and Its Applications, A S T M STP 381, ASTTA, American
Society for Testing and Materials, April 1965.
[4] Wessel, E. T., "State of the Art of the WOL Specimen for K~c Fracture Toughness
Testing," Engineering Fracture Mechanics, Vol. 1, No. 1, Jan. 1968.
Fracture Toughness," Materials Research and Standards, MIRSA, American Society for
Testing and Materials, Vol. 7, No. 6, June 1967, p. 262.
[6] Fisher, D. M. and Repko, A. J., "Note on Inclination of Fatigue Cracks in Plane Strain
Fracture Toughness Test Specimens," Materials Research and Standards, MIRSA,
American Society for Testing and Materials, Vol. 9, No. 4, April 1969.
[7] Jones, M. H., Bubsey, R. T., and Brown, W. F., Jr., "Clevis Design for Compact
Tension Specimens Used in K~o Testing," Materials Research and Standards, MIRSA,
American Society for Testing and Materials, Vol. 9, No. 5, May 1969.
[8] Heyer, R. H. and McCabe, D. E., "Evaluation of a Method of Test for Plane-Strain
Fracture Toughness Using A Bend Specimen," Research Laboratory, Armco Steel Corp.
Middletown, Ohio.
[9] Fisher, D. M., Bubsey, R. T., and Srawley, J. E., "Design and Use of a Displacement
Gage for Crack Extension Measurements," NASA TN-D-3724, NASNA, National
Aeronautics and Space Administration, 1966.
1.6275
t.620
9/64 DRILL FOR

6 - 3 2 NC SCREW
0.375
0.041
0.039
0.020
~
R I"= o.~
re) m
/ ~ _ 0.065
"\\ / ~o'-~\ 0.060
\,o l I
I O.OZi J
/
NotrE--Dimensions are in inches.
METRIC EQUIVALENTS

U n i t s , in. mm U n i t s , in. mm
0.004 . . . . . . 0.10 0.060 ...... 1.52

0.006 . . . . . . 0.15 0.065 . . . . . . 1.65
0.010 . . . . . . 0.25 9/64 ...... 3.6
0 019 . . . . . . 0.48 0.186 . . . . . . 4.72
0.020 ...... 0.51 0.188 . . . . . . 4.78
0.021 . . . . . . 0.53 0.370 ...... 9.40
0.025 . . . . . . 0.64 0.373 . . . . . . 9.47
0.030 . . . . . . 0.76 0.375 . . . . . . 9.52
0.039 . . . . . . 0.99 1.620 . . . . . . 41.15
0.041 . . . . . . 1.04 1.625 . . . . . . 41.28
FIG. AI--Beams for double-cantilever displacement gage.
TENTATIVEMETHOD OF TEST 269
5/64 DRILL ~ - - ~ ~ 8 ~
z HOLES--~II ~ I o.19o
0'166 /~ / /
~ I , o.o93
o. oo
I IYi "
" ~--f --UNDERCUT
0,050
0.045 45~ CHAMFER
9/64 DRILL FOR /

6--32 NC SCREW
N o T E - - D i m e n s i o n s a r e in inches.
METRIC EQUIVALENTS

U n i t s , in. mm U n i t s , in. mm
1/32 ....... 0.80 0.195 . . . . . . 4.95

0.045 . . . . . . 1.14 0.205 . . . . . . 5.21
0.050 . . . . . . 1.27 0.373 . . . . . . 9.47
5/64 ....... 2.00 0.375 . . . . . . 9.52
0.087 . . . . . . 2.21 0.376 . . . . . . 9.55
0.093 . . . . . . 2.36 0.378 . . . . . . 9.60
0.125 . . . . . . 3.18 0.400 . . . . . . 10.t6
9/64 ....... 3.60 0.402 . . . . . . 10.21
0.186 . . . . . . 4.72 0.490 . . . . . . 12.45
0.188 . . . . . . 4.78 1/2 ....... 12.70
0.190 . . . . . . 4.83 0.500 . . . . . . 12.70
FIG. A2--Spacer block for double-cantilever displacement gage.

STP 463-1970

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STP 463-1970

Uploaded by

Copyright:

Available Formats

REVIEW OF DEVELOPMENTS

IN PLANE STRAIN FRACTURE

W. F. Brown, Jr., editor

ASTM SPECIAL TECHNICAL PUBLICATION 463

List price $18.25

AMERICAN SOCIETY FOR TESTING AND MATERIALS

NATIONAL AERONAUTICS AND

Printed in LutherviUe-Timonium, Md.

C r a c k T o u g h n e s s Testing o f High-Strength Steels--E, A. STEIGERWALD. . . . . . 102

Fatigue Cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

This Special Technical Publication brings together a set of papers that

a part of the Aerospace Material Document series issued by the Society of

Progress in Fracture Testing of

REFERENCE: Kaufman, J. G., "Progress in Fracture Testing of Metallic Mate-

ABSTRACT: Progress in the development of fracture-toughness testing techniques

KEY WORDS: fractures (materials), toughness, test techniques, fatigue pre-

In 1958, A Special ASTM Committee on Fracture Testing, now known

To preserve the continuity, it is necessary to present the review in rather

Initial Procedural Recommendations

Thickness Emerges Clearly as a Problem

r/I~~~, /11~'/// ////~///

As a result of the close attention to crack initiation a "pop-in" concept

FIG. 4--Representative load-deformation curves from fracture-toughness tests, showing the

at the behavior of metals in the presence of cracks or metallurgical flaws,

Emphasis Shifts to Thick Sections

Specimen Thickness, in.

F I G . 5--Values of KQ (candidate value of K i t ) ,from tension tests of single-edge-notched

Evolution of Notched Bend Specimens

result of a blunt crack, and, as mentioned previously (Fig. 4), fatigue-

The Compact Kic Specimen

FIG. 8--Representative load-deformation curves from notch-bend fracture toughness tests o f

specimen achieved this greater applicability through a reduction in the frac-

upon service experience. This point is emphasized solely to illustrate that

Back to the Thin-Section Problem

that once a recommended fracture test procedure has been adopted it is

[7] ASTM Special Committee on Fracture Testing of High-Strength Metallic Materials,

Evaluation of a Method of Test for Plane

REFERENCE: Heyer, R. H. and McCabe, D. E., "Evaluation of a Method of Test

ABSTRACT: A task group of ASTM Committee E-24 conducted an interlabora-

KEY WORDS: interlaboratory tests, fractures (materials), toughness, fatigue

There were several important steps in the development of a linear elastic

As experience with the practice broadened, changes were made in the

TABLE 1--Materials and mechanical properties.

2219-T851 18Ni Maraging 4340--500F 4340-800 F

Yield strength, ksi . . . . . . 51.2 276 238 206

2219 4340 18-Hi 4340

1. Specimens were lost in fatigue cracking (about four percent of the

TABLE 2--Summary of information of fatigue cracking. c

TABLE 3 - - T e s t results, Kit, ksi in.1/L 8c

NOTE--1. Specimen broke in fatigue cracking.

FIG. 2--Photographs of crack fronts in 4340 steels.

l FIG. 3--Bend test fixture design.

NOTE: SLOPE OPs IS EXAGGERATED FOR CLARITY

2219 4340 18Ni 4340

2.5 (Ki~/~rvs) ~. . . . . . . . 1.0l 0.09 0.09 0.38

FIG. 5--Typical load-displacement records.

Analysis of KIe Data

2 --7 --4 _=~3

2219 4340 18Ni 4340

+2aKi. + mean Kic • 100

Average replication for all laboratories improves as KIo increases. The

00 00o0o 2219 T851

AverageKic- KSl In1/2, ~ +-~gp,

A e r o n a u t i c s and Space A d m i n i s t r a t i o n , Lewis R e s e a r c h Center; T h o m p s o n ,