H. W. L1U Fatigue.

Crack Propagation and Applied

Senior Research Fellow,
Graduate Aeronautical Laboratories,
California Institute of Technology,
Pasadena, Colif.
Stress Range---An Energy Approach
Using the saturation of hysteresis energy absorption as a criterion for fatigue crack
propagation, the rate of crack propagation in a thin and wide metal sheet under repeated
axial loading is found to be proportional to the square of the applied stress range. The
local stress and strain fields adjacent to the crack tip are employed to compute the energy
density. The analytical results correlate very well with the experimental results on
2024-T3 aluminum alloy.

I Introduction correlation of the analytical with the experimental results is very

good. Experiments with better environmental control will con-
EVER since 'Vohler found that fatigue or progressive tribute more valuable information.
fracture was caused by initiation and propagation of one or more
cracks, fatigue has been the subject of numerous engineering and II Stresses.Strains,andHysteresisEnergyAbsorption
Near
scientific investigations. However, the advantages of separating
the phenomena of crack initiation and propagation have been
a Tipof a Crackin a PlateUnderRepeatedLoading
recognized only comparatively recently. After the crack or In order to calculate the absorbed hysteresis energy in the
cracks are formed, the stress and strain fields near the crack tips plastic zone, the elaatoplastdc solution of a plate containing a
are intensified. It is these intensified stresses and strains which central crack under axial load is needed. Since only limited in-
are responsible for subsequent fatigue damage, crack propagation, formation of this problem is known, therefore the elastic solution
and final fracture. Therefore the gross macroscopic stresses and will be modified and used in this investigation.
strains are only remotely related to the crack propagation. A wide thin sheet of material containing a. crack of length
In a previous investigation i1],1 the author derived the follow- 2a (Fig. 1) is loaded axially by a combination of a constant range
ing expression for crack propagation in a thin, infinitely wide of repeated stress t.1.S and a constant maximum stress Sm'- A
sheet specimen under any combination of stress range tiS and gross nominal stress S is shown acting on the specimen in Fig. 1.
maximum stress Sma",. In an infinite plate, the elastic stresses at a point P(r, 8} near the
crack tip for the case of plane stress were investigated by Wester-
d(2a)
C(2a) (la) ~Any two of the four variables, amplitude of stress, mean stress,
dN maximum stress. and minimum stress, provide an equivalent and.
therefore, a complete specification of the repeated load.
or
5
(2a)
In (2a,) = C(N - No) (lb) y

where
2a crack length
N number of cycles of load p (r. 9)
C crack propagation factor
For a given combination of stress range and maximum stress, C 9
,--..I--(.__ .L-_-,--+-x
is a constant. It was also shown [1] that stress range is the
dominant factor in determining the magnitude of C. In this In- 1-20 --11-£>0
vestigation, the crack propagation factor C is related to the ap-
plied stress range by using the saturation of hysteresis energy
absorption as a criterion for fatigue crack propagation whereby
the local stress and strain fields adjacent to the crack tip are em-
played to compute the energy density. It is found that the
I Numbers in brackets designate References at end of paper.
Contributed by the Metals Engineering Division of THE AMERICAN
SOCIETY OF, MECHANICAL ENGINEERS and presented at the AWS·
5
AS:l\.'1EMetals Engineering Conference. Cleveland, Ohio, April 9-13.
1962. Manuscript received at ASME Headquarters, January 2, Fig. 1 Schematic diagram of a section of an ,axially loaded sheet
1962. Paper No. 62-Met-2. specimen with a central crack

---Nomenclalure---------------------------
a half crack length G shearing modulus of elasticity TO octahedral shear stress
Poisson's ratio Toy octahedral shear yield strength
r, 0 polar coordinates I'

.
S gross sectional stress ~TDY octahedral shear yield strength
rp plastic zone size range
AS gross sectional stress range
C crack propagation factor Srnax maximum gross sectional stress ~ normal strain
N number of cycles of load (J normal stress 'Yo octahedral shear strain
Young's modulus of elasticity T shearing stress 'YoY octahedral shear strain at yield
E

116 I MARCH 1963

Transactions01the ASME
gaard [2], Irwin [3], and Williams [41. The principal stresses and 3 After each reversal, if L\7"OY is exceeded, plastic deformation
strains at a point P arc will continue at a constant value of stress, Le., ideally plastic
deformation is assumed.
8 ~a-cos-
_/_
~2 T
e [ l+sin-
2 .
OJ
2
(2a) Suppose that a constant range of repeated stress 11S is applied.
The ranges of plastic strains at a point P due to an applied stress
range AS are the differences between the. total ranges of strains

J2~~co'f[I-'infJ (2b) and the elastic ranges of strains at the point. The total ranges of
strains experienced at point P due to the applied stress range AS
are assumed to be the differences between the maximum and the
Ua = 0 (20)
minimum strains as given by Eq. (3). Hence
and

V28 E .t: cO'"2 ° [0 - + (1 + fJ,) sin "2OJ (3a)

fl',~:: cos: [0 -
E ~~ fJ,) + (1 + fJ,) sinf J (5a)

fl"- ~ .G. cosi

E, = ,,-; fJ,)

V2EY;: [(1 - fJ,) - (1 + fJ,) siniJ (5b)

e, ~ ~ E ~~ cosf [0 - fJ,) - (1~ fJ,)sinfJ (3b)

r 2

.1~3= - Ii! J~cos !!.- 2~ (5c)

8 '~ ti· V2E", 2
fa = - .y2E ~-;: cosZ2~ (30)
The ranges of plastic strains are the differences between the
total ranges of strains given by Eq. (5) and the elastic ranges of
where
strains. After each reversal of the sense of load and deformation,
E elastic modulus there is an elastic deformation range Li'YoY. Within this range,
j.I Poisson's ratio the strains at a point are linearly proportional to the stresses at
this point. The range of the applied stress LiS. for elastic defor-
The octahedral shear strain "Yoand the octahedral shear stress TO mation after each reversal of the sense of load can be found by
are considering A:")'oy first. The total strains were assumed to be
given by the elastic solution, Eqe. (3) and (4a). Suppose that the
'Yo 8.ra
30"--;C08Z °[ 1+3 (8)'ZJ'1>
sin (4a) specimen is loaded to some gross sectional stress S~, then un-
loaded to Si' At the maximum stress Smax the octahedral shear
strain is 'Ymax. When the stress is unloaded to Si. the octahedral
TO = O'Yo (4b)
shear strain is reduced to "Yo- The change in octahedral shear
where strain o"Yoisthe difference between 'Yma::rand 'Y. given by Eq. (4a).
Therefore,
G = shear modulus

'When the octahedral shear stress reaches the yield strength

TW the material yields. Therefore the plastic zone is defined by
an lao-octahedral shear stress line of the magnitude ToY. Neglect-
08 ,,-;
3G . t: cos"2e [ 1 + 3 (8 '2 )'J';'
sin (6)

ing stress and strain redistribution because of the plastic defor-

mation, the extent of the plastic zone rp' can be obtained by set- where
ting7"~equal to TOY in Eqs. (4a) and (4b). 08 = Smu. - S,
Therefore the elastic range of octahedral shear strain at P is

The maximum Tp is in the direction of. () = 70.6°. A'YOY ~. ~~' ~~ co. f [1 +3 (,in f)T' (7a)
Within the plastic zone, the octahedral shear stress is assumed
to be the yield strength TcY; outside the plastic zone, the octahe- A"YoYis assumed to be a constant and is defined schematically
dral shear stress is less than the yield strength. .The total strain in Fig ..2. Similar analysis can be made for the case where load is
at any point is assumed to be given by Eq. (3). The above
discussionof the stress and strain fields near the crack tip and the
plasticsoneaiae is for a material under a monotonically increasing
load. For the real material under repeated load, the problem is
greatly complicated by the repeated plastic deformation near the
cracktip and the history effect on the behavior of the material.
However, this complication will be simplified by considering the
total cyclic strain range as given by the elastic solution together
withan assumed simple yielding behavior under repeated loading.
Aschematic diagram of the simplified yielding behavior under
repeated loading is shown in Fig. 2. A hypothetical loading path
oo.bcdeju, is shown to demonstrate the important characteristics r.
ofthe assumed yield behavior under repeated loading. The major
characteristics can be summarized as follows: .
1 Immediately after each reversal of the sense of deformation,
the slope of the stress-strain diagram bas the same value as at the
beginning(0) of the first loading.
d c
2 After each reversal, the material will deform elastically
until the yield strength range _ATnY is exceeded. The yield Fig. 2 Schematic diagram showing yield behavior under repeated
8trengt;hrange is assumed to be constant. loading

Journal01 Basic Engineering

MARCH .1963 '/ 117
 reversed after the minimum stress is reached. Since A-yo¥ is in I1Ep_
the elastic range, it is related to the elastic range of octahedral cycle
shear stress by

LiToY = A'YovG (7b) 3I1T,Y[(1 - 1') + (1 + 1') (sin~)'J

( O)'J'I' -
(11 )
Solving Eq. (7a) for AS. and writing in terms of dTOY
E [ I +3 sin 2" 6
I1S ~ M.T,,· (7e)
when ..' f3 was defined previously.
• ~ cos~ [1+ 3 (Sin f)T' When the octahedral shear stress exceeds the yield strength
To}' under the monotonically increasing loading, the material
It should be noted that /lS, depends upon the position of the flows. The size of the plastic zone is given by Eq. (4c). During
point. Upon substituting AS. into Eq. (5), the ranges of elastic the unloading half cycle of the repeated loading, if AS is not high
strains can be found as enough to cause the octahedral shear stress range at a point to
exceed drOYI there will be no pla-stic deformation taking place
3!J.To,[(1-I') + (I + I')Sin~J during the unloading half cycle although it may yield during the
loading half cycle. Therefore there is no hysteresis energy ab-
11". ~ V2E[1 + 3 (sin fYT' (8a)
sorbed. The effective plastic zone size for the hysteresis energy
absorption must consider the applied stress range and is then given
by

r, ~~ C~~~J
f)' [I+ (cos 3 (Sin f)'J (12)

Substituting Eq. (12) into Eq. (11),

(Sin ~YJ
(8b)
I1E, 9I1T'O" [(I - 1') + (1 + 1')
3/lTOY

V2E [ 1 + 3 ( sin '2e )'J'I' 21'

(8e)
cycle E [I+ (Sill~)']3

The plastic strain ranges are the differences between Eq. (5) and {~; - I} (13a)
Eq. (8).
At a point P(T, 0) along the crack path, the hysteresis energy ab-
11,,p -.!-
v2E
[(1 - 1') + (1 + 1') siniJ2 (9a) sorption per cycle is

I1Ep 9( I - 1')I1Tov' {.I rp I} (13b)

11", ~ .,g
~ _

v2E
[(1 - 1') - (1 + 1') sin iJ2 (9b) cycle E ~ r
Eq. (13) indicates that, at the homologous points, the geometri-
{3 cally similar points within the plastic region, i.e., at the points
~€3P = -----=-- 2fJ. (ge)
of the same values of Tp/r and 8, the hysteresis energy absorption
V2E
per cycle is the same. Both the stress range t1S and the crack
where length will change the size of the plastic zone r e-

{3 ~ I1S ~~ cos ~ - [
1+3
3t
OY

sm-
0yr
2
III Saturation of Hysteresis Energy Absorption as a
Criterion lor a fatigue Crack Propagation [5, 6, 7, 8]
As the load is repeated. the localized macroscopic plastic ~e-
After each reversal of loading, the material at a point will de- formation reflects the cumulative effect of the movement of in-
form elastically for a gross sectional stress range, AS., At first, dividual dislocations. The hysteresis energy absorbed is partly
the stresses of a point increase with S until !J.S is equal to AS and dissipated because of the movement of dislocations an.d
ATo at the point is equal to Llroy. Then, it is assumed, with any partly stored in the crystals. Eq. (13) gives the hy~teresls
further increase in AS, the material at the point will deform energy absorbed per cycle The rate of crac~ propa~atlOn can
plastically without any change in its stresses. Therefore the be related to the applied stress range LiB by using the input hys-
ranges of stress at the point can be obtained by eubsntutin« LiS teresis energy as a measurement of fatigue damage ".
for S in Eq. (2). 0 •
When the hysteresis energy absorption at a P01~t reaches a
critical amount the saturation point, new surface will form and
311Toy (1 +sin f) the crack will propagate through this material. It is ho?ed that,
in the future the critical saturation point of hysteresis energy
(lOa) absorption fo~ crack propagation can be exp.lained i.n grea.ter de-
tail in terms of the movement and the formation of dislocations.
The rate of the absorption of hysteresis energy .will b~ con-
sidered for homologous incremental crack lengths m spectmene
under different applied stress ranges. Along the cr~ck paths,
(lOb) IJ = 0, of two different specimens under repeated loading of two
different stress ranges LlS] and t1S2, the incremental crack lengths
.L\a, and t1a2 are homologous if
(lOe) .1al Aat = k" (14)
The hysteresis energy absorption per cycle of loading at point P is T,,\ . Tp2

118/ MARCH 1963 Transactions 01 the ASME

Assuming the material within Aat and Aa2 has absorbed the same b Smox: 40 000 psi
a.mount of hysteresis energy due to previous cycling, the addi-
tional amount of hysteresis energy needed to propagate the cracks o Smox. 36 000 psi
-3.6
through .6.al and Aa2 must be the same. Therefore according to "i- Smox" 32000 psi
Eq. (13), the additional number of cycles!1N necessary to propa-
v Smcx> 28000
psi
gate a crack through Aat and Aa2 will be the same. Therefore -3. 71-'-="'-'--r-=~:---I-~~--1
Eq. (14) can be written as

da,
-- =k' (15) -3.8f----+---A--,'~---1
rp2/:1N

Dropping the subscripts and writing Eq. (15) in differential

-3.9 f------+--.,/-,/,riL-------j
form: .0
d(Za) ~
o
(100)
dN -'
-4.0 f---------,4-,L.-./_
d(Za)
dN ~ k(Za)dS' (16b)
-4.1 f¥~,L__+L------J------J
Upon integration and use of the relation of Eq. (12), Eq. (16b)
becomes
Za
in Z"" = kdS'(N - No) (i6c)

Eq. (16) indicates that the fatigue crack propagation rate is -4.3·i-?v,£---1-----!--------1
proportional to .182 •
. Equation (16) was derived for plane stress condition, however, 4.40 4.50 4.60
It can also be derived for plane strain condition, where 173 = Log AS
p.(Ut + 172) and €3 = O. The derivation of Eq. (16) is based on Fig. 3 Correlation between crack prapagalion factor and stress rarwe
several idealized assumptions, and therefore its validity seems to
be restricted. However, the derivation and thereby the validity 2024-T3 aluminum alloy against the stress range fJ.S. The ex-
of Eq. (16) depend upon only three conditions: the plastic zone perimental results were reported in detail previously [I}. Straight
size is proportional to the square of the applied stress range /:18 lines of slope 2 in Fig. 3 correlate very well for experimental re-
the strains at homologous points along the crack path are the sults of equi-maximum stress. The slope of 2 is predicted by
same, and the homologous points remaining homologous ail the Eq. (17). The maximum deviation of the experimental value
cracks propagate. from that predicted is only 15 per cent. The possible explanation
The first condition, i.e., T p is proportional to /:182, is satisfied for the good correlation through maximum stress is that the local
reasonably well if the plastic zone size is small. This is true crack propagation is controlled by the Griffith process, i.e., the
under ordinary fatigue loading. It is reasonable that the second elastic strain energy released by crack propagation is equal to the
condition, i.e., the strains at homologous points along the crack dissipated energy. The elastic strain energy released is a function
of maximum stress. Therefore maximum stress should give
~aths arc the same, is also satisfied. However, elastoplastic solu-
tion to the problem of a plate containing a central crack under reasonable correlation.
axial loading is necessary to verify these two conditions rigorously.
!he. third condition, homologous points remaining homologous; V Concluding Remarks
IS discussed in detail in the Appendix. The maximum error of After the crack or cracks were initiated by repeated load, the
thi? assumption for the' results reported in this investigation is stresses and strains near the crack tips were intensified. It is
estimated to be five per cent. Therefore, in spite of the fact that these intensified stresses and strains that are responsible for
the assumptions of the idealized stress-strain relation and the subsequent fatigue damage, cracking propagation, and final
strain distribution in the plastic zone may seem to be very re- fracture.
st~ctive, the final result, Eq. (16), is applicable to real materials, As the load is repeated, the material adjacent to the crack' tip
which strain harden by plastic deformation. will undergo repeated plastic deformation. With every cycle
of load, the material absorbs hysteresis energy. Saturation of
IV Experimental Results and Discussion hysteresis energy was hypothesized as a criterion for fatigue
crack propagation. Local stress and strain fields adjacent to
In a previous investigation [I), an expression, Eq. (1), for
the crack tip were used to compute the hysteresis energy ab-
crack propagation was derived, based on the concept of geometric
similarity of crack configuration, for an infinitely wide sheet sorption rate.
The fatigue crack propagation factor was found to be proper-
specimen subjected to repeated loads consisting of a constant
tional to the square of the applied stress range. Experimental re-
stress range f:J.8 and maximum stress Bronx.: .'
sults on 2024-'1'3 aluminum alloy support this analysis. The
d(Za) maximum deviation of fatigue crack propagation factor from the
-- ~ C(Zal (Ia) predicted value is 15 percent, which is small compared with the
dN .
usual scatter of fatigue data. Experiments on other materials,
where C is the stress dependent crack propagation factor. Com- under better environmental. controls will contribute additional
parison of Eq. (10) arid Eq ..(16b) valuable information.
d(Za)
dN ~ k(Z~)dS' . (16b) APPENDIX
indicates that the crack propagation fact~r C of Eq. CIa) is re-
The Effect of Crack Propagation on the Condition of
lated to the stress range fJ.8 by . Geometrical Homology Within the Plastic Region Along the
C ~ kdS' (i7) Crack Path
Fig. 3 is the logarithmic plot of crack' propagation factor C of . In the course of derivation of the crack propagation equation,

MARCH 1963/119
Journal of Basic Engineering
Eq. (16), the homologous points along the crack paths of two an equal amount. If the effect of increase of crack length is small
specimens were assumed to remain homologous after crack in comparison with the effect of decrease of the distance r and is
propagation due to the same number of cycles of loading. AB.~ thereby neglected, then homologous points along the crack paths
Burning P1(rl, 0) and P2(r2, 0) are homologous along the crack remain homologous as the crack propagates, therefore the as-
paths of two specimens under repeated load of stress ranges of sumption ofEq. (19) is true. In other words, the assumption that
aSt and AS'l, therefore, homologous points remain homologous as cracks propagate is
equivalent to ..the assumption that the plastic zone size Tp is small
(ISa) in comparison with the half crack length a. This is the case for
most of the fatigue tests.

Upon eubatitution of r"l and ';>2, Eq. (18) becomes

Acknowledgment
r, T2
This investigation was conducted while the author was Re-
(ISb)
aJ~.s12 = llt.L\S22 search Assistant Professor at the Department of Theoretical and
Applied Mechanics, University of Illinois, Urbana, JIl. - It was
If homologous points remain homologous after crack propagation conducted as a part of the work of the Engineering Experiment
of llal and .t:1tl2 then Station, University of Illinois, under OSf...F Contract No. AF
r/ 33(6161-5153, Project No. 7021, T""k No. 70627. Grateful
(19) acknowledgment is due Professors G. M. Sinclair, project super-
r,,' visor, and H. T -. Oorten of the University of Illinois for their
where helpful assistance and suggestions. The research was admin-
istered under the direction of the Metallurgy and Ceramic Lab-
oratory, Aeronautical Research Laboratory, Air Force Research
Division, Wright-Patterson Air Force Buee, Ohio. with Dr. Harry
A. Lipsitt acting as cognizant scientist. Acknowledgment is due
R. E. Price and Mrs. G. Koepke for their assistance to this in-
vestigation.

References
1 H. W. Liu, "Crack Propagation in Thin Metal Shcct Under
Repeated Loading," WADC Technical Note 59-383, December, 1959;
Upon substitution, Eq. (I9) becomes also JOURNALOF BASIC ENGINERING. Series D, TRANS. ASME, vol.
83, 1961, pp. 23-3l.

(1 - -fla,)
2 H. M. Westergaard, "Bearing Pressures and Cracks." Journal

r, r, (20)
of Applied MechanU::8, TRANS. ASME. vol. 61. 1939, pp. A-49-53.
3 G. R. Irwin, "Analysis of Stresses and Strains Near the End of
a Crack Traversing a Plate." Journal of Applied Mechanics, vol. 24,
TRANS. ASME. vol. 79, 1957, pp. 361-364.
4 M. L. Williams, "On the Stress Distribution at the Base of II.
Stationary Crack," Journal of Applie4 Mechanics, vel. 24, TR.ANS.
With the relations of Ega. (14) and (18), Eq. (20) can be ASME, vol. 79,1957, pp. 109-114.
5 C. E. Feltner and JoDean Morrow, "Microplastic Strain
simplified to
Hysteresis Energy as a Criterion for Fatigue Fracture," JOUR.NAL OF

1 1 BASIC ENGINEERING, Series D, TRANS. ASME, vol. 83, 1961, pp.

(21) 15-22. '
6 N. Enomoto, "On Fatigue Tests Under Progressive Stress,"
Proceeiinoe, ASTM, 1955. p. 903.
7 N. P. Inglis. "Hysteresis and Fatigue of Wohler Rotating
Cantilever Specimens." The Metallurgist, February, 1927, pp, 23-27.
If .1al and .t:1~ are infinitely small, the condition of Eq. (21) is 8 R. F. Henstock, "Damping Capacity, Strain Hardening and
true. However, l:i.a is finite and must be smaller than the plastic Fatigue," Proceelinae, PhY8ical Society, vol. 59, 1947, PP. 275-287.
zone size, since when homologous points are outside the plas-
tic region there is no hysteresis energy absorbed by the materials
at these points by stress cycling. Therefore, evaluating Eq. (21) DISCUSSION
and taking rpI and Tr/l and .L\aI .and I:i.~ for the extreme experi- D, S. Dugdale'
mental conditions reported in this investigation, the error of the
The analysis presented is a useful attempt to explain observed
ass~mption of Eq. (19) ca~ be estimated. Eq. (12) gives rate of crack growth in terms of saturation hysteresis energy.
r.
-;; ~ 9"
1 (tIS)'
AToy
(22)
Much experimental work on crack propagation has been done in
England in the past few years [9, 10].4, Plastic-elastic analysis has
also been attemped [11]. Several variants of the idea of satura-
The maximum and the minimum stress ranges for the experi- tion energy have been suggested [12]. The theory of Feltner and
mental results reported in this investigation are 38,000 psi and Morrow is attractive as it bolds promise of explaining why cracks
22,000 psi. The octahedral shear yield strength range ATOY for refuse to grow if applied stress range is insufficient, but its physical
the material, 2024-T3 aluminum alloy, is taken as 2Tor, which is basis is not completely satisfactory. It predicts that when a homo-
equal to 48,000 psi. Upon substitution of these values and Eq. geneous crack-free specimen reaches the saturation energy, it is
(22) into Eq. (21), the maximum error of the assumption of ho- ready to fracture suddenly at all points simultaneously. This has
mologous points remaining homologous as cracks propagate, is never been demonstrated, as fatigue fracture is always preceded
estimated to be 5 per cent. by gradual growth of a crack. The theory cannot be disproved
The condition of homologous points remaining homologous as either, unless Borne way is found of preventing crack formation
cracks propagate can be looked at from another point of view. while plastic strain cycles are continued. There is some evidence
As indicated by Eq -.(11), while the crack propagates, the rate of from the field of fatigue testing that endurance .can be extended
hysteresis energy absorption by the material at a point increases B Research Professor, University of Sheffield, Sheffield, England.
due to the two effects. First, the distance r decreases by .an 4,Numbers in brackets designate Additional References at end of
amount of lla, and, second, the half crack length a increases by this discussion.

120 I MARCH 1963

. Transactions of the ASME
by interrupting the test'to'remove a layer of surface material in d(2a ) (AKj>
(23)
which defects have formed. An alternative idea which the author dN ]I
could examine is that a crack root constitutes some kind of drain
for imperfections in the crystal lattice. With increasing stress and where M is a constant and K is Irwin's [17J crack tip intensity
increasing crack length these imperfections are generated at' an factor, or
increased rate' in alarger volume of metal, to produce the increased AK - (Au)a'l> (24)
rate of crack growth observed in macroscopic tests.
for the configuration of a crack of length, 2a, in an infinite plate
Additional References aubjectedto a uniform stress o at infinity. The experimental
9 N. E. Frost end K. Denton, "Effect of Sheet Thickness on the data strongly suggest 'Y = 4, as may be seen-in-Fig. 4 which im-
Rate, of Growth of Fatigue Cracks in Mild Steel," Journal of Me- plies, from Equations (23) and (24), that:
chanwal Engineering Science, vol. 3, 1961, pp. 295-298.
10 D. S. Dugdale, "An Experimenta.l Study of the V-Notch d(2a)
Fatigue Test," Journal o/the Mechanics and Physics of Solids, vol. 7. (25)
1959, pp. 282-287. dN
11 D. S. Dugdale, "Yielding of Steel Sheets Containing Slits,"
Journal of the Mechanic8 and Physics of Solids, vol. 8, H160, pp. 100- Upon comparing equation (25) with its various counterparts in
104. '..
Liu's paper the disagreement of Liu'e crack propagation laws
12 P. P. Benham and H. Ford, "LowEndurance Fatigue of a
Mild Steel and an Aluminum Alloy," Journal 01 Mechanical Engi- with the broad trend of the data is clear. Under the circum-
neering Science, vol. 3, 1961, pp. 119-132. stances no further comments seem appropriate.

AddlHonal References
Paul C. Paris' and Fazil Erdogan' 13 H. W. Liu, "Crack Propagation in Thin Metal Sheet Under
The Theories of Fatigue Crack Propagation presented by Liu Repeated Loading," TRANS. ASME, Series D, JOURNAL OF BASIC EN-

in the paper and in his earlier work [13F seem to be in disagree- GINEERING, vol. 83. 1961, p. 23. .
14 A. K. Head, "The Growth of Fatigue Cracks," The Philo-
ment with the work of Head [14, 15] and Frost [161. That is to 80phical Magazine, vol. 44, series 7,1953.
Bay that Liu's Equation (16b), in which the crack propagation 15 A. K. Head. "The Propagation of Fatigue Cracks," Journal of
rate is proportdonaltto the crack length and the.square of the Applied Mechanics, vol. 23, TRANS. ASME., vol. 78,1956, p. 407.
applied stress range, does not agree with Head and Frost who in 16 N. E. Frost and D. S. Dugdale, "The Propagation of Fatigue
Cracks in Sheet Specimens," Journal QI the Mechanics and PhY8ics 01
fact disagree with each other. Therefore one might be prone to Solid:!, vol. 6, 1958. '
suggest comparison of 'these laws with existing experimental 17 G. R. Irwin ... Analysis of Stresses and Strains Near the End
data. ' , of a Crack Traversing B Plate," Journal of Applied Mechanics, vol, 24,
TRANa. ASME, vel. 79, 1957, p. 361.
Fig. 4 of this discussion shows data from fou~ sources on the
crack propagation rates in 2024-T3 aluminum alloy sheets.
Since the data on this figure appear to form a straight line, the Author's Closure
relationship suggested by it is: The author thanks Professor Dugdale and Professors Paris and
,.
Erdogen for their enlightening discussions. The concept of the
'Assistant Professor of Civil Engineering, University of Washing- increasing rate of the generation of imperfections is consistent
ton (on leave 1960-1961, 1961-1962), Instructor of Mechanics, Lehigh
University, Bethlehem, Pe.: and Consultant to' The Boeing Airplane with the concept of hysteresis energy absorption. The hysteresis
Co., Transport Division, energy is the input energy which damages the metal, and the im-
6 Associate Professor Mechanical Engineering, Lehigh University. perfections are the end results of the damaging processes. The
Mem.ASME. ., . rate of the generation of imperfections is related to the rate
1 Numbers in brackets designate Additional References at end of
this discussion.
of hysteresis energy absorption, and the imperfections infested

,
5 .,.

/0
.. I I ~
6 _0 ~A~A
.1Jf
.. ~
/
4 _e L1U
c---
3
;2
-tJ.

-x
MARTIN

BO~ING
a SINCLAIR
, /r. ~ --- °
os
. x ~X
.. -
b
<l 4
I!l LIU
.<"0
/
-, XI ((I'
l!!
"
10
B
6 .... e»
00
c~
1-
2.024-131 ~
r cr
t

sc
<l ___ -:-- :>-a:: :A y I
4 ~ // I I <rr'a.k
3 ~
..- --0 . ..- /'
z 0 .....-- ./ 01 (2") =
(f;,.k{
..- .--- ./
dN M
3 /'
/0
/
/'
-1 -6 -5 -4
2 3 4 6 B 10
-3 2 3 4 6
10 z 3 4 6 B 10 2 3 4 6 8 /0 2 3 4 6 B '0
d(2o.) ( in/cycle)
dN
Fig. 4

Journal 01:BasicEngineering MARCH 1963/121

volume is related to the volume of the plastic zone. Therefore ranges of ASl and A82, respectively. At the geometrically similar
the imperfections are generated at an increased rate in a larger points of these two specimens, the hysteresis energy absorption is
volume of metal with increasing stress and increasing crack same as indicated by Equation (13). After a certai nnumber of
length as indicated by the proposed model. cycles, 6.N, the densities of the dislocations and the microcracks
The propagation of fatigue crack is intermittently slow and within the plastic zones of these two specimens reach the same
rapid [I8, 19, 20].8 Mter a rapid propagation of the crack or critical values, and the major cracks of both specimens are ready
"jump" as it is called by Christensen, the material ahead of the for another jump. If the intensity of the plastic energy absorption
crack tip is soft, and the crack remains dormant for a number of at geometrically similar points, during the jumping, is same, the
cycles. Further cyclic plastic deformation of the material during average intensity of plastic energy absorption I is same for both
the dormant period increases the densities of dislocations and specimens. If the materials within the plastic zones of both speci-
microcracks. The material is hardened by the additional disloca- mens are work hardened to the same critical point by the same
tions, and the strength of the material is reduced by the micro- number of cycles 6.N if the average intensities of plastic energy
J

cracks. When the densities of dislocations and microcracks reach absorption 1 of both specimens are same, and if the equality of
the critical values, the major crack will take another jump. As Equation (30a) holds for both specimens, Equation (30b) indicates
the crack jumps, the elastic energy of the specimen is reduced. that Slrnar. is equal 82max• Therefore the maximum stress should
Part of the released elastic energy becomes the surface energy, be kept same in order to correlate the crack propagation factors
and part of the energy is dissipated to enlarge the plastic zone with the applied stress ranges. If the mean stress or the minimum
ahead of the crack tip. stress is kept same, the exponent in Equation (16b) will no longer
The released elastic energy by the propagation of a crack be 2. The above discussion is for a wide specimen. Its validity
through Aa is depends on the assumption that the plastic deformation at
geometrically similar points is same. The assumption has to be
(26) compared with the elaatoplastic solution of a centrally cracked
where k' is a const·ant [3 J. The total plastic energy absorbed plate. 'The exponents of 2 in Equation (16b) would be modified,
ahead of a crack tip can be written as should further theoretical consideration and experimental results
justify. It is hoped that the sketch of the physical model will
s, ~ A!!](r,O)rdOdr (27) serve as a connection between the macro-phenomenological
fatigue investigations and the micro-mechanistic fatigue in-
where Itr, (J) is the intensity of the absorbed plastic energy, and A vestigations.
is the area of the plastic zone. The area of the plastic zone has The information given in Fig. 4 is of immense practical value.
been estimated to be [21] It covers a wide range of crack lengths and applied stresses,
Fig. 4 certainly points out the importance of a qualification of
(28) Equation (1) as discussed in reference [IJ. Equation (1) is valid
only if the mechanism of fatigue crack propagation remains same.
where kif is a constant. Let 1 be the average intensity of the ab-
Weibull [22J reported that the mode of fracture transits from the
sorbed plastic energy. Then the amount of the absorbed plastic
flat, to double shear, and to single shear as the crack propagates.
energy by the propagation of a crack through Aa is equal to
As the mode of fracture changes, the mechanism of fatigue crack
(29) propagation is no longer same. Therefore Equation (1) and,
consequently, Equation (16) ere no longer applicable.
If the surface energy is small and negligible, the released elastic
energy is equal to the absorbed plastic energy. Therefore Additional References
18 W. A. Wood, "Some Basic Studies in Fatigue of Metals,"
(30a) Conference on Fracture, Technology Press of Massachusetts Institute
of Technology, April, 1959.
or 19 H. A. Lipsitt, F. W. Forbes, and R. B. Baird, "Crack Prcpaga-
k' tion in Cold Rolled Aluminum Sheet," Proceedinae. ASTM, 1959.
(30b) 20 R. H. Christensen, "Cracking and Fracture in Metalsand Struc-
Smax' = 2k"1 ture," Symposium on Crack Propagation, Cranfield, England,
September, 1961.
Two specimens containing cracks 2al and 2a2long are loaded by 21 M. L. Williams, "A Review of Recent Research at GALCIT
fatigue loading of maximum stresses Slma.x and 82mru. and stress Concerning Fracture Initiation," ARDC, ARL TR-l, January, 1961.
22 W. Weibull, "The Effect of Size and Stress History on Fatigue
a Numbers in brackets designate Additional References at end of Crack Initiation and Propagation," Symposium on Crack Propega-
this closure. tion, Cranfield, England, September, 1961.

122/ MARCH 1963

Transactions of the ASME