TALUG AND AEIFSNIDEA ON SMALL FLAWS 95

boring grains that are relatively unaffected by the singular field unlike larger
ilK stress states. Finally. grain boundaries and the impurity and other
crysta\lographic disturbances associated with them would be expected to
be very effective crack stoppers (Re( 9). 11 would appear, then. that such an
argument is consistent with the existen~ of a physical property 6.K.... below
which cracks do not propagate.
Our earlier implication was that the concept of IlK_ could be related to
the endurance limit. That connection follows from the definition of the
stress intensity factor. For general purposes let us consider the form
0
K :::: 0 J 2Tro ( 12)

where 00' (1::: remote applied stress and 0 :::: crack length. If K is set equal to
IlK_ and rl is set equal to the ultimate strength. 0 should be the maximum
dimension of a flaw which cannot be propagated by any application of stress
which does not fail the material by gross fracture. For the material con-
sidered previously. that dimension is about 10-1 mm as before. If we set 0 0
equal to the endurance limit. we imply that flaws of length 0 preexist in the
material. For our example material. that result is about 4 X 10- 1 mm which
compares well to a common grain size. for example. If the maximum dimen-
sion of the preexisting flaws and ilK.. are known, a fatigue endurance limit
can be predicted, and so on. The choice of grain size as a characteristic
dimension is only a convenience. Characteristic impurity inclusion dimen-
sions, machining scratches. etc., could have been used instead. However,
for a fixed ilK,.",,, Eq 12 implies that a smaller grain size (smaller 0) will
allow a higher fatigue endurance limit , an effect which is observed in most
metals, other things being relatively constant [10]. Inclusions, however,
although more difficult to characterize, are more commonly associated with
initiation of small flaw growth (JI).
Finally. it should be noted Ihat the method of solution employed here
involves an established stress function that satisfies the governing equation
and a part of the boundary conditions of the problem. The remaining
boundary conditions are only satisfied at discrete points. It is, therefore,
expected that the accuracy of the solution would improve as the number
of boundary stations is increased. It should be noted that. if a direct 'elimina-
tion type of solution scheme is employed, a practical limit to the number
of boundary stations exists. Possible alternatives for the solution method
would be an iterative scheme for solving the linear system of equations or
making use of a least squares approach in applying the boundary conditions
rather than the point collocation method.
Rdu~nces

(/] Ctark. W. G.. Jr .. M~fQls ~i,,"rjf16 Qullmrly. A...... 1974, pp. 16-22-
[1] Conle. A.. Nf:Wllck. H.. Ind HlulC llmlnn. D.. -TIle Effed of Enginterin, Approx-

pr( do der
N. E. Dowlingl

Crack Growth During Low-Cycle

Fatigue of Smooth Axial Specimens

REFERENCE: Dowling, N. E., ~Cnoct Growth Durl.. Low-CJdoe Fad,OII of

S.ooth Adal Sped_ns,~ C,..:lir SIr~D-Slrain and P/(wi( lNjormDlion Asp«u
0[ FQllt~ C'lU"k G.o .... /h. ASH., STP 6J7. Am erican Socie!~ for Te!!in, aod Ma-
lerials, 1971, pp. 97-121.

ABSTRACT: CRck ,roWlh data a~ reponed for small 1lt~lIy loaded smooth
spttime ns of A533 8 it«l subjccted to stR;n eyel;n, fatiaue It IItle plasUe main,.
Surface crack lenl !hs ....ere monitored using cellulose acetate replicas, Ind occasional
specimens wcre broken open to determi oc cRck depth . Experimental crack ,,0,,"1.h
notes for different stRin le vels are ooTTCII!ed in fnocture m«lumic. fashion by Ihe
J inlelral COl"ICept, with J values beinJ eSlima!ed from sttUl-CtR;n hys!c-resis loops .
The crack I roMh ri le data of th is invcsliplion Ire compated wit h previous data
for tM lime materiaL obtained from linear ell. tic fraclure mechanics tests. I! is SUI-
Flted that resean:h On rhe behavior of small cracks i. fundamenlallo a belle. unde .-
lta ndinl of the fllilUC process.

KEY WORDS: "resses, strairu, fa!i,ue , cract s, deformalion

Fatigue crack initiation has been studied in considerable detail from the
viewpoint or dislocations and metallurgy to determine the mechanisms in-
volved. In such stUdies, the size scale of interest is usually of the order of
the crystallographic grain size or smaUer. At a size scale on the order of
I in., the growth of fa tigue cracks has been studied intensely from the view-
point of linea r elastic fracture mechanics.
Less attention has been given to the specifi c nature of fatigu e damage
at intermediate size scales. In particular, relatively few slUdies have been
made of the growth and behavior of small cracks, those of sizes on the order
of IO- l to 10- 1 in. However, cracks in this size range are of major im-
portance. Many components in service and many test specimens in fatigue
studies probably spend the major portion of their lives with such cracks
being present.
This paper reports an experimental st udy of crack growth in small axially
loaded smooth specimens subjected to cyclic plastic derormation. Such tests
are usually thought of as measuring the resistance of a material to macro-

' ~nior enBiIlttT. Wcstinchouse Researeh. Pillsbur,lI, PI . 1!l2J!I.

97

pr( ido pot oor

98 FATIGUE CRACK GROWTH

cracking. The total number of cycles to failure and the stress-strain behavior
are the principal information sought, and crack growth data generally are
not obtained. In contrast to the tests normally employed in fatigue crack
growth studies. those reported here involve specimen widths one order of
magnitude smaller, crack sizes one to three orders of magnitude smaller.
and plastic rather than elastic load versus deflection behavior. An attempt
nevertheless will be made to make a fracture mechanics type of inter-
pretation of the data obtained.

In this section of the paper, the material and test specimens employed are
described first. Test procedures are then discussed. followed by presentation
of test results.
Material and Specimens
The material tested was A533B pressure vessel steel having the following
mechanical properties: 70 ksi yield. 91 ksi ultimate. 187 ksi true fracture
strength. 69 percent reduction in area, and 95 0 F Charpy fracture appearance
transition temperature.
Cyclic stress-strain and low-cycle fatigue data for this material are shown
in Figs. I and 2. An equation fitted to the cyclic stress-strain data is given
in Fig. I. In Fig. 2, intercepts at 1/ 2 cycle, tf and 0;.
and also slopes. are
given for lines fitted to both plastic strain versus life and elastic strain versus
life. These may be used to obtain an equation relating total strain and life
as described in Ref J and ASTM Recommended Practice for Constant-

/
M~'~k

---- ----- -
I(\, __ _ --------
I -
I
I
I
I Cycl k: -
G IIG. I65
.' , (,,)
t," Z'I. OCO + m
-
"~__L-
' ~~ ' __L-~~__L -__~~
o G.G I O.Ol G.Ol
511,ln AmpmUde t,.
FIG. I-Cydi( imd mo n% n k JIUu-slrain t;Un·~s.

M pr K;lo pot del

 OOWUNG ON CRACK GR OWTH DURING lOW-CYClE FATIGUE 99

.. ~
.\5HII SINI

.. ~

i• 0.01
I

""." '. OOJ

Gj · 126 h i

. .. .
" I

0.001

IQ' 10' I~
" N. C)'tln tofi '"m

Amplitude Low-Cycle Fatigue Testing (E 606-7IT). The data in Figs. I and

2 were obtained during the course of this investigation, as described in de-
tail in the following section.
Threaded-end specimens with reduced-diameter. cylindrical test sections
were employed. the test sections being 0.55 in. long and 0.30 in. in diameter.
This type and size of specimen is typical of those used in strain-cantrolled.
axial low-cycle fatigue testing (ASTM Method E 606-77T and [1]).1
Material for these specimens was obtained from broken fracture speci-
mens of large size. which had in turn been taken from a 12-in.-thick plate.
Two inches of material on both surfaces of the original plate was avoided,
as was any region of plastic deformation in the fracture specimens. The
tension axis for the small axial specimens was perpendicular to both the
rolling direction and thickness of the original plate.

T~sl Procedur~s

All tests were conducted on a closed-loop electrohydraulic testing system.

A Wood's metal grip similar to that described in Ref 1 or ASTM Method
E 606-7IT was employed. To prevent specimen buckling and to ensure
that alignment was maintained . it was necessary to stiffen the hydraulic
actuator against lateral motion.
It was desired to follow crack growth while measuring strain in the test

lThe italic numben in brackets refer 10 IlK lisl of references appendtd 10 Ihis paper.

pr( ido por oor