Nonlinear Fracture Mechanics: John W. Hutchinson
Nonlinear Fracture Mechanics: John W. Hutchinson
l
JOHN W. HUTCHINSON
a course on
NONLINEAR
FRACTURE
MECHANICS
DEPARTMENT
OF
SOLID MECHANICS
THE TECHNICAL UNIVERSITY OF DENMARK
i
PREFACE
A COURSE ON NONLINEAR FRACTURE MECHANICS
The notes contained within were prepared in conjunction with a
set of lectures delivered during the fall of 1978 at The Solid
Mechanics Department of the Technical University of Denmark
under the auspices of the Danish Center for Applied Mathematics
and Mechanics. Starting more or less from scratch in the sub-
ject, but assuming at least a limited knowledge of plasticity,
I have tried to bring the reader right up to some of the topics
of current interest, from both engineering and research points
of view. The section headings are listed on the next page.
John W. Hutchinson
January 5, 1979
l.
2.
3.
4.
5.
6.
7.
ii
SECTION HEADINGS
Solutions to linear elastic crack problems ....... .
Energy release rate, compliance analysis, and rela-
tion to stress intensity factors
Energy methods for estimating G and K ......... .
The concept of small scale yielding .............. .
Initiation of crack growth in s.s.y.
Crack growth and stability in s.s.y.
Other applications of linear elastic fracture
1
c
10
12
14
19
mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 7
8. The J-integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
9. The Dugdale-Barenblatt model ...................... 35
10. Crack tip fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
11. Elastic-plastic solutions in s.s.y. ............... 48
12. Analysis of large scale yielding ........ .......... 5S
13. J-integral testing .. .............................. 63
14. Configuration dependence in l.s.y. and limitations
of single parameter crack tip characterizations ... 70
15. Relationship between ot and J under J-dorninated
conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 7
16. crack growth and stability under J-controlled con-
ditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
17. Source of stable crack growth ..................... 91
LINEAR ELASTIC FRACTURE MECHANICS
1. SOLUTIONS TO LINEAR ELASTIC CRACK PROBLEMS
References: P. Paris and G. Sih, ASTM STP 381, 1965
H. Tada, P. Paris and G. Irwin, The Stress Analysis
of Cracks Handbook, 1973.
H. Liebowitz (editor), Fracture, Academic Press
(7 vols.), 1971.
Isotropic, homogeneous elastic materials are assumed with E
as Young's modulus, v as Poisson's ratio. Let K = 3- 4v in
plane strain and K = ( 3- v) I ( 1 + v) in plane stress. The
shear modulus is G = E/[2(1+ v))
Common to all solutions of plane problems is the
dominant singularity at the crack tip.
(i) Symmetric fields with respect to crack tip in plane
stress or plane strain (Mode I)
u
a
(1.1)
(1. 2)
where a;
2
(6= 0) : 1 . The oo.f3(6) and u(8) are given in
almost all texts on fracture. Note that ahead of the crack
(1. 3)
2
KI is called the stress intensity factor (in Mode I).
v
33
= 0 in plane stress. o
33
= v(o
11
+ o
22
) in plane strain.
(ii) Anti-symmetric fields with respect to crack tip in plane
stress and plane strain (Mode II)
0 =
ai3
KII II
Oaa (8)
u
a
where o ~ (6 = 0) - 1
fiTir "
(iii) Anti-plane shear (Mode III)
u =
3
K
III 'f (S)
127fr a
2KIII fr sin(6/2)
G .j 2ii
(1. 4)
(1. 5)
0)
(1.6)
(1. 7)
3
where 'f
1
= -sin(S/2) and :r
2
= cos(8/2) .
In general, in any plane problem the crack tip singulari-
ty fields are a linear superposition of Mode I and Mode II.
For a 3 dimensional problem, the singular stress fields, at
any point along the crack edge, will be a linear superposition
of Modes I and II (plane strain) and Mode III.
A large catalogue of stress intensity factors is now
available. Most basic (and some not so basic) elastic crack
problems seem to have been solved, either analytically or
numerically. Perhaps the most complete source is the Stress
Analysis of Cracks Handbook. Some examples are listed below.
Finite crack in infinite plane in tension
(1.8)
Finite crack in infinite plane in shear
(1. 9)
Finite crack in infinite body and edge crack in half space in
Mode III
(1.10)
4
Edge crack in half plane in tension
Line of cracks in tension (also Mode III)
K
1
= oa> tan t
I 2 b I
I I
t=::::t c::::::::o
I I
I
I
c::::::::o
I
I
Cracks in finite width strips in Mode III
"'r--[2b (na)
Kill = 1: tna na tan 2b J
(l.ll)
(1.12)
(1.13)
=
2a
Standard ASTM compact tension specimen
P = load/unit thickness
(h= .6b, h
1
= .275b, D= .25b, c= .25b, thickness=b/2)
( 1.14)
p
h
a
b
18
14
12
see Tada. et. al.
1973
0.4 0.5 Q6 alb
p
Edge crack in strip in bending
moment/thickness
Mb -}/
2
f (a/b)
(1.15)
(a/b 1 , f 3.95)
( a/b ... o , K ... 1 .12 2 o lrra)
Penny shaped crack. Near crack edge fields are identical to
Hode I plane strain
6
(1.16)
2. ENERGY RELEASE RATE, COMPLIANCE ANALYSIS, AND RELATION TO
STRESS INTENSITY FACTORS
References: Tada, 1973.
consider prescribed load problems
in Mode I. Let a be length of crack
and P be the load per unit thickness.
Let PE denote the potential energy of
the system (per unit thickness) and let
be the load-point displacement
through which P works. Denote by G
the energy release rate per unit thick-
ness, i.e.
G = _ (<lPE]
aa P
Now for prescribed P ,
PE SE - Pll .!.p[l - Pll
2
G
1 a I P
2 ()a(Pll} p m 2 aa p
P. A
a
(2.1)
(2.2)
p
A
Define compliance C of body
c U/P
where c depends only on geometry, E and v
P dC .. G
da
Next consider a cracked body loaded
under compliant conditions (see figure)
in Mode I. Let CM be the compliance
of a spring (eg. testing machine) in
series with the body. Let AT be the
total displacement which will be
regarded as prescribed.
Now the potential energy is
PE
1 1 2 l -1 2
2
C- 6 + } CM (AT - 6)
and the energy release rate is
7
( 2. 3)
(2. 4)
( 2. 5)
8
G
(2. 6)
Thus the energy release rate does not depend on the nature of
the loading system. Note that eM ~ oo corresponds to dead load,
i.e. the previous case, and eM ~ 0 corresponds to prescribed
displacement, i.e. a rigid loading system. Eq. (2.6) is
frequently used in the experimental determination of G . The
compliances of two specimens, identical except for a small
difference in crack length, are determined experimentally and
in this way de/da can be estimated.
The relation between G and K in Mode I is now deriv-
ed. Let the body be subject to prescribed displacements (it
has just been shown that G is independent of nature of load-
ing) . We will calculate the energy released during a small
advance of the crack lla . As depicted in the figure, before
the crack is advanced the stress acting across the plane
y = 0 is o
22
<x,O) and for lla small (we will later let
t:.a ~ 0)
o
22
(x,O) 0 < x < lla (2. 7)
where K(a) denotes the stress intensity factor under the
prescribed displacement at crack length a . The energy
release during the extension, Glla , is the negative of the
work done by the traction o
22
<x,O) and this is
Glla
where V{x,O+)- V{x,O-) is the separation of the crack faces
1n the final position when the crack is at a+lla. From (1.2)
(dnd any of the standard references) for lla small
+ - (1 + K)
1
lla- X
V(x,O )-V(x,O) = K(a+lla)--G--J2'if
where K (a+ lla) is the final value of K .
IY
I
~ ~ ~
I
I x
IY
I
I
~
I
9
""'"'"' ~ -
~ X
vlx.o1- vlx.0-1
{after)
Thus,
Glla 14:GKK(a) K(a+ lla) J ~ a jlla;_ xdx
1 + K
!lG"" K (a) K (a+ lla) lla
and therefore G = ( 1 + K) K
2
/8G Using the expressions for
one finds
K
G
1- ,_,2 2
--E-K (plane strain) {2 .8)
(plane stress) (2. 9)
In a 3-D body with a crack edge subject to KI , KII and
KIII , the energy release rate per unit length of crack edge
is
10
3. ENERGY METHODS FOR ESTIMATING G AND K
References: Tada, et al. 1973.
J.R. Rice in Fracture, vol. 2, 1971.
As a first example consider a double cantilever beam specimen
(DCB).
P. !J./2
a
P. 6/2
Treat each arm of the specimen as a cantilever beam of length
a so that
Using (2.6) and (2.9) for plane stress gives
and ( 3 .1)
As a second example consider a semi-infinite crack in an
infinite strip held in rigid grips and subject to a prescribed
separation b
/
The strip is clamped in the unloaded state so that
xx 0 as lxl . Let the crack extend by an amount da
The stress and strain fields are unchanged if one the
origin in the positive x-direction by an amount da . Denote
the strain energy density for x + by (SED) and note
that the strain energy density vanishes for x - oo The
strain energy per unit thickness released is (SED)
00
2bda .
Thus, since dPE = dSE ,
11
G 2b (SED) ( 3. 2)
For example for plane stress as x "' ,
XX
and
0 '
yy
(SED)"' .!.(o + o )
2 XX XX yy yy
From (3.2) and (2.9),
and K
1 El>
0
XX
\!0
yy
( 3. 3)
These are exact results. Note that the general expression (3.2)
applies for a nonlinear elastic solid as well.
As a final example consider stress relief due to many
cracks as shown in the figure. Assuming b/a << 1 and that
the cracks extend together, show that for plane stress
where
that
G
Eb(;y)
2
2 (1- v
2
)
and K (3.4)
"'
= 6/w is the strain far ahead of the cracks. Note
G is the energy release rate per crack tip and that it
is independent of crack length.
12
4. THE CONCEPT OF SMALL SCALE YIELDING
Linear elastic fracture mechanics is based on the concept of
small scale yielding (s.s.y.). In words, s.s.y. holds when the
plastic zone at the crack tip is sufficiently small (compared
to the crack length and other relevant geometric length quanti-
ties) such that the elastic singularity fields of Section 1
still give a good approximation to the actual fields in an
annular region surrounding the tip. This is an asymptotic
condition which is increasingly violated as the load increases.
The condition is depicted in Figure (4.1).
plastic zone
I
/
I
/
I
...... "
\
\
\
I
\, \, ___ / 1)
region 1n which K- field holds1
Fig. (4.1).
13
Under conditions of s.s.y. the stress intensity factor K
(for a given mode) provides a unique measure of the intensity
of the strain fields at the tip, independent of other aspects
of geometry and details of loading. Considerable effort, both
theoretical and experimental, has gone into the delineation of
s.s.y., and this will be one side product of our later study
of large scale yielding. Roughly speaking, in most instances
s.s.y. appears to be a reasonable assumption as long as the
applied load is below about one half of the load at which full
plastic yielding occurs (i.e., the limit load for an elastic-
perfectly plastic solid) .
Later we will be investigating the behavior in the
plastic zone under various conditions. Here we will record by
way of the accompanying figure the simplest approximate shape
and size of the plastic zone in s.s.y. for three Mode I condi-
tions. In plane strain it is essential that
t is the thickness, and in plane stress
plane strain plane stress
!diffuse)
Fig. (4.2).
r
p
rp << t , where
>> t .
plone stress
I Dugdale)
With a
0
denoting the tensile yield stress the results
commonly used (see Tada, et aZ. pg. 1.17) are
14
r (l/311) (K/a )
2
plane strain (4 .1)
p 0
r ( 1/11) (K/a )
2
plane stress (diffuse) (4. 2)
lJ
0
r (11/8) (K/a
0
)
2
Dugdale model (4. 3)
p
Thus if L is the smallest relevant length (eg. crack length
or uncracked ligament), s.s.y. requires rp << L
5. INITIATION OF CRACK GROWTH IN S.S.Y.
References: J.F. Knott, Fundamentals of Fracture Mechanics,
chp. 5, Butterworths, (1973).
D. Broek, Elementary Engineering Fracture
Mechanics, chp. 6, Noordhoff, (1974).
J. Carlsson, Brottmekanik, Stockholm, (1976).
Griffith's energy criterion
Consider a "perfectly brittle" material which requires a
surface energy T (energy per unit area) for the creation of
new surface. A necessary condition for crack propagation (in
Mode I, for example) is G 2T Griffith proposed that
fracture would occur when G attains the value Gc = 2T
Thus the criterion for fracture initiation (and for continued
quasi-static propagation for a perfectly brittle material) can
be written as
G = G
c
(5 .1)
where G is regarded as being applied and therefore a func-
tion of geometry and load. On the other hand, Gc is regarded
as a material parameter. Using either (2.8) or (2.9) we can
rewrite this as
where Kc
K K
c
(5. 2)
. .. I
O<l.c fLOW ncatT
-
. ' l
....
-
[IP(biUTAI.
T
llj)OO
,..
,...-
-
/
SlAAl 01 we< (Xl[IISI3tl
L
0
t
'
"
.. ..
A
Odi1,.AC.MU,T (-I
Fig. (13.1). J-integral vs.
load-line deflection for
specimen T52 .
!
, ... ,
..
..
E"
i 10
I
I
I
I
I
--:
..
...
...
0-1
.. ..
.llq Ql.(t. (Jlllii$H'Iil 11
WI
*
!
Fig. (13.2). Load-line displace-
ment vs. crack extension for 4T
compact specimen T52.
65
66
.
.
..
Fig. (13.3). J-resistance curves
for A533B material 2 tested, at
93c- 4T side-grooved compact
specimens.
..
0
>u o
,.,, 0
!E :
...
"""' . .....
Fig. (13.4). COD-resistance for
A533B material 2, tested at
93oc T4 side-grooved compact
specimens.
... ,
p
. ..
10 "
f:. t.OAO LU;I( DlSfUC(W{:tTII!O!
Fig. (13.5). Applied load vs.
load-line displacement for 4T
compact specimen T52, 25% side-
grooved, (W- a ) : 86 mm
(3.385 in.).
0
67
68
1.#--
.. 1..., ,
...
0 ... -- 4 ...
IMU-11 lr'
"
.. ....
0
Fig. ( 13.6) . J vs. crack-growth
plots for C-shaped specimens.
Fig. (13.7
[EJ - v2]
spetlmen s
Kic and
for various
zes.
""
r
l
j
T
I
t
i
I
i
I
t
I
1
69
behavior. This region is on the order of the crack tip opening
displacement, i.e. on the order of J/a
0
As discussed by
Paris (1977), the requirement for bend-type specimens, which
appears to be adequate, is that the thickness and uncracked
ligament c both satisfy
c > 25J/a
0
( 13. 3)
For center-cracked specimens in tension (13.3) is not adequate,
as will be discussed in the next section.
Other studies along these lines can be found in the
literature. The general conclusion is that the J-method does
give fracture toughness measurements which are in agreement
with those obtained from valid KIC tests. Most of the
studies have used bend-type specimens, compact tension
specimens being considered in this class. Begley and Landes
(see, for example, Int. J. of Fracture, 12, pg. 764, 1976)
used center-cracked tension specimens as well as bend-type
specimens. Their measured values of JIC using the extrapola-
tion method just described gave similar results for the vari-
ous specimen types. On the other hand the slope of the
resistance curve (J vs. differed by almost a factor
of three. Furthermore, the crack in the tension specimens
advanced at 45 to the initial plane of the crack, while the
crack advanced in the plane in the bend-type specimens. Other
tests have found that JIC and measurements using center-
cracked specimens in tension are as much as two times the
values obtained from bend-type specimens even when (13.3) is
satisfied. The reason for this is taken up in the next section.
70
14. CONFIGURATION DEPENDENCE IN L.S.Y. AND LIMITATIONS
OF SINGLE PARAMETER CRACK TIP CHARACTERIZATIONS
References: R.M. McMeeking, J. Mech. Phys. Solids, 357,
19 77.
R.M. McMeeking and D.M. Parks, "On Criteria for
J-Dominance of Crack Tip Fields in L.S.Y.", to be
published in ASTM-STP.
C.F. Shih and M.D. German, "Requirements for a One
Parameter Characterization of Crack Tip Fields by
t.he HRR Singularity" G.E. Technical Report, Oct.
1978 (to be published).
On pages 47 and 4& of the notes we have already mentioned that
the idea of a dominant singulali ty uniquely tied to J or 6
tor all configurations involves the assumption of strain
hardening. With strainhardening, it is also essential that
the size of the region dominated by the singular fields, as
determined from the small strain plasticity theory, be large
compared to the region of large strains governed by finite
strain plasticity. In relatively ductile metals the fracture
process zone is comparable in size to this zone of "large"
strains.
line fields for three rigid-perfectly plastic plane
strain configurations are shown in Figs. ( 14 .1)- ( 14. 3) These
ctnd others can be found in the article by McClinock in Fracture
vol. 3, 1971. (Also displayed there are pictures of actual
straining patterns in steel specimens obtained by etching)
The near-tip stress field of the edge-cracked strip in tension
in Fig. ( 14 .1) is the same found for the limit of the
fields as n..., which is shown in Fig. (10.5).
Tlns Prandtl field is characterized by the high triaxial
stress, akk , ahead of the crack (10.8) as well as a high
non:1al stress ahead of the crack. Small scale yielding finite
element solutions in plane strain for elastic-perfectly plastic
give strong support to the attainment of the Prandtl
field at the tip of the crack (e.g., D.M. Tracey, J. Engr. Mat.
t
Fig. (14.1). Fig. (14.2).
,/'
Fig. (14.3)
71
Tech., 98, pg. 146, 1976). For the bend specimen in Fig. (14.2),
the singular strains are concentrated along a lens shaped arc.
According to the rigid-perfectly plastic solution, the triaxial
and norrr-al stresses ahead of the crack do not quite attain the
high levels found in s.s.y. or in the edge-cracked strip. How-
ever, numerical solutions reported below for elastic-perfectly
plastic materials, as well as for strain hardening materials,
indicate very little difference near the crack tip between this
case and s.s.y. or the edge-cracked strip at corresponding
levels of J .
The near-tip behavior for the fully yielded, rigid-
perfectly plastic center-cracked strip in Fig. ( 14. 3) is
entirely different from the other cases. Here intense shear
deformation is confined to slip planes eminating at 45 from
the tensile direction. A state of plane strain tension exists
ahead of the crack in the 90 wedge where the normal and
triaxial stresses are given by
(14.1)
Thus the stress levels near the crack tip in the rigid-plastic
fully yielded center-cracked strip are much below the levels in
the other cases. At the same value of J , the strain levels
will be correspondingly higher on planes of approximately
i 45 to the tensile axis eminating from the tip. Thus, the
rigid-perfectly plastic solutions for fully yielded specimens
72
1ndicate that the crack tip fields are configuration-dependent
and cannot be characterized by a single parameter such as J
or 6 . Strain hardening can substantially alter the rigid-
perfectly plastic fields. We now discuss some very recent
work, most of which is not yet published, which is directed to
the conditions for J-dominance of crack tip behavi-
or.
Work of McMeeking (1977) and McMeeking and Parks (1978)
based on finite strain, finite element methods has shown that
finite strain effects are important over distances of about 2
or 3 times crack opening displacement. In plane strain s.s.y.
the crack tip opening displacement 6t is given by
where the coefficient d
fairly strong function of
(14.2)
is a weak function of o
0
/E and a
n varying from about .6 for very
large n to about .3 for n = 3 . (Detailed results for d
will be given later). Beyond 2 or 3 times 6t the stress and
strain fields essentially coincide with the results from a
small strain formulation. McMeeking also found that J was
essentially path-independent for all contours which fell out-
side a radius of 36t (a finite strain, J
2
flow theory was
used in the calculations) .
McMeeking and Parks (1978) calculated the stresses and
near the crack tip in an edge-cracked bend specimen
and in a center-cracked strip in tension from s.s.y. to fully
plastic yielding. These they compared with the s.s.y.
distributions normalized in a proper way. If the fields are
J dominated, plots of stresses and strains near the tip
against r/(J/o
0
J should be independent of J . This was
found to be the case for the bend specimen even with no strain
hardening. On the other hand, they found significant deviation
from the normalized s.s.y. distributions already in the inter-
mediate yielding range for the center-cracked strip. Their
results for the center-cracked strip are sketched roughly in
Fig. (14.4). Here, for a material with n = 10 and
..
3
2
/ s.s 'j.
--
- .......... ":::- / ':>3'7
....... "' .............. -....... /
- ...... ...... 286
Yl
I
j-
2
'......,
..... _,77
3
..... --- 61. 60 c
J
s
Fig. (14.4)
73
o
0
/E = l/300 , the normal stress ahead of the crack is plotted
against x/(J/o
0
l . The solid line is the s.s.y. result. The
dashed lines correspond to various levels of intennediate
scale yielding as determined by J/(o
0
c) where c is the
uncracked ligament. The lowest dashed curve for o
0
c/J = 64
corresponds to fully yielded conditions. Similar plots for
strains on 45 planes indicate an intensification above the
normalized s.s.y. results which becomes expecially pronounced
for
that
sion
co
0
/J < 200 . McMeeking and Parks tenatively proposed
the size requirement for the center-cracked strip in ten-
(plane strain) should be
c 200 J/o
0
( 14. 3)
assuming the material does not have high strain hardening.
They find that the original proposed (13.3) for bend-type
configurations is adequate for essentially any level of strain
hardening.
Shih and German (1978) calculated stresses and strains
near the crack tip using a finite element method based on a
small strain formulation of J
2
flow theory for a deeply-
cracked bend specimen (CBB) and a center-cracked strip in ten-
sion (CCP), both in plane strain. Employing a normalized plot
similar to Fig. (14.4), they compared the calculated stress
and strain distributions directly with the HRR singularity
74
che Shlh and German paper. Shown is the normal stress ahead
of the crack oyy at varlous load levels (eg. values of
rdnging from essentially s.s.y. to fully plastic
ileld1ng. For both n = 3 and n = 10 , the stress distribu-
tion (and strain) of the bend specimen is well approximated
by the HRR singularity field for distances which are always
yreater than 2 or 3 times J/o
0
, i.e. always greater than
about 4 to 6 times the COD, as long as (13.3) is satisfied.
Th1s is not the case for the center-cracked strip (CCP) . For
the low s train hardening case, n = 10 , the results of Fig.
(14.6) are conslstent with the size restriction (14.3). For
!1igh strain hardening (n = 3) (14.3) can be relaxed some-
Whcit.
More work is needed to tie down the circumstances under
wtnch a single parameter such as J or 6 can be used to
characterlze the crack tip fields. What is clear however is
that the restrictions are a strong function of the configura-
t1on-type when large scale yielding occurs. The condition
(14.3) for center-cracked configurations may effectively mean
that the analysis of initiation and growth in this configura-
tion cannot be related to s.s.y. or bend-type data for inter-
mechate strength, high toughness metals. For these materials
200 J/o
0
can be on the order of 100 mm at initiation, and
this would limit testing and application to very large
spec1mens and cracked bodies. On the other hand, for the other
types of confrgurations it appears as if the weaker require-
ment (13.3) 1s adequate to ensure correspondence with s.s.y.
data. Furthermore, (13.3) does not seem to be strongly
dependent on the level of strain hardening.
Analogous studies in plane stress have not been made. It
rs possible that the configuration-dependence under fully
plastic conditions not be as strong as in plane strain,
primarily because the perfectly plastic limit solutions do
not display as wide a variation at least as far as the
stresses are concerned.
"'
i
..
I
'
6.0
'3
6.0 -
HRR fiELD
C"o/J600
41>
!J1.
.. ,
2.0
1.0
CBB
--
0
0 20 40 w 60 100 120 140
X/(J/"
0
)
7
i
I 3
6
\.
HRR FIELD
\
C<ro/J60
5
\
4
\
.. YJ '\:
.. 0
"
3
"
'-
2
X/(J/"o)
51
"o
7.0
A ' 3
6-0
IIRR fl(lD
C"o/ J 200
51>
4,
2.0-
w-
X/(J/"ol
12.0 a 3
10.0-
HHR fl ELD
C"o!J 30
8.0
2.0-
\.
\
\
" ' .
'-...._ _,..css
----
X/(J/"ol
75
Fig. (14.5). Normal stress ahead of crack in CBB and CCP from
obtained to fully plastic conditions with n = 3 , o
0
/E :
0.002 and a/b = 0.75 .
76
'.J,/
"'-
o8
0
0
7
.... 0 0
.... -cr
I
. "'
bo
c X ..,
I obo
"':::;
::::
><
2
2
0
0
0
..;
....
flbo
0
A. m
0
8
I
:!
C>
...1
"'
...
I
;:;; ..: ..,
2
.....
Ct:
bo
I
:.
"'
0 :It
2
I
oP
I
o .....
.......
' I ""
I
0
..
0
....
0
0 0
0
....
...;
.... CD
(,.) /CD
<.>/o
C> 0
-'
...,
... .
..: ..,
I
2
.....
"' .
""
I
.. X 0
I
I
I
I
.
I
I
. /
/
_....-"'
..
0
...;
-.9b"
f>
CD
10
u u
oo
I -'"'
... .
..:
2o: :::::
.. 0 . Ct:
I
"'
u
I
I
. I
I
.;
0 0
0
.,;
""
... ...;
-.9 ....
..,
,._
s
0
.. "
...
....
(<l
""'
..,_
{.) 0
If' uo
....
.. 2 'tl
'
.... 1::
>< olo
.....
Ill
Ill " {.)
N 1::
1::
....
.c
..>(.j.)
0..-<
0
Ill )
..
0 0 Ill
1::
""0
0..-1
.....
'tl..-<
ol'tl
C!) 1::
.co
Ill 0
Ill 0
11)..-1
::! C!).j.)
1-<11)
.j.)tGLI'I
en ....or--
0.
..... 0
tG>t
2
bo
O::t.Q
:z; .... ,
tG
....
K
0
"'
_....,
ID 'tl
't) 1::
... Q).,
..
....
"'"'
.. .j.)O
0'1::0
..-<O
0 c...uo
0
"'
15. RELATIONSHIP BETWEEN 6t AND J UNDER
J-DOMINATED CONDITIONS
References: D.M. Tracey, J. Eng. Mat. and Technology, 98,
1976' 146.
C.F. Shih, "Relationships between Crack
and Growth Parameters Based on the J-Integral and
the Crack Opening Displacement", Oct. 1978, G.E.
Tech. Information Series, to be published .
The separation between the
opened crack faces accord-
ing to the J-fields (10.6)
and (10.10) will be denot-
ed by o(r) . The relative
displacement of the faces
to the tip in the x-direc-
tion is ux
given by
These are
Fig, (15.1).
( 15 .1)
Only in the limit n oo is there a nonzero value of 6 as
r 0 . Rice (see Tracey, 1976) has suggested a definition for
ot to be used for both hardening and nonhardening materials.
As depicted in Fig. (15.1) ot is taken as the opening at the
intercepts of the two 45 lines drawn back from the tip of
the deformed profile, that is where
(15. 2)
78
Solvlny \15.2) and (15.1) for this value of 6 gives
(15. 3)
where
(15.4}
Shih (1978} has computed d for both plane stress and
plane strain. Plots are shown in Figs. (15.2} and (15.3) where
Lo is identified with o
0
/E . Note that a = 1 in these
plots. In both cases the influence of strain hardening is
...
0.4
0.2
.!!!.I
0.001
0.004
o.ooz
0.001
.
llfl/t
Fig. (15.2}. Variation of dn
with n and o
0
/E for plane
stress with a = 1
0.8
..
lit II
Fig. (15.3). variation of dn
with n and o
0
/E for plane
strain with a = 1 .
large. In plane stress the limit of the HRR singularity as
n oo appears to be 6t = Jjo
0
which is the same as the
Dugdale result (9.5), although there is no obvious reason
why these should be the same. In plane strain the limit as
n according to Shih's HRR singularity results from Fig.
(15.3) is 6t = .78 J/o
0
. There is considerable variation in
the published finite element results for the relation between
ot and J (or K} in s.s.y. for the elastic-perfectly
plastic case. Rice and Sorensen (J. Mech. Phys. Solids,
1978, pg. 163} give a comparison of results, not including
Sh1h's results presented here. Table (15.1) taken from Shih's
!
"
'
'
'
79
o
0
/E
!L:-1.
.001 .13 .27 .46 .78
Froa Eq. 5
.002 .17 .31 .50 .78
(HRR Singulu-ity)
.004 .21 .36 .53 . 78
Flow Theory .002 .18 .29 .48 .63-.66
Defor.ation Theory .002 .20 .32 .52 .70
'l'racey-Pu-ka [31) .001 .30 .47 .65
Sorensen [31) .001 .27 .47 .66
Hc:Heeking [31) .003 .27-. 30 .41-.44 .55-.67
Cracked Bend Bar .002 .19 .49 .58-.65
Fully Plastic
Center Cracked Panel .002 .22 .64 .82-.87
Fully Plastic
Table 15.1.
Comparison of Values of ot/ (J/oo} for Range of o
0
/E and n
for Plane Strain *
paper compares results for plane strain from various sources -
reference [31) in the Table is the Rice-Sorensen paper. As
expected, the finite element results for the fully plastic
center-cracked panel are not in line with the other results
because J-dominance was not acheived. For the other cases
there is reasonable consistency among the results, except for
the elastic-perfectly plastic limit where d ranges between
about .55 and .8 .
Further comparisons of the results from (15.3) and l.s.y.
finite element analyses can be found in the paper by Shih.
Some comparisons between the theoretical predictions and
measured 6t- J relations are also given in this paper and
reasonably good agreement is found. (The materials considered
have hardening exponents ranging from n = 5 to n m 13).
*
Unless otherwise indicated, Ot/ {J/a
0
) is determined from finite element
calculations for small scale plasticity. In Ref. [31). employ
J
2
flow theory. The fully plastic analyses for the cracked bend bar,
center cracked panel also employ J2 flow theory.
80
16. CRACK GROWTH AND STABILITY UNDER J-CONTROLLED CONDITIONS
References: Paris, Tada, Zahoor and Ernst, "A Treatment of the
Subject of Tearing Instability", U.S. N.R.C.
Report NUREG-0311, August 1977.
Hutchinson and Paris, "Stability Analysis of J-
Controlled Crack Growth", to be published in an
ASTM-STP, 1978.
In general, the J-integral cannot be used to analyse crack
growth in the l.s.y. range. (In the s.s.y. range J could
equally well have been used instead of K in the analysis of
Section 6}. crack growth involves elastic unloading and non-
proportional plastic loading which is not properly modelled
by a deformation theory of plasticity on which J is based.
However, under restricted circumstances discussed below, small
amounts of crack growth and its stability can be analysed. The
approach is based on experimental data from a J-resistance
curve and it closely parallels, and includes as a special case,
the s.s.y. analysis of Section 6.
Conditions for J-controlled Growth
We consider materials with a J-resistance curve,
such as that in Fig. (16.1) which involves increasing
l.ncreasl.ng a . For many
intermediate strength
metals increases of JR
several times the initia-
tion value Jc occurs
after only a millimeter
or two of crack growth
see, for example, Fig.
(13.3). As depicted in
Fiy. (16.1), let D be
the amount of growth
necessary to double J
above Jc calculated
blunting
D
Fig. ( 16 .1)
JR(lla) ,
J with
do
81
using the initial slope following initiation
D = J /(dJR/da)
c c
( 16 .1)
Note that in s.s.y. D is exactly the -quantl.ty
introduced on pg. 25.
We now consider
the conditions under
which the region of
elastic unloading and
non-proportional
plastic flow will
continue to be imbedded
in J-dominated fields.
The conditions sought
are those which will
guarantee the situation
depicted in Fig. (16.2).
With R denoting the
size of the region in
which the J-fields of
deformation theory have
elashc
unloading
/
I
\ \I I/
)<, //
/nearly propor!1onal load1ng
controlled by J- fields
Fig. (16.2).
dominance, one condition is obviously that
6a << R ( 16.2)
The second condition follows from the requirement that
nearly-proportional plastic loading occur within R but not,
of course, right up to the tip. Consider the strain field
(10.10} from the deformation theory solution, i.e.
n n
kn
J
n+l r- n+l
ij ij
(16. 3)
where kn is a dimensional constant. The increments in the
strains calculated from (16.3) for simultaneous increments in
J and a are
82
dl
lJ
(16.4)
~ _E_l cose - e
3
-
"ij : n+ E:ij + Sl.n ae E:ij
uilU wllere the change due to a l.S calculated as - da (a ( ) I ax)
s.lnce the crack tip is slufting ahead in the x-direction. The
f2rst term ln the bracket in (16.4) corresponds to exactly
L>roportlonal loading while the second, arising from da , is
ulstinctly nonproportional. Since ~ l J and i
1
j are of
comparaGle magnitude, nearly proportional loading will occur
wttere
da/r << dJ/J
Usiny the material-based length quantity D from (16.1) for
growth JUSt following initiation, this condition can be re-
wrltten as
D << r (16. 5)
The second condition is therefore
D << R (16.6)
since this insures that there exists an annular region, as
depicted in Fig. (16.2), D << r << R , in which nearly pro-
portional loading holds and in which the J-fields control, or
dorninat,;, the actual behavior.
For a fully yielded configuration R will be some frac-
t1on of the uncracked ligament (or other relevant characterist-
lC lenyth) c . (As discussed in Section 14, R may be very
small for certain configurations, eg. the centered-cracked
strlp in plane strain). The (16.6) can be written as
i
I
'
'
or
c
w - o
D << c
c d JR
--- >> 1
JR d a
83
( 16. 7)
(16. 8)
Only limited quantitative results are presently available for
a more precise specification of the two conditions (16.2) and
(16.8). Calculations by Shih, et al. (1978), shown in Figures
(13.1) - (13.5), were carried out for the compact tension
specimen of A533B steel. The calculations were carried out
using J
2
flow theory. Nevertheless, J was found to be
essentially
as Smm or
path independent for the crack advancing as much
6% of the uncracked ligament, corresponding to
increases in J above Jlc of more than a factor of 5 . In
addition, the calculated values of J agreed well with the
experimentally measured values of J using the deformation
theory formula (13.1). For their material and specimen w; 40.
Further work by Shih and Dean (to be published) indicates that
values of w as small as 10 may insure J- controlled growth
in bend-type configurations, although this is a tenative
result. Values for D in Table (16.1) introduced later range
from a small fraction of a millimeter to centimeters, but for
many of the intermediate strength metals D is on the order
of a millimeter or less and thus (16.8) may often be met.
Stability of J-controlled Growth
Consider the general specimen under compliant loading in
Fig. (8.1) where 6T is taken to be prescribed. Given the
resistance curve of Fig. (16.1) the condition for continued
crack growth, with 6a having already occured, is
( 16.9)
With (16.9) satisfied the condition for stability of crack
growth is
84
r aJl
(3aJ
6
T
dJR
< --
d a
(16 .10)
'-' t cz l. ( 19 77) introduced nondimensional quanti ties
(16 .11)
and called TR the tearing modulus of the material. Stability
is then insured if
(16 .12)
and instability sets in when equality is acheived. Values of
TR (T in the Table) and D are given in Table (16.1) for a
number of materials based on values of d JI!da just following
initiations. This Table is taken from Appendix II of Paris,
et aL. 1977. Note that TR ranges from .1 to over 200
(These values should only be regarded as preliminary estimates).
For later purposes, we now derive a general expression
(3J/3a)A to (3J/a)P Referring to Fig. (8.1),
T
relating
let
AT ., A + eM P Regard A and J as functions of a
and
P Thus, with AT fixed,
Then from
and eliminating
(<lJ1
taaj
6
T
dP we obtain
- raJ) r c + (2E.) J-
1
3P a 3a pl M 3P a
This generalizes (6.8).
(16 .13)
MATERIAl
ASTMA469 (7]
(6. E.)
rotor steel
(tMoV)
ASTM-A470 [1]
rotor steel
(CrMOY)
ASTM-A471 [2]
rotor steal
(Nt-Cr-MoY)
AJSI-403 [3]
( 12Cr-
sutnhss)
rotor steel
ASTM-A217 [4]
(2J
1
CrlMo)
ca t stel
ASTM-A453 (2]
(A216)
(Dlsea1oy)
stotnhu stu1
ASTM-A453 14)
stinless stet
(tu tontoton
arc welds)
ASTM-AS40 [Z)
(AISI 4340)
steel
lnco l.f.A.
( expan-
sion 11Joy)
sol'n treted
41nd ged
[4]
1. c. t. 1 s
100"
125"
1so
175
zso
3oo
1
11
-c.t. Joo
soo
soo
1"-c.t. so
1so
zso
2so
3oo
soo
aoo
1"-c. t. 75
zoo
Joo
soo
8oo
1"-c.t. -so
7s
3oo
soo
8oo
1"-c.t. -452
]5
4oo
soo
1"-c.t. -uz
c.t. -1oo
c. t.
so
Js
2so
3oo
sso
-4sz
1s
95.9 (760)
95.3 1260
94. 5
93.7 1340
93.0 1050
(88.0st.)
80.0 720
90.0 470
90.0 413
85.9 493
146.1 860
134.0 594
128.5 670
138.0 575
12S.O 500
1\S.O JZS
104.0 JZS
109.0
103.0
96. 3
94.5
86.2
79.5
71. 9
67.0
64.5
62.8
167.0
135.0
119.0
112.0
572
425
460
417
372
905
860
666
433
333
815
692
600
517
107.8 1666. 1
(IS3.5est.) 600
146.4 800
145.5 6S2
143.0 730
142.5 720
133.0 725
187.2
137.2
194
263
3. a x 1 o
3.0 110'
3.8 liO'
3.6 110'
3. 4 x 1 o
2.8 110'
6.9 XIO'
7. 0 X 10
1
1. 2 XI O'
1. 45X 1 O'
3.25110
3
6.54l10'
2.03XIO'
1.47X10'
9.SijX10
3
8.75110
1
7. 1 X 10 I
7.9 X10
1
7.3 l10'
5.83X10
3
4.08110
3
2.5 110'
1.89X10'
3.3)110'
2.40110'
Z.JJX10'
J.9H10'
2. 0 X 1 O'
1.2 110'
9.3 C10
3
3.54110"
3.7 X10'
2.1SX10'
2.10110'
1.75X10'
1. 0 X 10'
1.45110'
SJX I O'
. on 1 o'
()
123.0
101.0
130.0
124.0
131.0
131.0
25.5
25.8
48.6
20.4
5.4
11. 9
31. 9
27.9
2\.6
24. I
17. 9
22.3
23.6
19. 6
16. 5
119.0
109.0
221.0
172.0
176.0
41. 5
32.8
25.6
22.4
91.4
47.1
30. 1
29.8
25.7
14.8
24.6
2. H
1.71
D
0 3
.OJ
0 3
025
.07
.06
.04
.Ob
. 18
.02
.05
.05
0 2
0 7
.oa
85
86
((HttJI 1 JGN*
t,_J0 _ li_l F_:_
S i
1
!: C I Mf. N
u +o 1
dJ
Td_,[ E
r Y"[ ll.HL
o
di" ;;z
-
tVt.l- 16 5 l t 3] 1"-c t.
Alumlnum
_. 1 o-... l4 J c.t.
stdlnltss steel
{<..hl('lded metal
arc elds)
I-eunt 1 [ 5 J VIM-VAR/ST
050 00-/ST
1-J-/STOA
AAH-VAR/STDA
VI11-/STOA
HIP--
HJP-/STOA
A1S!-3105
[6] STQ
sta1nless
sens.1t1zcd
'i teel
( 6] ln ( L- t)
'ittlinless CI<(T-l)
l s TQ ( r -L)
[ 4] G TAW
CW/GTAW
GTAW/CN
GTAW/CW/AN
GTAW/CW/AN..
lcon1(4)VJH-EFR/STDA
706 V!HVAR/STDA
Pyroruet (4) GATW
538
(21Cr-6N1
;l"'nj
stalnle;s
s 1
c.t.
c t.
c.t.
c.L
t:.L
"thlck
ct.
no
75
-452'
-452'
-4 52
-452'
-4 sz
-452.
-452.
-4S2"
-452'
-452'
-452"
-4 ';.2
-452
-45 z
7';, 0
-452'
-452.
oenotes follow1ng mater1al condltions
inductlOfl melted
followed by Vdtuum arc remelt.
AAM-VAH-Air arc me1ted by
YdCUJ:i -'r': r(;Pi!lt
- :;elttd.
arc
Shtelded metal .lrc welds.
4 3. 25
152.
120.4
127.0
139.0
184.0
213. 7
1 72. 6
185. 3
I >4. 0
152 0
227. 3
216.0
175. 0
181.4
199.9
214.2
199.0
I 75. 7
211 . 5
207 0
166 0
209.1
1 37. 3
98.4 4.92X10' 2.79
336.> 1.03X10' 13.27
320 4.64X10' 10.2.
240 1.72X10' 3.6
138 0. 9
135 5.09X10' 50.4
500 2.06X10' 15. 1
465 4.57X10' 51.4
233 1. 89x 1 o 18.4
!bOO 1.2 X10' 0.15
675 7. 5 X10' 97. 1
452 3.0 X10' 1.74
388 1.0 X10' 6. 43
1250 2.0 XI0
3
I. 96
615. 9.3 X10' 8.48
725 9.17X10
3
6. 89
216.7 2.71X10
3
1.77
429 1.9SXIO' 14.8
812 3.62Xl0' 35.19
530 1. 21X1 O' 8.13.
660 1. 94X10' 13.6
357 1.88X10' 20.45
166 5. 19X10
2 0. 35 .
862.3 3.27XIO' 51.69
Denotes heit
ST - Solution tre4t.
STDA - Solution treat and double
a9ed.
CW - Cold worked. JC% reduction
1n th1C: t!!>S.
STQ - 2000f-1 hour- water quencl
....Q_
2
.03
.07
13.3
.15
.62
.04
32
Next we will present
some specific results for
the deeply-cracked strip in
three-point bending shown
in Fig. (16.3). An analysis,
essentially identical to
that in Section 12, gives
b
J
cr Pd b
0
cr
J =
c
(16.14)
where t. = t.nc + t.cr This
result is not strictly cor-
rect under increasing crack
length. As shown in Hutchinson and
expression for J under increasing
b
87
cl
Fig. {16.3).
Paris (1978) the correct
a is
fo
cr P
r
J .. 2 -db (16 .15)
c cr
ao
where
ao
is the initial crack length. For small amounts of
growth with (/.) >> 1 , the difference between ( 16 .15) and
(16 .14) will usually be small.
Carrying out an analysis similar to that in Section 12
by making use of a nondimensional function similar to {12.2)
for deeply-cracked strips, one can show
(
aJ) = 4P
2
[
06
c) _
aa P c2 aP a c
Using the above expression in (16.13) and after some rearrange-
ment, one can show that (see Hutchinson and Paris)
(16 .16)
where C .. CK + Cnc is the combined compliance. Up to and
including initiation, all quantities on the right hand side of
(16.16) can be measured directly from a load-displacement
record; C is assumed known. The quantity {3P/<lbcr) is
a
88
at initiation but cannot be measured directly once
grow takes place. In (16.10), (3J/3a)ll from (16.16) provides
a check on stability at initiation whicfi can be assessed
directly from the experimental record.
For a fully yielded strip of a nonhardening material P
the limit load and (3P/3ll ) = 0 . Then (16.16) reduces
cr a
to
(16 .17)
The strong role of the combined compliance of the system on
stability is obvious from this formula. The limit load for the
strip of Fig. (16.3) is P = 4Ao
0
c
2
/L where A a .364 in
plane strain and A = .268 in plane stress. Regarding the
uncracked strip as a beam gives C = L
3
/(4Eb
3
) . Then, from
nc
its definition in (16.11),
(16.18)
Paris, at al. (1977) have tested a number of specimens
under three-point bending in series with a beam of variable
compliance as depicted in Fig. (16.4). The compliance of the
test machine is negligible compared to the combined compliance
of the specimen and the beam so the relative displacement of
the machine heads LIT is prescribed. Using (16.16) and data
directly from the experimental load-deflection record, the
value of T (T ) at initiation ranged from about zero
applied
when the beam was absent to a value of just under 50 for the
longest beam. (Note that the notation for b and L in Fig.
( 16.4) is different from that in Fig. ( 16. 3)) . The value of
TR (Tmat) just following initiation was 36 and the w
parameter in (16.8) is about 15 . For each specimen the point
(T, TR) is plotted in the Figure and it is shown as a solid
point if fracture initiation was observed to be stable and as
and open point if unstable. The solid line T = TR is the
theoretical dividing line for stability and instability from
( 16 .12) .
e :s. LtQUIVOlcnt 5 76
11
ASTI.I- 471 ROTOR STEEL
(N C Mo V)
Tmot' '36
0
... .b. <IJmot. 15
Jlc (10
2 (Topplied)
01
,,.tiolioo S. 45
10 I /1
AT t.r
... -- -.;- ..
Fig. (16.4).
89
The significance of the large TR values for inter-
mediate strength metals (TR 30 , say) is that in many situa-
tions the T-value will be considerably smaller than TR so
that small amounts of crack growth can be tolerated. For
example, the s.s.y. result for the finite crack of length 2a
in an infinite body under tensile stress
J =
and
00
o has
Thus, T for this problem in s.s.y. will neyer exceed about
l . Next consider a finite crack of length 2a in an infinite
body under fully plastic conditions. From fully plastic solu-
tions such as those in Section 12, where in plane strain ten-
sion c/Ey = (o/oy)n ,
n+l
oo n+ l '"' n
J "" oyy a h (n) (o /cry) = oyy a h (n) (c /cy)
where h 3 for n = 10 - see Hutchinson, et al. (pg. 59).
90
Under these conditions
T
n+l
h(n) ~ / ) n
y
and T wLll only reach values as large as 30 , say, for
~ 10 c Of course, when a crack extends across a
y
signLficant fraction of a cross-section and the loading condi-
tions approach dead loading, then T will be large as il-
lustrated by the analysis of the 3-point bend bar.
'
91
17. SOURCE OF STABLE CRACK GROWTH
References: F.A. McClintock and G.R. Irwin, ASTM STP 381, 1965,
pg. 84.
A.D. Chitaley and F.A. McClintock, JMPS, 19, 1971,
pg. 147.
J.R. Rice, in Fracture, Volume II, 1968, pg. 277.
J.R. Rice and E.P. Sorensen, JMPS, 26, 1978, pg .
16 3.
In this section the problem of a crack growing quasi-statical-
ly in an elastic-perfectly plastic medium is considered. We
start by contrasting the steady-state problem in Mode III with
the stationary problem, both in s.s.y. Then the transient
problem is treated with emphasis on the tearing resistance of
the material immediately following initiation. Lastly some
recent work on the s.s.y., plane strain problem will be dis-
cussed.
Steady-state crack growth in s.s.y. Mode III
Yl
I
I
actiw plastic zone
X
~ e /lr
wake of residual --1 ,;.. a __
plastic strans
Fig. (17 .1).
We consider the steady-state situation depicted in Fig.
(17.1) where the crack has grown under s.s.y. conditions such
that to an observer traveling with the crack tip the stress
and strain fields no longer change. Far from the tip (except
tn<= wak<-1 the strt;sst:s are given by the elastic singularity
tields,
The material is asswned to be elastic-perfectly plastic with
the y1eld condition
The analysis given below is due to McClintock and Irwin (1965).
The present version is largely taken from the fracture article
by Rice (1968), us1ng results from the more recent paper by
Chitaley and McClintock (1971), but contact will be made with
the results of the last few sections.
Let denote the material derivative with respect to
the increase in crack length, i.e. a( )/aa If f(x, y)
denotes the instantaneous spatial variation of any quantity in
steady-state, then
f = - at;ax (17.1)
In the active plastic zone ahead of the tip equilibrium and
yielding, plus a focussing of the strain-rate, requires the
same fan of slip lines as in the stationary problem (see Fig.
(11.1)) with
(17.2)
1\'Hh Y
as the strain-
rdte vector which in steady-state is given by
y ( 17. 3)
the constitutive law is
..
..
93
e p
y + r
T/G + :\(r, 8)-r i;c + >. ,
0
( 17. 4)
Now,
1 (17. 5)
Using (17.3) and (17.5) in (17.4) and equating components, we
find (with Yo = 1
0
/G
a [aw) sinS
<lr ax = Yo r
1 a (awl
r ae lax) = - ).
1
0
(17.6)
The general solution to the first of (17.6) is
Yx = aw;ax = Yo sine tn r + F(6) ( 17.7)
To determine F(6) note that on the elastic-plastic boundary
!: 1/G (since i:p = 0 ) , i.e. for r = f(6)
!: = Yo[-sin6 +
Thus on c yx - y
0
sine so that
F = -y
0
sin6(l+R.nf(6))
and therefore in the active zone
yx = -y
0
sine[l+ R.n(f(6)/r) J (17. 8)
Since a yy/ ax = a y x;ay , from ( 17. 8) we obtain for 8 = 0
ahead of the crack tip (<l( )/ax a <l( )/<lr and a( )jay:
r-
1
a< J/a8J
0
'
1 a Yx Yo r I
- - - - 1 + in f ....)J
r ae r l r
L
(17 .9)
where rp = f(8 0) Finally integrate (17.9) subject to
Ye = Yo at r = rp with the resulting strain distribution
directly ahead of the crack in the active plastic zone
94
(17.10)
This result can be contrasted with the corresponding
stra1n distribution directly ahead of the crack in the sta-
tionary problem from (11.14)
(17 .11)
where rp is the distance to the elastic-plastic boundary in
the stationary problem. Chitaley and McClintock (1971) have
carried out a detailed numerical analysis of the s.s.y.,
steady-state problem in Mode III. They find that to within
about 5%
(17.12)
and in the subsequent discussion we will take (17.12) to hold.
The singularity in the stationary problem is much
stronger than for the growing crack. At the same value of K
(or J} the strain near the tip (i.e., r/rp << 1) is con-
siderably larger in the stationary problem. In Fig. (17.2) a
plot is given of
(y e> s .s.
(y a> stat.
(17 .13)
The lower strain levels in the growing crack problem
result from the distinctly nonproportional plastic deformation
which occurs in the active plastic zone. An elastic-plastic
material has considerable resistance to nonproportional
straining as compared to a nonlinear elastic solid. This is
the source of the stable crack growth. For example, note from
(17.4) that the plastic part of the strain-rate is constrained
to have only an ~ E I component. For a true deformation theory
of plasticity there is no difference between the strains in
the stationary problem and the growing problem at the same
value of J .
..
..
95
ll'els.s .
1.0
tl'alstat.
.8
.6
.4
.2
0 r/rp
.01 .05 .1 .5
Fig. (17.2).
To further explore the implications of the steady-state
solution vs. the stationary solution, McClintock and Irwin
proposed the following "microscopic" critical strain condition
for initiation and continued growth in Mode III
at (17.14}
in the plastic zone ahead of the crack. (Here we use total
strain; McClintock and Irwin used plastic strain). For initia-
tion this implies from (17.11) and (17.12)
7f c
J c "' 2 r c Y c 1' 0 and r p
ar
c
where a = yc/y
0
. For steady-state growth the criter1on
implies, using (17.10),
(17.15)
Solving for r ~ r c in the above and using (17.12), one finds
(17.16)
96
1000
Js.s
T
100
10
Fig. (17.3).
Thus,
J
55
lexp[l2a-l- 1)
-:r;;a
( 17 .17)
A plot of this ratio is shown in Fig. (17.3). For
larger values of a , Jss a = 10 , Jss ; 3Jc For even
becomes very much larger than Jc . Note that for relatively
1 < a < 10 , Jss is not substan- brittle" materials where
tially above J
c
for
Initiation and subsequent growth in s.s.y. Mode III
We now consider the transient problem starting from the
crack prior to growth and analysing the subsequent growth. At
an appropriate point the fracture criterion (17.14) will again
be invoked. The distance to the elastic-plastic boundary
r ( e , a) is now a function of a and 6 . Again, let ( )
denote the material derivative with respect to a . Eq. (17.4)
still stands and is still given by (17.5) since
=
in the plastic zone. Since
..
97
Eq. (17.4) implies
- y
0
siner
-1
and (17.18)
The first of these equations can be differentiated to give
- y
0
cos6r
-1
which then integrates to
-1 -1
Y
6
= - y
0
cos e r R.n r + r F ( o)
With <Yale denoting the value of y
6
at the current location
of e , this can be rewritten as
-1
y
6
= y
0
cose r .tn(f/r) + (f/r) <Yale ( 17.19)
Directly ahead of the crack (6 = 0 , r = rp) this specializes
to
(17.20)
To calculate the last term note that on e 1
since X
. I.et oy
6
/Dt denote the derivative of
y
6
following the elastic-plastic boundary. Since (Yelr is
p
the material derivative of y
6
, ahead of the crack these two
derivatives are related by
Dye
-- .. - 0
Dt
where in the convective term d(a+rp)/da l+rp. Now,
;rf .. = Ye + ; aayer
so that ( Cly el Clr) .. - y olrp
(since y
6
Yo and Yr 0 on
98
Uslng thls ln (17.20) glves
( 17. 21)
or
(17.22)
'fhe ilrst term in (17.21) or (17.22) is exactly the same
as the steady-state contribution - see (17.9) - with no change
rp ; while the second term can be thought of as that due lD
to a stationary crack with increasing rp - see (17.11).
Again we invoke s.s.y. and assume (17.12) provides a
good approxlmation to
21, 1973, pg. 337 and
1978, pg. 485. Since
rp - see for example H. Andersen, JMPS,
E.P. Sorensen, Int. J. Fracture, 14,
drp = 2dJ/(rry
0
ol , (17.22) becomes
We now adopt the near-tip fracture criterion (17.14).
Consider initiation and the first increment of growth. Again
with u = ic/Yo ,
ar
c
(17.23)
(17.24)
Assume (17.23) and (17.24) pertain such that initiation is
possible. Regard the strain ahead of the crack as a function
of x , which is fixed to the material points and chosen to
cou1cide with r before growth and a , i.e. y e (x , a) . The
99
condition which is currently met is
increment of growth da we require
This is
Y
6
<rc, a) yc ; after an
y
6
( r c + da , a + da) = y c .
y
6
( r c + da , a + da)
or
where the last equality follows from (17.24).
Using (17.21) in the above with (17.23) we find
rp = a- l- in a (17.25)
or, from (17.12),
[ ~ ~ =
c
Tf
2 1:
0
y
0
(a- 1 - in a) {17.26)
This provides the tearing modulus at initiation
TR,._9_(dJ) .,.!(a-1-ina)
2[da 2
'o c
(17. 27)
A "perfectly brittle" material with a = 1 corresponds to
TR "' 0 ; while for large a , TR ~ va/2 . Similarly the
material-based length quantity, D , introduced in (16.1) is
given by
J
D .. .,..-:;-;.:;c-.-- = r a
(dJ/da)c c a- 1- ina
(17.28)
~ rc for a > 10
Thus for a material with a large tearing modulus (a > 10 ,
TR > 15) the distance D needed to double J above Jc is
essentially rc , which can be thought of as the size of the
fracture process zone in the present context.
The model suggests certain implications relating macro-
100
scopic fracture resistance to features of the fracture process
zone. In particular, note that the ratio, Jss/Jc , in (17.17)
and the nondimensional tearing modulus TR in (17.27) depend
only on a = rc!r
0
. Furthermore, for large a Jss/Jc in-
creases exponentially while TR increases linearly in a .
Note that for a = 60 , TR ~ 100 and Jss/Jc ~ 1000 . For
larger values of a the small strain assumptions will certain-
ly be violated for typical values of r
0
at the point where
r = rc . But the model does suggest the source of the large
values of TR which are observed experimentally in plane
strain. The very large values of Jss/Jc for large a result
from the considerable resistance an elastic-plastic material
otfers to nonproportional straining, as has already be noted.
This effect is undoubtedly overestimated by the simple smooth
yield surface of Mises (and Tresca in Mode III) used in the
present analysis. In this sense the values of Jss/Jc for
large a may be considerably in excess of observable values.
Starting from (17.22) it is possible to set up the
problem for calculating the entire resistance curve J R ~ a )
we have only determined its initial slope and its asymptote.
See McClintock and Irwin (1965) and Rice (1968) for an approxi-
mate determination of the whole curve.
Rice and Sorensen (1978) have considered the more diffi-
cult Mode I, plane strain problem in s.s.y. Qualitatively the
findings are similar to Mode III and several features of the
analysis are closely analogous. The authors assume that the
near-tip stress field is still the Prandtl slip line field
under growing conditions. This is not obvious in plane strain
but they present numerical results which suggest the plausibi-
lity of the assumption. While the criterion (17.14) is sensible
in Mode III, a critical condition cannot be taken to be met
ahead of the crack in plane strain Mode I since the strains
are nost intense above and below the tip in the small strain
solution. Instead, Rice and Sorensen used an alternative
criterion which is essentially an integration of the near-tip
strains. They require the crack opening displacement to reach
a critical value at some fixed small distance back behind the
tip. By making contact with numerical results they are able to
101
obtain an approximate integration of the equations relating
the crack opening displacement, the crack advance and J
Resistance curves are determined. Large tearing resistance is
found, typical of observed values, with realistic choices for
the near-tip fracture criterion.