Crack Propagation and Arresting
Crack Propagation and Arresting
Crack Propagation and Arresting
1971
<-..
SHIP STRUCTURE COMMITTEE
AN INTERAGENCY ADVISORY
COMMITTEE DEDICATED TO IMPROVING
THE STRUCTURE OF SHIPS
to the
by
under
Department of the Navy
Naval Ship Engineering Center
Contract No. NOO024-68-C-5073
The specimen and the plastic zones produced prior to crack ini-
tiation are first described and compared with analytical and experimental
results in the literature, for purposes of calibration. The mechanism of
cleavage crack propagation was then investigated. It was found that the
fast moving crack bypasses some of the grains as it grows, leaving behind
unbroken ligaments. Rupture of these ligaments consumes a large amount
of energy locally and this process can account for crack propagation re
sistance values estimated from these experiments.
ii
CONTENTS
PAGE
INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Section 1. PLASTIC ZONES IN Fe-3Si STEEL DOUBLE-CANTILEVER-BEAM sPECIMENS .
ABsmAcT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1. INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . 4
11. EXPERIMENTAL PROCEDURES . . . . . . . . . . . . . . . . . . . . . . 5
111. KESULTS AND DISCUSSION. . . . . . . . . . . . . . . . . . . . . . 7
Iv CONCLUSIONS. . . . . . . . . . . . . . . . . . . . . . . . . . . 12
v. REFERENCES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
ACKNOWLEDGMENTS, . . . . . . . . . . . , . . . . . . . . , . . . . . !s1
iii
SHIP STRUCTURE COMMITTEE
The SHIP STRUCTURE SUBCOMMITTEE acts for the Ship Structure Committee
on technical matters by providing technical coordination for the determination
of goals and objectives of the program, and by evaluating and interpreting the
results in terms of ship structural design, construction and operation.
iv
LIST OF ILLUSTRATIONS
INTRODUCTION
1 SCHEMATIC REPRESENTATION OF THE ENERGY CHANGES IN A
MRGE PLATE UNDER UNIFORM TENSION FOLLOWING THE
INITIATIoN OF A CRACK IN A BAD REGIoN Km ITS ARREST
BY THE BASE METAL SURROUNDING IT .. ., ., .. .. .. 2
Section 1
Section 2
1 STRESS-STRAIN CHARACTERISTICS OF THE Fe-3Si STEEL AT
OCCOMPARED WITH IOOoC .. .. .. .. .. .. .. .. 17
2 SCHEMATIC DRAWINGS OF THE PLASTIC ZONE PRODUCED BY:
(a) Stationary Crack and (b) Growing Crack .. .. .. .. 17
3 ARRANGEMENT USED IN WEDGE LOADING OF A DOUBLE-CANTI-
LEVER-BEAM (DCB) SPECIMEN .. .. .. .. .. .. .. .. 17
4 PLASTIC ZONE FOR SPECIMEN 3P-17 FOR SLIT CUT IN UNDER
LOAD CORRESPOi?U)INGTO =
K 20
0.6.&i. .. .. .. .. .. ..
ay
5 PLASTIC ZONE OF SPECIMEN 3P-18 FOR SLIT CUT IN UNDER
LOAD CORRESPONDING TO K
=0.7G. .. .. .. .. .. .. 21
Uv
6 PLASTIC DEFORMATION Associated WITH CLEAVAGE FRACTURE
ON THE SURFACE OF Fe-3Si SPECIMEN 3Q-16 .. .. .. ,. .. 22
7 PLASTIC DEFORMATION ASSOCIATED WITH CLEAVAGE FRACTURE
ON THE PLATE MIDSECTION OF Fe-3Si SPECIMEN 3q-16. 150X .. 23
v
LIST OF ILLUSTRATIONS (Cont.)
EE?E w
Section 3
1 THE DCB TEST SPECIMEN . - .- - - 28
Section 1
1 SUMMARY OF RESULTS .. .. .. .. .. .. .. .. .. 8
2 ONSET OF THE CRACK-TO-BEAM-ZONE T~SITION .. .. .. .. 12
Section 2
1 SUMMARY OF RESULTS .. .. .. .. .. .. .. .. .. 19
Section 3
1 COMPOSITION AND YIELD STRENGTH PROPERTIES .. .. .. .. 28
2 SUMMARY OF DCB TEST RESULTS .. .. .. .. .. .. .. 34
3 TABULATION OF K K=, AND Rd VALUES FOR A-517 AT -196C .. 147
Q
vi
INTRODUCTION
With the trend toward higher strength steels and more highly stressed ship
hulls, more precautions must be taken to guard against fracture. Existing methods
already make it possible to identify safe stress-level/flaw-size combinations, pro-
vided the flaw is imbedded in material of standard quality. However, this LS not
the only source of failure. A more likely source is a crack located in an atypical
region of low toughness (like the HAZ of a weld). Such a flaw will become unstable
and begin to propagate at even lower stress levels. The question then is: will the
crack emerging from the bad region be arrested when it reaches the good (standard
quality) material that surrounds it? The concept of designing for crack arrest is
not new. Pellini and his coworkers at NRL(l) have for some years advocated a crack
arrest philosophy. What is needed now is a more precise description of steels
arrest capabilities--something like Pellinis FAD (Fracture Analysis Diagram), but
in terms of the length of running crack that can be arrested at different stress
levels by the base metal and by continuous weld-affected regions.
In overall terms, the study encourages the view that the crack propagation
characteristics of steels can be measured, analyzed, and ultimately used to control
fracture arrest. This is illustrated in Figure 1 by an example, the case of a large
plate loaded in tension. The example presupposes that a crack in a bad region of
the plate becomes unstable, propagates a short distance, and then is arrested by base
metal characterized by a relatively large value of the average energy absorption rate,
Ed. Stress levels and the size of bad regions that can be tolerated can be esti-
mated in this way. Crack arrest may thus offer a more conservative approach to
fracture safety, both from rhe viewpoints of design, inspection, and repair. However,
both the experimental and analytical techniques need to be developed more extensively.
In addition, the metallurgical factors governing arrest toughness need to be under-
stood for the purposes of alloy development.
Reference
; (basemetal
)
Fig.1. SCHEMATIC REPRESENTATION OF THE
ENERGY CHANGES IN A LARGE PLATE
UNDER UNIFORM TENSION FOLLOWING
THE INITIATION OF A CRACK IN A
GA
BAD REGION AND ITS ARREST BY
THE BASE METAL SURROUNDING IT.
HERE U REPRESENTS THE ELASTIC
STRAIN ENERGY STORED IN THE
PLATE .
(bud
(31C regi
on
R(bedreglon
H
L /
-v
Un$toble
propagation
Crack
Length
[o)
-3-
SECTION 1
by
ABSTRACT
1. INTRODUCTION
* Also referred to as the wedge-opening loading (WOL), compact tension (CT), and
crack line loaded specimen.
** Wilson offers comparisons at a stress level corresponding to nominal stress-to-
yield stress ratio o/ay = 0;36. The differences among the zones should disappear
as G/oy + 0. \3)
-5-
yla
0.15 ~~
0
I
Present
Clark
P
measurements
A(~)
H/w=o.50/
0.06
/-
1
{
WW=O.35
\\
0.10
L A = 0.25; results
u forplates under
.5
-uniaxial tension, /}1 o
L
<
k]
Xlo
0.05 (K/+ I/@ a/W H/W
A=0,13 Calculation 0.45 0.5 0.25,0.50
Experiment 0.41 0.54 0.35
The results reported here are for DCB Specimens having the geometry shown
in Figure 3 (Series P), and for two specimens with the same length and thickness but
a beam height of 0.5 in. (Series A). These two configurations have beam height to
length rations, !! = 0.35 and 0.125, res.pectively, compared to the ratio~ = 0.45
employed by Clar~. Another d~fference ;sthatthe specimens used here!ontained
spark machined slits cut with 0.005 in .-diameter wire which produced a slit root
radius Y 0.005 in., rather than sharp cracks.*** While the deformation within a
root radius of the slit tip is therefore not characteristic of the zone of a crack,
portions of the zone at a distance greater than 2-3.root radii from the slit tip are
*** This was done to make the zones comparable to those produced in simulated crack
growth experiments (7,8) discussed in the next section Of the report.
-6-
r
).37
60
0.20
1.5C .
~o.o Io
1 u
3.80
I
I
J_
1
I I I
\
I I
4
0.90 L\
Thread 6-40
Depth 0.20 in.
o 2 4 6
Fig.3. DOUBLE-CANTILEVER-BEAM Fig.4. STRESS-STRAIN CHARACTERISTICS OF LOT
SPECIMEN P Fe-3Si STEEL AT 100C
.
probably indistinguishable from those of a sharp crack. The zone size to thickness(7)
ratios examined favor a close approach to plane strain conditions and this is discussed
more fully in the next section.
The DCB specimens were fabricated from Fe-3Si Steel, annealed, aged, polished,
and etched according to procedures described in detail in Reference 9. Series A speci-
mens were tested at room temperature and displayed a yield stress of 65,000 psi. Series
1? specimens were tested at 100C to facilitate comparisons with crack growth experi-
ments made at this temperature to avoid cleavage. Lot P displayed a yield stress of
53,500 psi at 100C and some evidence ofstrain aging, but not enough to significantly
affect the stress-strain curve (shown in Figure 4) during @e time the specimens were
under load. The spectmens were stressed.in a horizontal loading device and both the
load and the slit opening were monitored (the latter with a displacement gage). Loads
were applied and released relatively slowly, at about 5,000 Ibs. per minute. In one
case, a number of zones were produced in a single specimen by extending the slit after
each load-unload cycle and then reloading to a different K-1evel. Compliance values
calculated from theKanninen.(~~ e~uation agree reasonably well with the actual
measurements as shown in Figure 5. Plastic zones generated at various K-1evels were
revealed on the DCB surface and midsection by etching.
I -7-
c=~~~x3H3~A3d3
[ ( )
,.~k2a2sinhkc CashAc -1.sinAc cosAc ~
sinllz
i. - Sinzkc
(SeeTable1 torexplamticm
ofsymbsls)
(See Table 1, PageR, for explanation of
symbols)
Examples of the plastic zones observed on the surface and in the interior
on the specimen midsection are presented in Figures 6-9. The general appearance%of
the zones are in the main, similar tO Chose df.splayed by zones fn tensile loaded
plates.(4) Values ofp, the furthest extent of the zone measured normal to the plane
of the slit, are about 25% larger in the interior of the DCB specimens (on the plate
midsection) than on the surface, a feature also displayed by the tensile loaded
plates(41. Table 1 shows thar this is not a question of specimen alignment since a
given zone extends about the same amount on opposite surfaces of the same specimen.
Values of /3, the specimen thickness co zone size index:<,are in the range
1.25 ~ ~ ~ 0.32 for the current series of experiments. The P-values are for the most
part larger than B = 0.4, whtch is a suggested upper limit for achieving plane strain
conditions in tensile loaded plates.(2) However , since the p-values for the DCB are
about half the value for tensile loaded plates, the limiting value of B could also be
larger in this case. In fact, Wessels results for HP-9-4-25 and A-302B show no
significant departures from the plane strain toughness vlaue below ~ w 0.7 which
corresponds to the limit I< < 0.55 ~. for plane strain in 0.5 in.-thick specimens.
q
K
On this basis, the DCB zones corresponding to = 0.6 ~., and 0.7 G. still repre.
Dv
sent a reasonably close approximation of plane &xain conditions. The zone obtained
by loading to ~ = 0.4 ~. probably deviated more from that which would be expected
for a crack bec$use its size fs only twice the slit root radius.
2
b=:
H ~
~y Y ield stress.
, where t is the specimen thickness, K the stress intensity, and
-8-
T K=2fi
Lt:slz
[(
h a
sinh<c+ sin2).c+
s inh2A
c - sin2A
c)
( sirhkc Coshkc - sinkc coskc
sinh2kc - sin%
)1 a- length
t~ Series
A, Uy = 65.0ksi slit
H- (heightof am) = 1.25 in.
SeriesF,Uy = 53.5ksi t- (thickness)= 0.5in.
P- (load)
Ttt Apparent I- (momentof inertia) =~
c- (widthofweb)= 3.6- 3
forSeriesP
k- (6)lt~b/H
(a) Surface (b) Midsection (c) Midsection
,,,
k! ,, , ,.:,,
~= 0.38 %. 0.57 0.70
q
Fig.8. PLASTIC ZONES OF SPECIMEN 3P-23, A NUMBER OF ZONES WERE PRODUCED BY EXTEND-
ING THE SLIT OUT OF RANGE OF THE ZONE BEFORE RELOADING TO A DIFFERENT STRESS
INTENSITY LEVEL.
(a) (b)
As shown in Figure 1, the p-values for the DCB specimens are consistent
with Clarks findings and are about 112 as large at comparable K-levels as the
values derived from edge and center notched plates loaded in tension. In the range
0.4 En. <~< 0.7 ~., the stress intensity dependence ofp can be ex@ressed
approximatei$:
(1)
H
A=O.13 - Etched, Series-P DCB Specimens ~ = 0.35
The p-values for specimens 3P-23 ( K = 0.8 fi.), A-4 and A-3 deviate noticeably
Oy
from Equation (1), and this may be connected with the transition in zone character
which is discussed in the next paragraph. Consistent with Wilsons(5) calculations,
values of J, the extent of the zone directly in front of the slit, are comparable to
those for slits in the tensile loaded plates. As a result, the ratio#H 0.2 (for
~ = 0.6-0.7 G.) is about twice as large in the DCB specimens.
q
K
Figure 2 compares the outline of the zone shown in Figure 8b, = 0.8 ~.
Y
with calculated enclaves involving nearly the same stress level and bracketing
g - ratios. The agreement is reasonably good considering the many imponderable:
w
the effects of plastic straining within the zone, strain hardening, and possible
departures from plane strain conditions, etc. Note that the etched zone does lean
back slightly, but not as much as might be expected from a linear interpolation
between the two calculated zones. Comparisons among Figures 6b, 7b, 8b, and 9
support Wilsons result, namely that the zones lean further back as the stress level
Ls increased and as the beam height is reduced. Tt is also apparent from Figure 9b,
that the zone finally arches over completely, thereby forming the plastic zone
characteristic of a partially yielded cantilever beam.
The process of tilting back and arching over can be regarded as a tran-
sition from a crack to a beam zone, and this is likely to invalida~e calculated K-
vaIues. The two zones reproduced in Figures 8b and 9a both show a small amount of
* a is the crack length, H the beam height, W the beam length, K the stress in-
tensity and CTy the yield stress.
-12-
backward tilt, and are in a comparable, early stage of the transition. The results
for these two specimens suggest a correlation between the onset of tilting (and
attending deviations from calculated K-values) and thezone size to beam height ratio
P
_% 0.09:
H
This ratio can be transla&ed into a minimum specimen beam height requirement by way
of Equa~ion (1), where A = 0.13:
(3)
It would appear from this and the earlier discussion that the DCB specimen beam height
and thickness requirements for plane strain may be - e similar. This view fS con.
sistent with the proportions recommended by Wessel( .f5~or ~-~ype specimens, i.e.,
H = 1.25t. However, it does raise a question about X-type specimens (H = 0.5
this case, departures in the plane strain toughness observed with increasing ~ I/ :U,d
[)
arise from an inadequate beam height before thelimiting thickness is exceeded.y In
addition, the difference between plastic zone size and shape in double-cantilever-beam
vs straight tension specimens is puzzling since measured Kc values on both specimens are
consistent. This suggests that the extent to which the plasttc zone reaches out may not
be a sensitive indic;~or of events at the crack tip. Th&e is a need to examine this
point as well as for more expertients involving systematic variations of thickness and
beam height to establish minimum specimen size requirements.
Iv. CONCLUSIONS
1. At low stress intensity levels , plastic zones in DCB specimens have the same general
appearance as the zones in tensile loaded plates. The furthe~t excen~of the DCB
zones on the plate midsection is about half the values reportel for tensile loaded
plates about 25% larger than on the surface.
2. The DCB zones begin to til~ back and undergo a transition from a crack zone tothe
zone of a partially yielded cantilever beam at high stress levels when the zone
size to beam height ratio~> 0.09. This transition probably invalidates calculated
K-values and places a low! ~imit on the specimen beam height which is tentatively
estimated as H> 1.5 Ki!io
q
()
-13-
V. REFERENCES
2. E. T. Wessel, State of the Art of the WOL Specimen for I{IcFracture Toughness
Testing, Eng. Fracture Mech., Vol. 1, p. 77, 1968.
3. W. G. Clark, Jr., Visual Observation of the Crack Tip Plastic Zone Developed in
a 3 Fer Cent Si-Fe Alloysf,Westin~house Scientific Paper, 66-1D6-BTLFR-P1,
September 27, 1966.
Cracks in Fe-3Si Steel Under Conditions Approaching Plane Strain, ship Structure,
CommiLtee Report-191, Novmber, 1968.
5. W. K. Wilson, ifGemetry and Loading Effects OJJ Elastic Stresses at Crack Tips~
Westin~house Research Report 67-ID7-BTLPV-R1 Proprietary Class 3, July 3, 1967.
Section 2
by
ABSTRACT
1, INTRODUCTION
With this in mind, we have made direct observations of the zones attending
cracks in steel foil (plane stress) . It was found that the zones produced by a
stationary and slowly growin crack under monotonic loading are similar (9), but differ
from the cyclic growth zone (Ye) and the zone attending anunscable shear crack.(g,~l)
The effect of crack growth was also studied in Fe-3Si Steel, where zones were revealed
byetchirig(12). There, crack growth was simulated in a controlled way by spark-
cutting slits in~o specimens under load (at constant stress intensity) , Under con-
ditions approaching plane stress, the zone was wider than the zone produced by con-
ventional loading and unloading without crack :rowth. Experiments were not ruccessfu~
under plane strain because cleavage cracks were initiated at room temperature during
the spark-cutting operation.
The cut slit was 0.010 in. wide with- 0.005 in.-root radius. Since spark
machining does not produce plastic flow on this scale, the tip of the slit was
essentially strain-free.
Since cleavage cracks are also initiated at room temperature when the spark cutting
is performed in C C14, we conclude that the earlier difficulties with cleavage were
probably not related to hydrogen embrittlement, but simply to the high transition
temperature of l?e-3Si.
80 -17-
I
,; 60
a
o
~
:
. 40
.
G
5 (a)
E
G
z 20
0
0 2 4 6
Elongation, percent
Fig.1. STRESS-STRAIN
CHARACTERISTICS OF THE
Fe-3Si STEEL AT OC
COMPARED WITH 1000C
(b)
111. RESULTS
Iv. DISCUSSION
While the present work shows no differences between the monotonic and the
stable growth zone for Mode I and plane strain , a question remains because the
observations involved unloading. Similar experiments have revealed that the zones
&i
~t r
LOAD SLIT LENGTH PLASTIC ZONE SIZE, D (tn~
SPECIMEN (lbs) (in) (Ksifi) (&) surface midsection
1 3b(l - V2) I ~ h-
b-
(height of arm) = 1.25 in
(Ehickness) = 0.5 in
(b) Corresponding with
cleavage arrest
t$ UY(lOO c, Series P) = 54 Ksi v- (Poissons ratio) = 1/3
P- (load)
uy(OO C, Series Q = 68 Ksi
I- (moment of inertia) = bh3
T
(b) midsection
Fig.4. PLASTIC ZONE FOR SPECIMEN 3P-17 FOR SLIT CUT IN UNDER LOAD
CORRESPONDING To: K = 0.6 ~.: (a) specimen surfaCe and
(b) Midsection.
Y
I
IQ
u
Fig.5. PLASTIC ZONE OF SPECIMEN 3P-18 FOR SLIT CUT IN UNDER LOAD CORRESPONDING TO ~= 0.7 m.
-22-
(a) initiation
(b) propagation
(c) arrest
_23-
,,
,.
.
*a.
.
.,, ,,
,i, ,
,. , ,
,.
%<,+.
,,
,< ! O*
c * v
.. .
*. 1 ,. $ . ,
,.
,
*. *,> .
r,,. . * !,
d *
., ,!, ,
B ffi:
,- _, , ,,. . . . ... . . .
of fatigue cracks are larger than expected(lO~ , possibly because thd yield stress is
lower after a stress reversal (Bauschinger effect). With a reduced yield stress,
more plastic deformation is produced during unloading and this may make it difficult
to see the differences that existed before the load was removed. The problemcould
be overcome by aging the test specimens under load and cooling prior to unloading,
a procedure that makes the unloading deformation transparent to the etchant.
The present results are in accord with observations of steel foil under
load(g-ll~, showing that crack growth proceeds with no major changes in the plastic
zone, provided the crack speed is reasonably slow. However, Che zone size and pre.
sumably the crack-tip displacement do depend on the fracture mechanism, the most
striking example being the marked difference between the zone artending cleavage
cracking and those attending other fracture modes.
Taken together, these three effects decrease the stress intensfty factor
of a rapidly moving cleavage crack in a DCB specimen by almost an order of magnitude,
compared to a stationary crack. Since the plastic zone size depends on (K/Y)2, it
could be thus reduced by a factor of 50-100. In fact, close examination at high
magnification away from the links of Figures 6 and 7 show that the zone extent is
less than 1/200 of the static value. The remaining zone shrinkage can easily be
accounted for by the high strain rates at the crack tip, and the corresponding yield
stress elevation. Eftis and Krafft(24) estimate a crack tip strain ra~e of 107sec-1
Unfortunately, reliable yield stress data exist only UP to 105se~1, for which strain
rate the yield stress of silicon iron is about 2-1/2 times its static value. (9) ln
any event, yield stress elevation can contribute at least as much to zone shrinkage as
any of the effects which decrease K. Furthermore, it is quite likely that the
stress at the tip of a running cleavage crack is on the order of the theoretical
strength of the lattice.
v* REFERENCES
5. J.A.H. I-Iolt
and F. A. McClintock, IX Int. Congress Appl. Mech. , Vol. 8, p. 51,
1.956.
12. G. T. Hahn and A. R. Rosenfield, Int. J. Fracture Mech., Vol. 4, p. 79, 1968,
15. l?.A. McClintock, Proc. Roy. Sot., Vol. A 285, p. 58, 1965.
18. J. Dvorak, Fracture 1969, P. L. Pratt, et al, eds., Chapman and Hall, London,
P. 3383 1969.
20. A. A. Wells and D. Post, Proc. Sot. Exptl. Stress Anal., VOI. 16, p. 69, 1958.
24. J. Eftis and J. M. Krafft, J. Basic En&., VO1. 87, p. 257, 1965.
25. J. R. Griffiths and G. Oates, Fracture 1969, F.L. PratL, et al, eds., Chapman
and Hallj London, p. 229, 1969.
-26-
Section 3
by
ABS!ITMCT
1, INTRODUCTION
TABLF1. COMPOSITION
AND YIELDSTRENGTH
PROPERTIES
Yield
Temperature Strength
Ma~erial c M P s Si Ni Cr Mo (Oc) (ksi)
4340 0.40 0.70 0.04 0. OLF 0.27 0.95 0.20 i22 240
-196 310
A-517 0.15 0.80 0.008 0.013 0.26 0.86 0.48 0.42 22 109
-135 160
-196 178
N o
I
1
Approxknate Specimen Dimensions
L Spark machined
Material I L ] H I B
slit. Length-H 1
4340 I 8.5 11.501.5001
A-517
12
q-~5 1.50 .375
Specimen
thickness a
Fig.1. THE DCB TEST SPECIMEN. THE SPARK MACHINED SLIT FACILITATES STARTING
THE CRACK, THE SLIT WIDTH IS A VARIABLE IN THE STUDY, IN THE
CASE OF THE A-517 TWO SPECIMEN LENGTHS WERE EMPLOYED.
In a series of fracture toughness measurements of a mild steel in which both
temperature ancl loading rate were var<ed, Eftls and Kra,fft(g) :Eound that the initi-
ation tOUghneSS, ~<lc of a mild Steel, decreased with increasing loadin~ rate to a
minimum value determined by test temperature and then increased at still higher loading
rates. From Ehese resulCs, they suggest that the energy absorption rate would follow
the same trend as the crack velocity increased. They substantiate thefr results by
estimating the stress intensity-crack velocity relation from wide plate test results
obtained by Hall~lO) and coworkers. These estimates TnclfcaEea rapid increase in
stress intensity with crack speeds in the range of 3000 fps to 5000 fps. There is,
however, an element of uncertainty in these estimates as they were not corrected for
dynamic effects, effects which Eroberg (11) S~loWSto be important in this speed range.
In addition, Dvorak(~2) suggests that propagation is discontinuous due to the formation
and rupture of unbroken links left behind as the crack grows. He further suggests that
plastic deformation of these links is the principal energy absorbing process during
propagation. On this basis, the processes of crack initiation and propagation could
be different.
In thfs paper we present the results of crack propagation tests which com-
pare initiation and arresk toughness for four steels having widely different strength
levels. A DCB specimen and a relatively stiff loading arrangement were used in an
attempt to minimize interactions between the loading system and specimen during propa-
gation. Skress levels during propagation were altered systematically by varying the
bluntness of the startfng crack. This technique made it possible to relate the arrest,
toughness to an average energy absorption rate and the stress level at initiation. The
presence of energy absorbing ligaments is demonstrated and a quanti~ative analysis of
their role in propagation is developed.
where 2~ is the wedge angle and # is the coefficient of friction between the pins and
the wedge. Suppression of the bending stress is controlled primarily by varying the
wedge angle. Initially, an 11 degree wedge angle was used. For the approximate speci-
men dimension given in Figure 1, this wedge angle was adequate for maintairiing straight-
forward propagation of cracks less than about 2 to 3 inches in length al~!nough longer
cracks turned to the side of the specimen. A 30 degree wedge angle was used later in
this study and was found to maintain the desired crack plane for cracks longer than 5
to 6 inches. It should be pointed out that even though ~his loading method may en-
courage straightforward advance of the crack at the initiation stage, several theore-
tical investigations have shown that the crack-tip stress field undergoes reorientation
with increasing velocity to the extent that branching or at least curvature of the
crack path is favored. This effect can also be suppressed by increasing the wedge
angle but tht required wed.qe angle would depend on the maximum crack speed.
-31-
lln important Eacet of this study involves the determination of the vari-
ation in the stress intens-ityat crack arrest with the stress intensity at ini~i-
ation . In order to increase the initiation stress intensity above the minimum
sharp crack value, l~lc, the specimans were provided with spark rnacl~inedslots of
VariOLISroot diameters, These starter slots were cut to depths of not less than H,
the beam height.
(2)
(3)
Since G = 1C2/.E,combining Equations (2) and (3) gives a first order differential
equation for compliance as
@ =% (:+o.7)3+@o {5)
(6)
The actual dependence of K on crack length departs from Equation (6) for very short
crack lengths and as the crack approaches the end of che specimen. Estimates(6~17)
of the range of applicability of Equation (6) give
:<; <1-:
-32-
0,4
H= 1.5 inches
0.3
:\
o I I I I I I I I
0.8 I,2 1.6 z .0 2,4 2.8 3.2 3.6 4.0
a/H
4 plot of Equation (6) is shown in Figure 3. Clearly, for a constant pin
displacement, y, the applied K decreases rapidly with increasing crack length, a
condition which favors crack arrest. However, truly fixed pin displacement as the
crack extends could be achieved only for an Infinitely stiff testing system. Relax-
ation of the loading system as the crack grows will produce an additional increase
in y. lf we let b represent the total displacement of the system required to produce
a pin displacement of y then for de/da < 0, the specimen-load train system is in-
herently unstable since the stress intensity increases as the crack extends. In a
detailed analysis of crack stabiliry in varfous fracture mechanics specimens,
Clausin~(18) includes the effect of the compliance of the loadfng system on the con-
dition for crack instability. From Equations (18) and (19) in Reference 18, it can
be deduced that for a DCB specimen db]da must be positive for
@L<2 @ (7)
where $~ is the compliance of rhe loading system. As long as Equation (7) is satis-
fied, c~ack growth is mechanically stable.
In this study, an lnstron testing machine with a 10,000 lb. load cell was
used . For this loading system, the compliance (including the wedge) was found to be
5.4x IO-6 in./lbf and, therefore, crack extension can be shown to be mechanically
stable for a/H> 1 on the basis of Equation (7). The UIOSt mechanically compliant
member of the load train is the load cell, the wedge contributing only 8% to the
rotal compliance. The inherent stiffness of the wedge-loading technique is an ad-
\ vantage since increased stiffness decreases the interaction between the testing
machine and the test specimen during rapid crack propagation.
-33-
Room kcmperature testswere conducted with the specimc;-,loaded vertically
as shown in Fi&ure 2 while the pin displacement was rEcorded continuously by memls
of the Lli]3~age. The gage used is similar to that described by Brown and Srawley, L19)
Ilowcver, the sage is attached to the specimen by conical-pivot seats mounted on both
sides of the slot on the end-face of the specimen. This seaEing arrangement enabled
a positive ccjntactbetw~en the ~age and the gage blocks thereby mi.ntii.zingthe effect
of the shock associatcilwith a rapiclpropagation evcnc. OEher arrangements proved
unsatisfactory a.s Lhe ~a~c position relative to the speci~en changed,or che ga~e
was completely clislo~ed during propagation. The gage calibration was repeatable to
i 0.0002 in. and was checked at frequent intervals during this study. While the load
was not used to determine K, the load-displacement record provided a convenient means
for rioting the initiation and arrest conditions since a drop in the applied load
signifies a propagation event. At test temperatures other than room temperature, the
specimen-displacement gage ass<inbly was either placed in a bath or, for intermediate
subzero temperatures, was surrounded by a helical tube which sprayed,at controlled
rates,cold N2 gas supplied by a licluidiX2 container. After one or two propagation
evenEs, the specimen is removed from the test fixture and th~ crack length measured
on the surface to an accuracy OF t 0.002 in. Some measurements of crack length were
made on both sides of Lhe specimen but generally this was unnecessary as the crack
Erace on both surfaces were very nearly equal.
111. RESULTS
DCB Test Results. A summary of the average initiation and arrest results
for the Iour s~eels are compiled in Table 2. Both plane s~raiu and non-plane strain
data are referred to here as initiation toughness or I<Q. The thickness criteria in
terms of th~ parameter E/(CQ/Y)2 which establishes the lower limit of the plane strain
regime is somewhat uncertain, but is approximately in the range 1.5 to 2.5. loughriess
values for which ~/(kQ/~)2 is grea~er than abouz 2.0 should, therefore, correspond Co .K~c,
In the 4340, propagation was quite stable at room temperature for crack
extension from both a 0.006 in..wide slot and an arrested crack. Increasing the slot
width to 0.026 in. resulted in a more than fivefold tncrease in the initiation
threshold and loss of plane strafn. In this case, the crack extended rapidly for
some distance down the length of the specimen and then turned to the side of the
specimen typical of a very unstable propagation in which Ka is considerably less than
KQ . The resulting crack path in this specimen is shown in Figure 4. At -196C, the
initiation threshold was elevated by a 0.006 in.-wide slot, although in this case,
_the co>~~i+ion of h. ~1nt- +<p-p~~s.<x.
+ Metallographic examination often revealed that crack extension was somewhat
nonuniform along the crack front when propagation occurred in small increments.
-34-
Shoemaker and
Rolfe notch
bend(z~) -196 sharp 35
(static)
22 sharp 28 26 2.5
sharp 34(C) -8 1.6
0.007 28 27 2.5
0.042 75 < 25 0.4
0 0.010 50 12 0.9
Project
Steel E -112 sharp 22 18 2.7
In the Fe-3Si and mild steel, slot widths great~r than 0.007 in. were
required to elevate the initiation toughness and produce unstable propagation. How-
ever, even for a sharp crack, an anomalous increase in KQ as, for example, occurred
in Fe-3Si at 22C results in unstable propagation and a low arrest toughness. In this
particular example, the increase in the toughness from 28 ksifi. to 34 ksi ~. is
the result of extensfve divisfon of the arrested crack tip as described in the follow-
ing section. It is also interesting to note that sharp elevation of F@ by increasing
the temperature from 22C to 100C did not promote unstable propagation but resulted
in very slow crack extension.
-36-
I.0
\ Legend
0,8 4340 :
-\ A-517 F g.5. A COMPILATION OF HE TOUGH-
Fe-3Si
-\ Mild st~el~ NESS DATA FOR THE FOUR STEELS
0.6 +\
ERd KIC SHOWING THE DEPENDENCE OF THE
ERd ~K;C - ARREST STRESS INTEN51TY, Ka,
-\J\D-,5 ON THE INITIATION LEVEL, KO.
4 -\\\ 22
Test temperatures in C
22 are indicated. Dashed lines
-lp<?-p,.,y
7
1 are theoretical predictions
0.2 22
\.>,\
based on Equation (13)
! I I I
0. 5
2 3 4
KQ
r,
DenoCing the m-inimum initiation threshold at which the propagation is essen-
tially stable by KIC, then the ratio? KQ/KIc, represents the amount of averstressing
achieved by blunting the starter sIo~. For KQ/K1c above unity,Ka for each of the
steels falls abruptly. This behavior is summarized in Figure 5 for the four steels
a~ each of the test temperatures where I(TCwas established. In this figure, the
stable propagation data for these materials nearly coincides to the coordina~es
KajKq = 1, Ka/KIc = 1, and so have not been plotted. Also shown in Figure 5 are pre-
dictions based on the condition that the energy absorption rate during propagation,
~, is constant and related to KIC. This point is discussed in more detail below.
(a)
(c) (d)
.:
!.,
,,, ! ~.
m
., ., , ...
,... *
%.
final arrest
Fig.7. CLEAVAGE FRACTURE PROFILES IN THE MILD STEEL. PHOTOGRAPHS OF THE MID-
THICKNESS POSITION OF A SPECIMEN TESTED AT -196C. CRACK EXTENSION
OCCURRED BY VERY SMALL JUMPS
~~ti~??<;+ $.:;
,,,,
,,,, ~.
.
,: ,, , ,,,
,,,
6, ,,
, .,
.,1: !,
v,..
;:
.+7?-/,
,,1,~
: -::
,.;
t?- , w. ,,
:
A ,:
Fig.8. CRACK PROFILES IN Fe-3Si SPECIMEN TESTED AT OC. THE AREAS MARKED A
AND B ARE EXAMINED IN FIG.9
-39-
., ,,
..-
0.0012
, <
b >-. ,.
*>
,.
,,, ,
...
0.0026
. .
,.,
P
.,, ,
0.0072
REGION A OF FIG.8.
., . .
ti~y-+
,,!. .)
.,
,. , ! $, ---.: .
~J ., ,-
*
0.0 0.0012
,, >. .
.. ..
.
*
0.0043
,-
~+7--&
,, ,,),
.,
;
.,, -.
,,
,,, ,,+ , ..-,
.,. . .
, .,
,- .., , ,,#, . ~,.
0.0050 0.0072
REGION B OF FIG.8.
!!!,,,!
,::, :!
),
:, ,,
FIG.12. ARRESTED CRACK TIP IN QUENCHED 4340 AFTER UNSTABLE PROPAGATION AT -196C
-41-
In Figure 7, it appears that several micro cracks are isola~ed and that the
~igaments perhaps form a single multiply connected network. Sections revealing the
crack profile in the Fe-3Si also have the same appearance as shown in Figure S. Here
again, apparen~ly isolaEed microcracks are common but become more numerous near Che
crack tip. J3y progressively sectioning in increments of about 0.002 in. (a total of
twelve sections were m~de) it was possible to obtain a three-dimensional concept of
the crack shape and better judge the degree to which the microcracks were inter-
connected. From the sectioning, it was found that in no instance was a microcrack
observed to be completely isolated. Two examples in which apparently isolated micro-
CraCICSwere observed to connecc Eo the main crack are ,sho~ in Figure 90 These re-
sults sugg~st at least for the Fe-sSi that the ligaments correspond to islands of
unbroken material. The crack itcself forms a continuous surface and truly isolated
microcracks are rare. Because of Lhe close similarity in the fracture topography
with the Fe-3Si, this result should also apply co the mild sEeel although this
material was not progressively sectioned.
In the A-517, which has a quenched and tempered microstructure, the crack
propagation mode was also primarily cleavage at -lg6C. However, in contrast cothe
mild steel and Fe-3Si, the cleavage facets shown in Figure 10 are smaller, more
irreeqlar, and generally not as well d~fined. Areas of ductile tearing were also
evident. In profile, Figure 11, the crack was found to exhibit occasional discon-
tirmities over most of its length. The crack shown in ~igure 11 propagated in a
relatively stable manner by very small but rapid increases in length. The final
propagation before the specimen was sectioned extended the crack approximately 0.10
fnches . Near &e crack tip, the crack path becomes quite discontinuous. Progressive
sectioning in very small increments ( - 0.0004 in.) revealed Lhat the crack surface
in the A-517, like the Fe-3Si, is continuous and surrounds the unbroken ligaments.
It would appear that in the case of A-517, these ligaments rupture at relatively
small crack opening displacements since they are confined primarily cothe crack-tip
region. Exactly the same behavior is evident in the Lk3ii0aS can be seen in Figure
12. Here the crack propagated in a very unstable manner (at -lg6C) for approxi-
mately 2 in. before arresting with a microcrack array at the tip.
In Figure 13, are shown a series of macrographs which illustrates the in-
crease with temperature of the extent of deformation on the surface of specimens
tested at -75C, OC, and 100C. The plastic zones extending on either side of the
crack are similar in size for both the OC and -75C test temperatures. At I009C,
the plastic zone sfze is considerably larger cons<s~ent with the increased 1~/Y.
Accompanying this transition to a more ductile fracture mode is the development of
shear lips and the tunneling of the crack along the miclthickness.
-42-
(a) -75C ,,
,,
(b) 00C
,,,
!!,,
(c) 1000C
(b) Midthickness
Fig.14. COMPARISON OF THE DEFORMATION AT THE SURFACE AND MIDTHICKNESS OF A
SPECIMEN TESTED AT 22 C. THE INITIATION-ARREST LOCATIONS ARE IN-
DICATED IN THE MIDTHICKNESS VIEW BY THE KAT WHICH PROPAGATION
OCCURRED
(a) 0.6 in. from crack tip 0.25 in. from crack tip
~=o.13:2
H
in the DCB specimen. Therefore, during rapid propagation K/Y < 0.09 ~. at the
crack t~p. Near the arrested crack tip where the uncertainty in the extent of defor-
mation due to etching is small, K/y < 0.04 ~. The plastic zone at the arrested tip
in Figure 15d provides the measure of the arrest stress intensity as 7.7 ksifi.
taking the static yield strength to be 62.4 ksi. However, the unbroken regions span-
ning the crack interface may support a portion of the applied load and, therefore,
the applied stress intensity couldbe larger than this.+
. --
+ The arrested crack length in this specimen is outside the range of applicability
of Equation (6).
45-
Iv. DISCUSS1ON
The ratio Ka/K~ div%des the DCB test results into two regimes. Thr property
of stable crack propagation has been attached to those results in which &/KQ is very
nearly unftyo In all other tesCs, Ka/KQ was about 0.4 or less and here the term un-
stable propagation was applied. Crack propagation behavior yieldiw lta/KQ values in
the range between these two extr~mes was not observed. From a macroscopic standpoint,
crack propagation inscabili~y appears to develop very rapidly as <Q increases A
simple energy balance offers a means of rationalizing these results and also provides
a way of estimating the energy absorption rate frm a comparison of l{awith KQ.
During propagation, the energy release from the elastic strain field in the
specimen is expended as kinetic energy and irreversible processes, the principal one
being plastic deformation. The resulting energy balance is a modified Griffith-
Orowan relation
(8)
where T is ~he kinetic energy and u is the strain energy in the specimen. The ~ett
side of Equation (8) represents the rate of work done on the specimen during propa-
gation. The crosshead travel is negligible during the short time period of propa-
gation and as the elastic relaxation of the loading pins, wedge, and testing machine
is also small, we can , as a first approximation, set dW/da = O. For consistency we
have used the symbol R to represent energy dissipated during cracking but in fact,
R and G can be used interchangeably in keeping with formal definftions(23~. In the
instant prior to the onset of propagation, while the kinetic energy is still zero,
the familiar condition for fracture rez.ul~s, i.eO,dU/da = - RL Gc. Also, note that
dU/da and R have upposite signs and fQr the magnitude of dU/da greater than the
magnitude of R the kinetic energy must increase with increasing crack length. The
interdependence between the three energy terms is shown schematically in Figure 16
where the dependence of R on crack length has been assumed. Actually, R is related
to crack length only indirectly since
R = R(v)
and (9)
v = v(a)
where v is the crack velocity. An important point made in Figure 16 is that the
kinetic energy in the system is not an energy sink term, i.e., kinetic energy may
also be expended indirectly as deformation. Therefore, at the second point where
g agatn equals R, Lhe kinetic energy is maximum and for a further increase in crack
da
length R>~, $<O. The kinetic energy supplies the energy deficit to continue
the expansion of the crack. The condition of arrest corresponds to T = O which from
Equation (8) yields
a
f faf
a
J o
dU=L~Rda
~
0
(lo)
-46-
Legend
$!4
4 . R
h-
Shudedarea @ is excess stroln energy
,/ mnve.?ed t. kine,lc energy
Fjg.16. SCHEMATIC REPRESENTATION OF THE ENERGY
After P kinetic energy wed CHANGES ACCOMPANYING PROPAGATION AND
ARREST IN A DCB SPECIMEN. The L#
K22A Area @ = Area @ at orr@st7
curve corresponds to crack growth under
fixed-displacement. conditions, The R-
curve is hypothetical
.~+_-..-._--..x+
---
Unstah:e propagoitcn
I 1~- - .J
L.Jhere
a. and af are the initial and fillaIcrack ~cngths, respectively. In arri.vinS
at Equatian (10), it is assumed that the system (the specimen and loadin~ mechanism)
is conservative (isolated) except only for heat lost through the crack plane during
propagation. This assumption essentially requires that all kinetic energy is dis-
sipated along the crack plane. To proceed further requires knowledge of the de-
pendence of R upon crack velocity. When viewed on a macroscopic scale, the etching
results show that the deformation is fairly uniform along the crack path. The defor-
mation accompanying crack propagation appears to show a dependence on crack velocity
only at the specimen surface when shear lips begin to form at the higher test temper-
atures . Therefore, as ~ reasonable first approximation, we can set
=G o
Q Q =
R= [ (11)
1d V>o
af
1
d= dU
af-ao !a
o
Substituting Equation {5) and recalling that G = KL/E, integration of 12 gives
or
d
.1
r
3 .GQ
1/2 + ~ 1/4 ~ 1/4 + c 1./2
Qa a 1[ Q
1/4 c 1/4
a 1
(13)
It can be recognized from EquaEion (1.2) that Rd falls between G and Ga. values of
Q
Rd calculated from Equation (13) and expressed in stress intenszty units for A-517
steel.are given in the following table.
-47-
TABLE 3.
ABuLATION F Q a ND d ALUES
FOR A-517 AT -196DC
*VL ....
Diameter (in.) KQ (ksi~n.) Ka (kSi&) ~ERd (ksiy~n)
Sharp 32 (KIC) 31 31
0.007 42 18 28
0.009 43 12 24
0.0.25 69 16 36
The data of Shoemaker and Rolfe for A-517 steel at -196C give
K1c(dynamic)/K1c(static) - 0.S, a value which is nearly equal to the ratio @id/KIc
for Rd = 2/3 GIC. This suggests that,within the assumptions employed in arriving at
Equations (11) and (13), the sharp crack dynamic toughness corresponds to the stress
intensity reqnired to sustain propagation. This does not mean, however, that for a
given flaw size, a stress level which provides a K ~@Rd !s incapable of causing
unstable catastrophic propagation in a structure. T-n ;~articular if tn~structure
is stressed such tha~ K increases with increasing crack len~tfi, tnsn a flaw originating
in a small brittle region may enter the tougher surroundings and continue to grow even
though its stress intensity at the instant it enters the matrix is < flRd . For ex-
ample, consider the case of a large plate dead-weight-loaded in ten~ion with a brittle
region at the center. If a small crack initiates within and grows rapidly through the
brittle zone, then at the instant the crack enters the surrounding good material,
iEs size and the applied stress level define a K. For the crackoto arrest, it is
easily shown that Rd for the good material must satisfy ~R~ > ,21{4, If this inequality
is not satisfied,continued propagation is energetically fa;ored.
At high magnifications, the etching results also show that Rd << Ka2/E
except around the ligaments. Therefore, it appears that the major portion of the
energy of propagation is expended within the unbroken ligaments behind the crack tip.
If plastic deformation during propagation is confined primarily to unbroken regions
behind the tip, then these ligaments must also support a significant portion of the
applied load. Hence, the stress intensity at the crack tip will be less than the
applied stress intensity. A solution which affords a useful estimate of this effect
is obtained in Appendix A for the case of an infinite center-cracked plate loaded
-48-
along the cr~ck plane by a fini~e number of point forces. l?rom this analysis, the
crack-tip stress intensity is given by
where I(A is the applied stress intensity, 1? is the average force exerted by each
ligament, m is the number of unbroken ligaments and d is Che fraction of the crack
length (2a) over which Lhe ligaments remain unbroken (cf. Figure A: 1). The function
G(m,d) is plotted in Figure A-2 for a range of values of m and d.
and
2F G(m,d) ~ 10 ksifi.
z. -
The results of the above model serve to emphasize the importance of the un-
broken ligaments on propagation behavior. If the ligaments act to produce a signifi-
cant difference between the applied K and the crack-tip K, then also the ligaments
must provide an important sink for energy dissipation during propagation. Therefore,
a major part of Rd can now be attributed to deformation and rupture of unbroken areas
as the crack extends, clirect evidence for which was available only in the case of tht
Fe-3Si etching results.
In summary, the etching results coupled with the energy balance and ligament
model analyses provide evidence that the source of the resistance to rapid crack propa-
gation is not in the crack-tip reg?on but rather unbroken regions which are bypassed
as the crack-grows . This conclusion seemsapplicable at least to the steels and
-.
-49-
testing conditions employed in this study. The source of the ligaments is not yet
clear although two possibilities exist: anomalously tough regions which are scattered
randomly throughout the steel, or a region of one or more grains whose cleavage planes
are poorly orien~eclwhile the surrounding grains are well oriented for propagation.
Finally, there s.remany asp~cts of rapid crsck propagation which have not
been considered here. In particular, elas~ic wave effects such as rtf~eCtiOIIS from
free surfaces back to the crack tip may play an important r~le in propa~ati,on. blhile
quasi-static analyses of crack propagation have been dev?lopcd, a fLl~~ydynamic treat-
ment of crack propagation would be enormously difficult. The evaluation uf such
effects is more realistically attacked from the experimental standpui,nt.
v. CONCLUSIONS
1. The inherent stiffness of the wedge-loaded DC13 test affords a means of propa~ating
s crack in a relatively stable manner such that the stress intensity Jt initiation
(KQ) and arrest (Ka} are very nearly equal. By increasing K throl-ghincreasing
the root radius of the st:lrL~r slut, Ka undergoes an abrupt fecre.;e to Ka/Kq <0.4
for steels with a wide range of yield stren~ths from 62 ksi to 310 ksi.
4. Deformation associated with the crack tip during cleavage propagation could not
be resolved. Thereforej propagation by cleavage> at least in the Fe-3Si steel,
is nearly ideally brittle with Rd(tip)/~-rc j 0.01. The deformation is confined
to the unbroken ligaments as the crack extends and the overall energ]- of propa-
gation is primarily absorbed in their deformation and rqpture.
VI . REFERENCES
,. Y. hkita and K. Ikeda, TransporEaCion Tech. Research Inst., Report 56, Tokyo,
.,.
.
December, 1962.
3. pl.Yoshiki, T. Kanazawa, and S. Machida, Proc. Seventh Jap. Cong. Test. Mat.,
p. 71, 1964.
9. J. Eftis and J. M. Krafft, Trans ASME, Series D, Vol. 87, p. 257, ~g64.
10. 1?.P. Videon, 1?.W. Barton, and W. J. Halls SSC Report 148, 1963.
12. J. Dvorak, Fracture 1969, P. L. Pratt, et al, eds., Chapman and Hall, London,
p. 338, 1969.
13. P. B. Crosley and E. J. Ripling, Trans ASME, Series D, Vol. 91, p. 525, ~96go
17. E. Gross and J. E. Srawley, NASA Technical Note, TN D-3295, Pebruary, 1966.
19. W. F. Brown and J. E. Srawley, Plane Strain Crack Toughness Testing of Hi#n
StrenSth Metallic Materials, ils~ STP 41O, 1967.
20. A. K. Shoemaker and S. T. Rolfe, Trans ASME, Series D, Vol. 91, p. 512, 1969.
220 G. C. Sib, Inelastic Behavior of Solids, McGraw-Hill, New York, p. 607, 1970.
ACKNOWLEDGMENTS
The authors are grateful to Ship Structure Commit~ee for Eheir support and
LO Ac~visoryGroup 111 Of the Ship Research Committee (S. T. Rolfe, Coordinator) and
to A. R. I.ytle and R. W. Rumke of the Academy for valuable advice and encouragemen~.
They are also indebtedto M. ~ightner of the U. S. Steel Corporation, who contributed
the heat of T-1 steel used in this study. Experimental results for 4340 steel are
derived from studies for the Army Research Office (Durham). The authors also wish to
acknowledge the experimental assistance of R. L. Stephenson, and fruitful discussions
with M. F. Kanninen. Valuable assistance in preparation of the manuscript was pro-
vi~eclby Miss C. Pepper.
-52-
APPF,NDIX A
o Xx+o = 2[$5(2)-F@(z)]
YY
We may consider the stress function @ as arising solely from the point force tractions
along the crack plane. Then, if the plate is loaded in some fashion along its boundary
to produce an applied stress intensity, KA, the net stress intensity is
trafisformsthe unit circle in the ~-plane into a slit of length 2a in the z-plane.
In the ~-plane, Muskhelishvili provides a solution for a distributed force over an
arc of the circle. By allowing the arc length to decrease to zero such that the
resulting point force, F, is finite, Muskhelishvilis solution reduces to
o(g) = -&An(o--g)
where o is the point along the hole boundary at which the force? F, is applied. For
a distribution of n point forces on the crack plane except at the crack tips, x = ~ a,
~~ The stress intensity based on the applied stress and crack length neglecting
tractions along the crack surface.
-53-
A-2
Y
Fn
n-1I
bwndory)
(slit
r 2(]---i
[cl)
F-n
ttttt+
1------
-1-
++HH --mm
8
p-l-
(b)
Fig.A-l. (a) CENTER-CRACKED SHEET LOADED BY POINT FORCES DISTRIBUTED RANDOMLY OVER THE
SLIT BOUNDARY. THERE ARE N FORCES NUMBERED IN A CLOCKWISE FASHION.
(b) SAME AS (a) EXCEPT THAT THE FORCES ARE EQUAL AND UNIFORMLY SPACED
-54-
A-3
sir@.
3
(A-2)
l-cose .
J
where ej is defined by Xj = cos@; and Xj is the location of the jth force on the
crack plane. Let us now examine the particular case in Figure A.lb of equal magnitude
forces equally spaced a distance 6. The first force is positioned at xi = 4 and,
making the subs~itution d = (a-A)/a, Equation (A-2) becomes
~m-[m(l-d)+ (j-l)d]
or more compactly
where m = n/4. The function G(m,d) is plotted in Figure A-2. As m+ , the force
distribution becomes continuous and & G(m,d) approaches cos-l(l-d). Therefore, the
term due to the ligaments on the rigR t side of Equation (A-3) becomes, for large m
(large number of ligaments)
w~ich is precisely the result obtained by Dugdale(2~ for a continuous stress Y = F/6
superposed along the crack surface near the crack tip.
References
A-h
moo
= 0.01
6@3 - / 0.05
400 1
), 0.1
/ , 0+2
0.3
~ ;~ 0.;
200
~,
$4; 0.9
.0
I
100
80
60
_40
w
E
m 77
20
--v-
10
8
6
,, ..
m
4
+
II
Gm=~~2..[m(,+,+(j.,)d]2
. .
,+
I
i 2 4 6810 20 40 60 80100 200 4m
m
Fig. A-2. THE RESULTS OF THE ANALYSIS FOR THE UNIFORMLY SPACED EQUAL FORCE MODEL
None
Se.cuntyClas51ficati0n ,.1,
I OHl GlN4Tl NG AC TIVITV (Corporate author) 23. REPORT 5CCu RlTY CL A5SIF1 CATION
ed
Battelle Memorial ~nstitute 2b. GROUP
Columbus. Ohio
3. REPORT TITLE
August, 1970 35 58
~ti. CONTF?4CT OR GRANT NO. 9a. ORIGINATORdS REPORT NUM13ER(5)
b. PRO,ECT NO,
NOO024-68-C-5073
SR-180
I SSC-219
c, 9b. OTHER REPORT NO(51 (AnY other number. fhat may be essigned
thzs report)
a,
10. DISTR[@UT!Or.I STATEMENT
I,
I
Naval Ship Engineering Center
13, ABSTRACT
Fracture Mechanics
Fracture Toughness
Crack Initiation
Crack Propagation
Crack Arrest
Structural Steel
Plastic Zones of Crack Tips
Double-Cantilever-BeamSpecimen
Cleavage Fracture of Steels
Elastic-PlasticAnalysis
Fracture, Rate, and Temperature Effects
None
(PAGP2) Security Classification
GPIY 917.317
SHIP RESEARCH COMMITTEE
Maritime Transportation Research Board
National Academy of Sciences-National Research Council
DR. H. N. ABRAMSON
DiFecto~, Dept. of Mechanical Sciences PROF. W. J. HALL
Southmst Reseamh Institute %ofesso~ of CiviZ Engineering
Univmsity of Illinois
MR. W. H. BUCKLEY
Chief, Structural Criteria and Loads MR. J. E. HERZ
Bell Aerosystems Co. Chief Structural Design Engineer
Sun Shipbuilding & Dry Dock Co.
DR. D. P. CLAUSING
Senior Scientist MR. G. E. KAMPSCHAEFER, JR.
U.S. Steel Corporation Manager, Application Enginee~ing
ARMCO Steel Corporation
MR. 1. E. COX
Senior Progmm Manager PROF. B. R. NOTON
Neoport Ne%s shipbuilding & DPy Dock Co. Prof. of Aerospace & Civil Engineering
Washington University
MR. J. F. DALZELL
Senio~ Research Engineer MR. W. W. OFFNER
Stevens Institute of Technology Consulting Enginee?
SSC-21O, Analysis of Sbnming Data from the /S,S. WoZvetine State by J. II.
Wheaton, C. H. Kane, P. T. Diamant, and F. C. Bailey. 1970.
AD 713196.
SSC-215, A Guide for the Synthesis of Ship Stwetures Part One - The Midship -,.
fieldof a Txvxnsve~seQ-l@amad DXJyCargo Ship by Manley St.Denis.
AD717357.
1970. -,
SSC-216, (To be published).
.- ------