A GENERAL MODEL TO PREDICT THE

ELASTIC-PLASTIC STRESS DISTR~B~ION AND

FRACTURE STRENGTH OF NOTCHED BARS iN PLANE
STRAIN BENDINGt

NOTATION
specimendepth
specimenwidth
notch depth
ligament depth
hkansk
root radius
limiting root radius
princippl fongitudinal stm!ises
nominal stress
bending moment
bending moment (elastic section only)
elastic stms concentration factor after Neuber
elastic stress concenbatioa factor at position x
elastic stress concentmtion factor at elastic-plastic interf%cz
elastic stress concentration factor of a hypothetical notch of depth c+ R aad root radius
P’W)
plastic stress concentration factor [I + In (I + R/p)]
stress intensity factor for a crztck
critical stress intensity factor for fracture
stress intensity factor for a finite radius
critical stress intensity fw for a finite radius for fracture
yield stress
critical fracture stress in bending
plastic zone size (measumd along x-axis)
plastic zone size at maximum constraint
exponent describing elastic stress gradient in elastic plastic case
shift ofneutral axis
root radius oftbe hypothetical nutch
critical plastic zone size to pmduce u,* locally and cause fracture
applied moment to cause fmcture
the critical plastic stress conce@ation factor to cause fracture
the center of force ofthe logarithmic spiral region measured from the notch root

tPrcsented at the National Symposium on Fractmz Mechanics, Lehigh University, Bethkhem, Pa,
June I9-21,1967.

191
192 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

a(4 the actual compressive stressatx = 0

u*(a) the hypotheticalcompressivestress at x = (Iwiththe plasticzOneremoved
@au maximum tensilestressaheadof a notch

1. INTRODUCTION
UNSTABLE(brittle) fracture occurs in cracked structures, at nominal stress levels uF
that are below the general yield stress, when a critical opening displacement is pro-
duced at the crack tip[l-4]. The crack tip displacement 2V(c) is accommodated by a
plastic zone size of ‘radius’ R which develops near the tip. When R is small compared
to the crack and net section dimensions, R is proportional to V(c). A critical crack
opening displacement at fracture, 2V* (c) , implies a critical plastic zone size at fracture,
Rp. The fracture stress oF is that stress which is required to spread the plastic zones out
to RF, in a structure containing a surface flaw of depth c. Consequently, a quantitative
understanding of the relationship between RF (or V * (c) ) , uF and c is a prerequisite for
the development of analytic fracture criteria in terms of these parameters [5].
The methods of linear elastic fracture mechanics have been used to calculate these
relations for certain types of specimen and loading geometry and various experimental
techniques have been used to obtain these relations experimentally[6]. Most of the
theoretical work reported thus far has dealt with (1) very sharp cracks, whose tip radii p
are small compared with R, (2) plane stress[7] or anti-plane strain loading[8] and (3)
‘infinitely wide’ structures of ligament depth d that is large compared with R. None of
the previous work has dealt with the general problem of the development of plastic
zones in thick structures where plane strain conditions prevail, and where both the
ligament depth and the root radius are finite.
Previously, some workers[9-121 have obtained theoretically, solutions to the prob-
lem of the elastic-plastic deformation of notched specimens. All these solutions are
approximate in that assumptions have to be made about the elastic-plastic distribution
and no rigorous solutions exist to date [ 111. A quantitative description of the stresses
around the notch as a function of the applied moment has been deduced from experi-
mental observations of the plastic zones for one geometry [ 131. To repeat this on many
geometries would be tedious; consequently, a general method for predicting the
elastic-plastic stress distribution from the externally applied moment is needed.
In the present work, an approximate theoretical method is developed to calculate the
size of the plastic zone, R, for an applied moment, M, in plane strain bending by equating
internal moments to externally applied moments. The method is based on elastic
solutions of the stress concentration factors of a notch as obtained by Neuber [ 141 and
the plastic solutions developed by Hill[ 151 using slip-line field theory. This new
approach considers the development of the slip-line field in various elastic stress fields
and also ways by which the plastic zone perturbs the elastic stress field. In this manner,
the plastic zone size and the elastic-plastic stress distribution near the notch are
directly related to the applied forces (bending moment) in terms of the yield stress and
geometrical parameters. These theoretical predictions are found to be in very good
agreement with experimentally measured values of the plastic zone size in high nitrogen
steel.
Having considered the quantitative relation between uF and RF, which is a macro-
scopic problem, it is then of some interest to understand the physical significance of RF
in terms of the microscopic processes of fracture that are occurring near the crack tip.
In certain materials, particularly BCC metals tested at relatively low temperatures,
 Elastic-plastic stress distributionand fracturestrength 193

fracture occurs when au unstable microcrack (or confi~~tion of microcracks) is

nucleated ahead of the crack tip, accelerates, and grows into a fast nmning macrocrack.
In certain instances, this occurs when the maximum longitudinal stresses ahead of the
crack, a~~~, build up to some critical value, a,_* For perfectly plastic solids loaded in
plane strain, Hill’s slip-line field solution indicates that u$‘#=increases with increasing
plastic zone size R. Consequently, RF is the value of R at which a”yy”reaches the
microscopic fracture strength af*.
In the second part of this paper, theoretical predictions of fracture behavior based
on the relation between applied bending moment and R, and R and o;mvtrax, are used to
determine the relation between macroscopic and microscopic fracture strengths UPand
a/*. These predictions are then compared with experimental data obtained from tests
on notched bars of mild steel at cryogenic temperatures.
2. THE STRESS DEVRIWTION NEAR THE NOTCH ROOT IN A NON-STRAIN
HARDENING SOLID
(a) Elastic defor~ut~o~
Elastic solutions of the stress distribution in notched bars indicate that the longi-
tudinal stress o,, is a maximum at the notch tip (X= 0, Fig. 1) and that v,, decreases as
the inverse square root of the distance x from the root. At the notch tip
ali,L- *Inax= KG., (x = 0) (1)
where K, is the elastic stress concentration factor and aiv is the nominal (net) stress.

Fig. 1.Gcometricai parametersused to define the shape of a notched bar.

Using the fact that the stress gradient at the notch root is given by
du
yY= &lax
dx
X= 0 (2)
P
as predicted by Neuber[l4], Weiss[l6] has shown that ahead of the root (X > 0) olr,,
hastheform \ i
\
uv,( 1 = Kc7*m(x)
1 J((,+“,,>B<x < $(K:-1)
where p is the root radius. In bending, the stress gradient is slightly steeper because
aLvitself
decreases linearly with x as

(4)
where a is the ligament depth.

EFY-hi
194 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

The Iocal stresses near the tip of an ‘irately’ sharp notch (i.e. a crack) may also be
defined by the stress intensity factor KI

(5)
Bueckner [ 1‘I] and Gross [ 181 have calculated K, values from the applied moment M
and find that

KI = 6$+J(cld) Bueckner [ 171 (6a)

&=2M
-&31’7(c/d) -~*8(c/d)*~Zll(c~d)3]i’2 Gross I:181 @b)

where c is the notch depth and d = a + c is the specimen depth. Equations (aa) and (6b)
give equivalent values of Kf for equivalent specimen geometries.
The elastic stress concentration factor K, is in turn determined by the geometry of
the notch depth c, radius p, flank angle o, the ligament depth a and the ratio (c/d), Fig. 1.
For a shallow notch (c/d small) K, depends only on c and p and a commonly used
approximation for c/p large is
K, = Ksb,mow= 1+ 2~/(hd. (7)
As the notch is deepened, the local stress field becomes increasingly influenced by
the free surface opposite the notch and, as c/d + 1,
Kc --+ Kiiw =fb, ~1. V-3)
Neuber[l4, 191 has shown that the stress concentration factor for a notch of
arbitrary depth may be approximated by averaging the two limiting cases represented
by (7) and (8) such that

(9)

He devised a simple geometrical circle construction to perform this averaging in the

form of Nomograms which were used in the present work. Following his earlier analy-
sis [ 191, Neuber extended the theory to incorporate the effects of flank angle o on
K,[ 141.The effects of relative notch depth (c/d) and flank angle on K, are illustrated in
Fig. 2 for notches with constant radius p = 0.25 mm in a 10 mm deep specimen. The
ideal shallow notch case given by (7) is also plotted; note that it is only valid for ex-
tremely shallow notches. K, increases with increasing depth (but less rapidly than in
the ideal case) reaching a maximum at c/d = 0.3 and thereafter decreases to unity as
c/d --+ 1. The effect of reduced flank angle is to increase K, and sharpen the peak, the
maximum still occurring at c/d = 0.3.
Values of K, for notches with constant radii @ = O-25mm) and flank angle (w = 36”)
in bars of varying depths lo,20 and 30 mm are presented in Fig. 3. Both c and c/d are
required to define the depth of a notch; therefore, a notch of given depth c will be rel-
atively shallower and thus more severe in a larger specimen (K, -+ Kshaw). Conse-
quently, there is an ‘elastic’ size effect which in~uences the magnitude of the elastic
stress field as calculated by Bueckner [ 171and Gross [ 181for sharp cracks (aa) and (6b).
 Elastic-plastic stress distribution and fracture strength 195

p - 0.25mm

c/d

Fig. 2. Elastic stress concentration factor, K,, as a function of notch depth, cld, for various
tlank angles o (atIer Neuber[ 141).d = 10 mm; p = 0.25 mm.

(b) Elastic-plastic deformation

When a notched bar is bent, it will deform elastically until the maximum longitudinal
tensile stress developed at the root, is equal to the tensile yield stress o,,,. Consequently,
the criterion for ‘local’ yielding is
ouu = a& = UM (10)
where uyr is evaluated at the appropriate strain rate. (A Tresca yield criterion is used
throughout the paper.) At this stage, a small plastic zone will appear at the notch root;
further bending will cause the zone to spread into the ligament so that the plastic zone
size R increases. Under plane strain conditions, the zone has the form predicted by
slip-line field theory (Fig. 4a) as shown in Fig. 4b.
For rigid perfectly plastic materials, the longitudinal stress along the net section
within rhe plastic zone (x d R), is given by Hill as

The increase in oyl) with increasing x results from the build-up of transverse stress a,,
and art below the root.
 196 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

8.0

6.0

1.0 C/d
0 0.5 1.0
Fig. 3. Elastic stress concentration factor rC, as a function of notch depth, cld, for various
specimen sizes d (after Neuber[ 141). p = O-2.5mm; o = 36O.

0 -a-_--_

ffw
P” (1 +ln(l+x/pfl

oytl*In(liqg/P)]=o~~l+a/2-u/2]

I
M
a/2_&

~/2-----_-
- - - _ --
LM

aii X

Fig. 4a Schematic elastic-plastic stress distribution for a notched bar in plane strain bending.

The maximum longitudinal stress (a?; ) occurs at the elastioplastic interface (x = R)

so that OFF increases with increasing plastic zone size. The parameter inside the square
brackets in (12) is a measure of the triaxiality and is known as the plusric stress
Fig. 4b. Fry’s etch pattern of a typical
plastic zone.

[Facingpage l%l
 Elastic-plastic stress distribution and fracture strength 197

concentration factor I(o(P)

K&j= [l+ln(l+R/p)]. (13)

Equations (11) and (12) are only valid, while the plastic zone remains within the points
of tangency between the notch root and the flanks. Beyond this stage, indicated by the
shaded region in Fig. 4a, the maximum stress remains constant at its maximum value,
which depends only on the flank angle.

By combining (12) and (14), we observe that this occurs when

l+ln (l+R/p)= [l+T-F]

or when

R=R,=p(exp(y)-1). UW

Consequently, the total stress distribution inside the plastic zone is given by (11) for
x < RB and by (14)for x > RB (Fig. 4a). For x > R, the stresses are elastic. They
decrease sharply ahead of the elastic-plastic interface as shown in Fig. 4a. In the com-
pression region, they are assumed to vary linearly with x from zero at x = a/2- Ax to a
maximum value o(a) at the surface x = a.

L
/- on = K,,(x). oy(x)
\

Fig. 4c. Schematic diagram to illustrate the effect of a plastic zone on the elastic stress distribu-
tion. The dashed curve represents the completely elastic distribution and the solid line the
elastic-plastic distribution for rlre same applied moment in a material with a lower yield stress.
Note the increase in stress gradient at x = R due to plastic deformation, i.e. K, (RI > K,, (xl.
198 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

3. THE QU~A~ RELATION BE~EEN PLASTIC ZONE SIZE R AND

STRESS INTENSITY FACTOR K, FOR PLANE STRAIN BEND LOADING
As the applied bending moment h4 and stress intensity factor KI increase, the plastic
zone size R will increase also. Prior to the present work, there has been no quantitative
treatment of the relation between R and K, for plane strain deformation, although some
approximate relations have been proposed [8].
The method used here is based upon the fact that the stress distribution beneath the
notch root is a unique function of the plastic zone size, according to (11) and (14).
Consequently, once the relation between u,, and K, is determined, the relation between
R and K, is determined also. The relation between a,, and K, is determined by the
condition that the sum of rhe internal bending moments resulting from the ~ongitudinaf
stress distribution u,,(x) must be equal to the applied bending moment.
The elastic-plastic stress dist~bution across the minimum section is shown
schematically in Fig. 4a. The distribution of stresses within the plastic region is given
exactly by (I 1) for x c RB and by (14) for x > RB. The distribution of elastic stresses on
the remaining section is unknown. For the purposes of this calculation, they are
assigned a specific distribution which is considered realistic on the basis of a photo-
elastic analysis [33]. The position of the neutral axis is not at a/2 but occurs closer to the
notch root at some position (a/2 -Ax) where AXwill be a variable. The distribution of
stresses in the compression region is assumed linear. Using this distribution, the total
moment is calculated by superimposing the moments due to the elastic and plastic
regions of the specimen.
(1) The elastic foment
For the purposes of calculation, we first remove the plastic region from the distri-
bution shown in Fig. 4a, leaving an elastic stress distribution shown in Fig. 5. This
unknown distribution results from the presence of a ‘hypothetical notch’ of radius p*(R)
and depth c+R. Consequently, from (1) and (2), the stress gradient at the ‘notch tip’
(x = R) is given by
du -2s: * K: (R I_ -2amax (16)
Z&R==
I P*(R) p*(R)
where K,*(R) is the elastic stress concentration factor of the hypothetical notch,
evaluated at its ‘tip’ (x = R). uf is the nominal stress required to produce this elastic
stress ~st~bution and thus

M* -_ 2(a-R)2 (17)

is the applied bending moment per unit thickness required to produce it.
Since the stresses are continuous at the elastic-plastic interface, the maximum
stress at the elastic-plastic interface is also given by (11) or (14). Therefore,
-2u.v* K,*(R) =---2dl+W1+RIdl tR < RBj (18)
p*(R) p*(R)
and substituting into (17) gives
[ 1 + In (1 + RI&] (a - R)*
UY~
M*= (R < RB) UfW
6K,*(R)
 Elastic-plastic stress distribution and fracture strength 199

(I?)* cr,[t* ln(l*fvp)]

o-R

O/~-AX---_---_-_ \
M*

Fig. 5. The elasticportion of the total stress distn’bution (x > R) similar to Fig. 4a but with
the p&tic zone removed, hence (r l (LX) < u(a).

Similarly, for R > RB

It should be noted, at this point, that the relation between M* and R can be deter-
mined in this manner without knowing the exactform offhe elusric stress ~jsfr~&~~iu~,
provided that the function &*(R) can be determined.
In the fully elastic condition, the stress distribution in front of the actual notch is
well approximated by (3) so that the elastic stress concentration factor at some point x
ahead of the notch K,(x) is

In the limit as R 3 0, K,*(R) --* K,(x) and hence for small values of R

11+ 4Rlp]-“*
KY*(R) = KY (1 _2Rja) - (2la)

However, as the plastic zone develops, this relationship will no longer hold. Consider a
notched bar under an applied moment M which is sufficient to produce a plastic zone
of depth R, Fig. 4c. If the specimen had responded in a completely elastic manner, then
the stress distribution would be given by the dashed curve. However, for the same
moment, the real distribution is the elastic-plastic one represented by the solid curve.
This distribution is governed by the balance of forces and the equivalence of moments
such that the elastic stress at the elastic-plastic interface (X= R) will always be greater
than for that predicted for the same moment but assuming the specimen is completely
200 T. R. WILSHAW. C. A. RAU and A. S. TETELMAN

elastic. Therefore, the stress concentration factor K,(R) decreases less rapidly for the
elastic-plastic case than K,(x) which is predicted by (2 la.)
Hence,

KM < R,*(R)
since
M*a”
KJR) = K,*(R) @lb)
M(a - RY ( 1 - 2R/a)

and for R Q a, K,,(R) = K,*(R). In addition, the deviation from the fall-off predicted
by (21a) will increase as the plastic zone size increases. With these facts in mind we
chose to represent the function K,(R) through K,*(R) by an expression

K,*(R) = K,[ 1+4R/p]-““ * (1 _ iR, ). (22)

(I.

This function is plotted in Fig. 6 for various values of g. The value of g = 2 represents
the fully elastic state, and values of g > 2, elastic-plastic states.

DISTANCE BEL(M NOTCH

3.0
I b R/p w x/p

-I/g
K:(R)=Ko[l*4R/p] 1
(I -2R/a)

p =‘0.25mm
a =8mm

) ELASTIC
- Fl_ASTIC

9’2 - ELASTIC

Fig. 6. Katie of the effective elastic stress concentration factor, K,(x) or K,*(R) at any point
below the root to that at the root in the fully elastic state, K,, as a function of the distance
below the root x/p (elastic) or R/p (elastic-plastic) for various values of g. for a = 8 nun;
p = O-25 mm.

2. The plastic moment

The second stage in this calculation of the total elastic-plastic bending moment is to
replace the plastic region which was removed to calculate M*. In order to maintain
equilibrium of forces, the tensile forces in the plastic region must be balanced by an
 Elastic-plastic stress distribution and fracture strength 201

equivalent force in the compression region. This final distribution is shown in Fig. 7.
For simplicity, it is assumed that the forces are redistributed as shown in the shaded

O/~-AX- ---

Fig. 7. The complete elastic-plastic stress distribution obtained by replacing the plastic zone
to Fig. 5. The shaded areas represent equivalent force, i.e. F, + Ft = F..

area and that there is no redistribution within the elastic tensile region. The maximum
compressive stress is now a(a) and has increased from a*(a) such that

l/2 [a(a) - a*(a)] (a/2 + Ax) = F8 = - (F, + F,) (23)

where
F, = ib”’ uW[lfln(l+x/p)]dx
and
I;;= (R-R,)a,[l+ln(l+RB/p)]

for R < Rgl RBis replaced by R so that Fz = 0.

The rorulmoment of the elastic-plastic dis~bution is, of course, equal to the applied
moment M required to produce it. M is simply M* plus the couple generated by the
forces F1, Fz and Fd,and is given by

M=M* G-; (a/2+Ax)} (24)

where R is the center of force of the logarithmic slip region (11).

Using (19), (22), and (23), (24) may be rearrangedto relate the plastic zone size,
R/p, and the maximum local stresses (&,+r& through (I 1) and (14) to the applied
moment, M/o,, (K,/v& through (6a) and (6b). This relationship depends upon geo-
metrical parameters, which define the specimen geometry and two, as yet, unspecified
parameters, Ax and l&,*(R). The relationship between the applied moment and the
plastic zone size was computed using various values of Ax in the realistic range
0 s Ax c a/6.
These results showed that Ax does not significantly influence (i.e. less than 2 per cent)
202 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

the magnitude of Mjrr, and hence an exact knowledge of the position of the neutral
axis is not critical in the present analysis. Consequentiy, I&r*(R) is the only unspecified
parameter in the theoretical calculation of the relation between plastic zone size and
applied moment. Figure 8 shows the computed variation of R/p with KJu, (using
(6)) or N/u,,, for various values of p and the parameter g which appears in (22). One
notes that the plastic zone size increases most rapidly with applied stress as the
radius of the notch decreases, and that the curves are sensitive to small changes in g,
the sensitivity increasing with R/p.

12

J
CT

I I I
0 01 02 0.3 04 05 06 07

Fig. 8. The theoretically predicted plastic zone size, R/p, as a function of applied bending
moment, M/u,, for various root radii, p, using either (i) g = 3 dashed curves: (ii) g = 2.5 solid
curves. Other parameters; D = 8 mm; c = 2 mm; w = 45”.

In order to obtain the value of g, notched specimens of high nitrogen steel having
varying geometries? were deformed at room temperature and the plastic zone size
was measured (by etch pitting) as a function of applied load. Figure 9 shows the varia-
tion of R/p with M/u, and K,fcr, for 2 mm deep notches of varying radii while Fig. 10
shows the variation of R/p with Mlo, for notches having constant (0.25 mm) radius but
varying notch depth. Within the range of logarithmic slip, (R -C Rg, R/p < 2-25 for
W= n/4) very good agreement between theory and experiment is noted for a value of
g = 2-s.
In Fig. 9, it is apparent that significant deviations from the theoretical curves occur
at smaller values of R/p as the root radius @/a, p/f) increases. For r = 0*25,0-50 and
0.75 mm the deviation begins at R/p = 2*0,1-75 and l-5, respectively. These devia-
tions are too large to be accounted for by changes in K,,*(R) and therefore must result
from a breakdown of the triaxial stress state. This breakdown of triaxiahty results from
relaxation of the utZ stresses [f(RIp, R/a)] and/or a,, stresses [f(R/p, R/f)].
tSee Appendix for details of the experimental observations.
 Elastic-plastic stress distribution and fracture strength 203

“iF
o.Oci2
0.005

0.010
0.020
0.030

/
t
-----TtiEORETUL (g* 2.5)

t
0 4
/
l
f /I //
/
t //
/
/
/ / /I
I /

GEMS&
YIELD
M/u, rn! x I$,

0 K) 20 30

KI (,a) 10, in”’

I I I I I I .
0 0.1 0.2 0.3 0.4 0.5 ’ 0.6 0.7

Fig. 9. The experimentally measured plastic zone size, R/p, as a function of applied moment,
M/u_, for various root radii p. The theoretical values computed using g = 2.5 as represented
by the dashed curves. a = 8 mm; c = 2 mm; o = 45”; p-varied.

3.0 -

-----TH.dETKAL (q= 2.5)

2.0 -

0
1 I
0. I 0.2
I I
0.3
I
0.4
I
0.S
I
0.6
I
0.7
K Lp) /u,
*
in.“*

Fig. IO. Comparison of the theoretically predicted and experimentally measured plastic zone
size, R/p. as a function of applied moment, M/u,. for various notch depths c. The theoretical
values were computed using g = 2-S and represented by the dashed curves. p = 0.25 mm;
d = 10 mm; o = 45’; c-varied.
204 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

4. QUANTITATIVE DETERMINATION OF THE FRACTURE LOAD AND K,,

VALUE FOR MILD STEEL AT LOW TEMPERATURE, AS A FUNCTION OF
ROOT RADIUS
There is now su~c~ent evidence to indicate that low tempe~ture brittle fracture
in notched bars of mild steel occurs when the maximum longitudinal stresses below
the root, *ax, build UP to a critical value, v,* [21-231. Since a,* is determined by the
microstructure, primarily gram size and the distribution of secohd phases, it is a
‘microscopic fracture strength’ and hence it is a material parameter, independent of
specimen geometry. From (12), we can write the criterion for microscopic and hence
total (macroscopic) fracture as
~ax=ai*=a,[l+ln(I+R/p)]
YY (RF < RB) (25)
where RF is the critical plastic zone size required to achieve the plastic stress con-
centration necessary for fracture initiation beneath the root. Rearranging this equation
gives
RF
-=exp[a,*/a,-13-l.
P

On a macroscopic scale, the fracture moment MP.and the fracture toughness for
a notch of finite radius, K&), are those values of M and Kr at which R/p = RF/p,
according to (24) and Fig. 8.
To investigate the validity of this approach, mild steel specimens of standard
Charpy V notch geometry? were fractured under a constant crosshead velocity of
0.1 in./min at temperatures ranging from -145 to -l%“C and the fracture loads were
recorded (Table 1). Tensile tests were also performed in the same temperature range
to obtain values of cm, at a strain rate (8 X 10W31sec)
corresponding to that which was
estimated to exist beneath the notch root. The experimental values of M,/a,, are
listed in the fourth column of Table 1. From Fig. 8, it is, in turn, possible to obtain
values of R,Jp for the O.OlOin. Charpy notch using the experimentally determined
value of g = 2.5. These values are given in the fifth column of Table 1; the sixth
column predicts values of Ka(p)r= a,*/~,,, according to (13). Finally, from the values of
out obtained at each temperature, Us* is obtained at each temperature. It is interesting
to note that over a wide range of temperature, cr,* remains remarkably constant at
195A 3 ksi. Consequently, the increase in PF, MF and K,,(p) with increasing temper-
ature does not result from an increase in or* with temperature, but rather from a
decrease into vu. with increasing temperature.
The results of a series of fracture tests, (on specimens of a different batch of
similar material) within the temperature range -1% to -17o”C, for notch radii varying
between 0,002 and 0,050 in. and constant notch depth = 2 mm, are presented in Fig. il.
Values of MF/cTYwere calculated from these fracture loads and the measured yield
stresses, for three temperatures, -196, -180 and -170°C. The associated critical zone
sizes for fracture, RF/p, were then obtained from the theoretical curves in Fig. 8. The
loci of these plastic zones were plotted on the zone size vs. moment curves for the six
radii, for each of the three temperatures in Fig, 12.
For p 2 O*OlO in., R,,/p was found independent of p $ and fracture occurs at all

tSee Appendix I for all experimentaldetails.

*For the largest values of p the results indicate that RF/p and hence o,* decreases with increasing p.
This phenomenon is not pertinent to the present discussion of sharp notches.
 Table 1. Fracture resulta

Fmcturet
$
lOad Yield strcssS
Temperature pF &lo;. K crw,= [I+hU+R/p)] Q=u~,K,,~) K,c( PI
9
(“C) (lb) iz, (in*X I@) RIP (ksi) (ksi (in.)““)
-150 2450 89.5 274 3.50 3
2.18 195 48.0
-160 2000 95.0 21.0 2-00 191 40.0 3
2.11
-170 lsoo loo*0 18.0 I.60 195 36.0 E.
1.95
-180 1650 107.5 15.4 1.25 1.81 197 33.0
-1% 1600 ll!bO 14.0 1.00 1.69 194 32.0 i
-1% 157s 120*0 13.2 0.84 1.64
197 31.5 8

tD&cctiou rate O-1 in./min. ;

SStrain rate 8 x lo-” SW-~.

_ _. _ .._._
206 T. R. WILSHAW. C. A. RAU and A. S. TETELMAN

three temperatures when the same critical tensile stress, v,*, is attained beneath the
notch root. However, for p 4 O-005 in. there is a very pronounced increase in RF/pas
p decreases. The specimens were ‘stronger’ than expected and there is evidently no
simple fracture criterion. Further evidence of this root radius size effect will be presented
below.
FRACTURE LOAD
A lb

x)0()-

o-
-200 -190 -180 -170 -60 ‘C
TEMPERATURE
Fig. 11. The variation of fracture load with temperature for specimens of varying radii
(c = 2 mm).

p=O.CCQin
-------ntEORETlcAL ga2.5
,

p =O.O05in

Fig. 12. Experimental values of M+./a,vs. theoretical RF/p for various notch root radii at
three different temperatures.
 Elastic-plastic stress distribution and fracture strength 207

At a given temperature (i.e. a given u,*/w~ and hence a critical (&Jo) ) the root
radius p will exert a large effect on MF and K,(p). This is apparent from Fig. 8, which
shows that the M/Q ratio required to achieve a given Rip decreases as p decreases.
Using either (6a) or (6b), it is possible to relate R/p to &(~)/a, for various values of
p, as shown in Fig. 8. As discussed above, fracture occurs when R = RF or K, ( p) =
l&(p). In the limit, as p -B 0, K,=(p) --+ Klc. which is the plane strain toughness
evaluated by sharp crack fracture mechanics.
The variation of fracture toughness KXp) with root radius was theoretically pre-
dicted at one temperature, -l%“C, using the measured values of u, and or*. (The
latter value was determined using specimens with p = O-010in.). Values of K,&) were
calculated by first evaluating R&p from 25 and then K,,(p)/u, from Fig. 8. These
results are presented in Fig 13 as a solid straight line and it is shown that K&) is
directly propo~ion~ to the square root of the radius. The actual experimental values of
K,&) are also presented on the same figure as the open circles and include the results
of fracture tests on notched bars which had been fatigued. These latter results represent
the limiting case of a sharp notch, i.e. a crack.
The critical maximum stress criterion of fracture represented by the straight line in
Fig. 13 predicts that K,,(p) + 0 as p + 0. However, the experimentally determined

Krc(PI
l/2
KSI-IN

L I f I I I b
0 .oo2 005 .a0 020 .Q3o P !N
Fig. 13. Variation of the critical stress intensity Klr( pf with the square root of the notch
radius (pf at -l%“c. The straight line was calculated from theory using g = 2.5 and the
experimentally determined values of a,* and CT,. The open circles are the experimental points.
208 T. R. WILSHAW, C. A. RAU and A. S. TETELMAN

value of Kre = 215 ksi (in.)ln using fatigued specimens in the present work and K,,
measurements on similar materials[25-271 shows that K,, does not approach zero for
sharp cracks. In fact, specimens containing sharp cracks are as strong as those contain-
ing sharp notches of finite root radii = O+IO2in. Other workers[4, 24, 341 have also
observed this phenomenon, that there is soye critical value, p = p,,, below which K,,(p)
is independent of p and achieves its minimum, K,,.
Cottrell[32] calls p. the effective limiting sharpness, which governs the scale of
microscopic size effects at sharp notches and natural cracks. p. will be determined by
some microstructural feature such as grain size or spacing of inclusions or slip or twin
bands. For fracture to occur, it is apparently insufficient to attain a critical tensile stress
u,* at a point within the specimen, but in addition, the stress must act ooer a finire
region. As the notch radius decreases, the stress gradient increases, and it is this gradient
coupled with the microscopic parameters, which will control the magnitude of this
microscopic size effect.
The model predicts that K&I) is proportional to d(p) for a given value of RF/p.
Well below the ductile-brittle transition, brittle fracture in mild steel occurs when the
critical tensile stress a,* is reached, since the first microcrack which forms propagates
unstably; RF/p is then given by (26). This is also the case for refractory metals, beryl-
lium and ionic and ceramic materials fractured below the ductile-brittle transition.
Within the transition range, stable microcracks are formed in mild steel and final
fracture occurs when they join together to form an unstable crack. Similarly for high
strength steels, ahuninum, and titanium alloys, unstable fracture occurs by the coales-
cence of voids. In both of these cases, the concept of a critical tensile stress criterion
for fracture has no physical significance since McClintock[35] has shown that in-
stability will occur when a critical strain, q*, is achieved at the notch root.
Cottrell[4] has shown that the crack opening displacement 2V(c) is approximately
equal to 2pe, where E is the strain at the notch tip. Consequently, if fracture occurs
when E = E,*, there will also be a critical value of V(c)/p = V*(c)/p and hence again a
critical RF/p, since V(c) is proportional to R. Based on this model, we would expect
that for p > po. K,, is proportional to d(p) for low energy tear fracture as well as for
cleavage. This dependence has, in fact, been observed experimentally for an H-l 1 steel
[34] anda aluminum alloy[36].
One limitation on the model should be noted. Since the model is based on a perfectly
plastic material, it becomes less realistic with increasing R/p as the strain hardening
rate of the material increases.
The present work has enabled a quantitative description of the dependenc,e of K,,
on both the root radius and the plastic zone size, thereby providing some physical signif-
icance to these parameters and others which have hitherto not been possible. In this
respect, it has been shown quantitatively how K,, depends on both a,* and a,, and how
the effects of changes in u,* and vu, may be evaluated independently. For example,
both neutron irradiation and certain tempering conditions (‘temper embrittlement’)
produce a decrease in K,, for medium strength steels. The irradiation embrittlement
results from an increase in my,, while u,* remains constantIS, 271, whereas the temper
embrittlement results from a decrease in a,* while uy, remains constantI301. Using the
model described here, these changes can be measured quantitatively using standard
Charpy specimens in slow bending, for example, or by instrumenting the standard
Charpy test to record fracture loads. If tests are run on a series of specimens containing
various root radii, then curves similar to Figs. 11 and 12 can be plotted and K,, can be
 Elastic-plastic stress distribution and fracture strength 209

obtained by extrapolation, if p. is known. Similarly, the effect of alloy additions and

thermomechanical treatment on fracture toughness can now be evaluated quantitatively
in terms of the variation of o,* and a,,{31]. This approach has recently been used to
determine the reason that nickel additions improve the toughness of low carbon steel.

5. SUMMARY AND CONCLUSIONS

theoretical model for predicting elastic-plastic, plane strain stress distribution
1. A
and the plastic zone size in notched bars under bend loading has been developed. The
basis for the solution is that the applied bending moment is equal to the internal bending
moment resulting from the longitudinal stress distribution ahead of the notch.
2. Plastic zone sizes were ex~~ment~ly determined in notched bars of high
nitrogen, mild steel, having various root radii and notch depths. The experimental
values were found to be in very good agreement with the theoretical predictions for
cases where the defo~ation is completely plane strain.
3. Whether unstable fracture initiates when the longitudinal stress ahead of the
notch builds up to a critical value o,*, or when a critical strain is attained at the notch
root, it is shown theoretically that this implies the existence of a critical ratio of plastic
zone size to root radius RF/p, and hence a critical value of fracture moment MF and
fracture toughness K,,(p) are relevent fracture criteria.
4. The fracture loads of notched bars of the mild steel were measured over a range
of temperature from -1% to -150°C. .At all temperatures, the fracture loads were in
excellent agreement with the theoretical calculations and indicate that (rr* = 195+ 3 ksi.
5. The theoretical model predicts that for mild steel and other materials where
fracture initiates at a critical value of the longitudinal stress, K,, (p) is proportional to
V(P) forp > po- Porp s P~JG&) = &,.
6. The theoretical model shows how a rational, physical significance can be attached
to the parameter K,, for those cases where fracture is initiated at a critical stress.
~c~~o~~~~ge~enr~-~e authors wish to express their )patitude to Ltoyd Kaechelc and Dave Harris for
many stimulating discussions. We would also like to acknowledge the financial support of the U.S. Army
Research Office, Durham, throughout the course of this work under Contract #DA3 I- 124ARO(D)25 I and
the Advanced Research Projects Agency for support through the CMR programs at Stanford University.

REFERIENCES
[ 1] A. A. Wells, ~ru~ei~Symp. on Crud Propagation, College of Aeronautics, Cranlield, England (1962).
f2] G. T. Hahn and A. R. Rosenfield Acta Met. 13,293 (1965).
[3] B. A. Bilby, A. H. Cottrcll and K. H. Swinden, Proc. R. Sot. A272.304 (I 963).
[4] A. H. Cottrell, Proc. R. Sot. A285; 34 (1965).
[S] A. S. Tetelman and A. J. McEvily, Fracture U~St~~ctffrffi Moteriais. Wiley, New York (1967).
I61 J. E. Srawky and W. F. Brown, Fracture Toughness Testing, A.S.T.M. Spec. Tech. Pubi. No. 381,
p. 133 (1965).
[7] D. S. DugdaIe,J. Mech. Phys Solids 8,100 (1960).
i81 F. A. McCIintock and G. R. Irwin, Fracture Toughness Testing, A.S.T.M. Spec. Tech. Publ. No. 38 1,
p. 84 (1965).
[9] N. P. de G. Allen and R. Southwell, Phil. Trans. R. Sot. A242,379 (1949).
[lo] J. A. Hendrickson, D. S. Wood and D. S. Clark, Trans. A. S. M. 51.629 (1959).
[l l] F. W. Barton and W. J. Hall U. S. Dept. Commerce Ship Structure Committee, Rep. No. SSC-147
(1963).
[ 121 N. P. Allen, C. C. Earley and J. H. Rendall, Proc. R. Sot. A285,120 (1965).
[ 131 T. R. Wilshaw and P. L. Pratt,.!. Me& Phys. Solids 14. I, 7 (I 966).
f14] H, Neuber, Theory of Notched Stress; Principles for Exact Calculation of Strength with Reference to
Structural Form and Material. A.E.C. Teclr. Rep. No. 4547 (1958).
[ 151 R. Hill, Mothemoticol Theory OfPlasticity. Oxford University Press, London (1950).

EFM-N
2fO T. R. WILSHAW, C. A. RAU aad A. S. TETELMAN

D61 V. Wds8,AS.M.E. puper No. 62.WA-270 (1962).

4171 H. F. Rucckmz .~rdercllcedbyD.H.WinaeaadB.M.Wundt,Trunr.AS.M.E.80,~613(1958).
I181 A. P. Gm, U~MisM workrefercuccd by J. E. Stawky aud W. F. Browu, in/6).
[ 191 H. Ntub~r,TIvoryofNotcliSfresses. David Taybr o&l Basin Tramtation (1946).
I201 V. Weirs and J. G. Suskr.A.5.T.M.&WC. Tech. Pu F1. No. 302, PP. 3-20 (f%2).
1211 T. R. Wilshaw aud P. L. Pratt, int. Fmcrure Co&. Sendai. Japan (1%5).
[22] J. F. Knott,Proc. Ry. Sot. A315.150 (1965).
1231C. A. Rau,Jr. andA. S.Tctclman,To be pubWed.
1241A. K.SkKMLer,Frectuntouebnejstesti~AS.T.M.Spec. Tech.Publ.No. 381,p.26(1%5).
[25] A. K. !&maker, Trw. AS.M.E. Paper No. 66.WA/Met. 8 (1966).
[26) E. T. WCSSCLW. G. Clark and W. K. Wilson.Westinghouse Rep. to the U.S.Army Tank-Auromotioe
Center, Warren, Mich. Contr. No. DA-30.069-AMC-602Cr, (1966).
1271 G. T. Hahn aud A. R. Rosenheld. Progr. Rep. on Project No. SR-164(1966).
{ZS] 0.Gat.ca.J. ImnSree~lntr. M6,991(1966).
1291 J. G. Y. Chow, S. 8. McRickard and D. H. Gwius&y, Radiation effects on metals and ucutron &simctry.
AS.T.M.Spec. 7ech.Publ. No. 34i,p. 162(1%3).
1301 F. L. Cam, J. Nunes and F. R. Larson,To be published.
(3 11 D. Wuliacrt and A. S. Tetelmn, To be gutbiishcd.
1321 A. H. Cottrclt, The Mechanical Properties qfhfatter. Wiky, New York (1963).
[33] C. A. Rau. Jr., To be Published.
1341 G. Irwin.Appi. Muter. Res. 3.2, (1964).
[35] F. A. McCliitock, To be Published.
[36] J. H. Hulherin, D. F. Atmiento and H. Markus,.!. Basic Engng. 709-717 (1964).

APPENDIX I
Experimental procedure
All the experiments on notched bars were performed on specimens which were
basically rectangular prismatic bars 10 X 10 X 60 mm ~n~~~ 45”notches of varying
depths and radii. They were machined from hot rolled plates of high nitrogen mild
steel. with a nominal chemical composition.

C Mn Si P S AI N*
0.14 O-24 O-048 co*01 0.020 <oXlo 0,013

This material was selected because of its macroscopic etch pitting properties. Follow-
ing machining, all specimens were annealed in vacua at 850°C for one hour to produce
a grain size of 30-40 & All notched tests were performed in three-Point bending and
under constant crosshead velocities.
Specimens with various notches were deformed (as above) up to various values of
A4!oy at room temperature. Following defo~ation, they were aged at 150X! for 2 l-us
to decorate the plastically deformed regions, cut along the Ed-lon~tu~n~ section
and finally etched in Frys’ reagent to reveal the plastically deformed regions. A typical
plastic zone is shown in Fig. 4b. The sizes of the zones were measured using the
Vernier scale on a travelling stage microscope to an accuracy of *@OS mm.
Specimens of the standard Charpy V-notch geometry, (e.g. 2 mm deep with varying
root radii were fractured under a constant crosshead velocity of O-1in./min within the
temperature range -145 to -196°C. Those temperatures were attained using the liquid
nitrogen meniscus technique [33].
Uniaxial tensile tests were carried out at room temperature at three strain rates,
3+3x lo-‘, 3.3 x 10msand 3.2 X lo-* set+. The lower yield stresses were measured and
used to calculate the M/u, values shown in Figs. 9 and 10. At room temperature, this
material exhibited a 2-3 per cent Liiders strain. Tensile tests were also performed in
 Elastic-ph~~tic stress distribution and fracture strength 211

the same temperature range as the fracture tests at a strain rate of 8 X 1OwJ
set-’ to
measure up, so that Mp/uyl and a,* could be calculated (Table 1).

(Received 29 March 1967)

PkDlJ - Un mu&k th&orique permettant de p&fire ks distriitions de tension pour une contraiote plane
l’&at Clastique-plastique sur da barres B eotailkrs. sous une force de pliage. a Cti &vebppC en combinant
les solutions Clastiques de Neuber avec ks solutions de Hill en champ de glissement lit&ire. Le modtie
sert B cakukr la grandeur de la zone plastique en tant que fonctioo de la charge exteme appliqu& (moment),
de la tension Iimite u,,, et des param&tres g6omdtriques tels que le rayon de la racine. la profondeur de
I’entaille et la profondeur du ligament. Les pr&ictions th&uiques s’accordent t&s bin avec ks valeurs
exp4rimentak.s de la grandeur de la zone plastique dans Its barres B entailles en acier B haute teneur de
nitroghre.
La r&stance B la rupture des barres it eotailles ayant ditHrentes catactCristiques gCom&iques, a Ctt?
mesur6e B des temperatures cryogeniques et la vakur de K,, &term&e. II est mot& que la vaktir est
conforme B uo crithe de rupture bask sur l’obtention dune force de tension critique of* pour un petit volume
en face de l’entailk. La rupture instabk se produit quand ks xones plastiques se sont &endues B une distance
critique telk que k niveau maximum de la force de tension darts la xoue plastique s’Clbve de CT,B u,*. Le
mod&le rattache done la r6sistance des entailks sur des barres comportant des encocbes var%es, aux
caractCristiques intrin.&ques de rupture du matCriau. On trouve qu’il y a un rayon d’environ 0.05 mm a la
racine qui commande la dun& de la rupture des Cprouvettes comportant des entailks pointues et des
cntquelures.

mm-Durch Kombination der elastischen Liisungen von Neuber mit den Schublinienfeld-
l6sungen von Hill ist ein theoretisches Model1 fiir die Voraussage der Spannungs verteilung im elaslisch-
plastischen fall bei ebenem Dehnungszustand in gekerbten Proben unter Biegelast entwickelt worden. Das
Modell tird zur Berechnung der GtiBe der plastischen Zone in Abhiingigkeit von der aussen angekgten
Last (Moment). der FlieRspa~ung or turd geometrischer Parameter, wk Wurxelradius, Kerbtiefe und
LigarnentGinge. venvendet. Es wird gute fjbereiostimmung zwischen den theoretischeo Voraussagen und
den experimentell bestimmten Werten der griifte der plastischen Zone in gekerbten Proben aus Stahl mit
hohem StickstotTgehalt festgestetlt.
Die Bruchfestigkeit von gekerbteo Proben verschiedener Form wurde bei Tieftemperaturen gemessen
uod der K,, Wert bestimmt. Es wird gezeigt. dass der Wett mit einem Bruchkriterum iibereinstimmt, das auf
dem Erreichen einer kritischen Zugbeanspruchungu,* in einem kkinen volumen vor der Kerbe beruht. Wenn
die plastischen Zonen eine kritische Gri%e erreicben haben, so dass der Htihstwert des Zugspa~ungs-
niveaus in der plastischen Zone von o,,, bis a&u,* erhiiht ist, werden die Brikhe instabil. Das Model1 btingt
daher die reduxierte Festigkeit von Probe0 mit Kerben verschiedener Form in eine Bexiehung zu den
spezifischen Brucheigenschaften des Werkstoffes. Es wurde festgestellt, dass ein wirksamer Minimal-
wunclradius von ca. 0,002 Zoll besteht, der die Bruchfestigkeit von Proben mit schhrferen Kerben und
Kissen bestimmt.