International Journal of Fracture, Vol. 10, No.

3, September 1974
Noordhoff International Publishing- Leyden 305
Printed in The Netherlands

Strain-energy-density factor applied to mixed mode crack problems

G. C. S I H
Professor of Mechanics and Director of the Institute of Fracture and Solid Mechanics, Lehigh University, Bethlehem,
Pennsylvania, 18015, U.S.A.
(Received May 4, 1972; in revised form September 14, 1973)

ABSTRACT
This paper deals with the general problem of crack extension in a combined stress field where a crack can grow in
any arbitrary direction with reference to its original position. In a situation, when both of the stress-intensity factors,
kl, kz are present along the crack front, the crack may spread in any direction in a plane normal to the crack edge
depending on the loading conditions. Preliminary results indicate that the direction of crack growth and fracture
toughness for the mixed problem of Mode I and Mode II are governed by the critical value of the strain-energy-density
factor, Scr. The basic assumption is that crack initiation occurs when the interior minimum of S reaches a critical
value designated S¢,. The strain-energy-density factor S represents the strength of the elastic energy field in the vicinity
of the crack tip which is singular of the order of 1/r where the radial distance r is measured from the crack front. In the
special case of Mode I crack extension So, is related to k~c alone as So, =(x - 1)kx2¢/8~. In general, S takes the quadratic
form a 11k 1 + 2al 2 k lk2 + a2z k z whose critical value is assumed to be a material constant. The analytical predictions
are in good agreement with experimental data on the problem of an inclined crack in plexiglass and aluminum alloy
specimens. The result of this investigation provides a convenient procedure for determining the critical crack size
that a structure will tolerate under mixed mode conditions for a given applied stress.

1. Introduction
The present day version of the theory of fracture mechanics of crack instability is based almost
entirely on the idea of Griffith [1, 2] originated more than fifty years ago. Advances made in the
methods of testing [3] have enabled a successful application of the theory for characterizing
the fracture resistance of the material. This is accomplished through a new factor called the
"fracture toughness" which measures the intrinsic property of the material, i.e., its value should
be independent of the crack geometry and loading. For a given applied stress, the fracture
toughness value should define the largest flaw that a structure can tolerate without causing
fracture.
Past studies on fracture toughness have been concerned mostly with the plane strain measure-
ments of specimens of practical size. For this purpose, it suffices to keep the crack geometry and
loading conditions as simple as possible. Thus, the existing test specimens deal only with cracks
loaded symmetrically so that it suffices to consider the critical value of Mode I stress-intensity
factor kit* alone. In practice, however, the stress state ahead of a crack is often of the mixed
type where both k 1 and k 2 a r e present. The factor k2 designates the amplitude of the crack-tip
stresses owing to skew-symmetric loading. Problems of this type arise in multi-phased materials
such as composites, bridge and aircraft structures. Because of the analytical difficulties as-
sociated with solving the branched crack problem and the lack of a realistic combined mode
fracture criterion, little attention has been given to the mixed mode crack problem.
A basic feature of mixed mode crack extension, which is inherently different from the classical
model of the Griffith crack [1, 2] is that the crack does not extend in a plane coincident with
the original crack. Hence the classical energy balance of Griffith can no longer be carried out
in a simple fashion. For a crack that changes direction, the mathematical details stating the
conversion of work done on the system by the applied stress to the energy required for the
formation of new fracture surfaces are lacking. The familiar expression relating the energy
release rate G 1 to kl for Mode I crack extension given by
re(K+ 1)k2 (1)
G1 - 8/~
* The stress-intensity factor k1¢ used by the author differs from that of ASTM by the number n ½, i.e., ~+ k lc =Klc.

Int. Journ. of Fracture, 10 (1974) 305-321

306 G. C. Sih

is obviously not valid if both kl and k2 were present. Moreover, it would be incorrect to obtain
the mixed mode energy release rate by adding G ~and G 2, being normally taken as zt(x + 1) k~/8#.
This relationship between G2 and k2 is a fictitious one in that it is derived on the assumption
that the direction of Mode I and Mode II crack extension is identical which is contrary to
experimental observations [4-7]. A true value of the mixed mode energy release rate can be
obtained by solving the problem of a bent crack. This presents a formidable task since it
requires the solution to a new class of boundary value problems involving the crack in its
extended configuration*. In equation (1), the elastic constant ~ctakes the value (3 - 4v) for plane
strain and (3 - v)/(1 + v) for generalized plane stress with v being the Poisson's ratio. The shear
modulus # is related to the Young's modulus E as E=2p(1 + v).
In 1960, Sih et al. [8] solved a number of key crack problems involving both k~ and k 2 and
proposed a mixed mode criterion of fracture in the form
f(k~, k2) = fer (2)
which states that the combination of Mode I and Mode II stress-intensity factors present will
cause crack initiation upon reaching some critical value. Although the stage had been set for
studying the mixed mode problem, there lacked a way for finding the functional relationship of
k~ and k 2. As a start, Erdogan and Sih [4] turned to a study of the initial direction of crack
growth in a combined stress field and reexamined the original hypothesis of Griffith that the
crack extends in a direction perpendicular to the local maximum tensile stress. A series of
fracture experiments on plexiglass sheets with an inclined crack was performed and the results
checked reasonably well with the criterion of maximum stress. One of the findings in [4] is that
under Mode II loading crack extension takes place in the direction which is approximately
70 degrees away from the plane of the original crack. Many of the results were later confirmed
by others [5, 6]. However, recent studies [9, 10] have shown that data on the direction of crack
initiation can vary substantially depending on a critical distance from the crack tip where
measurements were made. This distance is strongly influenced by the sharpness of the crack tip
radius.
In this paper, a new theory is proposed in which attention is focused on the singularity
strength of the strain-energy-density field around the crack tip. This energy field possesses a
1/r-type of singularity whose strength or amplitude will be designated as the "strain-energy-
density factor", S, and is direction sensitive. Unlike the stress-intensity factor k which is a
measure of the amplitude of the stress in a local region, the strain-energy-density factor S
represents the energy density in an element near the crack tip. The difference between k and S
is analogous to the difference between a scalar and vector quantity. In this sense, the critical
value Scr provides a knowledge of mixed mode crack extension in that it specifies the direction
of crack initiation and the fracture toughness of the material. The analytical results will be
shown to agree well with the experimental data on crack growth under mixed mode conditions.

2. Strain-ener~j-density field

Generally speaking, there are two approaches to the study of fracture mechanics, the energetic
treatment and the stress parameter approach. In the Griffith theory [1, 2] the crack system is
treated as a whole by assuming that cracks will propagate if the elastic energy released by their
growth is greater than the energy required to create new fracture surfaces. Alternatively, Irwin
[11] emphasized the stress conditions in the vicinity of the crack tip and used the critical
intensity of the local stress field as a material constant, commonly referred to as klc. The ap-
plication of the foregoing concepts however, is limited to Mode I crack extension where the
direction of crack propagation along the axis of symmetry is known as an a priori. In the
general case of combined loading, the initial angle of crack growth is an unknown and a scalar
quantity such as G 1 or kl alone cannot give a complete picture of the mixed mode crack ex-
tension behavior.
* It should be kept in mind that the crack does not extend in a self-similarmanner under mixed mode conditions.

Int. Journ. of Fracture, 10 (1974) 305-321

A strain-energy-density factor 307

The search for a quantity that provides a description of the direction of crack growth as well
as a measure of the intrinsic property of the material leads to a consideration of the energy
concentrated in the crack tip region where fracture takes place. To this end, the strain-energy-
density distribution around the crack tip will be computed and examined in detail. For an
elastic material, the stored energy in the crack tip region depends only on the final state of stress
as given by [12] and [13]
ar = 2(2r) -~ [k1(3 - c o s 0) cos(O/2)+k2(3 cos 0 - 1) sin (0/2)] + ...
a0 = 2(20 ~ [kl(1 +cos 0) cos(O/2)-k2(3 sin 0) cos (0/2)] + ...
zr0 = 2(2r) -~ [kl sin 0 cos(O/2)+k2(3 cos 0 - 1 ) c o s (0/2)] + ... (3)

dA= rdOdr

0
./ ----

~. ~ r . = radiusof core region

Figure 1. Cracks in general stress field.

where at, ao and r,0 are the stress components referred to as the cylindrical polar coordinates
r, 0 in Fig. 1. Those terms that remain finite as r--4) can be ignored since their contribution
for r ~ (than a characteristic length) is negligible in comparison with the r -½ terms. The cor-
responding displacement components in the radial and circumferential direction are [121
r½
ur = 4(2/~)~ {kl [ (2~c- 1) cos (0/2)- cos (30/2)]

- k 2 [ ( 2 x - 1) sin (0/2)- 3 sin (30/2)] } + ..~

r½
vo = ~ {kl [ - ( 2 x + 1) sin (0/2) + sin (30/2)]

- k 2 [(2x+ 1) cos(0/2)- 3 cos (30/2)] } + . . . . (4)

With attention focused on the conditions close to the crack tip, the strain energy stored in the
element dA = rdOdr in Fig. 1 for a two-dimensional stress system can be found from

dW=½ a,~-r + o~ r + - - f f ~ ) + 8-0 + 8r dA. (5)

Upon substitution of Eqns. (3) and (4) into (5) and after a considerable amount of tedious
algebra, the local strain-energy-density field can be expressed in the quadratic form
dW 1
- (alak~+2a,zklk2+a22k~)+... (6)
dA r

Int. Journ. of Fracture, 10 (1974) 305-321

308 G. C. Sih

in which the coefficients aq (i, j--- 1, 2) stand for

1
all = ~ [(1 + c o s 8 ) ( ~ - c o s 8)]

1
al2 = ~ sin 8[2 cos 8 - ( ~ - 1 ) ] (7)

1
322 = ~ [ ( K + I ) ( 1 --COS 8) "~ (1 +COS 8)(3 cos 8 - 1 ) ]

and they depend on the elastic constants r and #. Note that the strain-energy-density function
defined as dW/dA is inversely proportional to the radial distance r measured from the crack
tip and hence it becomes unbounded as r-*0. The magnitude of this energy field will be denoted
by S and called the strain-energy-density factor:
S = a11k2+2a12klk2+a22 k2 . (8)
This factor depends on 8 through the coefficients a 0 and therefore gives a description of the
local energy density on any radial plane intersecting the crack tip.
Since S in Eqn. (8) represents the intensity of dW/dA for the interior element rdSdr, it
ceases to be valid as r becomes indefinitely small. Hence, a critical distance r 0 as shown in
Fig. 1 or a core region surrounding the tip of a stress free crack must be observed*. Moreover,
it should be understood that the element rdSdr occupies only the interior region and none
of its sides shall coincide with the free crack surface as 8 = _+7r corresponds to the trivial case
S=0. Under these considerations, the following hypotheses on crack initiation in a two-
dimensional stress field will now be made:
(1) The initial crack growth takes place in the direction along which the strain-energy-density
factor possesses a stationary (minimum) value, i.e.,
BS
O--0 = 0 at which 8 = 8o (9)

where 0 o is between - ~r and ~: - ~r< 8o < 7r.

(2) Crack initiation occurs when the strain-energy-density factor reaches a critical value, i.e.,

attk2+2a12klk2+a22 k2 = Sor for 8 = 8o (10)

where 8o marks the angle of crack extension.
The stipulation here is that the crack tends to grow in a plane along which $ is an interior
minimum. In the case of a crack tilted at an angle with the tensile axis, the energy density
associated with volume change along this direction is found to be greater than that of distortion
or change in shape. In Eqn. (10), the critical value So, is postulated to be a material constant,
and hence it could be used as a measure of the fracture toughness of the material under mixed
mode conditions. In fact, in the case of Mode I crack extension, i.e., k2 = 0, $cr is directly related
to klo a s Scr=(Ic-1)k2c/8~t. It should be emphasized that although S~r is mathematically
connected with kl~, the k- and S-concept is basically different. The S-criterion assumes fracture
to occur when the energy density in a volume element near the crack tip reaches a critical value.
The classical fracture mechanics theory is based on global energy balance and energy released
for a segment of crack extension as k I is related to G t, the energy release rate.

* One way of determining the size of r o or the core region is to assume the crack to have a very small but finite radius
of curvature p such that the predictions based on the S-criterion at r = ro will match those for a perfectly sharp crack
with p = 0. The concept of a critical distance is also necessitated by the fact that the idealized continuum solution
must stop short at the core region within which the properties of the material and the nature of deformation may
differ significantly from those on the outside.

Int. Journ. of Fracture, 10 (1974) 305-321

A straimenergy-densityfactor 309

~ (Tarlsion) ~ (Compression)

Y
+0.

-0o

+0.

• (Tension) z (Compressi
U x on)
(o) (b)
Figure 2. Inclined crack: (a) in tension and (b) in compression.

3. Initial direction of crack growth

As mentioned earlier, the first study of the initial direction of crack growth in the presence of
both k i and k 2 were made in [-4] for the problem of a crack of length 2a inclined at an angle fl
with the loading axis as in Fig. 2(a). The stress-intensity factors k I and k 2 for this problem
are [8]
k l = aa½ sin2fl (11)
k 2 = a a ~ sin fl cos fl
where a is the applied stress. It was assumed that the crack will start to extend in the plane
which is normal to the maximum circumferential stress a o. Setting Oao/O0 = 0 or, as is the equiv-
alent, z,0 = 0, in Eqns. (3) gives the condition
k I sin 0o+k2(3 cos 0 o - 1) = 0 (12)
for determining the initial angle of crack growth 0 o. Eliminating k 1 and k 2 in accordance with
Eqns. (11) yields
sin 00+(3 cOS0o-1) c o t f l = 0 , fie0 (13)
which contains no elastic constants. This result implies that for the crack configuration and
loading condition given in Fig. 2(a) the initial angle of crack growth is independent of the
elastic properties for an isotropic material.
Turning now to the strain-energy-density criterion given by Eqn. (1), the following examples
will be considered:

3.1. Uniform extension o f a central crack

Consider the trivial case of the Griffith crack configuration consisting of an infinite body with
a central crack of length 2a subjected to uniform stress a at infinity (Fig. 3(a)). Inserting the
result
k 1 = a a ½, k2 = 0

Int. Journ. of Fracture, 10 (1974) 305-321

310 G. C. Sih

o"
"T"

r r Plane Strain
T
I
I I
•-'--- I
z,' =0.3

..._l"--I I e.=o" - " 1

o - ~ o tO----- x 0" "t'!
_ 0 ~---~-'-" X ,=
-~t, l- o-r P
I
I-" t
I_._ I JL--____
o" T
(a) (b)
Figure 3. Line crack: (a) in tension and (b) in shear.

into Eqn. (8) renders

- ~ (I +cos 0 ) ( x - c o s 0)
S = 1a2a (14)

For S to have a stationary value, it is necessary that OS/O0= 0 or

sin 01-2 cos 0 - ( K - l ) ] = 0, 0o = 0 (15)
which can be satisfied by either letting sin 0 = 0 or cos 0 = ( x - 1)/2. Since the elastic constant
x < 3, the minimum value of S corresponds to an initial angle of crack extension 00 = 0 meaning
that the crack extends along the x-axis as expected and

Smin __ (X-- 1)tr2a

8~t
The portion of Smin which is associated with change of volume, denoted by Sv, is larger than
the portion due to change of shape given by Sa where Smi. = So + Sd. For plane strain x = 3 - 4v
and
Sv - (1 + v)(1 - 2v) (1 - 2v) 2
tr2a, Sa= tr2a.
6/~ 2#
This is consistent with the classical concept that brittle fracture is more likely to take place-in
the plane where S o > S~.

3.2. Central crack under in-plane shear

Referring to Fig. 3 (b), the Griffith crack is subjected to a uniform shear ~ at infinity. For this
problem
k 1 = O, k 2 = za ½
and Eqn. (8) gives

z2a [(K+ 1)(1 - c o s 0)+ (1 +cos 0)(3 cos 0 - 1 ) ]

S -- 1-~ (16)

Applying the condition stated in Eqn. (9) yields

sin 0 [ ( x - I ) - 6 cos 0 ] - 0 , 0o = - a r c c o s / ~-- .

Recall that K= 3 - 4v for plane strain, the angle 0 o corresponding to Smi. is a function of the

Int. Journ. of Fracture, 10 (1974) 305-321

A strain-energy-density factor 311

Poisson's ratio v, Table 1 shows the different values of 0 o for v ranging from 0 to 0.5. With
cos 0o = (K- 1)/6, Eqn. (16) becomes
Stain = (18K-- 1 -rc2)r2a
192p

TABLE 1

Fracture angle under pure shear

v 0.0 0,1 0.2 0.3 0.4 0.5

0o -- 70.5 ~ - 74.5 ° - 78.5 ° - 82.3 ° - 86.2 ° - 90.0 °

Before comparing data on the fracture angle 00 obtained from the maximum stress and
strain-energy-density criterion, special precaution should be given to a near tip critical distance
at which 0o is measured. The crack tip notch effect plays an essential role here [9, 10] and cannot
be ignored*.

3.3. Uniaxial extension of an inclined crack

The stress-intensity factors k I and k2 for the problem in Fig. 2 are given in Eqn. (11) and
the strain-energy-density factor is
S = a2a(aal sin213+2a12 sin/3 cos fl+az2 cos2/~) sin2fl (18)
where the coefficients aij are those shown in Eqns. (7). Differentiating Eqn. (18) with respect to 0
and setting the result to zero, the fracture angle 0 o for a given position of the crack specified by/~
can be calculated from
(~:- 1) sin(0o-213)-2 s i n [ 2 ( 0 o - 1 3 ) ] - s i n 200 = 0, /3 ~ 0. (19)

3.3.1. Uniaxial tension. The numerical results of Eqn. (19) for negative values of 0o and positive
are shown in Fig. 4 which is a plot of the fracture angle Oo versus the crack angle 13from 0 ° to 90 °.
The curve based on the maximum stress criterion which is dotted agrees well with Eqn. (19)
for large values of 13and represents a lower bound for small values of 13. In general, it can be
taken as an average curve.
The validity of these predictions can be checked with the results of a series of experiments 1-4]
performed on the specimen in Fig. 2(a). Plexiglass sheets of approximately 9 " x 18"x 3 pt
were used with a central crack of approximately 2" in length positioned at angles of 13from 30 °
to 80° in increments of 10°. The initial fracture angle 00 at both ends A and B were measured
and will be designated by (0o)A and (0o)B, respectively. The experimental data for four sets of
tests are given in Table 2 with (0o)av~ being the average fracture angle of all the measured
values. The last two rows give the theoretical calculations of Eqn. (13) for the maximum stress
criterion and Eqn. (19) based on the strain-energy-density criterion with v= ½. As it can be
seen the agreement between theory and experiment is good. Additional experiments are needed
to verify the influence of Poisson's ratio on the fracture angle 0 o as predicted by the solid
curves in Fig. 4.
* It can be s h o w n from a m o r e refined a n a l y s i s [14] (in w h i c h the c r a c k tip r a d i u s of c u r v a t u r e effect is included)
t h a t the fracture angle 0o can v a r y s u b s t a n t i a l l y with the r a d i a l d i s t a n c e ro m e a s u r e d from the c r a c k tip. In this analysis,
the c r a c k is a p p r o x i m a t e d by a n a r r o w ellipse h a v i n g a r a d i u s of c u r v a t u r e p. To q u o t e a n u m e r i c a l example, c o n s i d e r
the same p r o b l e m of a c r a c k u n d e r p u r e shear but w i t h a b l u n t e d tip ofp/a = 0.0001. F o r a P o i s s o n ' s ratio ofv = 0.25, the
n o t c h sensitive S-theory predicts a fracture angle of 0o = - 7 0 ° for ro/a= 0.0004 a n d 0 o = - 8 0 ~ for ro/a= 0.0096.
N o t e t h a t the first angle 0 o = - 70~ agrees well with the m a x i m u m stress criterion while the s e c o n d angle 0o = - 80 °
agrees m o r e closely with the s t r a i n - e n e r g y - d e n s i t y t h e o r y of a perfectly s h a r p c r a c k w h i c h gives 0 o-- - 8 0 . 4 ° for
v = 0.25 a n d the s o l u t i o n is i n d e p e n d e n t of r o. It i s a p p a r e n t t h a t e x p e r i m e n t a l d a t a on fracture angle s h o u l d be a n a l y z e d
in t e r m s of r o, p as well as v.

Int. Journ. of Fracture, 10 (1974) 305-321

 312 G. C. Sih

=,=0.5
- 0.4 Or

- 0.3
t
90* -0.2
-0.1
~~'0o
LI.I
_J
¢.0
Z
80* - 0

uJ
rr Or
l
60*
Plane Strain
r~
It.
uJ
-~Ma)t Stress
Criterion

uJ
Z
20* ~Stroin- Energy -
D e n s i t y Criterion
I
0
QD
i I I I I I
0 20* 40* 60* 80" 90*
/ 9 - CRACK ANGLE
F i g u r e 4. C r a c k a n g l e versus f r a c t u r e angle in tension.

TABLE 2

Measured and calculated values of the fracture anole

fl 300 40° 5 0° 60° 7 0° 8 0°

(0o)A 1 -64 ° -55.5 ° -50 ° -40 ° -29 ° -170

2 -60 ° -52 ° -50 ° -43.50 -30.50 -18 °
3 -63 ° -570 "--53° -44.5 ° -- -15.50
4 -- -570 -52 ° -43.50 -- --

(00)B I -65 ° -58 ° -50.5 ° -44 ° -31.5 ° -18.50

2 -- - 53 ° - 52° - 400 - 31 ° - 17.5 °
3 -60 ° -550 -51.50 -46 ° -31.5 ° -170
4 -- - 570 - 5 0° - 43.° -- --

(0o)=v= -62.4 ° -55-6 ° - 5 1 . 6o -43 .l° -30.7° -17.3°

eq. ( 1 3 ) - 6 0 . 2° - 5 5 . 7° - 5 0 . 2° - 43.2° - 3 3 . 2° - 19.3 °
eq. ( 1 9 ) -63.5 ° -56.7 ° -49.5 ° -41.5 ° -31.8 ° -18.5 °

3.3.2. Uniaxial compression. There exists another set of solutions of positive 0o to Eqn. (19)
which will also yield a minimum on S. Physically, the positive values of 0o correspond to
the problem of Fig. 2(b) in which, the inclined crack is now under uniaxial compression.
Since S depends on a 2, Eqn. (18) contains both the solutions of uniaxial tension (+ a) and
compression ( - a ) . A plot of the positive fracture angle 0o against fl is given in Fig. 5 for
different values of the Poisson's ratio. In contrast to tensile loading, where the crack tends to
become horizontal, the crack path under uniaxial compression is towards the direction of
loading. Such a phenomenon has indeed been observed by Hoek and Bieniawski [15] who
have tested a number of glass plates with an inclined crack under uniaxial compression.
Unfortunately, they did not report the initial angle of crack extension so that a comparison of

Int. Journ. of Fracture, 10 ( 1 9 7 4 ) 3 0 5 - 3 2 1

A strain-energy-densityfactor 313

O"

i
j
180 °

170*
Plane Strain /¢ I
w
i
I
150'
!
I
w
tr,
:D
I
I-
130~ O"
I
I
U_
I
w
_> I10~ I
I--- I
o.I
0 I
0.2 I
I 90*
0 0.5 I
0.4 I
L_ o5 I
70*
I
I I I I J I
0 20* 40* 60* 80* 90*
/3 - CRACK ANGLE
Figure 5. Crack angle versus fracture angle in compression.

the theoretical results in Fig. 5 with experiments cannot be made at this time. Nevertheless,
they did publish the values of the applied stress aor to initiate crack growth for different positions
of the crack. It will be shown that their results on the variations of tr~ra~ with fl are indeed
predictable from the present theory.

4. Stationary values of density factor

In order to become familiar with the strain-energy-density factor S, further considerations will
be given to the inclined crack problem. For the case of uniaxial tension (Fig. 2 (a)), the crack
will spread in the negative 0-direction in a plane for which the S attains a minimum. Figure 6
illustrates the variations of 16#S/a2a with the negative polar angle - 0 . Note that for each
value offl, there exists a minimum value of S. For instance, at fl = 40°, Smi. equals to 0.68o -2 a/161~
and the crack is predicted to initiate in the direction 0o = -56.7 ° for v =~. Figure 7 shows a
plot of 16#SmiJa2a against fl for v varying from 0 to 0.4 inclusive. In general, the quantity
16#Smi./aZa increases with the crack angle fl reaching a maximum along the axis of Mode I
crack extension. As Smi. will be used as a material constant, the above statement implies that
the lowest value of the applied stress acr to initiate crack propagation occurs at fl = n/2 for a
material with low Poisson's ratio. A similar graph for uniaxial compression is shown in Fig. 8.
An interesting point to be observed here is that the quantity 161~Smi,/a2afirst increases with
the crack angle fl reaching a peak and then decreases in magnitude. The peak value is a function
of the Poisson's ratio. This suggests that given 16btSm~n/a2a=constant there exists a critical
angle fl0 at which the critical applied compressive stress is a minimum.
Additional insights into the nature of the mixed mode problem may be gained by taking a
glance at the relative magnitudes of the individual terms in Eqn. (8) whose dimensionless
values as a function of fl for v=0.3 are displayed in Fig. 9 and compared with the resultant
curve labeled 16#S~,~,/a2a.It should be pointed out that the interaction effect between Mode I
and Mode II is appreciable for intermediate values of the crack angle fl and becomes less
signMcant as the crack approaches the axis of load symmetry, i.e., fl = n/2. The discrepancy
between the mixed mode theory and the one using Mode I stress-intensity factor k only can be

Int. Journ. of Fracture, 10 (1974) 305-321

 314 G. C. Sih

n-
O a"
I-

° t Plane Strain
~- ~ = 0.333
z

, 2.oF- ° ',

g
'
0.5
o

m 0 20 ° 40 ° 60" 80 °
:k
-0- NEGATIVE POLAR ANGLE
m

Figure 6. S t a t i o n a r y values o f n o r m a l i z e d strain-energy-density factor.

O"
n--
o v=O
I--
L)
<
LL

3.5
I-
z
i./J
o Cr
i
>..
n-- P~ne Str0in
LU
Z
1,1.1 2.5
i
_z
n,.
I.-
(/3 2.0
£3
ILl
N
.J 0.3
1.5
fie
o
z
I
o
0.4
%
I
0.5
E I
tO
I
::L I
to
J - i I I I
0 20 * 40* 60* 80* 90 °

/~ - CRACK ANGLE
Figure 7. Variations of density factor with crack angle for tensile loading•

Int. Journ. of Fracture, 10 (1974) 305-321

A strain-energy-densityfactor 315

u=O
o"
8
t--
0
,,< 0.3 1
>-
I--

Z
W
123
i
>-

n~
W
Z
W
I
z
<
t'r"
I-- Plane Strain
03
t/)
03
W
.-.1
Z
O
t.O
Z
W
5: 0.1
D
i

ff
b
)
E
O3
::L
tO
I
0 50 ° 60* 9()*
- CRACK ANGLE

Figure 8. Variations of density factor with crack angle for compressive loading,

as much as 138.5 ~o on the applied stress for a crack inclined at fl = 40 °. Hence, the interaction
effect cannot be neglected in general unless the crack lies in a plane which is nearly normal to
the direction of maximum principal tensile stress.

5. Intrinsic property of strain-energy-density factor

One of the principal aims of fracture mechanics is to characterize the behavior of materials
in the presence of flaws or cracks.,This requires a clear distinction of the difference between S
and Scr. The strain-energy-density factor S is simply the leading coefficient of the series ex-
pansion of dW/dA about the crack point r = 0 in Fig. 1 and it varies as a function of the polar
angle 0. Once S has attained a critical value Scr at the point of incipient fracture it may be regarded
as a crack extension force which should be independent of loading conditions and crack
configuration. If verified experimentally, Scr can be used as a material constant that serves as
an indication of the fracture toughness of the material.
The theoretical results will be compared with the experimental data obtained on D T D
5050-5½ ~o Zn aluminum alloy in [6]. All the tests were carried out in precracked specimens of
the type shown in Fig. 2(a). The measured values of tr,a ~ for different crack sizes and failure
loads are plotted against the crack angle fl in Fig. 10. The solid and dotted curves represent
the predicted values for the aluminum alloy with k1¢ equal to 28.2 ksi in ~ and 29.2 ksi in ~,
respectively. The agreement is good. The same data is given in Fig. 11 with the critical stress-
density factor Scr normalized with respect to its value (S¢r)~/2 corresponding to Mode I crack
extension and Scr remained essentially constant.
The fracture mechanics of an inclined crack under compression is basically different from
that of extension. For glass with a Poisson's ratio of v = 0.25, the theoretical curve predicts a
critical angle of flo ~- 37° at which the applied stress to initiate fracture is a minimum. In [15],

Int. Journ. of Fracture, 10 (1974) 305-321

316 G. C. Sih

n- 1.6
O
I--
0
1.4
I-
Plane Strain
z
hi I,=0.3 ®
c~ 1.2

(.9. ( D 16P" Smin./O'2o

n-
Z 1.0
bJ (~) 16F oll/sin4/~
!

z
<Z (~) 8P" 0=2/sin3/~ c o s ~
""
I-- 0.8
(/} (~) 16P" o2~/sin2/9 cos/~
I-..-
z
LIJ
z 0.6
0
0.
®
0
(") 0.4
13
Z
<
I--
z 0.2

..J
go
"tz' 0 I I I
20* ~ N ~ O * 60* 80'
a B - CRACK ANGLE /
Ld
N
J
<Z -0.2
=E
t~
0
z
-0.4 -
Figure 9. Resultant and component density factors.

compression tests on 6" x 6" precracked glass plates were performed. The critical loads were
then measured for cracks inclined at various different angles with respect to the axis of loading.
The experimental curve in Fig. 12 is dotted and indicates a critical angle flo-~31 °. Before
comparing the theoretical and experimental results, a few remarks on the crack closure problem
is in order. For cracks under compression, there is a tendency for the crack surfaces to come
into contact. McClintock and Walsh [16] have attempted to include friction into the Griffith
calculation assuming that the crack surfaces rub against one another. However, their model
predicted an unrealistically high value of the coefficient of surface friction in order to show
compressive fracture stress a c in excess of ten times the tensile fracture stress a t. Recall that the
Griffith's original result of ac/a t = 8 did not include the effect of surface friction which does not
appear to play a major role in this problem. The strain-energy-densitycriterion offers a possible
explanation for aJtr t ranging from 10 to 1000 times in the fracture of many brittle materials.
This variation of trc/at is attributed to the position of the average flaw in the material relative
to the axis of loading as measured by the angle fl in Figs. 11 and 13. The crack tip radius has
also been known to have a significant influence on fracture data of compression tests. A notch
sensitive theory of cracks under compression is also needed. These and other effects are beyond
the scope ofthe present paper and require further study. Thus, the important point to be made
here may not be in the quantitative agreement of theory and experiment but in the trend of the
failure stress variations with crack angle for fracture under compression.

Int. Journ. of Fracture, 10 (1974) 305-321

A strain-energy-density factor 317

O-cr

55-

5C

O'cr

45 I~ [ ------ ( °'cr'J-a" ) ~r/2= 29"2

(¢rcr Vr~)~2 = 28.2

¢-
40

~ ~
{
* Straight Cracks [6]
I r~ Slanted Crocks [6]

35-
),° XX
\
Plane Strain ~_\ a

30-
16P- Scr=l.O6xlO 9 Ib2/in.3

I I I I I
0 20* 40* 60* 80* 90*
/~- CRACK ANGLE
Figure 10. Critical tensile stress versus crack angle.

,. 2 o

oc,-- O-cr

1.4 /-(O'cr ~6") 7r~ = 28.2

1.2
¢M

1.0
U
Or) L (°'cr ~r6")~r/2 = 29.2
0.8

03
0.6 • Straight Cracks [6]
0.4 n Slanted Cracks [6]

0.2
I I I I I
0 20* 40* 60* 80* 90*
/~ - CRACK ANGLE
Figure 11. Critical density factor as a m~ltcrial constant.

Int. Journ. of Fracture, 10 (1974) 305-321

318 G. C. Sih

>-
I-

z O"
I.iJ
I- I
z 8.5
I
i Theoretical Curve
go
(/)
LLI
I I
rr 7.5
t--
co
I I
(/)
(¢)
I
w
j 6.5 I
z
o_
u) \ I o-
z
w 5.5 x J
Q \ !
J
'~ \ / Plane Strain
E 4.5 \" ~/ / =,=0.25 (Glass)
co \
::L
to
" E x p e r i m e n t a l Values [10]
a 3.5
b I I I I I I I
b 0 I0 ° 20 ° :30 ° 40 ° 50 ° 60 ° 70 °
- CRACK ANGLE
Figure 12. Critical compressive stress versus crack angle.

6. Mixed mode fracture criterion

Having discussed the possibility of using Set as a material constant, a mixed mode fracture
criterion will be proposed. The critical values of k 1 and k2, i.e., k1¢ and k2e in a given problem
will lie on a curve in the k~, kE-plane determined by the hypotheses stated in Eqns. (9) and
(10). The theoretical values of k~ and k 2 may be determined from these two equations for a given
material, i.e., a given value of S¢,. From the reported values of k~c in 1-6] on the aluminum alloy,
the values of 4.8#Sc~= 28.2 and 29.2 are used and the theoretical plots of k 2 versus k 1 are given
in Fig. 13. The third curve represents the prediction based on a criterion of maximum stress [4].
The curves based on the strain-energy-density theory appear to be closer to the experimental
results. The same observation can be made on plexiglass plates tested in [4], i.e., the measured
points of (k~c, k2e) lie outside of the klk2-curve of the maximum stress criterion.
The k~k2-curve governing the mixed mode fracture of cracks under remote compression is
basically different from that of tension and is shown in Fig. 14. First, the curve does not
intersect the k 1-axis which implies the obvious fact that Mode I crack extension does not exist
in compression. This can be easily verified by solving Eqns. (9) and (10) for kl and k 2 with the
constraint that the crack angle and fracture angle satisfies the relations dictated by the curves
given in Fig. 6. For a glass with v = 0.25, the theoretical prediction gives a slanted curve in the
k l , k2-plane. Again, the qualitative feature of the solution is in agreement.

7. Concluding remarks

A theory based on the concept of a strain-energy-density factor S has been presented. The
stationary values of this density factor can predict the direction of crack growth under mixed
mode conditions. As it is well known that the classical theory of Griffith cannot be conveniently
used to predict the direction of crack propagation. The critical value So, is proposed to be used
as a material parameter for measuring the resistance against fracture and hence it can be regarded
as independent of the crack geometry and loading. Using the example of an inclined crack, the
Int. Journ. o f Fracture, 10 (1974) 305-321
A strain-energy-density factor 319

(7"

t
| PlaneStrain ~.-~/~/
3o~- ~: o.333 I/fT.

20

c~, .Straight Cracks !6! ~ \~

o Slanted Crocks [B]I ~ \X
,o , ,,<,\ .

- =29.2 \XX °

" (._°'cr~'~°)TO2: 28"2 '~\

• Max. Stress Criterion ~\
I I I I I _ktl
0 I0 20 30
INTENSITY- FACTOR k, in ksi-v/qE".
Figure 13. Mixed mode fracture criterion of klc versus k2c.

4O

Theoretical Curve
z~

-~ 30
._

o
~- 2O
<[
U-

l--- Plane Strain I

O3
Z
W
v =0.25 (Glass) o-
Z
Experimental Values [10]

= I , I L I I
O IO 20 30 4O

INTENSITY FACTOR k I in ksi ~ .

Figure 14. Mixed mode criterion for cracks under compression.

theory predicts two basic solutions, one for tensile loading and the other for compressive
loading. The agreement between the theoretical and experimental results for both of these
loadings with the available experimental data on crack extension in a combined stress field of

Int. Journ. of Fracture, 10 (1974) 305-321

320 G. C. Sih

Mode I mixed with Mode II can be imprbved by considering other effects such as crack tip
radius, the size of core region, etc. These studies are being carried out and will be reported in
future communications. Nevertheless, it has been shown that serious errors on the prediction
of the critical applied stress could be encountered if the influence of mixed mode is ignored.

Acknowledgement

This work was partially supported by the Office of Scientific Research under Contract No.
AFOSR-74-2586 with the Institute of Fracture and Solid Mechanics, Lehigh University.
The autlaors also wishes to thank Dr. E. P. Chen for his valuable assistance in carrying some
of the numerical calculations in this work.

REFERENCES

[I] A. A. Grillith, The phenomena of rupture and flow in solids, Phil. Trans. Royal Society, Vol. A221 (1921) 163.
[2] A. A. Griftith, The theory of rupture, Proc. Ist Int. Conor. Applied Mech., Delft (1924) 55-63.
[3] J. E. Srawley, M. H. Jones and W. F. Brown, Determination of plane strain fracture toughness, Materials Research
.Standard, 7 (1967) 262-266.
[4] F. Erdogan and G. C. Sih, On the crack extension in plates under plane loading and transverse shear, J. of Basic
Engr9., 85 (1963) 519-527.
[5] B. Cotterell, Notes on the paths and stability of cracks, Int. J. of Fracture Mech., 2 (1966) 526-533.
[6] L. P. Pook, The effect of crack angle on fracture toughness, J. of Engro. Fracture Mech., 3 (1966) 205-218.
[7] T. T. Wang, T. K. Kwei and H. M. Zupko, Tensile strength of butt-joined epoxy-aluminum plates, Int. J. of
Fracture Mech., 6 (1970) 127-137.
1-8] G. C. Sih, P. C. Paris and F. Erdogan, Crack-tip stress-intensity factors for plane extension and plate bending
problems, Int. J. of Fracture Mech., 29 (1962) 306-312.
[9] J. G. Williams and P. D. Ewing, Fracture under complex stress--the angle crack problem, Int. J. of Fracture
Mech., 8 (1972) 441-446.
[10] G. C. Sih and M. E. Kipp, Discussion on "Fracture under complex stress--the angle crack problem", Int. J. of
Fracture (in press).
[11] G. R. Irwin, Fracture mechanics, Structural Mechanics, Pergamon Press, London, England (1960) 560-574.
[12] G. C. Sih and H. Liebowitz, Mathematical theories of brittle fracture, Mathematical Fundamental of Fracture,
Academic Press, New York (1968.) 67-190.
[13] M. L. Williams, On the stress distribution at the base of a stationary crack, J. of Applied Mechanics, 24 (1957)
109-114.
[14] G. C. Sih, Application of the strain-eneroy-density theory to fundamental problems infracture mechanics, presented
at the 10th Annual Meeting of the Society of Engineering Science, Raleigh, North Carolina, November 1973.
115] E. Hoek and Z. T. Bieniawski, Fracture propaoation mechanics in hard rock, Technical Report--Rock Mech.
Div., South African Council for Scientific and Industrial Research (1965).
[16] F. A. McClintock and J. B. Walsh, Friction on Griltith cracks in rocks under pressure, Proceedinos of the 4th
U.S. National Congress of Applied Mechanics (1962) 1015-1021.

R]~SUMI~
L'article a trait au probl6me g6n6rai de l'extension d'une fissure dans un champ de contraintes combin6es de fagon
teUe qu'une fissure peut s'6tendre clans toute direction arbitraire par rapport ~ sa position d'origine. Dans une situation
oh les deux facteurs d'intensit6 d'entaille /q et k2 sont pr6sents le long du front de la fissure, celle-ci peut s'6tendre
dans toute direction situ6e dans un plan normal ~tce bord de la fissure, en fonction des conditions de charge.
Des r6sultats pr61iminaires indiquent que la direction d'accroissement de la fissure, et la t6nacit6/l la rupture dans un
probl6me mixte oh se retrouvent ~ la lois le Mode Iet leMode II, sont r6gies par la valeur critique Sot du facteur de
densit6 d'(~nergie de d6formation.
L'hypoth6se de base est que l'amorqage de la fissure survient lorsque S passe par une valeur critique Set. Le facteur
de densit6 d'6nergle de d6formation S repr6sente la force du champ d'6nergie 61astique au voisinage de l'extr6mit6 de
l'entaille, lequel est une fonction singuli6re de 1/r, oh r repr6sente ia distance radiale depuis le front de fissure.
Darts le cas particulier du Mode I d'extension de fissure, Scr ne d6pend que de klo suivant:
S¢, = ( x - 1)k~,/8F
En g6n6rai, S prend la forme quadratique
atlkt + 2alzklk2 +a2.2k2
dont la valeur critique est suppos6e 8tre une constante du mat6riau.

tnt. Journ. of Fracture, 10 (1974) 305-321

A strain-eneroy-density f a c t o r 321

Les pr6dictions analytiques sont en bon accord avec les donn6es exp~rimentales pour le probl~me de la fissure
inclin~e d a m des 6prouvettes de plexiglass ct d'alliage d'aluminium.
Le r6sultat de eette recherche fournit une proe&lure commode pour la d~termination de la dimension critique d'une
fissure qu'une structure peut tol6rer sous des conditions de modes mixtes d'extension de fissure, en pr6sence d'une
contrainte appliqu6e donn6e.

ZUSAMMENFASSUNG
Man behandelt das allgemeine Problem der Riflausdehnung in einem zusammengesetzten Spannungsfeld wo ein
RiB sich in irgend eine beliebige Richtung ausdehnen kann beziiglich auf seine Anfangsposition. Im Fall wo beide
Spannungsintensit~itsfaktoren k 1and k2 an der RiBspitze vorhanden sind kann sich der RiB in einer beliebigen Richtung,
in einer normal zum RiBrand liegenden Ebene ausdehnen, abh~ingig von den Belastungsbedingungen. Die vorl~iufigen
Ergebnisse zeigen dab die Richtung der RiBausdehnung und die Bruchz~ihigkeit ffir das gemisehte Problem des ersten
und zweiten Modes durch den kritischen .Wert des Faktors der Verformungsenergiedichte Sot regiert werden. Die
Grundhypothese ist dab die RiBeinleitung stattfindet wenn das innere Minimum yon S einen kritischen Wert erreicht
bezeichnet durch Scr. Der Faktor der Verformungsenergiedichte stellt die St~irke des elastischen Energiefeldes in der
N~ihe der RiBspitze dar welches eine Besonderheit der Ordnung 1/r hat wo der Radialabstand yon der Front des
Risses ausgemessen wird. Im Sonderfall des Modes 1 ist die RiBausdehnung Scr mit kl~ allein verbunden da Scr=
( x - 1)k~c/8/~. Im aUgemeinen nimmt S die quadratische Form a llk I + 2a12 k 1k 2 + a2z k 2 an, deren kritischen Wert
als eine Unver~nderliche des Materials angenommen wird. Die analytische Voraussagungen sind in guter Ober-
einstimmung mit den Versuchergebnissen des Problems eines schr~igen Risses in Proben aus Plexiglas und Aluminium-
legierungen. Die Ergebnisse dieser Untersuchung ergeben ein gutes Verfahren zur Bestimmung der kritischen Riflgr6ge
die ein Bauelement unter gemischten Moden Bedingungen zulassen kann unter einer gegebenen Spannung.

Int. Journ. of Fracture, 10 (1974) 305-321