J INTEGRAL AS FRACTURE CRITERION FOR SHORT

FIBRE COMPOSITES: AN EXPERr~~NTAL APPROACH

B. D. AGARWAL, B. S. PATRO and PRASHANT KUMAR

Department of Mechanical Engineering, Indian Institute of Technology, Kanpur 208 016, India

Abstract-Fracture behaviour of randomly oriented short glass fibre reinforced epoxy resin has been
investigated. Fracture tests were conducted on single edge notched specimens and J integral evaluated
using energy rate interpretation. Its value is found to be independent of crack length when crack length to
specimen width ratio (a/w) is larger than 0.35. For smaller cracks, general material damage away from the
crack tip also influences the energy absorbed significantly. An extrapolation method has been developed to
separate the crack tip energy from the energy absorbed due to general material damage. The J integral thus
obtained is independent of crack length and specimen length and its critical value is the same as obtained
for a/w >0.35 without extrapolation. It also agrees well with the critical stress intensity factor obtained
using R-curve approach in an earlier study. Approp~ate modelljng of the stress-strain curve and obser-
vations on damage mode also justify the use of J integral to characterize fracture of short fibre composites.

INTRODUCTION
MOST OF the research work on fracture of composite materials has, so far, centred around linear-elastic
fracture mechanics concepts employing elastic analysis of the crack tip region[f-121. There are practical
di~culties in accurately analyzing the crack tip region even for homogeneous isotropic materials and
more so for heterogeneous composites. A characterization of the crack tip area by a parameter
calculated without focussing attention directly at the crack tip would provide a more useful method for
analyzing fracture. The path independent .I integral proposed by Rice[l3] is such a parameter. Its value
depends on the near tip stress-strain field. However, the path independent nature of the integral allows
an integration path, taken s~cien~y far from the crack tip, to be substituted for a-path close to the
crack tip region. Therefore, the I integral can be calculated using numerical methods more accurately
compared to the stress intensity factor. Also, an experimental evaluation of J integral can be ac-
complished quite easily by considering the load deflection curves of identical specimens with varying
crack lengths.
The use of the J integral as an elastic-plastic fracture criterion has been discussed by Broberg[14]
from an anafytic s~ndpoint. A jus~fication for choosing this parameter as a fracture criterion comes
from a consideration of the Hutchinson-ice-Rosengren (HRR) crack tip model[lS, 161 where the
product of plastic stress and strain is shown to have a l/r singularity; r is a near tip crack field length
parameter. For a deformation plasticity theory, McClintock[17] has demonstrated, through the crack tip
plastic stress and strain equations expressed from the HRR singularity, the existence of a singularity in Y
whose strength is J integral. Thus, f integral may be chosen as a parameter to characterize the crack tip
environment because it can be evaluated experimentally and calculated with less di~culty than the
plastic stress and strain intensity factors.
Begley and Landes[l8-201 discussed various aspects of using the J integral as a failure criterion for
metals. They demonstrated applicability of .I integral for the case of large scale plasticity at the crack tip
through experimental results on an intermediate strength rotor steel for which the J integral at failure for
fully plastic behaviour was found to be equal to the linear elastic value of strain energy release rate (G)
at failure for extremely large specimens. Thus, the J integral approach eliminates the necessity of testing
very large specimens.
A major limitation of the approach arises from the fact that the .I integral is path independent only
when the stress-strain relation is unique. It is truly path independent for linear and nonlinear elastic
stress strain laws and also for elastic-plastic behaviour under situations of monotonic loading. This rules
out its application to mate~~s which exhibit si~i~c~t subc~tica1 crack growth prior to fracture since
any crack extension necessarily implies unloading near the crack tip.
It is well known that in composite materials, microcracks at the fibre matrix interface appear at very
low loads due to the stress concentrations produced by the fibres lying perpendicular to the load. It is
probably this unavoidability of microcracks that has deterred researchers from exploring applicability of
the f integral as a fracture criterion for composite materials.
61.5
676 B. D. AGARWAL et al

In the present paper, the J integral is being developed, as a fracture criterion for composite materials,
based on test results. The evaluation of J is not tied down to the analytical limitations. A new set of
limitations is being developed around the analytical limitations but not strictly adhering to them. The
results reported in this paper on short fibre composites are very promising. Similar developments for
laminates are currently underway.

EXPERIMENTAL DETAILS
The present studies were performed on randomly oriented short glass fibre reinforced epoxy resin. A
chopped strand mat of glass fibres having a mass of 0.6 kg/m2 and an average fibre length of 50 mm was
used as the reinforcement. The matrix material was Ardaldite CY 230 epoxy cured with hardener
HY 951. The composite plates (3 mm thick) were cast in the laboratory and cured at room temperature
for at least 10 days. The cured plates exhibit a fibre volume fraction of about 36%. The single edge
notched specimens were 25 mm wide and the length between grips was at least 3 times the specimen
width. The initial notches were machined using a 0.2 mm thick slit cutter and their length was varied
between 1.25 and 17.5 mm.
The fracture toughness tests were performed on a 10 Ton MTS machine. Load and load point
displacement were recorded in a X-Y recorder. All the tests were conducted in a displacement
controlled mode. The data was analysed using J integral approach.

RESULTS AND DISCUSSION

Typical load displacement (at load poipt) curves for specimens with different initial crack lengths are
shown in Fig. 1. The tests were conducted under displacement controlled conditions so that the load
displacement behaviour beyond maximum load is also clearly indicated. Specimens with small cracks,
fracture suddenly causing an abrupt drop in load whereas the specimens with larger cracks show a more
gradual fracture process beyond maximum load. The behaviour is similar to that observed in metals [ 181.
This is because the strain energy stored during loading in a small crack length specimen is sufficient to
cause catastrophic failure. It is not the case for longer crack length specimens.
The observed fracture load is plotted against crack length in Fig. 2 along with the fracture load that
would be expected if the strength was unaffected by the crack, i.e. the fracture load obtained by
multiplying the net cross sectional area and the unnotched strength. The observed fracture load is

Expected fracture load

2.5 -

O----k . 1
10.0
I
15.0
I
17.5

Displacement, mm Crock length, mm

Fig. I. Fig. 2.

Fig. 1. Load displacement curves for different initial crack lengths.

Fig. 2. Observed and expected (based on net cross-section area) fracture loads for notched specimens.
 J integral as fracture criterion for short fibre composites: an experimental approach 677

smaller than the expected indicating that the crack reduces the fracture load far greater than can be
accounted for by the reduction in cross-sectional area. The extent of this influence is illustrated in Fig. 3
through the ratio of observed to expected fracture loads. The decreasing ratio indicates the increasing
influence of cracks which stabilizes for cracks larger than 7.5 mm.
The fracture process becomes unstable at a displacement beyond which the load decreases mono-
tonically. This displacement may be referred to as the critical displacement. It is plotted vs the crack
length in Fig. 4. Initially the critical displacement decreases with incwase in crack length and remains
constant for cracks larger than 7.5 mm. The initial variation in critical displacement occurs due to the
significant deformations away from the crack plane because of large loads. This point will be further
explained later in this section. The critical value of J integral is obtained corresponding to the constant
critical displacement of 0.96 mm as shown in Fig. 4.
The load displacement curves can be used to obtain the value of J integral experimentally through its
energy interpretation as follows [ 191

J,_aU
aa I constant displacement

where U is the potential energy per unit thickness and a is the crack length. It may be mentioned that
when displacement is kept constant for evaluating J the potential energy, U, reduces to the area under
the load deflection record and is equal to the strain energy[l9]. Thus, area under load displacement
curves is first obtained and plotted against crack length for several displacements (Fig. 5). For a given
displacement, energy absorbed by a specimen decreases as the crack length increases (Fig. 5) because
smaller loads are required. The variation in energy absorbed is less for cracks shorter than 8.75 mm
compared to that for longer cracks because in specimens with longer cracks the energy absorbed is
essentially in the vicinity of the crack tip and is thus strongly influenced by the crack length.
The J integral is obtained from eqn (1) through slopes of the energy curves in Fig. 5. The J integral is
independent of crack length for cracks larger than 8.75 mm since the energy curves are straight lines in
this range. The variation of J with displacement is shown in Fig. 6. The critical value of J corresponding
to the critical displacement of 0.96 mm is 51.8 kJ/m2. For smaller cracks, the value of J integral depends
upon the displacement as well as the crack length because the slope*of the energy curve changes with
crack length (Fig. 5). The variation of J for cracks smaller than 8.75 mm has not been shown because it

01 5.0
Crack length
IO0
,mm
15.0 I7.5 .T 5 IO 15 115

Crack length, mm

Fig. 3. Fig. 4.

Fig. 3. The ratio of observed to expected fracture load as a function of crack length.

Fig. 4. Variation of critical displacements with initial crack lengths.

 B. D. AGARWAL et of.

01 I
I I II 1
5 IO 15 17.5 0 0.2 0.4 0.6 0.8 1.0 I.1

Crack length, mm Displacement, mm

Fig. 5. Fig.6.
Fig. 5. Strain energy per unit thickness of specimen for different displacements.
Fig. 6. J integral as function of displacement.

is not unique. However, in view of Fig. 5 it may be stated that in this region .I will be smaller for a given
displacement but defined for a greater range of displacement. Its apparent critical value is also expected
to be larger in these cases. The applicability of J integral in this region is further discussed later in this
section.
From preceding discussion, it appears that when the crack is larger than 8.75 mm or when a/w > 0.35,
the fracture behaviour is governed essentially by the crack tip environment resulting in a constant
critical displacement and a unique value of J integral. For these crack lengths, the fracture load is small
which does not cause any general material damage away from the crack tip region. On the other hand
when cracks are small (a < 8.75 mm or a/w < 0.35) the .l integral and critical displacement depend upon
the crack length indicating that in addition to the crack tip environment, the region away from it also
influences such quantities as the energy absorbed and displacement at fracture. This may be attributed to
the fact that the fracture loads are high enough to cause general material damage.
In order to study the influence of the general material damage in specimens with smaller cracks,
additions specimens with varying specimen lengths were tested, The length between grips was varied
from 3 to 6 times the width of the specimen. Displacement at fracture {critical displacement) is plotted
against specimen length in Fig. 7. As expected the critical displacement increases with specimen length
for all crack lengths. However, critical displacements for specimens with 10 and 12.5 mm long cracks are
the same which is consistent with Fig. 4. The total displacement of the specimen is the sum of the
displacement in the crack tip region, which may be expected to be independent of the specimen length
and displacement in the region away from the crack tip which should be a function of specimen length.
The intercept on the ordinate obtained through extrapolation of a straight line in Fig. 7 may be regarded
as the displacement in the crack tip region alone. Interestingly, all the straight lines in Fig. 7 intercept the
ordinate at the same point. This common intercept may be regarded as a critical displacement due to the
presence of the crack and whose value is independent of crack length and specimen length.
Variations of energy absorbed upto fracture are shown in Fig. 8 for different crack lengths. The total
energy absorbed may also be thought of as the sum of the,energies absorbed in the crack tip region and
the region away from it. The energy absorbed in the crack tip region should depend upon the crack
length but not on specimen length whereas the energy absorbed in the region away from the crack tip
does depend upon the specimen length. It is observed that when the crack length is 10 or 12.5 mm, the
energy absorbed is independent of the specimen length signifying negligible energy absorption in the
region away from the crack tip. For crack lengths of 5 and 7.5 mm, the total energy absorbed increases
 J integralas fracture criterion for short tibre composites: an experimental approach 679

I I I
0 50 100 150

Specimen length,mm

Fig. 7. Variation of critical displacement with specimen length for different initial crack lengths.

3.0 -

5
0
2.5 - /
E 20-
: /
1 /’
/’
B /’
& 1.5 - ,’
I=I /
/’
__---
_---

_---__-__----
.E __--
; 1.0 .-_---_--_---- _ IO_ 4
0

0.5 I a = 12.5mm

050 50 100

Specimen length, mm

Fig. 8. Variation of strain energy with specimen length for different initial crack lengths.

linearly with the specimen length indicating a significant energy absorption in the region away from the
crack tip as well. These observations are further supported by visual observation of the specimens. The
damage in the specimens with 10 and 12.5mm cracks is confined to the crack tip region whereas in
specimens with smaller cracks the material damage is all over. This is illustrated in Fig. 9 through a
photograph of two fractured specimens. The photograph was taken in a bright light background and
therefore, the damage (opaque to light) is indicated by dark areas.
The intercept on the ordinate obtained by extrapolation of a straight line in Fig. 8, may be regarded as
the energy absorbed in the crack tip region. Energy absorbed thus obtained is plotted in Fig. 10. It was
680 B. D. ~~~~~~~ et Qi.

explained with respect to Fig. 7 that the critical displacement due to the presence of a crack alone is
independent of crack length. Thus, it may be argued that the energy absorbed for different crack lengths
(Fig. 10) correspond to the same critical displacement and therefore, the slope of the straight line may be
used to obtain the critical value of 1, independent of crack length. The J critical thus obtained is
50.7 kJlm2 which is close to the value 51.8 kJfm* obtained earlier in Fig. 6. This is a very significant
observation. This shows that the energy absorbed at the crack tip may be isolated from that absorbed in
the region away from it; Thus, a parameter independent of testing variables (i.e. crack length and
specimen length) is obtained which may be used as a fracture criterion for the material.

FURTHER DISCUSSION
As discussed earlier, a justification for choosing .I integral as a fracture criterion comes from a
consideration of Hutchinson-Rice-Rosengren (HRR) crack tip model[l5, 161 where the product of
plastic stress and strain has been shown to have a l/r singularity; r is a near tip crackfield length
parameter. For plasticity behaviour which can be modeled by the Ramberg-Osgood relation

where Crand g,, are equivalent stress and equivalent strain, respectively, Cr,is a constant and n is the
strain hardening exponent. McClintock[17] has shown that the cracktip plastic stress and strain
equations can be expressed from the HRR singularity

_ j Mn+lt l

q=fli
( ) cril, pi@m c;,(e)
I/(n+l)
J
Eij =
(d” >
7 & 4j(e)

where I, is a function of n and mode of crack opening[21]. These plastic crack tip stress and strain
equations demonstrate a singularity in I where J is the strength of this singularity.
The stress strain behaviour of the composite studied here can also be modeled by the Ramberg-
Osgood relation (eqn 2). Figure 11 shows a typical stress strain curve and superimposed on it a curve

1.25r

1.0 -
E
2
5 0.75 -
$
i:
.g 0.5 -
L
z

01.5
Crack length, mm Strain ,%

Fig. 10. Fig. il.

Fig. IO. Strain energy at the crack tip for ditt’erent initial crack lengths.
Fig. Ii. Experimental and idealized stress-strain relations for randomly oriented short glass fibre reinforced
epoxy.
 J integral as fracture criterion for short Cbre composites: an experimental approach 681

Fig. 9. Transmitted light photograph of two specimens with different crack lengths.

Fig. 12. Scanning electron microscope photograph of a fibre bundle near crack tip shows debonding crack
but no fibre breaks.
 J integral as fracture criterion for short fibre composites: an experiments approach 683

(broken line) representing the following relation

0.83

A= ( 1
ii%
(5)

where u is the composite stress in MPa and t is the strain in percent. The constants 25.75 MPa for stress
and 0.225% for strain correspond to the end of linear portion of stress strain curve. The two curves in
Fig. 11 are quite close to each other indicating that the stress strain behavior of the composite may be
closely approximated by the Ramberg-Osgood relation. This representation of the nonlinear behaviour
of the composite provides a justification for exploring the possibility of using J integral as a fracture
criterion for composite materials. The development of the criterion in the preceding section is based on
test results. Further justification for this empirical development is given in the following paragraphs.
During fracture tests on composite materials, damage at the crack tip occurs primarily due to matrix
cracking and debonding at the fibre-matrix interface resulting in unloading of the matrix locally but not
necessarily of fibres for two reasons. One, the Scanning Electron microscope observation of the
specimens prior to fracture do not show evidence of any significant fibre failures even in the crack tip
region. A photograph taken in the Scanning Electron microscope shows (Fig. 12) only debonding cracks
and no fibre breaks in a fibre bundle which is about 1 mm away from the crack tip. Two, even in the case
of short fibre composites like the present one, the fibres are long enough (average length 50 mm) so that
major part of their length is embedded in the matrix away from the damage zone. Therefore, debonding
at the interface and few fibre failures do not significantly inffuence the load carried by the fibres. Further,
since fibres carry major portion of the toad in fibrous composites, this crack tip damage does not
constitute material unloading to the same extent as a crack extension does in metals, Afthough the
requirement of no unloading is not completely met, its influence may be quite small and ignored in an
empirical development of the criterion.
It has been shown that J, is related to the parameters of linear elastic fracture mechanics. For plane
stress case it is related to critical stress intensity factor in mode 1, K,, by the following relationI

where E is the modulus of elasticity, The present material has an average elastic modulus equal to
11.5 GPa. Therefore, eqn (6) gives I(, equal to 24.4 MPaV’m. This value of critical stress intensity
factor agrees very we11with the Kc of 24.82 MPaV’m obtained by Agarwal and Giare [3f for similar
glass fibre composite. This demonstrates that the present method of characterizing fracture toughness is
consistent with the R-curve method. However, the J integral method is a lot simpler for experimental as
well as analytical (computational) evaluation.

CONCLUSIONS
Fracture behaviour of a short fibre composite has been investigated. J integral has been evaluated
using the energy rate interpretation. its value is found to be independent of cracklength when the ratio of
crack length to specimen width (a/w) is larger than 0.35. For smaller crack lengths general material
damage away from the crack tip also inffuences the energy absorbed significantly. However, an
extrapolation method has been developed through which the crack tip energy may be separated from the
energy absorbed due to genera1 material damage. The J integral thus obtained is independent of crack
length and specimen length and its critical value is the same as obtained for a/w > 0.35 without
extrapolation. Further, it also agrees well with the critical stress intensity factor obtained using R-curve
approach in an earlier study.

Acknowledgements-This research work was sponsored by the Aeronautics Research and Development Board (Structures Panel),
Government of India. The authors would also like to thank Shri B. R. Somashekar. Prof. K. Rajaiah and Dr. K. N. Raju for their
interest in the work.

REFERENCES
[I] S. K. Gaggar and L. J. Broutman. Crack growth resistance of random fibre composites. J. Composite Matls, 9. 216227 (1975).
[?] S. K. Gaggar and L. J. Broutman, Strength and fracture properties of random fibre polyster composites. Fihre Sci. Tech. 9,
205-224 ( 1976).
684 B. D. AGARWAL et al.

[3] B. D. Agarwal and G. S. Giare, Crack growth resistance of short fibre composites: I-influence of fibre concentration, specimen
thickness and width. Fibre Sci. Tech. 15(4),283-298(1981).
[4] B. D. Agarwal and G. S. Giare, Crack growth resistance of short fibre composites: II-influence of test temperature. Fibre Sri.
Tech. 16(l), 19-28 (1982).
[S] B. D. Agarwal and G. S. Giare, Effect of matrix properties on fracture toughness of short fibre composites. Mat. Sci. Engng 5X2),
139-145(1982).
[6] B. D. Agarwal and G. S. Giare, Fracture toughness of short fibre composites in Modes II and III. Engng. Fracture Mech. 15U-2),
219-230(1981).
[7] G. C. Sih, P. D. Hilton, R. Badalliance, P. S. Shenberger and G. Villarreal, Fracture mechanics for fibrous composites. Analysis of
the Test Method for High Modulus Fibers and Composites, ASTM STP 521, 98-132. Philadelphia, (1973).
[8] J. F. Mandell, S. S. Wang and F. J. McGarry The extension of crack tip damage zones in fiber reinforced plastic laminates. J.
Composite Mat. 9(2), 266 (1975).
[9] Y. J. Yeow, D. H. Morris and H. F. Brinson, The fracture behaviour of graphite/epoxy laminates. Experimental Mech. 19(9),1-8
(1979).
[IO] D. H. Morris and H. T. Hahn, Fracture resistance characterization of graphite/epoxy composites, Composite Materials Testing
and Design (Fourth Conference), ASTM STP 617, 5-17 Philadelphia (1977).
[I I] J. Awerbuch and H. T. Hahn, K-calibration of unidirectional metal matrix composites. J. Composite Mater. 12, 222-237(1978).
[12] C. Bathias, R. Esnault and J. Pellas, Application of fracture mechanics to graphite fibre-reinforced composites. Composites 12
195-200(1981).
[I31 J. R. Rice, A path independent integral and the approximate analysis of strain concentration by notches and cracks. J. Appl.
Mech. 35, 379-386(1%8).
[14] K. B. Broberg, Crack growth criteria and non-linear fracture mechanics. J. Mech. Phys. Solids 19, 407418 (1971).
[I51 J. R. Rice and G. F. Rosengren, Plane-strain deformation near a crack tip in a power law hardening material. J. Mech Phys. Solids
16. l-12 (1968).
[I61 J. W. Hutchinson, Singular behaviour at the end of a tensile crack in a hardening material. J. Mech. Phy. Solids 16, 13-31 (1968).
[I71 F. McClintock, Plasticity aspects of fracture. Fracture Chap. 2, Vol. III (Edited by H. Liebowitz) Academic Press, New York
(1971).
[I81 J. A. Begley and J. D. Landes, The J integral as a fracture criterion. Fracture Toughness, ASTM STP 514, l-20 Philadelphia
(1972).
[I91 J. D. Landes and J. A. Begley,‘The effect of specimen geometry on JI,. Fracture Toughness, ASTM STP 514,24-39 Philadelphia
(1972).
[20] J. D. Landes and J. A. Begley, Recent developments in JI, Testing. Developments in Fracture Mechanics Test Methods
Standarization. ASTM STP 632, 57-81, Philadelphia (1977).
[21] P. D. Hilton and J. W. Hutchinson, Engng. Fracture Mech. 3,435-451 (1971).
[22] J. R. Rice, Mathematical analysis in the mechanics of fracture. Fracture (Edited by H. Leibowitz), Vol. II, Chap. 3, pp. 191-311.
Academic Press, New York (1%8).

(Received IS February 1983;received for publication 8 April 1983)