1270 AIAA JOURNAL VOL. 28, NO.

Mixed-Mode Bending Method for Delamination Testing

James R. Reeder* and John H. Crews Jr.t

NASA Langley Research Center, Hampton, Virginia 23665

A mixed-mode delamination test procedure was developed combining double cantilever beam (DCB) mode I
loading and end-notch flexure (ENF) mode II loading on a split unidirectional laminate. By loading with a lever,
a single applied load simultaneously produces mode I and mode II bending loads on the specimen. This
mixed-mode bending (MMB) test was analyzed using both finite-element procedures and beam theory to calcu-
late the mode I and mode II components of strain-energy release rate GI and GH, respectively. A wide range of
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

GI/GH ratios can be produced by varying the load position on the lever. As the delamination extended, the
GI/GH ratios varied by less than 5°7o. Beam theory equations agreed closely with the finite-element results and
provide a basis for selection of GI/GH test ratios and a basis for computing the mode I and mode II components
of measured delamination toughness. The MMB test was demonstrated using AS4/PEEK (APC2) unidirectional
laminates. The MMB test introduced in this paper is rather simple and is believed to offer several advantages over
most current mixed-mode test.

Nomenclature Introduction
a
b
= delamination length, m
= specimen width, m
F AILURES in composite structures often develop as de-
laminations between plies. Typically, such delaminations
initiate and propagate under the combined influence of normal
c = position of applied load on lever, m and shear stresses. Therefore, tests of delamination resistance
En = lamina longitudinal modulus, GPa should account for the effects of combined stresses. The pre-
£"22= lamina transverse modulus, GPa sent study addresses delamination testing with combined ten-
G = total mixed-mode strain energy release rate, J/m 2 sile normal stress (mode I) and sliding shear stress (mode II).
Gc — total mixed-mode delamination fracture toughness, Various approaches have been used to develop test specimens
J/m 2 with such combined normal and shear stresses on the delami-
GI = mode I strain energy release rate, J/m 2 nation plane. Unfortunately, however, several different types
GH = mode II strain energy release rate, J/m 2 of specimens are often needed to generate delamination tough-
GIC = delamination fracture toughness for mode I loading, ness data over a desired range of mixed-mode combinations.
J/m 2 For example, Fig. 1 shows interlaminar fracture toughness
GH C = delamination fracture toughness for mode II loading, curves measured using three different specimen types. 1 The
J/m 2 pure mode I values for delamination fracture toughness G Ic
G™ =mode I component of Gc for mixed-mode loading, were obtained using a split unidirectional laminate loaded as a
J/m 2 double cantilever beam (DCB). The pure mode II values GHC
G17c = mode II component of Gc for mixed-mode loading, were found using the same type of specimen but subjected to
J/m 2 three point bending; this type of test is called an end-notch
Gn = lamina longitudinal shear modulus, GPa flexure (ENF) test.2'3 However, the mode I and mode II com-
GO = lamina transverse shear modulus, GPa ponents of mixed-mode fracture toughness (G™ and G™c, re-
h = specimen half-thickness, m spectively) were generated using cracked-lap shear (CLS) and
k = stiffness of elastic foundation, N/m 2 edge-delamination tension (EDT) specimens.4 The use of dif-
L = specimen half-span, m ferent test configurations can involve different test variables
P = applied load, N and analysis procedures that can influence test results in ways
PI — mode I load, N that are difficult to predict. The purpose of this paper is to
Pu = mode II bending load, N
d = load-point displacement, m Specimen Type

X = elastic foundation parameter, 1/m D -Double cantilever beam (DCB)

T6000/F185 O -Cracked lap shear (CLS)
A -Edge delamination tension (EDT)
O -End notch flexure (ENF)

Received Sept. 19, 1988; revision received Aug. 7, 1989. Copyright

© 1990 American Institute of Aeronautics and Astronautics, Inc. No
copyright is asserted in the United States under Title 17, U.S. Code.
The U.S. Government has a royalty-free license to exercise all rights
under the copyright claimed herein for Governmental purposes. All
other rights are reserved by the copyright owner.
*Research Engineer. Materials Division, Mechanics of Materials
Branch.
tSenior Scientist. Materials Division, Mechanics of Materials Fig. 1 Delamination toughness for mixed mode I and mode II
Branch. loading.
 JULY 1990 MIXED-MODE BENDING METHOD FOR DELAMINATION TESTING 1271

introduce a new test apparatus that can be used to measure In the Arcan test configuration, part c of Fig. 2, a split
delamination toughness over a wide range of mode I/II ratios unidirectional laminate is bonded between two metal fixtures
as well as pure mode I and mode II. that can be loaded to produce various mixed-mode conditions
First, the current methods for mixed-mode delamination at the delamination front. 7 But, as with the CLS and EDT
testing will be briefly reviewed. Next, the proposed mixed- tests, the mode I/II ratio must be determined by a numerical
mode bending (MMB) test is described. Then the total strain analysis. Also, bond failures can limit the Arcan test use,
energy release rate G and its mode I(GO and mode II(Gn) especially for tough laminates.
components will be evaluated for the MMB test specimen using The asymmetric DCB test, proposed by Bradley and Co-
a finite-element analysis. In addition, closed-form equations hen,8 avoids most of the problems found with the first three
for GI and Gu will be developed using simple beam theory with methods. As shown in part d of Fig. 2, this approach involves
modifications to improve their accuracy. Finally, the MMB loading the arms of a unidirectional DCB specimen with two
test method is demonstrated by testing graphite/PEEK different loads. The loads can be selected to produce the full
(APC2) specimens over a wide range of Gi/Gu ratios. Delam- range of mode I/II ratios. Equal and opposite loads produce
ination toughness data are presented in terms of the mode I a pure mode I delamination, and equal loads produce a pure
and mode II components of delamination fracture toughness. mode II delamination. Unfortunately, the asymmetric DCB
approach requires a complex loading system to control the two
applied loads simultaneously.
Current Mixed-Mode Delamination Tests The mixed-mode flexture test, proposed by Russell and
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

This section briefly reviews current approaches for mixed- Street,9 is shown in part e of Fig. 2. This test specimen is
mode delamination testing. This review provides background similar to the CLS specimen but is loaded in three-point bend-
for the new approach presented in the next section. Combined ing. Unfortunately, different arm thicknesses are required to
mode I and mode II delamination fracture toughness tests produce different mode I/II ratios. This requires that speci-
usually employ a specimen containing an artificially intro- mens be fabricated with the delamination starter at different
duced delamination. The specimen is loaded until the delami- ply interfaces. Also different arm thicknesses can influence the
nation grows. Measured load and delamination length can stress distribution ahead of the delamination10 and, therefore,
then be substituted into strain energy release rate equations to may influence toughness measurements.
calculate the delamination toughness. The variable mixed-mode test in part f of Fig. 2 was pro-
A sketch of the CLS specimen is shown in part a of Fig. 2. posed by Hashemi et al.11 A pure mode II condition is created
Uniaxial loading is applied to one arm of a split unidirectional when the delamination tip is centered under the midspan ten-
laminate. The load transfer to the other arm causes interlami- sion load. The mode I/II ratio increases as the delamination
nar normal stresses (mode I) and interlaminar shear stresses extends toward the left load point. A pure mode I condition
(mode II). Although the CLS specimen can be tested in con- exists when the delamination is under the left load point. As a
ventional tension test machines, it has several serious limita- result, the full range of mode I/II ratios can be produced.
tions. First, the mode I/II ratios cannot be calculated by sim- However, the ratio changes as the delamination grows. This
ple closed form stress analyses, and, therefore, a numerical could complicate the data analysis—especially for large incre-
analysis is required. Further, because large rotations can result ments of unstable growth. Also, when the delamination tip is
from the load eccentricity at the delamination front, a geomet- near either load point, simple closed form equations for GI and
rically nonlinear numerical analysis may be required to evalu- Gn will not account for the complex effects of load concentra-
ate GI and G n . 5 Also, different ply layups are required to tions or loading fixture stiffness.
create different mode I/II combinations, and only a rather
narrow range of ratios is attainable.
The EDT specimen, shown in part b of Fig. 2 was developed Mixed-Mode Bending Test
by O'Brien.4 A Specimen with a layup such as ( ± 35/0/90)5 is The MMB test simply combines the mode I DCB and the
loaded in tension, and the mismatch in the Poisson's ratios of mode II ENF tests. This is achieved by adding an opening-
the plies causes high edge stresses at the 0/90 ply interfaces. mode load to a mid-span loaded ENF specimen as shown in
The load-induced mode I and mode II stresses at these inter- part a of Fig. 3. This additional load separates the arms of the
faces can initiate edge delaminations. Unfortunately, however, split unidirectional laminate as in a DCB test. The relative
hygrothermal interlaminar stresses also exist at this interface magnitudes of the two applied loads determines the mixed-
and can seriously reduce the measured delamination tough- mode ratio at the delamination front. By applying these two
ness.6 Also, numerical analyses are required to calculate the loads through a lever and hinge apparatus as shown in part b
interlaminar G! and Gn components in the EDT test. of Fig. 3, the test can be conducted by applying a single load.

Arcan, Hashin. (a) Test specimen and loading.

& Voloshin6
Loading lever
(a) Cracked lap shear. (b) Edge delamination tension. (c) Arcan.
Hinge
Specimen

Apparatus base

Hashemi. Kinloch,
Bradley & Cohen Russell & Street & Williams10
11 f t t 1111 t t 11 11 i ) i i t i i
(d) Asymmetric DCB (e) Mixed-mode flexure, (f) Variable mixed-mode. (b) Schematic diagram of apparatus
Fig. 2 Mixed mode 1 and mode II delamination test specimens. Fig. 3 Mixed-mode bending specimen and test apparatus.
 1272 J. R. REEDER AND J. H. CREWS JR. AIAA JOURNAL

Finite element
40

c=41 mm
20 a=25 mm
L=50 mm

0.00 0.25 0.50 0.75 1.00 1.25

Load point displacement 6 (mm)

Fig. 6 Comparison of computed and measured load-displacement

curves.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

Fig. 4 Photograph of MMB test apparatus.

3.1 mm (not to scale)

-*— [—|—j—|—i—T—i—i—i—i—i—i'—
—1——rT|

t Mr
50 mm

110 Elements
411 Nodes
819 Degrees of freedom
8 Noded quadrilateral element
0 20 40 60
Fig. 5 Finite-element model for MMB specimen.
Load position c (mm)

Fig. 7 Load position for mixed-mode testing.

The loading position c determines the relative magnitude of the
two resulting loads on the specimen and, therefore, determines
the mixed-mode delamination ratio. Pure mode II loading thickness, an effective modulus was calculated from a three-
occurs when the applied load is directly above the beam point bend test.13 Measured midspan displacements and loads
midspan (c = 0). Pure mode I loading can be achieved by re- were used with beam theory to calculate the longitudinal mod-
moving the beam and pulling up on the hinge. ulus En for the laminate. An effective ^n of 116 GPa was
A photograph of the MMB test apparatus is shown in Fig. determined by this approach, compared to the uniaxial tension
4. The loading lever is an aluminum I-beam weighing only 6 N, value of 129 GPa from Ref. 14. A transverse modulus £22 of
which was assumed to be a negligible weight. The lever is 10.1 GPa, a Poisson's ratio j>12 of 0.329, and shear modulus
several orders of magnitude stiffer than the specimen and, G 1 2 of 5.5 GPa were used.14
therefore, was assumed to be rigid. The lever load, the The finite-element modeling was evaluated by comparing
midspan load, and the left support reaction are applied the computed load-displacement results with measured results.
through bearing-mounted rollers to reduce frictional forces. The solid line in Fig. 6 represents the computed finite-element
The right end of the specimen is loaded through high-quality, results, and the symbols show measurements taken during
extruded aluminum hinges bonded to the specimen arms. The loading and unloading of the MMB apparatus. The load-point
specimen in this photograph is a 24-ply graphite/PEEK unidi- displacement was determined from the crosshead position.
rectional laminate, 102-mm long, 25-mm wide, and 3.1-mm The solid curve agrees with the test data very well in the lower
thick. The apparatus is mounted on a thick steel base. load range and slightly overestimates the displacements in the
upper range. The discrepancy may be caused by geometric
nonlinearity, which was not accounted for by the finite-ele-
Stress Analysis ment analysis. However, the correlation in this figure does
This section presents the stress anlysis of the MMB test suggest that the finite-element model is sufficiently accurate to
specimen and focuses on the calculation of strain energy re- analyze the MMB specimen. Crack-tip forces and displace-
lease rates. The G^Gn ratio is needed to resolve the measured ments were used in the virtual crack closure technique15 to
mixed-mode delamination fracture toughness Gc into its mode calculate Gj and Gn for the MMB specimen.
I and II components, Gg and GH C , respectively. First, a finite- As just mentioned, the load position c determines the ratio
element analysis was used to provide a basic understanding of of mode I and mode II. To establish a relationship between c
the strain-energy release rate during MMB testing. Next, a and the Gi/Gn ratios, the finite-element analysis was repeated
more convenient beam-theory analysis is presented, and its for several values of c. This relationship is shown in Fig. 7,
results are compared with those from the finite-element analy- where the symbols are the computed values and the solid line
sis. Finally, the simple beam theory analysis was modified to is a best-fit curve. The test data shown previously in Fig. 1
improve its accuracy. suggest that a mixed-mode toughness curve could be estab-
lished reasonably well using five test cases: three mixed-mode
and the two pure mode toughness tests. In the present study,
Finite-Element Analysis mixed-mode ratios of 4/1, 1/1, and 1/4 were selected. As
The MMB specimen was modeled and loaded as shown in shown in Fig. 7, the three corresponding c values are 95, 41,
Fig. 5. Eight-noded quadrilateral elements were used with the and 27 mm, respectively. Although the present study was lim-
MSC NASTRAN finite-element code.12 To account for the ited to three mixed-mode ratios, the MMB apparatus can be
effect of an uneven fiber distribution through the specimen used to measure any G\/G\\ ratio from zero to approximately
 JULY 1990 MIXED-MODE BENDING METHOD FOR DELAMINATION TESTING 1273

five. Notice that G\/G\\ is zero for c less than about 18 mm. Fig. 7 to establish the load position c for each of the three
Below this value, the mode I loading is not large enough to mixed-mode cases.) Within the useful test range (delamination
overcome the crack-face normal stresses produced by the lengths from 25 to 45 mm), the G\/G\\ ratios deviated from the
mode II loading. Hence, the delamination does not open, and nominal values by only about 5%. This small deviation shows
Gj must be zero within this range despite the nonzero c values. that the Gi/Gn ratio can be assumed to be constant during
As just mentioned, the mode II tests were conducted with c delamination growth for the MMB test.
equal to zero.
Finite element analyses of the MMB specimen were con-
Beam-Theory Analysis
ducted to determine the variation of the strain energy release
rate during delamination. A half-span length of 50 mm was Although the finite-element analysis provided an accurate
used, and analyses were conducted for delamination lengths strain energy release rate analysis of the MMB specimen, a
from about 20 to 45 mm. Figure 8 shows computed values of closed-form analysis offers many advantages in setting up
total strain energy release rate G for the pure mode I and mode MMB specimen tests and evaluating test results. A closed-form
II cases plotted over a range of delamination lengths. The two analysis also shows the functional relationships among the test
curves in this figure represent limits for the mixed-mode cases. parameters, which improves the basic understanding of the
For convenience, G values are normalized by the square of the MMB test. This section presents strain energy release rate
load-point displacement. If the curves are interpreted as G equations based on beam theory. The MMB loading was repre-
variations during delamination growth under constant dis- sented by a superposition of simple mode I and mode II load-
ings, equivalent to those used with DCB and ENF tests, respec-
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

placement <5, then they can be compared with toughness Gc to

predict if delamination growth is unstable (G continues to tively. Thus, strain energy release rate equations from the
equal or exceed Gc as the delamination grows) or is stable (G literature on DCB and ENF tests could be combined to obtain
falls below Gc) requiring additional loading for subsequent the desired equations for the MMB test.
growth. The negative slope of the mode I curve shows that Part a of Fig. 11 shows the MMB loading expressed in terms
delamination growth in a DCB test should be stable. In con- of the applied load P, the loading lever length c, and the
trast, the peak of the mode II curve at 35 mm suggests delam- specimen half-span L. As shown in part b of Fig. 11, the mode
ination growth would be unstable for delamination lengths less I component of this loading is
than this value but stable for longer lengths.
The G /d2 curves for the three mixed-mode loading cases are
shown in Fig. 9. The case where mode I is dominant (G\/
Gn = 4/l) has a curve with a negative slope, which indicates
stable delamination growth as in the pure mode I case shown Simple beam theory analysis of the DCB specimen leads to
in Fig. 8. The 1/1 and the 1/4 curve both indicate a region of
unstable growth below delamination lengths of 25 and 30 mm, (1)
respectively. As expected, a higher mode II component in the
mixed-mode test results in a larger region of instability. Substituting for P/ leads to the following equation for G/ of the
Ratios of GI/GU are plotted vs delamination length in Fig. MMB test.
10. Preferably, this ratio should be constant throughout the
test range of delamination lengths. The horizontal lines in Fig. (2)
10 represent the 4/1, 1/1, and 1/4 mixed-mode ratios and pass
through the computed values for the 25 mm delamination
length. (Recall that a 25 mm delamination length was used in

c = 41 mm

c = 27 mm

0 20 30 40
Delamination length a (mm)
Fig. 10 Gi/Gn ratio for mixed-mode loading.
30 40 50
Delamination length a (mm) Applied load
Fig. 8 Stain energy release rates for pure mode I and mode II load-
ing. Loading lever
0.06
JrH
0.05

0.04
t-H
(GJ/m4)
0.03 Specimen

0.02

0.01

0
20 30 40 50
Delamination length a (mm) (a) MMB specimen loading, (b) Mode I loading, (c) Mode II loading.
Fig. 9 Strain energy release rates for mixed-mode loading. Fig. 11 Superposition anlysis of loading on the MMB specimen.
 1274 J. R. REEDER AND J. H. CREWS JR. AIAA JOURNAL

a = 25 mm
Finite element Finite element //'
Gj/Gjj = 1/1 Finite element analysis analysis analysis

(kJ/m2) (kJ/m2)

Beam theory.
equation(4)
Beam theory, equation (6)
0.0
0.00 0.02 0.04 0.0 0.5 1.0 1.5
P2 (kN)2 P2 (kN)2
(a) Mode I (b) Mode II
0.0 0.1 0.2 0.3
Fig. 13 GI and Gn components from beam-theory and finite-element
P2 (kN)2
analysis.
Fig. 12 Comparison of beam-theory and finite-element results.

Part c of Fig. 11 shows the mode II portion of the MMB Modified beam theory (elastic foundation
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

loading. Note that the right-end loading has been divided ' and shear deformation)
equally between the two equal-stiffness arms of the specimen.
This is equivalent to the conventional loading of the ENF test.
For the ENF test, the mode II bending load is

1/4

as shown in part c of Fig. 11. The following equation for Gn

of the ENF test was presented in Ref. 3.
0.0
(3) 0.0 0.1 0.2 0.3
I6b2h3£n
P2 (kN)2

Substituting for Pn, the corresponding equation for Gn of the Fig. 14 Comparison of G results for mixed-mode loading cases.
MMB test is

His analysis of an isotropic DCB specimen was extended to an

-(c+L) 2 (4) orthotropic DCB specimen by replacing E with Eu and E22.11
l6b2H3L2E}
By dividing Eq. (2) by Eq. (4), the G}/GU ratio for the MMB \2P2 2a
(7)
test can be expressed as
4f(3c-L)l2 L where
(5)
~3L(c+L)J

Notice that G^Gn is only a function of load position c and

half-span length L. The Gj/Gn ratio is zero for c =L/3 (c = 17 and
mm for the present study), and Eq. (5) is invalid for smaller c
values because this model does not account for contact be- = 2bE22/h
tween the two arms of the specimen. The total strain energy
release rate for the MMB test is obtained by adding Eqs. (2) The beam theory equations for strain energy release rate can
and (4). be further improved by accounting for the shear deformation
energy associated with bending. Adding the shear deformation
3a2P2 component of strain energy release rate18 to Eq. (7) leads to the
G= [4(3c-L) 2 + 3(c+L) 2 ] (6) following modified beam theory equation for G/ in the MMB
l6b2h3L2En
test.
Equation (6) is compared with the finite element results in
Fig. 12. These results for a 25-mm delamination with G^Gn 3P 2 (3c-L) 2 2 2a J_ h2En
+ (8)
equal to unity show that Eq. (6) underestimates G by about X + X 2 + 10G 1 3
15%. To analyze this discrepancy, Gj and Gn were calculated
using Eqs. (2) and (4) and are compared with the correspond- Similarly, adding the shear deformation term from Ref. 19 to
ing finite-element results in Fig. 13. The G! values are 18% low Eq. (4) results in a modified beam theory equation for Gn in
and the GH values are only 6% low. This indicates that most the MMB test.
of the error in Eq. (6) can be attributed to the Gj component
and, therefore, to the beam theory equation for the DCB test. 9P 2 (c+L) 2 Q.2h2En
(9)
Kanninen 16 introduced an improved beam theory equation \6b2h3L2Eu
for the DCB test. He recognized that simple beam theory did
not properly model the interaction between the two arms of the Assuming the unidirectional composite specimens used in the
DCB specimen. The two arms are not fixed against rotation at present study are transversely isotropic, the shear modulus G13
the delamination tip as assumed in simple beam theory. In- in these two equations can be replaced by G\2.
stead, they rotate slightly due to the elastic support that they The total G values for the MMB test were recalculated using
provide one another. To account for this, Kanninen assumed the modified beam theory of Eqs. (8) and (9). Figure 14 com-
that each arm was a beam supported by an elastic foundation. pares the G calculations for the three-beam theory equations
 JULY 1990 MIXED-MODE BENDING METHOD FOR DELAMINATION TESTING 1275

with the finite-element results. For the earlier case of Gi/Gu mens. These specimens were 25-mm wide and contained a
equal to unity, G calculated with the elastic foundation correc- Kapton film delamination starter at one end of the specimen.
tion (dash-dot curve) agreed with the finite-element values The Kapton film was 25-mm wide, 25-mm long, 13-^m thick,
(solid curve) with an error of about 8% compared to the 14% and located at the specimen midplane. To produce an initial
error for the simple beam theory (dashed curve). The addi- delamination, each specimen was precracked to about 30 mm
tional modification for shear deformation led to the dash- using the 4/1 mixed-mode ratio. No significant fiber bridging
double dot curve, which is within about 6% of the finite-ele- was observed at this mixed-mode ratio. The G\/G\\ mixed-
ment results. The corresponding errors for the 4/1 and 1/4 mode ratios of 4/1, 1/1, and 1/4 were produced by selecting
cases were also only about 6%. the load positions from Fig. 7. Pure mode II tests were con-
Next, Eqs. (8) and (9) were used to recalculate the G\/G\\ ducted by applying the load at the specimen midspan (c = 0).
ratios for the three mixed-mode test cases. Recall that the For the pure mode I case, the loading lever was removed, and
values of c for the three Gi/G n ratios of 4/1, 1 /1, and 1 /4 were an upward load was applied directly through the upper hinge
selected earlier using finite-element results for a 25-mm delam- on the specimen. Each specimen was loaded under displace-
ination (see Fig. 7). The recalculated G\/G\\ ratios are shown ment control with a rate of 0.5 mm/min until the delamination
as the solid curves in Fig. 15 and are compared with the dis- grew. The maximum load was recorded, and the delamination
crete finite-element values shown by symbols. These curves length was measured visually at the specimen edges. The edges
agree very well with the finite-element values. Within the 25- to had been coated with white water soluble typewriter correction
45-mm range of delamination lengths, the modified beam the- fluid to make the delamination easier to see.
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

ory ratios vary by about 3,5, and 8% from the finite-element The delamination growth was usually stable, and delamina-
results for the 4/1, 1/1 and 1/4 cases, respectively. tion growth increments were about 5 mm. The tests were
Compared to the finite-element analysis, the modified beam stopped when the delamination reached 45 mm in length, and
theory of Eqs. (8) and (9) appear to provide acceptably accu- specimens were split apart to examine the markings usually
rate values of Gj and Gn for the MMB test. Before testing, produced by the loading-unloading sequence. These markings
Eqs. (8) and (9) can be used to select the loading positions that provided accurate measurements of delamination length and
produce desired G\/G\\ test ratios. After testing, measured where used to verify the measurements taken during testing.
values of delamination length and the corresponding MMB The recorded values for load and delamination length were
specimen loads can be substituted into these equations to cal- used in Eqs. (8) and (9) to calculate Gg and Gff c , respectively.
culate the mode I and mode II delamination fracture toughness Test results are presented in Fig. 16 with G™ plotted against
components G™ and G™c. Because both the finite-element and Gffc • Each symbol represents a toughness measurement corre-
beam-theory analyses presented here are linear, they do not sponding to growth from the precrack. The data exhibited
account for possible effects due to geometric nonlinearities. considerable scatter, but this is somewhat typical of delamina-
Such effects should be the subject of future research. tion testing. The solid curve fitted through the data can be
viewed as a delamination failure criterion for mixed-mode
loading. The curve is nearly horizontal in the region where
Test Results G 1 7/G I 7 C >1 indicating that the toughness is nearly indepen-
As mentioned in the preceding discussion, tests were con- dent of GH, and therefore, delamination growth was con-
ducted with 24-ply unidirectional AS4/PEEK (APC2) speci- trolled by mode I loading. In the region where Gi/G n < 1, the
curve is sloped indication that both mode I and mode II load-
ing influence the delamination toughness.
Fracture toughness was also measured when the precracks
were initiated from the Kapton insert using the 4/1 loading
condition. The average toughness from the precracking dif-
fered by only 2% from the data shown in Fig. 16 for the 4/1
O , D , A - Finite element analysis case. This suggests that precracking may not have been neces-
sary.
Modified beam theory, ratio of
equations (8) and (9)
Concluding Remarks
A MMB delamination test procedure has been presented for
c = 27 mm a split unidirectional laminate. The mixed-mode loading was
-O——————O- created by combining the mode I loading for the DCB test with
0 20 30 40 50 that for the mode II ENF test. This combined loading was
Delamination length a (mm) produced using a loading lever, and the ratio of mode I to
mode II was varied by changing the load position on the lever.
Fig. 15 Comparison of G\/Gu results. Both finite-element analysis and beam-theory analyses were
conducted to determine the mode I and mode II components
of strain energy release rate GI and G n , respectively. The MMB
test procedure was demonstrated by measuring the mixed-
mode delamination fracture toughness of AS4/PEEK (APC2)
unidirectional laminates.
AS4/PEEK (APC2) Finite element analyses were conducted to determine the
(kJ/m2) J loading lever lengths necessary to produce the desired mixed-
mode ratios of 4/1, 1/1, and 1/4. The finite-element analysis
showed that these ratios varied by less than 5% over a 20-mm
1/4 test range of delamination lengths. Therefore, the GI/GU ratio
can be assumed to be independent of delamination length.
Beam-theory equations from the literature for DCB and
ENF tests were used with a superposition procedure to develop
equations for GI and Gn for the MMB test. These equations
were then modified using elastic foundation and shear defor-
Fig. 16 Delamination toughness measurements from the mixed- mation analyses. The resulting modified equations for GI and
mode bending test. GH were within about 6% of the finite-element results. Mea-
 1276 J. R. REEDER AND J. H. CREWS JR. AIAA JOURNAL

7
sured delamination lengths and loads from the MMB tests Arcan, M., Hashin, Z., and Voloshin, A., "A Method to Produce
were substituted into these equations to determine the mode I Uniform Plane-Stress States with Applications to Fiber-Reinforced
Downloaded by UNIVERSITY OF NEW SOUTH WALES on July 13, 2022 | http://arc.aiaa.org | DOI: 10.2514/3.25204

and mode II components of delamination toughness during Materials," Experimental Mechanics, Vol. 8, April 1978, pp. 141-146.
8
mixed-mode delamination. Pure mode I and mode II tests were Bradley, W. L., and Cohen, R. N., "Matrix Deformation and
Fracture in Graphite-Reinforced Epoxies," Delamination and
also conducted by simplifying the MMB test to produce DCB Debonding of Materials, edited by W. S. Johnson, American Society
and ENF loadings, respectively. This approach provided de- for Testing and Materials, Philadelphia, ASTM STP 876, 1985 pp.
lamination fracture toughness data over a wide range of G\/G\\ 389-410.
ratios using identical test specimens and procedures. 9
Russell, A. J., and Street, K. N., "Moisture and Temperature
The MMB test is a rather simple and direct combination of Effects on the Mixed-Mode Delamination Fracture of Unidirectional
DCB and ENF tests and seems to offer several advantages over Graphite/Epoxy," Delamination and Debonding of Materials, edited
most current mixed-mode delamination tests. Many of the by W. S. Johnson, American Society for Testing and Materials,
data reduction procedures that have been developed for the Philadelphia, ASTM STP 876, 1985, pp. 349-370.
10
DCB and ENF tests should be applicable to the MMB test Crews, J. H., Jr., Shivakumar, K. N., and Raju, I. S., "Factors
Influencing Elastic Stresses in Double Cantilever Beam Specimens,"
because of its similarities with these pure-mode tests. Also, NASA TM-89033, Nov. 1986.
DCB and ENF studies of test parameters such as insert thick- H
Hashemi, S., Kinloch, A. J., and Williams J. G., "Interlaminar
ness and precracking may be applicable to the MMB test. Fracture of Composite Materials," 6th ICCM & 2nd ECCM Confer-
Therefore, it should be relatively easy to use MMB testing ence Proceedings, Elsevier Applied Science, London, Vol. 3, July
beyond the initial procedures of the present study. 1987, pp. 3.254-3.264.
12
MSC NASTRAN User's Manual, Version 64, The MacNeal-
Schwendler Corp., Los Angeles, Nov. 1985.
References 13
O'Brien, T. K., Murri, G. B., and Salpekar S. A., "Interlaminar
Johnson, W. S., and Mangalgiri, P. D., "Influence of the Resin on Shear Fracture Toughness and Fatigue Thresholds for Composite Ma-
Interlaminar Mixed-Mode Fracture," Toughened Composites, edited terials," NASA TM-89157, USAAVSCCOM TM 87-B-9, Aug. 1987.
14
by Norman J. Johnston, American Society for Testing and Materials, O'Brien, T. K., "Interlaminar Fracture Toughness Testing of
Philadelphia, ASTM STP 937, 1987, pp. 295-315. Composites," (to be published); supporting data available from
2
Barrett, J. D., and Foschi, R. O., "Mode II Stress Intensity Fac- ASTM Headquarters, request RR D30.02.02.
15
tors for Cracked Wood Beams," Engineering Fracture Mechanics, Raju, I. S., "Calculation of Strain-Energy Release Rates with
Vol. 9, No. 2, 1977, pp. 371-378. Higher Order and Singular Finite Elements," Engineering Fracture
3
Russell, A. J., "On the Measurement of Mode II Interlaminar Mechanics, Vol. 28, No. 3, 1987, pp. 251-274.
16
Fracture Energies," Defence Research Establishment Pacific, Victo- Kanninen, M. F., "An Augmented Double Cantilever Beam
ria, Canada, DREP Materials Rept. 82-0, Dec. 1982. Model for Studying Crack Propagation and Arrest," International
4
O'Brien, T. K., "Mixed-Mode Strain-Energy-Release Rate Effects Journal of Fracture, Vol. 9, No. 1, March 1973, pp. 83-92.
17
on Edge Delamination of Composites," Effects of Defects in Compos- Shivakumar, K. N., Personal communication, Analytical Services
ite Materials, American Society for Testing and Materials, Philadel- & Materials, Inc., Hampton, VA, 1988.
18
phia, ASTM STP 836, 1984, pp. 125-142. Aliyu, A. A., and Daniel, I. M., "Effects of Strain Rate on
5
Johnson, W. S., "Stress Analysis of the Cracked-Lap-Shear Spec- Delamination Fracture Toughness of Graphite/Epoxy," Delamina-
imen: An ASTM Round-Robin," Journal of Testing and Evaluation, tion and Debonding of Materials, edited by W. S. Johnson, American
Vol. 15, No. 6, Nov. 1987, pp. 303-324. Society for Testing and Materials, ASTM STP 876, 1985, pp. 336-348.
6 19
O'Brien, T. K., Raju, I. S., and Garber, D. P., "Residual Thermal Carlsson, L. A., Gillespie, J. W., and Pipes, R. B., "On the
and Moisture Influences on the Strain Energy Release Rate Analysis of Analysis and Design of the End Notched Flexure (ENF) Specimen for
Edge Delamination," Journal of Composites Technology & Research, Mode II Testing," Journal of Composite Materials, Vol. 20, Nov.
Vol. 8, No. 2, Summer 1986, pp. 37-47. 1986, pp. 594-604.