NASA TECHNICAL

MEMORANDUM

104210

AN EVALUATION
OF MIXED-MODE
DELAMINATION
FAILURE CRITERIA
#

J. R. REEDER

FEBRUARY

1992

National Aeronautics and

Space Administration
Langley
Research
Center
Hampton, Virginia 23665

(NASA-TU-I042]_)

MIX_D-MDOE
(NASA)
52

AN

_CLAMINATInN
[_

EVALUATION

FAILUR_

OF

N92-1Q947

C_ITLRIA
CSCL
G_/24

Uncles
0071_07

ABSTRACT
Many different
delamination

toughness,

over the full mode

(MMB)

failure

criteria

but few sets of mixed-mode

I opening

to mode

test was used to measure

a state-of-the-art

composite

over the full mixed-mode

responses

of the different

examined.

An evaluation
was performed,

analyzed.
failure

mechanism

were compared

the bilinear
criterion.

provide

and the range

criterion
Since

provides
general

of a brittle

criteria

which

surfaces

have been

based

tested.

were best modeled

with the new bilinear

failure

testing

a critical
guidance

differed

evaluation

with the more

to characterize

of the mixed-mode

for selecting

an appropriate

in the

criterion

criteria

criteria

The

well with
linear

response,

a composite
failure

in the

criterion.

simple

failure

was

The responses

could be modeled

in their mixed-mode

will be needed

reported

The different

of the three materials

the materials

were also

on a change

response

but could also be modeled

failure

by each criterion

surfaces.

composite

epoxy

into the different

modeled

was developed

from the delamination

bending

and a tough thermoplastic

fracture

of responses

criterion

The mixed-mode

To gain insight

failure

of the tough thermoplastic

mode delamination
paper

range.

composite,

of several

to the failure

response

epoxy

the delamination

failure

data exist that are consistent

toughness

materials,

observed

of the two epoxies

failure

toughened

A new bilinear

for mixed-mode

II shear load range.

the delamination

composite,

literature

have been suggested

material.

failure
mixedThis

and should

for other materials.

INTRODUCTION

Delamination
Delamination
measured

is a primary

toughness

under mode

with the double

Several

fracture

toughness.

needed

to measure

mode

delamination

1. Unfortunately

mode bending

(MMB)

to measure

avoid geometric

nonlinearities

test allows

almost

mixed-mode

data.

Once delamination
will be required
paper

is to evaluate

of mixed-mode

toughness

response.
combines

delamination

of test specimens

Recently

were

however,

test

the mixed-

the DCB and ENF loadings,

toughness

when testing

criteria

was

and then redesigned3

tough composites.

inconsistencies

have been suggested

but these criteria

to

The
to be tested

present

in

of delaminations
accurately,

failure

data obtained

criteria

for

were often based on inconsistent

that accurate

the safety of composite

mixed-mode

II

over the full range of mode I and

Therefore,

failure

can be predicted
to ensure

mixed-mode

in

data sets can be avoided.

It is important

so that the extension

I and mode

of mode I and mode II loading

configuration.

growth,

are not just loaded

of mode

kinds

delamination

mixed-mode

delamination

sets of toughness

failure

encountered

toughness

Many different

developed

a mixture

flexure

it was unclear what effect the different

any combination

with the same test specimen

predicting

delaminations

toughness

test 2, which simply

mixed-mode

materials.

test and the end notch

different

fracture

had on the measured

previous

however,

In the past, several

configurations

MMB

(DCB)

II but grow under

composite

load and mode II shear load can be

types of tests have been used to measure

II combinations

developed

beam

In structures

I or pure mode

loading.

mode of laminated

I opening

cantilever

(ENF) test, respectively.

pure mode

failure

mixed-mode

in structures

failure

by comparing

using the MMB test.

be

can be predicted.

fewer component
structures.

criteria

and full scale tests

The purpose

of this

them to consistent

sets

The redesigned mixed-mode bending (MMB) test was used to measure

delamination
epoxy

toughness

composite,

range.

failure

surfaces

responses

criteria

have been

responses

failure

modeled

criterion

in the literature

was introduced.

were compared

to the different

criterion

for each material

was selected.

The redesigned
delamination
of testing

MMB test apparatus

toughness

over virtually

Consistent

of three different

In addition,

response

failure

and the
a

of the three

criteria,

and the best

TESTING

was used to measure

composite

the entire mixed-mode

sets of data with which

was evaluated.

mixed-mode

into

of delamination

was performed,

The failure

TOUGHNESS

to gain insight

Next, an evaluation

by each criterion

materials

toughened

over the full mixed-mode

were examined

materials.

reported

a state-of-the-ann

composite

of the composites

of the different

which

range of material
new bilinear

composite,

and a tough thermoplastic

The fracture

the failure

of a brittle epoxy

the

to compare

materials.

the mixed-mode
This test is capable

range with consistent

proposed

failure

test conditions.

criteria

therefore

were obtained.
The materials
toughness

properties.

IM7/977-2
strength

used in this study were chosen

AS4/3501-6

is a state-of-the-art

toughened

IM7 fiber and an epoxy

thermoplastic

additive.

graphite/thermoplastic

is a commonly

matrix

which

The AS4/PEEK
composite

epoxy

and therefore

epoxies.

The elastic

1. These

properties

were used in the calculation

calculation

properties

is very sensitive

used brittle
composite.

epoxy

radically

of a high

with a
is a tough

different

of these three materials

of fracture

of

composite.

It consists

has been toughened

to the longitudinal

a wide range

(polyether-ether-ketone)

thermoset

toughness

to represent

from the
are listed in Table

toughness.

modulus

E11,

Because
it was

the

measured using a 3-point bend test with a 3 inch span length.

bending

because

shown

for composite

to differ significantly

is a bending
transverse

E22

2h.

specimen

specimen

the lever
ratio

load position,

Gi/Gll.

which

the MMB

apparatus

Each material
the two pure-mode
1/4 cases which

in Figure

panels.

The panels

b, and nominally

were

were

0.12 in.
of the

insert was used in the epoxy

Hinges

I, so that the starter

shown

open resulting
bends

c. The value

provided

were
a 1 inch

in Figure

1 uses a lever to apply

The load on the top hinge

in mode I loading

the specimen

creating

similar

a mode II loading
is controlled

by

determines

the mixed-mode

uses a saddle

mechanism

to hold the loading

mid-plane

and on either

to drastically

when testing

to that of the

of c therefore

apparatus

has been shown

can develop

these properties

The ratio of the mode I to mode II loading

The redesigned

This configuration

A Teflon

to a split beam specimen.

rollers just above the specimen

errors

starter.

The load at the fulcrum

to the ENF test.

to the

a 0.5 mil thick insert at the midplane

MMB test apparatus

tends to pull the delamination

similar

test

a.

I and mode II loadings

DCB test.

Therefore,

were 6 in. long, 1 in. wide,

as shown

length

The redesigned

is not as sensitive

G13.

has been

4 and the MMB

insert was used in the PEEK specimens.

to the specimens

initial delamination

modulus,

into 24 ply unidirectional

contained

while a Kapton

modulus

in

6.

to act as a delamination

specimens
bonded

Each

which

measured

measured

calculation

and shear modulus

were made

cut into test specimen

mode

from an axially

from the literature3,5,

The materials

thick,

a flexuraUy

type of test. The toughness

modulus

obtained

materials

was measured

Ell

reduce

side of the test specimen.

geometric

tough materials3.

nonlinearity

The half span length

L of

was 2 inches.
was tested
cases.

in at least three mixed-mode

The three

corresponded

mixed-mode

to c values
4

ratios

ratios tested

of 3.83 in.,

(GI/GII)

and at

were the 4/1, 1/1, and

1.66 in., and 1.09 in.,

respectively.

The AS4/3501-6 material was also tested at a 1/20 ratio with a c

value 0.85 in. while the IM7/977-2 material was tested at the 1/2 ratio with a c value
of 1.30 in. The pure mode I toughness was tested using a standard DCB
configuration.

The pure mode II toughness was tested using the MMB apparatus

with a c value of 0. This is equivalent to an ENF configuration.

All the tests are

consistent since the mixed-mode test is simply a combination of the pure-mode tests.
Five tests at each test configuration were performed on the epoxies while only 3 tests
were performed for the PEEK material.
The specimens were loaded in displacement control at a rate of 0.02 in/min at
the lever loading point. The load-displacement response was recorded, and the
critical load used in Gc calculations was taken as the load where the loaddisplacement curve deviated from linearity.

The tests on the epoxy specimen

measured the toughness required for delamination to initiate from the insert. The
PEEK specimen which were tested in a previous study 3 were first precracked under
a 4/1 mixed-mode loading to a delamination length a of 1.25 in. and then tested.
Although Murri et. al. 7 showed that delamination toughness measured from a
precrack could be significantly different from values measured from an insert, a
study involving a 4/1 type precrack showed good agreement with insert initiation
values.8 The critical applied load Pc for all tests were taken as the load where the
load displacement curve deviates from linear. The PEEK toughnesses presented
here is slightly lower than that given in Reference 3 because that data was calculated
with Pc equal to the maximum applied load. The edge of the specimen was coated
with a white water soluble typewriter correction fluid so that the delamination could
be observed more easily with a 7x magnifying scope. The delamination extension
was observed at approximately the same time as the nonlinearity in the loading
curve. The delamination length, a, was determined by breaking the test specimen
open after the test and measuring the length of the initial
5

delamination.

For the

epoxy specimen this was the length of the insert while for the PEEK specimen, a
was the length to the marking on the fracture surface left by the precrack.
The mixed-mode fracture toughness,
AS4/PEEK

specimen

were calculated

64L2bEII 1

(3c2+6cL

+3L

64L2bE

I
11

1[

These equations

The fracture
The complication

(1)

bh 3

and

I-

and deformation

at the delamination

12

theory

and include

due to the rotation

toughness

calculation

arose because

two materials

to account

of the specimen

of the AS4/3501-6

the toughness

resulting

lb), and therefore,

how the lever weight

equations

corrections

for

cross section

tip 2.

in smaller

the lever weight

was included

for mixed-mode

fracture

critical

applied

is small

loads

toughness

in Appendix
are:

compared

to

Pc. The critical

as the lever weight

could not be ignored.

is presented

was more complicated.

of this material

loads in some cases were of the same magnitude

Pc=10

+ 10G,-----_

Gl3

are based on beam

shear deformation

equations:

"2h2Ell

6E 22

Z,=--ff4_ Etl

and

h2 j

a2+

G llc

the other

a2 + _

)Pc

where

using the following

2[
22[]

24c_____L
+ 4L 2)P c

(36c 2-

Gk:=

Gllcm,of the IM7/977-2

G mlcand

Pg (Pg=l.85

A derivation

A, andthe

resulting

of

lb,

(36c

- 24cL

G_=[

+ 4L

)Pc

(72CCg-24cL

(3c2+

6cL

+ 3L

-24CGL2+
8L2)pcPg
(36c g- 24CgL

+ 2 2/
+ 4L )Pgj

(3c2

lea',
'h2E"
1

G13

+ 6c gL + 3L2)p

64L2bE

The equations

in the appendix

are written

energy

rates,

GII, while Eqs. 2 are written

release

load

Gl

and

Pc and the fracture

criteria

for deciding

toughnesses,

when

The pure mode

(2)

+ 6c gL + 6L2)pcPg

(6cc g + 6cL
2 2

_+ 1_13
I

)Pc +

G lie =

a2 + 2a+
lk_.___X2
64L2bEll

the lever

I toughness

in terms of an applied

and

Gm
Ic

weight

111

load

in terms

GmIIc" Appendix

corrections

was calculated

Pa and strain
of a critical

A also includes

are needed.

using Equation

3 for all materials.

(3)
pc2 I [ a 2 +---ff
2a
bEll

Gic-

This equation

+-_ l

h2Ell.]
+ 10Gl----_

was used in developing

with the mixed-mode

equations.

the mixed-mode

equations

The pure mode II toughness

and is consistent
Gnc can be calculated

from the equations

Gicm

equations

delamination

for G lie in Equations

give erroneous
surfaces

results

1 or 2 where

at this

do not separate.

c is set to 0. Notice

c value because

The contact

forces

for

across

c _;0.67,

that the

the

the delamination
m

surfaces

which

are not modeled

do not effect the

delamination

(SEM).

1 and 2 cause the true

G _ to be 0 but

Gm
nc values.

The delamination
microscope

by Equation

fracture

surfaces

were examined

SEM photomicrographs

insert for the epoxy

composites

where

using

a scanning

taken just beyond

and just beyond

electron

the

the precrack

marking

for the PEEK composite. Therefore, the fracture surfaces show the first increment
of delamination growth which correspond to the measured fracture toughnesses.

TOUGHNESS

TEST

failure

of the three composite

The mixed-mode

responses

RESULTS

materials

are

presented
mode

by plotting

II component,

The failure

Gm
lie "

response

AS4/3501-6

to increase

as the

diagram

The AS4/PEEK
mode ratios

are quite

is even tougher

diagram

composite

model

in References

system

than the iM7/977-2

but the
the

decrease

lIc may be somewhat

and very similar

2.

G m
Ic

to zero

surprising,

9 and 10. The

in shape to the

studied

in Reference

material

1.

at all mixed-

response

is quit

G t: decreases

systems.

are so different,

For both epoxies,

near pure mode II and the shape of the failure

very similar

composite

in shape,

G m

different.

similar

in Figure

and then slowly

G_c with

is convex

G t: vs. the

are plotted

than the IM7/977-2.

for a brittle-epoxy

material

except

composites

is introduced
Gm
llc

response

toughness,

diagrams

can also be seen in data presented

shape of the failure

mixed-mode

of fracture

mixed-mode

The rising

Gilt.

but this phenomenon

overall

These

is more brittle

approaches

Gm
llc

I component

of the two epoxy

material

values appear
as

the mode

almost

linearly

to those presented

with

Gi1c which

in Reference

Since the shape of the failure

all delamination

failure.
parameters

One criteria

materials

if arbitrary

the data.

If this does not work then different

different

materials.

so that the criterion

would

tough

G1c and

be able to model

criteria

a mixed-mode

of the different

based on just

might

can be changed

1 for different

responses

it is clear that no Single criterion

produces

materials
GIIc will

the different
can be fit to

have to be used for

SEM photomicrographs show that the fracture surfaces of the different

materials change dramatically with mixed-mode ratio. As discussed earlier, these
photomicrographs were taken just after delamination initiation.

Figure 3 shows the

delamination surfaces at different mixed-mode ratios for the AS4/3501-6 composite.

The photomicrographs were taken at a magnification

of 1000x. At pure mode I the

fracture surface is very flat indicating a brittle cleavage fracture which would
explain the low mode I fracture toughness. As mode II is added, the fracture
surfaces becomes rougher as seen in the 4/1 ratio case. Troughs have appeared
where fibers have been pulled away from the matrix indicating interfacial failure.
The side of the fracture surface which did not contain a large percentage of fibers is
shown here because the texture of the fractured resin is easier to see in this view.
The increase roughness could explain why the mode I component of fracture
toughness rises as mode II is introduced. At the 1/I ratio, hackles 11 have appeared,
and there is little difference between the 1/1 fracture surface and the pure mode II.
The similarity between the 1/1 and pure mode II fracture surfaces indicates a single
failure mode through this region. The difference between these fracture surfaces
and those at the pure mode I and 4/1 case may indicate a change in the failure
mechanism around the 1/1 ratio.
Figure 4 shows the fracture surfaces of the IM7/977-2 composite at several
mixed-mode ratios. As seen from the figure, the fracture surfaces of IM7/977-2 are
very similar to that of AS4/3501-6.

The mode I fracture is cleavage, fiber troughs

appear at the 4/1 ratio, and hackling begins around the 1/1 ratio and continues
through the pure mode II condition.

The change in fracture surface between the 4/1

and 1/1 ratio is even more clear for this material than for the AS4/3501-6, and again
may indicate a changing failure mechanism. The similarity in the fracture surfaces
of these materials may explain the similarity between the shapes of the failure
responses of these two materials which can be seen in Figure 2. No indication of the
9

increased toughness of IM7/977-2 over AS4/3501-6 was noted by observing the

fracture surfaces.
The fracture surfaces of the Peek composite shown in Figure 5 are noticeably
different from that of the epoxies. All the PEEK fracture surfaces have cusps
caused by the extensive yielding of the matrix.

The larger strain to failure created

by the yielding gives this material a larger fracture toughness than the epoxies tested.
The ridges and valleys seen in the figure are due to the fibers which have been
pulled out of the valleys. The ridges form because the larger volume of matrix
material found between the fibers can deform more creating the ridges of cusps.
The primary difference between the fracture surfaces at different mixed-mode ratios
is the orientation of the cusps. The DCB fracture surface has cusps that are
randomly pulled in different directions.

In the 1/l fracture surface the cusps tend to

be drawn slightly toward the top of the micrograph due to the mode II action.
Larger amounts of mode II draw the cusps more and more, and in the ENF fracture
surfaces, all the cusps are extensively drawn upward. No indication of hackling or
of changing failure mechanism was observed on the AS4/PEEK fracture surfaces

REVIEW

OF

Many attempts
failure

response

of composite

a materials

release
resistance

have been suggested

DELAMINATION

have been made at describing

or strain near the crack

strain energy

MIXED-MODE

materials.

tip, crack

rate.

Strain

opening
energy

to delamination
can be written

fracture

toughness.

Delamination

loading

and pure mode II loading

Failure

extension
in terms
fracture

the mixed-mode
criterion

displacement,
release

delamination

have been based

stress intensity

rate seems

of a critical

toughness

strain energy

testing

on stress

factor,

or

to be a good measure

and most of the failure

is well established.

l0

CRITERIA

criteria
release

of
that

rate or

under pure mode I

Delaminations

in structures

are often subjected

important.

to mixed-mode

so mixed-mode

Since the mode I and mode II fracture

available,

the mixed-mode

toughnesses,

Gic and

delamination,

criteria

presented.

failure

criteria

suggested

of the fibers.

Although

and delamination,

are so similar,

one would

The mixed-mode

failure

expect
response

data is readily

interest

and ply cracking

the fiber-matrix

is growing

geometry

responses

of a material

is also

in temas of the pure'mode

a crack

the failure

toughness

the primary

for both delamination

Because

fracture

toughness

will be written

GIIc, when possible.

In both ply cracking

in the direction
cracks

loading

will be
in the matrix

of both

to be similar

can be described

here is

types of
as well.

by plotting

the mode
Gm

IIc"

I component

An accurate

this mixed-mode

failure

each criterion

assuming

E11/E22.

For comparison

component

the mode II component,

the material

response

response

3. Some criteria

when

of different

plotted

materials

first by looking

at the

each criterion

will be

For comparison

this ratio will be assumed

also involve

to be 10.

Later criteria

will be

assume

that either the mode I component

toughness

criteria

14 will stay constant

12, the mode

as the mixed-mode

are respectively:

(4)

G ill

(5)

Hc = _3it,:

+G

m
llc

=G

Tc

can

the ratio of

GIc =Glc

G_

on

data.

criteria

These

Giic=

vs.

will be evaluated

can model.

13, or total fracture

ratio changes,

G lc

will match

criterion

G[c = 1 and

toughness

The simplest

toughness,

Since the the material

each failure

evaluated

fit to measured

criterion

diagram.

be quite different,
range of shapes

of fracture

=G

lc

=G

llc)

11

(6)

II

Tb,es:e criteria

are plotted on the mixed-mode

first criterion

assumes

de!amination

growth

thatonly
Gic.

which

6 more,reasonable

an d_mode

processes

.The fracture

mode rati_ which

means

to be quit different

....."_

next

for materials

by its pure'mode

G Ic )

connecting

i
!

GI

and

is assomed
equal
the

mode
:)(i

J/

if either

GTc reaches

and

some

of mixedGIIc

are

_;i31-5

:_,_

known

6) was modified

L_3-JiiGi

The third criterion

GI). Sinee.;Glc

(iOii_ii-i:_

_ :-

that the mode

I and mode II toughnesses

i:i

Gn_ :,:_

not to be. a .function

GT_ criterion(Eq.

an infinite

occur

o.feach.other.

if the total strain energy

simply

IIc

to

as seen in the

.jii-_-.'i:_i_

L--?_

-2-

.........

of fracture

toughness

_:

,r_,i_.,.,..-_:_;,.._:.:c: ;>:_::.a; ?<_(71

_ on,I_e,._i_ec!:_od_,

fracture

I and pure _ode

6. This linear criterion(Eq.

to in literaturel,

law function

each component

the pure mode

referred

normalizes

value as given by 14

A fifth criterion
power

toughiiess

G._)t;iii_;'.]:'.)

criterion

The res:u!ti_ngcp_.

often

will extend

with different

the

growLh would

independently

.........

criterion.
The fourth

Figure

occur

that Glc should

.],_.-4 ];_]_[

and therefore

then models the assumption

for most materials,

i"h.....Ll_h_.t,t,,i-">_

be to combine

controls

The

GIIc. q_e _sec_nd.assu_es

is important

that de!amination

This criterion

that.a de!amination
valuo,

account

of loading

Gn) in Figure 6.

of!oading

assumes_ a13 iofi0ite

critefi.on.wo.u!d

5 is met.

!! failure

assumes
critical

therefore

4 & 5) by assuming

EquatJo.n4.or

(Gi vs,

that only the mode I compone_nt

the mode II component

criteria(Eqs.

diagram

toughnes_diagra.m

![ fracture

7) is perhaps

toughnesses

the mixed-mode

is a .!.'.m.e...,
::..
as show_
criterion

on ,,_,:
most

10,15,16,17

is obtained

by generalizing

12 as follows:

12

the linear criterion(Eq/7)

as a

By choosing
shown

and

in Figure

13, a wide range of material

7. When

o_=13>1 the failure

responses

curve is convex

the curve is concave.

If 13>o_ then the curve is skewed

convex

I axis.

near the mode

criterion(Eq.

7) several

Besides

(0.5,1) 14, (1,1.5) 19, (1.4,1.8)

given material
Another
would

can be found
criterion

+
The polynomial
adjusting
convex

failure

lc

by assuming

choice

by the epoxy
is unable

of o_ and

experimental

13 for a

data.

that the total fracture

a wide variety

8. This criterion

toughness

I as given by

to model
and

lc

of the mode
of

which
p=l.5.

by

both concave

and

can allow

and therefore

intensity

G m_c and

Gmllc as

the

This material

of the present

low mode I to mode

investigation.

II ratios.

Either

this criterion

the
ratio

is an

criterion.

by assuming

I stress

13

discussed

systems

mixed-mode

was developed

can model

responses

GmIIc will go to zero as the mixed-mode

is realistic,

as a general

of material

as seen when

composite

Gm

response

criterion

here in terms

value

(9)

as mode II is introduced

inappropriate

written

through

9) can model

gets very large or both

linear function

fitting

It is also the first criterion

this criterion

Another

have been used such as (2,2)18,

curves.

Neither

is more

the linear

of the ratio of mode II to mode

x as ,seen in Figure

is decreased.

reproduces

o_=13<1

(o.m/

is exhibited

However,
Gm

function21

mode I to increase
behavior

by curve

criterion(Eq.

p and

which

as

while when

so that the curve

20, (.64, .8) 20. An optimum

was developed

be a polynomial

a=13=1

(c_,13) combinations

can be modeled

factor22

the fracture
KIc'

toughness

This criterion

to be a
has been

GmIc + G m
llc = Gllc

If Gic=Gnc
Gic<<Gnc

the

-(G

Klc

then the

where

o_=.5 and

values

and produces

criterion(Eq.

10) reduces

to the linear criterion(Eq.

[3=1. This criteria

is completely

a concave

failure

which

envelope

was developed

The hackle
is a measure

terms of the pure-mode

constant

(10)

lc

10) reduces

formation23.

l+(Ku_:_)

Kic criterion(Eq.

The next criterion

hackle

lie -GIc)_/G_c/G

fracture

to the power
defined

If

law criterion(Eq.

by the

as seen in Figure

by modeling

7).

8)

Gic and Guc

9.

delamination

growth

through

criterion(Eq.

11) was based on a linear function

of the hackle

angle.

toughnesses,

The criterion

modulus

values,

is written

of

here in

and an arbitrary

Z.

m m

G k-t-G

The hackle

lie---(G

hz-

criterion(Eq.

can be chosen

to model

X)-I- _

j O7,/E,,
G _,._/

1+

11) is plotted
a variety

(11)

E 22

on the mixed-mode

of material

responses,

diagram

including

in Figure

10. Z

an increasing

G rn
Ic

with

Gllc

as seen when

GTc criterion(Eq.
so this criterion
A second

6), the hackle

toughness

criterion

hackle
values

However,

except

criterion(Eq.

is also an inappropriate

this time it is related

exponential

z=l.

for when

criterion(Eq.

predict

as a general

mixed-mode

choice

instead

GI c+Gllc=(Glc-Gllc)eY(1-N)+G,lc

constant

N=

14

gives the

an infinite

Guc

criterion.

angle parameter

JI+(K,/K_)

ofa linear functionl0.

The

12) can also be written

and only one arbitrary

which

11) will always

was based on the hackle

as an exponential

Z=0

, but

in terms of the pure-mode

7 as follows:

1*

G----_]

E2---"
_

(12)

This exponential hackle criterion(Eq. 12) has the advantage that it does model a
finite Gltc as seen in Figure 11. The T constant can be chosen to model a great
variety of material responses. The failure envelope can be concave or convex and it
Ill

can model

an increasing

Another
mixed-mode

mixed-mode
stress

here in terms

criterion

intensity

factor

of strain energy

Figure

KIc/KIIc

exponential

hackle

criterion(Eq.

criterion.

crack opening

A delamination

crack opening
derive

for

the following

displacement

(COD)

t? extend

is reached.

on the mixed-mode

diagram

modeled

jog in the failure

in

by the

curve

near the

was the basis of yet another

when the mode

A Dugdale

plasticity

I or mode

model

II critical

was used to

criterion25:

Glcm

Glc
m

E_III

( GIic_2

assumes

(14)
GIIc
m
G

a critical

mode

I crack opening

assumes

a critical

mode II displacement.

extend

if either

criterion

is met.

both

r I.

TI<I.

Gk: j

in Figure

constant

the same types of responses

second

diagram

has been written

(13)

12), but a strange

G,,'-"-_"
= _'a/'-ff'u-11_G_,

The first equation

of the

Gtc

13) is plotted

is assumed

displacement

function

The criterion

rates and the arbitrary

rlJG I_/G tl_

can model

mode I axis is introduced

Critical

ratio 24 Klc/Kllc.

criterion(Eq,

12. This criterion

Glk

was based on an exponential

release

G m G m
lc +
IIc=(GIIc-GIc)

The exponential

increasing

G lc
m with

This criterion

13. The COD criterion(Eq.

10 and 1. The higher

ratio corresponds

15

displacement

The delamination
is plotted

while the

is assumed

to

on the mixed-mode

14) is plotted

with

to the ratio of global

Ell/E22
stiffnesses.

equal to
Since

this criterion is based on crack opening displacement, a localized phenomenon, a

local stiffness ratio corresponding to a crack in a isotropic matrix material and
therefore the lower ratio, may be more appropriate.

With either stiffness ratio, the

mode I criterion, which produces a concave failure response, is critical for most of
the GutGII values.
Another criterion was developed based on a mode I-mode II interaction
parameter

126.

,/Io.
m

,c
The arbitrary

interaction

and mode II loadings

be describe
modes

are independent

those modeled

material

and the criterion

responses

modeled

by the power

law are equal.

different

and convex

Values

in Figure

responses.

14, this criterion

the

Gic and

is the same as the linear

law criterion(Eq.

seem to be almost

8) when

can

If _c=0 then the fracture

is the same as combining

by this criterion

the mode I

the arbitrary

GIIc

criterion(Eq.
identical

constants

of _c from 0.26 to 3.1227 have been suggested

to

of the
for

materials.

The interaction

criterion

was made

more versatile

parameter

to be a linear function

of

The iihear

interaction

interaction

criterion(Eq.

responses

criterion(Eq.

GI c

Gllc

m]( -q-U,,j
m 1
.,,, m L JL
0

16) can model

including

16

the interaction

as given by 28

m
)1 (
('J It: + G IIc

15), but by adjusting

can be obtained

by allowing

ic/(G lc + Gltc)

GI c

_) _[
t(G_____)(GG____
, , ,ic_
llc
.,

more

of how much effect

As shown

4 & 5), and if _c=2 the criterion

The material

power

_c is a measure

have on each other.

both concave

criteria(Eqs.
7).

parameter

all the responses

the arbitrary

an increasing

parameter

(16)
of the simpler
(p, many

mode I with mode II. 0e,(P)

values of (3,-4) and (4,-3) were suggested for different materials. The linear
interaction criterion is a rather complicated implicit function of Gic and Giic which
could make this criterion difficult to use.

BILINEAR

The delamination
mechanism

toughness

failure

mechanism

another
epoxy

surfaces

data of the epoxy

does change

regions

composite

indicated

data around

of both the AS4/3501-6

failure

criterion

pure-mode

linear

the 1/1 ratio.

one might

expect

diagram.

different

Shifting

the 1/1 mixed-mode

response

depends

further

The mixed-mode

failure

criteria

from one criterion

response
ratio.

as observed

use, a reasonable

delamination

supports

on the arbitrary

failure

this assumption.
parameters

If the

to hold in
to
in the

Since the linear

would be linear in each region.

and IM7/977-2

which

in failure

also reach a peak at this ratio.

and has seen widespread

be that the failure

rather

around

that a change

result in a peak in the toughness

7) is simple

appear

CRITERION

composites

of the mixed-mode

could easily

criterion(Eq.
would

fracture

may take place in the epoxy

fracture

different

FAILURE

assumption
The two regions

data in Figure
The resulting

_ and

2 do
bilinear

4, as well as the two

toughnesses.

m
m
Glc = _ Gllc+

Glc

(17)
!11

Gtc=

and
17).

_GII c-

_GII c

_ are the slopes

As shown

in Figure

and it can model

If _=_=-GI.ffGH_,
criterion(Eq.

of the two line segments

16, this criterion

an increasing

mode

then the bilinear

can model

I fracture
criterion(Eq.

7)

17

used in the bilinear

concave

toughness

or convex

component

17) would

criterion(Eq.

reduce

responses,

with mode II.

to the linear

CRITERION

EVALUATION

The fact that so many radically

suggested

and used indicate

different

The true test of a failure

of the material

of interest.

These

different

data sets will be used to evaluate

that produced

material

responses

optimize

the curve fits.

distance

between

curves

experimental

responses
data.

G_c, Giic,
observed

Since

analysis

so no attempt

it cannot

composite
increase

model

data.
would

polynomial,

at least produce

the

not able to model

The exponential

in

a material
hackle,

The shapes

13, & 17) were used to model

the

4, 5, & 6) are not even close to

to the

7) was only close to

G Uc observed

for either
response

of the

even

in the epoxy
to model this

model for the material.

and the COD criteria(Eqs.

The

9, 10, 11, & 14) were

to that seen in the experimental

Klc/Kllc,

all three material

18

to

of the failure

was not that large, and failing

Gm
lc

the exponential

2.

G to with

similar

in Figure

by minimizing

8) was fit to each material

response

Consistent

was performed

to use this criterion

a conservative

Kxc, the hackle,

presented

was

was made to fit these criteria

was made

the increasing

The increase

inconsistent.

was conducted
curve.

though

was available

analysis

the shape of the linear criterion(Eq.

law criterion(Eq.

these

shape close to that of one of the

GTc criteria(Eqs.

data, no attempt

The power

the response

criteria.

A least squares

The least squares

by the

where

the different

a general

was fit to the data.

the shape of PEEK

epoxies.

materials

each data point and the failure

produced

the material

and that which

tests and was therefore

data for three different

Each criterion

about this

is how well it models

data was available,

from several

sets of mixed-mode

criterion

have been

In the past, there was no good way to evaluate

Little mixed-mode

often obtained

criteria

that there is still much to be understood

phenomenon.

criteria.

mixed-mode

and the bilinear

responses.

data.

criteria(Eqs.

The simple

interaction

12,

criterion(F_,q. 15) was not fit to the data bccausc

interaction

criterion(Eq.

The results
given

in Table

materials

criterion(Eq.

criterion(Eq.

fit of each criterion

2. The best fit curves

are shown

other criteria.

power

the linear

the coefficients

law criterion(Eq.

and AS4/PEEK

of variation

R 2 given in

fits the data, and as

16 & 17)

and IM7/977-2.

criteria

but since the curve does not model the rising

are believed

the bilinear
number

criterion

of arbitrary

criterion(Eq.
m

to be better choices.
model

16) makes

with

GHc,

the other

criterion

and

well and with the same

of the linear interaction

to work with since it is an implicit

function

of

G ic and Gllc" The bilinear

therefore
of failure

easier

to use.

for these

The PEEK

composite

19.

smallest

experimental

coefficient

that was almost

is based on very simple

the bilinear

equations

is believed

and is

to be the best choice

materials.
was modeled

The coefficients

but they are all noticeably

are fewer

criterion

For this reason

criteria

in Figure

there

The complexity

it difficult

GIc

Both the linear interaction

the epoxy based composites

constants.

The

R 2 value for the AS4/3501-6

m

material,

better than the

criterion(Eqs.

to 1 for both AS4/3501-6

8) also had a very low

The linear

the epoxies

is to 1 the better the model

and the bilinear

data is

line and the bilinear

to model

by the coefficients

interaction

closest

IM7/977-2,

17, 18 and 19, respectively.

by the solid line, appear

This is also indicated

seen in the table,

to the experimental

by the heavy dashed

Table 2. The closer the coefficient

produced

for AS4/3501-6,

in Figure

16) shown

17) shown

case of thc linear

was fit to each set of data.

of the least square

response

interaction

16) which

it is a specialized

of variation

smaller

for this material

than the coefficients

points.

of variation,

fairly well by all the criteria

The power

because

8) produced

7) produced

variables.

as seen

the same,

of the other materials

but the linear criterion(Eq.

19

are all about

law criterion(Eq.

as good and with two less independent

tested

the

a curve

For this reason,

the

linear model is believed

bilinear

criterion(Eq.

PEEK

material

be consistent

the response

composite,

be subjected

mode

better

loading

testing

be included

response

the best failure

criterion

has been used for similar

failure
whether

for different

classes

is preferred

if it models
failure

that mixedOnce

presented

here.

When

consider
choice

which
of failure

such as a linear criterion

used.

the material

for

When choosing

of arbitrary

variables

A simpler

criterion

a
and
with

as well as a more complicated

for the material

data can be performed

is chosen,

to optimize

a least

any arbitrary

of the criterion.
CONCLUDING

Many
mode fracture
consistent

response

the shape of the

a standard

the number

criterion

will often

of a material.

composites.

can be easily

from that of

failure

one should

Hopefully

for epoxy

delamination
toughness

failure
responses

sets of mixed-mode

criteria

REMARKS
which

predict

have been reported

data exist with which

20

to

based on just the pure-

it is important

criteria

of material

also consider

fit to the experimental

constants

materials.

be chosen

different

the mixed-mode

failure

the

Since delaminations

for a given material

is in a form which

After the appropriate

squares

criterion

has been determined,

and a bilinear

one should

criterion

fewer variables
one.

criterion

composites

criterion,

was quite

the characterization

to the different

it models

7) and might

toughnesses,

during

of a material

choosing

will emerge

and because

from the pure-mode

can be compared

thermoplastic

composites

will be able to model all materials.

response

criteria

of freedom,

Since the

composites.

it is clear that no one failure

be determined

the mixed-mode

for this material.

degrees

used for the epoxy

of the epoxy

to mixed-mode

toughness

two extra

criterion

than the linear criterion(Eq.

with the criterion

mode toughnesses

cannot

17) contains

slightly

Because
the PEEK

to be the best failure

a wide variety

of mixed-

in the literature,

but few

to compare

these criteria.

The

MMB test was used to measure the mixed-mode delamination toughness of three
different classes of material. A common brittle graphite/epoxy composite
(AS4/3501-6), a state of the art toughened graphite/epoxy composite (IM7/977-2),
and a tough graphite/thermoplastic composite (AS4/PEEK) were tested. The MMB
test is a combination of the pure mode I, DCB test and the pure mode II, ENF test,
and can measure fracture toughness at virtually any mixed-mode ratio using a single
test specimen configuration.

The toughness data are plotted on the mixed-mode

diagram (Gt vs. GI_). The delamination surfaces were examined and a possible
change in failure mechanism was observed in the epoxy composites.
Criteria which have been suggested by other investigators were reviewed and
the range of material responses modeled by each criterion were explored.

A new

bilinear failure criterion was also developed in an attempt to model the possible
change in failure mechanism observed in the epoxy composites. The different
criteria were compared to the failure response of the three materials tested. The
new bilinear failure criterion was considered the best choice for the two
graphite/epoxy composites because it modeled the material responses well and
because it is relatively simple. The AS4/PEEK composite, which did not show signs
of a changing failure mode, was modeled well with either the bilinear or a simpler
linear criterion.
Since the response of the epoxy composites was quite different from that of
the PEEK composite, it is clear that no one failure criterion based on just the puremode toughnesses will be able to model all materials. Because delaminations will
often be subjected to mixed-mode loading and because the mixed-mode failure
response cannot be determined from the pure-mode toughnesses, it is important that
mixed-mode toughness testing be included during the characterization of a material.
Once mixed-mode toughness testing has been conducted the evaluation of the

21

different failure criteria provided in this paper should provide general guidance for
selecting an appropriate failure criterion.

REFERENCES

Johnson, W. S., and Mangalgiri,

P. D., "Influence
on the Resin on
Interlaminar
Mixed-Mode
Fracture," Toughened
Composites,
ASTM STP
937, Norman J. Johnston,
Ed., American Society for Testing and Materials,
Philadelphia,
PA, 1987, pp. 295-315.
Reeder, J. R., and Crews, J. H., Jr., "The Mixed-Mode
Bending Method for
Delamination
Testing," AIAA Journal, Vol. 28, No. 7, July 1990, pp. 12701276.
Reeder, J. R., and Crews, J. H., Jr., "Redesign
of the Mixed-Mode
Bending
Test for Delamination
Toughness,"
Composites
Design, Manufacture,
and
Application,
ICCM/8 Conference
Proceedings,
SAMPE, July 1991, Stephen
W. Tsai and George S. Springer, Ed., pp. 36-B-1 - 36-B-10.
O'Brien, T. K., Murri, G. B., and Salpekar,
S. A., "Interlaminar
Shear
Fracture Toughness
and Fatigue Thresholds
for Composite
Materials,"
Composite
Materials: F.atigue and Fracture, Second Volume, ASTM STP
1012, Paul A. Lagace, Ed., American
Society for Testing and Materials,
Philadelphia,
1989, pp. 222-250

5.

Hercules Composites
Products Group, "Hercules Prepreg Tape Materials
Characterization
Data Package," Hercules Co., Magna, Utah, February 1989.

6.

Smith, Donald L., and Dow, Marvin

B., "Properties
of Three
Graphite/Toughened
Resin Composites,"
NASA TP3102, September

1991.

Murri, G. B, and Martin, R. H., "Effect of Initial Delamination

Mode II lnterlaminar
Fracture Toughness
and Fatigue Fracture
NASA TM 104079, May 1991.

on Mode I and
Threshold,"

J. H., Jr., and Reeder, J. R., "A Mixed-Mode

Bending
Delamination
Testing," NASA TM 100662, August 1988.

Apparatus

Crews,

Hibbs, M. F., Tse, M. K., and Bradley, W. L., "Interlaminar

Fracture
Toughness
and Real-Time
Fracture Mechanism
of Some Toughened
Graphite/Epoxy
Composites,"
Toughened
Composites,
ASTM STP 937,

22

for

Norman J. Johnston, Ed., American Society for Testing and Materials,

Philadelphia, 1987, pp. 115-130.
10. Donaldson, S. L., "Fracture Toughness Testing of Graphite/Epoxy and
Graphite/PEEK Composites," Composites, April 1985, pp.103-109.
11. Morris G. E., "Determining Fracture Directions and Fracture Origins on
Failed Graphite/Epoxy Surfaces," Nondestructive Evaluation and Flaw
Criticality
American

for Composite
Materials,
ASTM STP 696, R. B. Pipes,
Society for Testing and Materials,
1979, pp. 274-297.

12. Whitcomb,

J. D., "Analysis of Instability-Related

Growth
Delamination,"
NASA TM 86301, September
1984.

Ed.,

of a Through-Width

13. Gillespie,

J. W., Jr.; Carlsson, L. A., Pipes, B. R.; Rothschilds,

R., Trethewey,
B., and Smiley, A., "Delamination
Growth in Composite
Materials,"
NASA
CR 176416, December,
1985.

14. Wu, E. M.; and Reuter Jr. R. C.: "Crack

plastics"

T & AM Report

extension
in fiberglass
No. 275, University
of Illinois, 1965.

reinforced

15. Spencer,

B., and Barnby, J. T., "The Effects of Notch and Fibre Angles
Crack Propagation
in Fibre-Reinforced
Polymers,"
Joumal of Material
Science, Vol. 11, 1976, pp. 83-88.

16. Jurf, R. A., and Pipes, R. B., "Interlaminar

Joumal

of Composite

Materials,

17. Mall, S., and Kochhar,

Composite

Bonded

Fracture
Vol. 16, September

N. K., "Criterion

Joints,"

NASA

of Composite
Materials,"
1982, pp.386-394.

for Mixed-Mode

CR 178112,

on

Fracture

in

May 1986.

18. Ashizawa,

M., "Faster interlaminar

fracture of a compressively
loaded
composite
containing
a defect," Paper presented at the Fifth DOD/NASA
Conference
on Fibrous Composites
in Structural Design, New Orleans,
Jan. 1981. (Available
as Douglas Paper No. 6994.)

LA,

19. Donaldson,

S. L., "The Effect of lnterlaminar

Fracture Properties
on the
Delamination
Buckling of Composite Laminates,"
Composite
Science and
Technology,
Vol. 28, 1987, pp. 33-44.

20. Hashemi,

S., Kinloch, A. J., and Williams, J. G., "Mechanics

of Delamination
in a Poly(ether sulphone)-Fibre
Composite,"
Science and Technology,
Vol. 37, 1990, pp. 429-462.

23

and Mechanisms
Composites

21. Yan, X. Q., Du, S. Y., and Wang, D., "An Engineering Method of
Determining The Delamination Fracture Toughness of Composite Laminates,"
Engineering Fracture Mechanics, Vol. 39, No. 4, 1991, pp 623-627.
22. Hahn, H. T., "A Mixed-Mode Fracture Criterion for Composite Materials,"
Composites Technology Review, Vol. 5, Spring, 1983, p. 26-29.
23. Hahn, H. T., and Johannesson, T., "A Correlation Between Fracture Energy
and Fracture Morphology in Mixed-Mode Fracture of Composites," ICM 4,
Vol. 1, 1983, pp. 431-438.
24. White, Scott, R., "Mixed-Mode Interlaminar Fracture of Graphite/Epoxy
Composites," Master Thesis, Washington University, Saint Louis, Missouri,
May, 1987.
25. Hashemi, S., Kinloch, A.

and Williams, J. G., "lnterlaminar

Fracture
Composite
Materials,"
6th ICCM & 2nd ECCM Conference
Proceedings,
Elsevier Applied Science, London, Vol. 3, July 1987, pp. 3.254-3.264.

26. Williams,
Strain

J. G., "The Fracture

Analysis,

Mechanics

of Delamination

Tests,"

Journal

of

of

Vol. 24. No 4. 1989.

27. Hashemi,

S., Kinloch,
Rate and Temperature
lnterlaminar
Fracture
Journal of Composite

28. Hashemi,

J.,

S., Kinloch,

A. J., and Williams, J. G., "The Effects of Geometry,

on the Mode I, Mode II and Mixed-Mode
I/II
of Carbon-Fibre/Poly(ether-ether
ketone) Composite,"
Materials,
Vol. 24, September
1990, pp. 918-956.
A. J., and Williams,

G., "Mixed-Mode

Fracture

in

Fiber-Polymer
Composite
Laminates,"
Composite
Materials: Fatigue and
Fract0re,
Vol. 3, ASTM STP 1110, T. K. O'Brien, Ed., American
Society
Testing and Materials,
Philadelphia,
1991, pp. 143-168.

24

for

APPENDIX
LEVER

WEIGHT

The MMB
beam

specimen

load,

Pg,

CORRECTION

FOR

MMB

test uses a lever to apply mode I and mode II loading

using only one applied

load,

Pa-

As shown

also acts on the lever at a point determined

also contributes

THE

to the mode

I and mode

The mode I and mode

II loading

II loading
were given

in Figure

by length

TEST

to the split
A1, a gravity

Cg, and this load

of the test specimen.

in Reference

2 as

P II = P a(-_-)

(ml)

but these equations

did not account

lever can be thought

applied

loading.

of as a second

The mode

for the weight

applied

I and mode

of the lever.

load which

II loadings

The weight

is superimposed

are therefore

of the

on the true

given by

(A2)
PII = Pa(_"_)

The equations
Equations

+ Pg\

for mode

I_

I and mode II strain energy

release

rate are given

A3.

Gl-

bEll I

+_

where
3P!
GI!-

64bE

Substituting
applied

by

a2 +
111

(A3)
1-

bh 3

Gl3

Eq. A2 into Eq. A3 gives an expression

load and lever weight.

25

12

for

in terms of the

(36c 2-

24cL

+ 4L

)P,, +

2a

a2 + ---_- +
GI=

(72 ccg -

24cL

- 24c gL + 8L2)p

aPg

Gll

Notice

6cL

+ 3L

(6_

)Pa +

g + 0eL

+ 6c gL + 6L2)PaPg

G13

64L2bE

111

for

G given

in Eq. A4

that account

for the weight

in Reference

2 if the terms

Pg are removed.

c is small (c_0.67),

transferred

across

so that a negative
negative,

lLa+
l

These equations

to the equations

Eq. A4 has been developed

When

64L2bElli

in Eq. A3, a cross term develops

load and the weight.

of the lever are equivalent

involving

(3c 2 + 6c gL + 3L2)p

the applied

1-_13J

.2h2Ell

that since the load is squared

between

--_

+ 41, 2 )Pg2

(36c 2 - 24cgL

(3c2+

-]
/

assuming

the delamination

the delamination

the faces.

faces are not in contact.

faces do not separate

PI (Eq. A2) is the load pulling

P_ is the load pushing

the faces are in contact

the delamination

and load is

the delamination
faces together.

so GI will equal zero, but

open
If P[ is

GII can still be found

using Eq. A4.

The errors

caused

by ignoring

the weight

of the lever are calculated

using

Eq.

A5.
(G1)

(G i)

% Error

GI=

(GI+GII)

x 100
o

(6
% Error

GII

(A5)

w- (G
(GI+GII)

x 100
O

The subscripts
respectively.
Gn,

w and

o indicate

The errors

are normalized

so that a small nominal

error just because

that the lever weight

by

was included

total strain energy

error in GI or Gu does not cause

that component

also happens

26

to be small.

or - neglected,

release

rate,

GI +

a large apparent

The error caused by ignoring the lever weight of the MMB apparatus used in
the present study is presented in Figure A2. The errors in GI and GII are plotted
as a function of lever load position for total strain energy release rate values from
0.3 to 3 in-lb/in. The redesigned MMB apparatus used in this study was found to
weigh 1.85 lb. The weight is partly due to the saddle mechanism which moves when
the lever load point is moved. Since the saddle mechanism is moved when the lever
load point is changed, the center of gravity of the lever assembly given by length Cg
also changes. The relationship between c and cg was found for this apparatus to
be Cg= 0.38 + 0.24 c.

The c values that produce the mixed-mode ratios used in

this study are also marked on the Figure. For the range of lever lengths tested in
this study, the largest error in GI was at the 4/1 mixed-mode ratio (c=3.83 in.) and
the largest error in Gii was at c= 0.66 in. The error in both GI and GH increases
rapidly as Gj
error is always
A simple
accounting
A5.

+ GII

below

goes below

+ 5% which

test was developed

for the lever weight

The error calculations

the end foundation

Higher

(;L=0)

(72cc
G I-=

Equation
before

GI! _

A6 is a function
testing.

for this type of testing.

to determine

when the added

complexity

First Eq. A4 was substituted

as shown

and shear correction

(G13 =oo)

the

of
into Eq.

in Eq. A6 by neglecting
terms found

in Eq. A4.

pg2 were also neglected.

-24cL
g
(39c 2-

(39c 2-

-24CgL+8L
18eL

+ 7L 2)

+ 6c gL + 6L 2)
18cL

2)

+ 7L 2)

by a function

for total strain energy

27

Pg
--xl00
Pa

(A6)

Pg
--xl_
Pa

of Pa, but the critical

Pa was replaced

G! + GI!. The expression

2 in-lb/in2

is negligible

is necessary.

(6cc g + 6eL
% Error

G1 + GII above

were first simplified

order term involving

%Error

1 in-lb/in 2, but for

value of Pa is usually
of the total strain energy
release

rate was obtained

not know
release

rate,

from Eqs.

A4 and simplified by neglecting the correction for lever weight, shear and end
foundation as discussed earlier.

Equations for the maximum

GI and GII errors

were created using the c values that gave thc worst errors in GI and Gu as seen
in Figure A2. Therefore c=3.84 in. and c=.66 in. were substituted into the GI and
GII error equations, respectively.

Since the largest GI error occurs near the pure

mode I condition and the largest Gll error occurs near the pure mode II condition,
Glc and Guc were substituted into the two equations respectively.

The resulting

estimates of the maximum errors due to lever weight are

Max % Error G m
Ic

aP g

(A7)
m

Max

% Error

G lie = 0.43

If the maximum

cg
--L--+
1)

aPg
G llc bE 1i I

error for both mode

then the weight

of the lever can be ignored.

additional

terms

in Eq. A4 involving

additi0nal

terms are neededai-onemixed-mode

mixed-mode

ratios tested

x 100

I and mode

II are below

If the error is larger

than + 5%, the

Pg should be used in calculating

for that material

28

ratio, they

should

say + 5%,

G. If the

be used at all

so that the data is consistent.

Table 1. Material Properties

Ma_fi_

Ell (measured)

AS4/3501-6

19.1 Msi

1.4 Msi

0.85 Msi

IM7/977-2

20.8 Msi

1.33 Msi

0.694 Msi

AS4/PEEK (APC2)

18.7 Msi

1.46 Msi

0.8 Msi

Table 2. Least Square

Material

curve fit parameters

Criterion

Power Law(Eq.
Exponential
AS4/3501-6

Exponential

8)

Hackle(Eq.

12)

K(Eq. 13)

Linear Interaction(Eq.
Bilinear(Eq.

Exponential
Exponential

16)

17)

Power Law(Eq.

IM7/977-2

E22

8)

Hackle(Eq.
K(Eq.

12)
13)

G13

for various

criteria

Constants

R 2

Glc

Giic

.5185

3.990

.4422

3.713

y= 0.1964

.9924

.5172

3.092

1"1=1.664

.9888

.4740

4.050

a= 0.0571
[3= 5.039

_= 1.279
q_=-4.905
_= 0.2039
4=0.2473
c_= 0.126

.9956

.9960

.4548

3.269

1.792

8.635

1.345

8.354

y= 0.304

.9962

1.730

7.833

11= 1.050

.9948

1.616

8.557

[_= 5.447

.9956
.9960

_:= 0.868
Linear Interaction(Eq.
Bilinear(Eq.

16)

17)

1.619
,

Linear(Eq.

7)

,,

8.106

.9974
9=-2.962
_= 0.2107
4- -0.371

.9976

4.745

7.147

.9787

4.753

6.649

4.538

6.734

7= 0.8701

.9797

4.509

6.733

1'1=0.4940

.9799

6.736

K:= 0.8679

4.778

.9803

t_= 1.662
Power Law(Eq.
AS4/PEEK

Exponential
Exponential

8)

ltackle(Eq.
K(Eq.

Linear lnteraction(Eq.
Bilinear(Eq.

12)
13)
16)

4.433

17)

6.861

13=0.7329

q_= 1.058
_= -0.1261
4= -0.7477
I

29

.9811

.9801

c_

O.
Q.
c_
,,omJ

Q_
c

7O
rCD
.JO

.J
cy

"10
O

E
X

E
"1o

"0
i,m

q_
r-

i-

/
I,,n

_9
Q.

IJ_

/
Tim

..J

30

[]

%
O0
0

VY

%
o

0
_

3_

ORIGINAL
BLACK

AND

WHITE

F-"h,GE
PHOTOGRAPH

C
Q)
_)
"10
fl)
I

0
LO

<

_6
0
c-

po
.0

E _.._..
o_
_.o

_)
I,..

._m
LL

qlMOJO
u0!leu!wele(]

.v

32

ORiGiNAL
BLACK

AND

PA{qE

WHITE

F'I-iOiOGRAPH

,,,I,-I

I-lh,--

"O
O4
I

OO
L O
._ .,-..-

O O
I,,--

,,#

Ii

qIMOJ_

uo!leu!welea

.._

"-

33

ORIGINAL
BLACK

AND

WHITE

P_,G_.
' '_"
PhOTOGRAP_

,-

70
v
111
UJ
O_
O)
0

_o

o
._oo
E

of A
"_.o

u'i

U_

LI]MOJ 0

,,_

u0p,eu!weleO

34

35

0
II
II

/
/
/
II

I_
/

I
/

/
/
/

/
0
0
GO

(,0

I,--=1

36

c_

E
O
cO
(m.

U3

c_

cL_

E
l_

c_

(If
"O
U)
U)

U_
O

L-

:D
O
(_.-.
O

ww,,-

m6"oW
O

E_-O
xS'c_

0
U_

(D _)
X
.c-

c6
(D
:D
O

c_
C_

O0

CO

"_"

C_l

37

.i

tt.

O
_pm

0
Inn
v

fO
,m
I,B

u_J

4mJ
L--

O
'+-

E
Tin

c_
I
(1)

I
C_
O
H
O

Tim

c_
lu

_
"0
Q_
X
Vm

O
c_
O

C_J

_D
O

38

c_

O
In

o
c_

o
_r)

l-

L_I

Q)
1--

"F-_
a)

i._

.m
I...

G)
O
(_
-r"

II
(/)
(/)

_r)
o
II

rj

_--

iii
0.9

E _U'_

(JE)

_r_

(3

39

E)

(M
o

G)

._x

LL

"E
Q

c_

L_

8.
x

Ln

(D

im

,ioor-

E_
0
x

_,@m

L_
LL
0
0
,Tin

(_
0

0
0

40

0
II

II
+

II

II

41

o
I1

u.T

I!

t_

..T

t_
t-

.u

"O

II

I
o
'_

II

IJ..I
"t"O

u.T

E-_

II

r3
O
"T""

O
O
O

CO

,5

42

I.i.

,ii,=-

II

II

E
C

43

t,D

o"

44

i._

CO

0
r_

.0
0
eLO
C_

L_
0

"0
0g
0g
(D

II

c_
0

u3

II

(D
o_z.
I._

CO
!

"OW
0 --"

E" 0

II

II

._
X

o
u3
0

II

0
0
C_

00
0

0
0

45

1.1-

"e"_-

cO

cO

_-- w

Ol

_8
__

g-

_o

6-

"_

_-

,.v

.'-

.'=

w0_ww-m
U

Rf"

//

. jy

/'.//

I I

/L,,
./

/
r_

_)

"1
.

_"

o _o _

"-

O0

-_
...

o
cO

c,O

",

Od

c_

g
I1

46

I"',I"0'_

;>
0
C
0
"__

LO

[]
0
!

or-4

[]

C
cO)

[]

0
"10
0

E
!

"0
0,J

0
1"

L_
0

oJ

I-.-4

II

t::::
v,--I

47

IJ_

cO

111
W
O_

0
s_

C
0

C
.C

"0
0

O0
0

eu

O
0

C3

48

In

c_

Q..
Q.
,4,--J

(/)

.i.m

rt-(I)
r'_

E
!

G)
X

0tQ.

E
C"
..4..w

0
"0
Q.
Q.
,,4..w

r-

1,--

(U
._I

I..

C)')
U_

49

tO')

II
I

>
0
,4,.e

o
I

c"

J
6

C
0
C
0
u)

U)

0
c_

o
0

>
,_1

G)
m

111

G)

11

JOJJ=1%

i
5O

! k

Form

REPORT

DOCUMENTATION

PAGE

,ill
i
i
i
P_bh_ reportln_
burden
for the, ,_c,lle_.ti_n
,ff mformahon
is _._t_m,_h.cl
r,J _veraqP
_ hour
g,lthetmg
_nd mamtainin,]
the data
need_'<t.
,rod completln_j
_nd r _v_.wln_j
ihP coltectiOn
collection
,)f Informai_on,
including
_ug_3eshon'_
for reduclncJ
the, t)urdPn,
tc, Washington
Daws Itigh_vay.
Suite
1204. Arlington,
VA 22202-4302.
and _() the' Olft_ e ()f Manaqement

Approved

OMe_o 070_ota_

=l

1. AGENCY

USE ONLY

(Leave

blank)

2. REPORT

per re.pon_,t,,
pnduding
the time
of inform,Jr=on
c,_nd omment_
Headquarters
,erwcc-S,
Directorate
and Budget.
P_perwork
Redu(tion

DATE

3. REPORT

TYPE

for reviewing
instructions.
"_earching
existing
data
sources,
re,larding
this burden
estimate
or any other
aspect
Of this
for mformatlon
Operations
and RepOrts.
1215
JeffefsoI1
Prc_ e,,t (0704-0188),
Washington,
OC 20503.

AND

DATES

COVERED

I.

........

4. TITLE AND

Februa_ 1992

Technical Memorandum
5.

SUBTITLE

FUNDING

An Evaluation of Mixed-Mode Delamlnation Failure Criteria

NUMBERS

505-63-50-04

6. AIJTHOR(S}

J. R. Reeder
J7. PERFORMING

ORGANIZATION'NAME(S)

AND

ADDRESS(ES)

8.

PERFORMING
ORGANIZATION
REPORT NUMBER

NASA Langley Research Center

Hampton, VA 23665-5225

ml

9. SPONSORING/MONITORING

AGENCY

NAME(S)

AND

| 0. SPON SORING / MONITORING

AGENCY
REPORT NUMBER

ADDRESS(ES)

National Aeronautics and Space Administration

Washington, DC 20546-0001

NASA TM-104210

i
11. SUPPLEMENTARY

NOTES
!
t
I

12a.

DISTRIBUTION

/ AVAILABILITY

STATEMENT

12b.

DISTRIBUTION

I
i ,

CODE

Unclassified - Unlimited

_= ,

Subject Category - 24
13.

ABSTRACT

(Maximum

200 words)

Many different failure criteda have been suggested for mixed-mode deiamination toughness, but few sets of mixed-mode
data exttat that are consistent over the full mode I opening to mode II shear load range. The mixed-mode bending (MMB)
test was used to measure the deiamination toughness of a brittle epoxy composite, a state-of-the-art toughened epoxy
composite, and a tough thermoplastic composite over the full mixed-mode range. To gain insight into the different failure
responses of the different materials, the delamlnation fracture sudaces were also examined. An evaluation of several failure
criteria which have been reported in the literature was performed, and the range of responses modeled by each criterion was
analyzed. A new blllnear failure criterion was analyzed. A new bilinear failure criterion was developed based on a change in
the failure mechanism observed from the delamination surfaces. The different criteria were compared to the failure
response of the three materials tested. The responses of the two epoxies were best modeled with the new bilinear failure
criterion. The failure response of the tough thermoplastic composite could be modeled well with the bilianar criterion but
could also be modeled with the more simple linear failure criterion. Since the materials differed in their mixed-mode failure
response, mixed-mode deiamlnation testing will be needed to characterize a composite material. This paper provides a
critical evaluation of the mixed-mode failure criteria and should provide general guidance for selecting an appropriate
criterion for other materials.
.!

14. SUBJECT
TERMS

15. NUMBER

Composite;Mixed-mode delamlnatlon; Toughness testing;Failurecriteria

17.

SECURITY CLASSIFICATION
OF REPORT

18.

SECURITY
CLASSIFICATION
OF THIS PAGE

19.

SECURITY
CLASSIFICATION
OF ABSTRACT

OF PAGES

51
16.

PRICE CODE

20.

LIMITATIOI_

,,

!/

#04
OF ABSTRACT
i"

I In,,,laeQiflm,I
7S40-01-280-5500

I In,.,l=e=;t;,,_l

' "
Standard
Pre_,cr_bed

Form
by

_N_I

298
_td

(Rev.
Z]g-IB

2-89)
1' ' '

"

NSN