You are on page 1of 11

Composites 26 (1995) 257 267

S UTTERWORTH
I N E M A N N 9 1995 Elsevier Science Limited
Printed in Great Britain. All rights reserved
0010-4361/95/$10.00

Prestandardization study on mode II


interlaminar fracture toughness test for CFRP
in Japan

Kiyoshi Tanaka*
Toyama Prefectural University, Toyama 939-03, Japan

and Kazuro Kageyama


Tokyo University, Tokyo 113, Japan

and Masaki Hojo


Kyoto University, Kyoto 606-01, Japan
(Received 4 April 1994;revised August 1994)

A three year long cooperative research programme has been carried out to establish a Japanese standard
for an interlaminar fracture toughness test method for CFRPs. For mode II fracture toughness tests, the
end-notched flexure test was employed. Two series of round robin tests (RRTs) were carried out using
conventional brittle epoxy CFRPs, a toughened epoxy CFRP and APC-2. The following points became
clear in the investigation: 1) crack shear displacement measurement gives precise information on crack
initiation; 2) the fracture toughness values at the start of non-linear behaviour showed a considerably large
scatter; 3) the toughness at the 5% offset points has a lower scatter and is still conservative; and 4) it is
necessary to take account of the increase in crack length for the toughness calculation after the 5% offset
point. Considering this information, together with other important observations obtained in the RRTs, a
proposal for the mode II fracture toughness test was made and it was established as Japanese Industrial
Standard K 7086 in March 1993.

(Keywords: CFRP; interlaminarfracture toughness test; Japanese Industrial Standard)

INTRODUCTION mode I fracture toughness test have been reported else-


where 1'2. With respect to the mode II test method, an
Interlaminar fracture is one of the most important frac-
investigation was run to study the effects of starter films,
ture modes in structures made of polymer composite
precracking, calculation methods, choice of the critical
materials. Unless adequate control measures are made,
points and control methods on the tests. The present
catastrophic fracture accidents can take place because
report summarizes this work and describes the results.
delamination damage in these materials is often invisible
from the surface. To avoid such a catastrophe, the resis-
tance of the materials against interlaminar fracture can P U R P O S E OF T H E S T U D Y
be studied by means of interlaminar fracture toughness
Although the E N F test has been carried out at m a n y insti-
tests. Standardization of the test methods is a topic of
tutions for several years 3,4, it is generally agreed that the
current interest in several major standards institutions in
test method requires experience and careful preparation
m a n y countries, such as the American Society for Testing
in order to obtain consistent results. Furthermore, there
and Materials, the European G r o u p on Fracture and the
is still discussion 5,6 from the fracture mechanics point of
Japan Industrial Standards (JIS) Committee.
view on the best method to determine the critical point
An investigation p r o g r a m m e was run from 1989 to
on the load v e r s u s deflection diagram. Hence, in the
1992 in the Japan High Polymer Center to prepare a
present research programme, investigations have been
draft standard of the JIS for an interlaminar fracture
carried out not only from the technical viewpoint, but
toughness test method for CFRPs. Double cantilever
also from an academic standpoint. The main subjects
beam (DCB) and end-notched flexure (ENF) specimens
considered in the study were as follows.
were selected and studied. Results of work regarding the
. To establish a consistent test method(s) that can be
applied for various purposes at a wide range of
* T o w h o m c o r r e s p o n d e n c e s h o u l d be a d d r e s s e d laboratories. This method should be applicable to a

COMPOSITES Volume 26 Number 4 1995 257


Mode II interlaminar fracture toughness test: K. Tanaka et al.
variety of materials, ranging from conventional specimens were carried out by the same company.
brittle epoxy matrix CFRPs to very tough matrix Laminates of APC-2 were press-consolidated by Fiberite
CFRPs like APC-2. Composites Materials, Arizona, USA.
2. To make the proposed method as diverse as poss- During fabrication of the laminates, thin plastic films
ible so that the method can cope with various situ- (starter films) were inserted at the midplane to produce
ations. For this purpose, new methods such as the the initial cracks. For the starter films, several materials
stabilized ENF test and new calculation methods were used to check the effects of material and thickness.
were investigated. These were 25, 12.5 and 7.5 Ixm thick polyimide (PI)
3. To clarify the limitations of the established test films and a 12.5 txm thick polytetrafluoroethylene
method from the viewpoint of scatter and other (PTFE) film. The 25 p,m thick Pt film was laid after it
statistical aspects, as well as from the fracture was folded, i.e. the total thickness of the starter film was
mechanics point of view. For this purpose, round 50 txm. Mould release agents were applied to the single
robin tests (RRTs) were organized and test data and PI films. Several materials, including liquid and spray
information were collected. Analyses were types, were used and compared with respect to release
performed on these data and it was determined characteristics.
whether a consistent result could be achieved by Figure 1 shows the ENF test specimen configuration
different institutions. used in the present investigation. The width of the spec-
imen, 25 ram, was selected taking into account the fact
In the course of RRTs the following items were taken that a tow of 12 K fibres spread 5-7 mm wide in prepregs
as the subjects for investigation. in ordinary cases. The specimen length and the bending
span were 140 and 100 mm, respectively. The overall
1. Effects of starter film materials, their thickness and starter film length was 45 mm and the initial crack length
mould release agents used. Precrack conditions were a was supposed to be around 25 mm. Precracks were
also compared. introduced by mode I fracture by means of wedge
2. Stability of the tests. The effects of the control opening for precracked tests, and an initial overall crack
methods, specimen preparation and the initial crack length of 25 mm was applied. The starter films were
length were examined. removed from the cracks during the precracking proce-
3. Effects of fracture toughness calculation method. dure. For the no-precrack specimens the starter films
Two sets of equations were compared. were also removed before the tests by opening the cracks
4. Benefits of the crack shear displacement (CSD) carefully so that no crack initiation took place. In some
measurement. cases, as reported later, the removal of the film was not
5. Effects of crack length adjustment after fracture possible because of insufficient releasing characteristics.
initiation.
6. Benefits of the stabilized test methods. Test methods
ENF test procedure. The three-point bend test
Items 4 and 5 in particular were viewed as important. method was applied in the investigation. Although other
By means of CSD measurement sensitive monitoring of test methods were possible it was thought that the three-
crack initiation and propagation was expected. It was point bend ENF test method was simple and sufficient
also thought that significant increments in crack length for obtaining necessary information. Before the tests,
had taken place, even at the maximum load point. and after removal of the starter film, a small coupon of
PTFE film of 0.15-0.30 mm thickness was inserted
between the crack surfaces to reduce the friction between
INVESTIGATION PROCEDURE the upper and lower cantilever beams during the shear
Materials and specimen preparation movement. The size of the coupon was 25 mm wide, the
The following four materials were used in the present same as the specimen width, and 5 mm long in the spec-
investigation. They included very tough matrix CFRPs imen longitudinal direction. It was placed just above the
as well as conventional brittle epoxy matrix CFRPs. bending support. The effect of this method was experi-
mentally checked.
9 Unidirectional (UD) CFRP with a conventional In the test, measurements were made of the load, the
(brittle) 120~ cure epoxy and 4 GPa class strength load point displacement and the crack shear displace-
CF (Besfight HTA). ment (CSD), where instrumentation was possible. Figure
9 Woven CFRP with a conventional 120~ cure epoxy 2 shows an illustration of the CSD gauge used 7. The main
and 4 GPa class strength CF (Torayca T400HB). body of the gauge is mounted on the upper sprit
9 UD CFRP with a 180~ cure toughened epoxy and cantilever of the specimen and the cantilever fixed to it
6 GPa class strength CF (Besfight IM-600). inspects the movement of the lower part. To obtain
9 UD CFRP of APC-2 (PEEK matrix with AS4 CF).
p
The conventional epoxy UD and the woven CFRPs were
fabricated from Q-1113 prepregs produced by Toho
Rayon Co., Ltd, and FJ-6343G prepregs produced by
Toray Industries, Inc., respectively. The number of plies
were 22 and 16 for each material, to produce a specimen
thickness of ~ 3 ram. The toughened epoxy CFRP was 2L
made of prepreg Q-C134, produced by Toho Rayon Co.,
Ltd. Fabrication of the laminates and machining of the Figure 1 End-notched flexure (ENF) test specimen

258 COMPOSITES Volume 26 Number 4 1995


Mode II interlaminar fracture toughness test: K. Tanaka et al.

~ rain~

f
Cantilever
beam
"-L§
+

Cr/ k
STcimen

t
z
886

788

668

580

4~3

9-, 338
2g~3
rl~x LOQD = 738 N

laa
Figure 2 Illustration of crack shear displacement (CSD) gauge / /-

o (~Jf- , _,__

compliance values to calculate the propagating crack D ISPLACEMENT(CS3), mm

length, an unloading procedure was applied at some (a) Load v s CSD (Crack Shear Displacement)
laboratories, where a control method for stabilizing the
crack extension was available.

Control methods applied. Although in most cases the


8~176
70~
MAX LOAD = 730 N

ENF tests were carried out under the condition of


constant crosshead speed, two other methods (detailed
600

508
i
below) were also applied in the present study. The latter
methods were applied to prevent unstable fracture prop- 400
agation, which is found in tests performed using the Q 30g
conventional method, and to stabilize crack propagation
in ENF tests. Using such control methods, the measure- 2~0

ment of fracture toughness during propagation, Gim, t 0~1 7 ~

became possible. The three methods are as follows. I


0--

1. Conventional method. The crosshead or the load DISPLP~ENENT(DTF), mm


point displacement speed is controlled and constant.
An unstable fracture tends to take place unless Gtm (b) Load vs.load point displacement
increases significantly after the initiation. Figure 3 Examples of records for tests using the CSD control method
2. CSD control method. A servo-controlled test
machine is used. The CSD output is input to the Method of analysis
feedback circuit of the machine and is controlled so
as to increase at a constant rate. This method was
Calculation methods. For calculation of the mode II
proposed by Kageyama et al 7. interlaminar fracture toughness Gnc, equations (1) and
(5) were applied. Equation (1) is derived from the load
3. Coordinate conversion control (CCC) method. A
point compliance equation, equation (2), obtained by
servo-controlled machine is used. The load output
Russell and Street 9
P~ and the load point (or crosshead) displacement
D~ are input to a simple circuit that gives an output 9P ZCa2
C~ = D , - o~P~. for feedback to the test machine. C'it = (1)
2B(2L3 + 3a 3)
When C,, is controlled so as to increase monotoni-
cally the crack propagation is stabilized. The 2L~ + 3a 3
constant ct is determined by experiments or by frac- c --- (2)
ture mechanics. This method was proposed by 8Er BH 3
Tanaka 8.
where P is the load; C is the load point compliance; a
is the crack length; B is the specimen width; L is the
Figures 3a and b show examples of test results obtained in
half span of the bending supports; EL is the longitudinal
the CSD control ENF tests on the conventional epoxy UD
elastic modulus of the specimen; and H is half the
CFRP. Figure 3a is the load versus CSD record and Figure
specimen thickness.
3b is the record for the load versus load point displace-
In the following calculation of the experimental data,
ment. The matrix epoxy of this material is brittle and an
two kinds of methods were applied: one using the initial
unstable fracture takes place after the maximum load
crack length a0 for a and the other using ap, which reflects
without exception under the conventional test method.
crack propagation. The propagating crack length ap can
However, crack propagation is stable as shown in Figures
be obtained using the following equation, derived from
3a and b when the CSD control method is applied. It
equation (2)
is found in Figure 3b that the load point displace-
ment decreased after the maximum load point as the crack
propagated. This movement of the test machine is not
possible in the conventional test method. Since crack
propagation was stabilized, an unloading procedure for
it, J
(3)

calculation of compliance was possible, as shown in the Calculation of ap can also be performed by means of the
figures. following relationship between the crack length and CSD

C O M P O S I T E S V o l u m e 26 N u m b e r 4 1995 259
Mode II interlaminar fracture toughness test: K. Tanaka et al.

compliance Z obtained through a preceding calibration 2008


flAX tOAD = 1672 N

Qp _-- + l'= (4)


1500

CSD5% ".-.
where r0 and fl~ are empirical constants.
1803 t NL
Gu~ can be calculated by another formula obtained by /
Kageyama et at. 7 /
5a0 / CS _
G ~ - 3 P 8 _ 3P2,t (5)
8BH 8BH
.~W ~
i r
where 6 is the CSD. This equation does not require the ~ .1 .3 .4 5 .6
crack length value. Here the effect of crack length DISPL~ICENENT(CSD), mm

increase is automatically taken into account because 6 (a) l.oad vs. CSD (Crack Shear Displacement)
and ,!. reflect the changes in specimen conditions caused
by crack propagation. The CSD compliance can be
obtained from the unloading trace when the unloading 2~00
procedure was applied in the test. It is also obtained from Na• LO~D = 1W2 r/
/
the slope of a line drawn between the start point of the
CSD versus load curve and a point of the interest on the 1500 ~.~/~'~- VD5%
record. The latter method is used when the unloading z

process was not applied. Gn~ values obtained by this


equation are designated G~'Ir in order to distinguish them 1~]00

from values calculated using equation (1).


S a
In the present investigation, calculation of Gm was
made for the following combinations of equations. 500

Method 1: equation (1) using original a, i.e. a0. @


Method 2: equation (1) with equation (3) or 0 1 2 3 4 5 6 7
equation (4) to calculate ap. DISPLI~EilENT(DTF), rnrn

Method 3: equation (5), which automatically takes (b) Load vs load point displacement
crack length increment into account.
F i b r e 4 Comparison of records of CSD and load point displacement
It is considered that Method 1 has been employed in for APC-2
most of the papers and reports so far, as only equation
(1) or similar equations have appeared there. Methods 2 toughness values CaR were used. GI[R
and 3 are employed here to check whether the effects of were calculated in the zone where the
crack extension can be ignored, particularly in cases with increase in crack length was 5-20 mm and
tough matrix CFRPs. the overall crack length did not exceed
45 nun.
Points f o r toughness calculation. Figure 4 shows one
of the test records for APC-2. The top figure is a trace R o u n d robin tests
of the load versus CSD output, indicating a considerably Two series of RRTs were run in the present study to
early start of non-linearity (marked by an arrow at the investigate the general applicability of the test method
lower position) compared to the record for the load and to obtain some basic knowledge of the reliability of
versus load point displacement shown in the bottom the data obtained using the method. The first R R T was
figure. These figures suggest the necessity of analysis for run by six members of the committee using the conven-
both records. Consequently, the following six points
were taken as the 'critical points' to calculate toughness Table 1 Details of round robin tests
values.
Number of institutions
Starter and specimen
VDNL: The point corresponding to the start of
Material film No-precrack Precrack
non-linearity in the load versus load point
displacement (vertical displacement) record. Brittle epoxy PI 25 ixm folded 4/20 7/48
VD5%: The point where the compliance has increased UD CFRP PI 12.5 p,m 2/10 2/10
PI 7.5 ~m 9/46 10/50
by 5% in the same record as above, as indi- PTFE 12.5 Ixm 3/18 2/10
cated by the higher arrow in Figure 4b.
CSDNL: The point corresponding to the start of non- Brittle epoxy PI 25 Ixm folded 1/2 2/10
linearity in the load versus CSD record. woven CFRP PI 12.5 i~m 2/10 2/10
The point where the compliance has increased PI 7.5 I~m 3/15 3/15
CSD5%: PTFE 12.5 ixm 2/10 3/15
by 5% in the load versus CSD record, as
shown by the higher arrow in Figure 4a. Toughened epoxy PI 12.5 p,m 3/21 3/15
PlTlax: The maximum load point. UD CFRP
PROP: Points in the crack propagation portion.
APC-2 UD CFRP PI 7.5 lira 224 2/5
An average of five or more propagation

260 COMPOSITES Volume 26 Number 4 1995


Mode II interlaminar fracture toughness test: K. Tanaka et al.

tional epoxy UD and woven CFRPs. The second R R T is, casy opening and detaching. Other materials resulted
was carried out by a wider selection of members consist- in quite poor characteristics. The crack surfaces hardly
ing of seven universities, three public research institu- opened, even when a considerable wedging force was
tions and 15 private companies, using the same CFRPs. applied. This problem took place only when the 7.5 Ixm
At the same time as the second R R T a small R R T was PI film was used. The results show that it is necessary to
also run on tough matrix CFRPs. Table 1 summarizes check in advance whether the mould release agent to be
the combinations of the materials and test conditions used presents sufficient release capability when applied
applied in the RRTs. Nearly 350 specimens were tested to the film and the prepreg to be used.
in the study. Table 2 shows the control methods applied Since a release agent with poor characteristics
by the R R T members. Test methods for stabilization happened to be used in part of the second RRT, diffi-
were carried out by a few members. culty in detaching the crack surfaces and removing the
film was experienced at several laboratories, it is consid-
TEST R E S U L T S ered that there was the possibility of insufficient separa-
Effect of release agents and films tion and unintentional precracking in tests with the
To reduce the friction between the upper and lower no-precrack specimens. Hence care has been taken in
surfaces of the crack, it is essential to detach the crack the data analyses of these test results. Additional tests
surfaces before the test. This can be done by opening the were also run for specimens with sufficient detaching
crack slightly with a knife. The starter films are removed characteristics and the results were used for the follow-
during this procedure. Care is taken not to initiate mode ing analyses.
I fracture in the case of no-precrack tests. When a single
film is used, the release characteristics of the film itself Effects of test technique and control methods
or the release agent used are important. This procedure In E N F tests it is always necessary to apply lubrica-
of detaching the crack and removing the film becomes tion to reduce friction between the crack surfaces in order
very difficult when insufficient release agent is used. In to obtain precise information on crack initiation. Figure
the present study, several materials were tried and 5 shows examples of load versus CSD relationships
checked in terms of releasing characteristics. obtained in tests on conventional epoxy U D C F R P using
Table 3 summarizes the results, which show that a the following lubrication methods. The initial defect
considerable problem seems to arise in the case of the PI applied was the folded PI film with a precrack.
film with 7.5 Ixm thickness. In this case only two mate-
rials out of the six agents offered sufficient results, that (a) No lubrication material in the crack.
(b) Single 12.5 txm thick P T F E film in the crack.
(c) Single 12.5 txm thick PTFE film together with
Table 2 Control methods applied by R R T members
P T F E lubricant powder sprayed in the crack.
Control Test speed Number of Measurement (d) A 150 txm or 300 ~m thick P T F E lubricant coupon
method (mm) institutions items placed in the crack just above the support jigs.
The size of the coupon was 5 mm in the specimen
Load point 0.5, 1,0
displacement (load point 25 longitudinal direction and the same width as the
control displacement) specimen, i.e. 25 mm.
Load, load point
Crack shear 0.03 displacement and All the tests were carried out using the CSD control
displacement (crack shear 2 crack shear
control displacement) displacement
method. Record (a) shows noise-like traces. These were
(when available) the records of quite frequent unloading that took place
Coordinate 0.5, 1.0 due to intermittent shear movement because of the
conversion (load point absence of lubrication between the crack surfaces.
control displacement)
Record (b) shows that a thin 12.5 txm P T F E film was

Table 3 Effects of release agents and starter films on release characteristics

Release
No. Film material Release agent Supplier Type Workability ~ characteristicsb Judgement

1 PI 25 t~m DF ND ~ Spray Good Good OK


2 PI 12.5 p~m DF ND" Spray Good Good OK
3 PTFE 12.5 g m Not applied - - Good OK
4 PI 7.5 p,m Not applied - - Very poor NR d
5 PI 7.5 ~ m DF ND c Spray Good Poor NR d
6 PI 7.5/zm DF ND ~ Liquid Difficult Very poor NR a
7 PI 7.5/xm B ND' Liquid Difficult Very poor NR d
8 PI 7.5/xm G N D '~ Liquid Difficult Very poor NR '1
9 PI 7,5/xm Frekote44 Hysol Spray Good Good OK
l0 PI 7.5/~m XK30 Releasomers Liquid Difficult Good OK

Difficulty in coating the agent and spreading film over specimen


b Difficulty in removing film from crack
" Not disclosed
d Not recommended

COMPOSITES Volume 26 Number 4 1995 261


Mode fl interlaminar fracture toughness test: K. Tanaka et al,

LOaD 9 ~11.6'7 N 661 N .nu 64~' N d

/it C
5gfl
Z
4~
=g
0 3~e
.J

Crack Shear Displacement

Figure $ Effects of crack surface lubrication methods on test results

not sufficient to give lubrication, in record (c) a PTFE 12.5 ~m PI film, a longer crack, ~ 0.5, was necessary to
powder was used as the lubricant on top of the thin achieve stabilized crack propagation. This is caused by
PTFE film. Although this improved the results, slight a higher fracture initiation toughness (in this case, the
unloading still took place intermittently. When a 150/am toughness at the maximum load) due to the thicker
thick PTFE film coupon was inserted between the crack starter film. As shown next, the toughness correspond-
surfaces, much better results were obtained, as shown in ing to the maximum load was almost the same as
(d). After this trial and other preliminary experiences it or higher than the propagation toughness obtained in
was decided that all tests should be conducted using tech- the stabilized tests. It is considered that the test mach-
nique (d). ines used were not able to react fast enough to stop
Although the CSD control method and the CCC crack propagation after fracture initiation with a high
method enable stable crack growth after the maximum maximum load, though stabilization became possible for
load, stabilization was not possible in some special eases. specimens with a longer initial crack.
Table 4 lists the conditions for stable crack propagation.
This information was taken from the data obtained in Effects of starter film thickness and precrack
the RRTs. When the conventional load point displace- In Table 5 the main test results from the first and
ment control method was employed, all the specimens second RRTs are summarized. The values shown are Gtxc
failed in an unstable manner. The CSD control and the values and coefficients of the variation. These values
CCC methods presented almost identical results. It was were calculated using the average GI~ values reported by
difficult to stabilize crack propagation when the crack the participating institutes based on the load point
length was small. For a specimen with a shorter crack, displacement measurement. In Figures 6-10, the test
the level of load point movement necessary to stabilize results for the C_rli e values are illustrated. The test results
the test for a unit length of crack propagation becomes for the conventional epoxy UD CFRP shown in Figure
larger. Hence the test machine has to react quicker as 6 indicate that the starter crack preparation has a very
the initial crack length becomes smaller. This may be the significant effect on the results. Not only the folded 25
reason for the difficulty in stabilization of specimens with ~m thick PI but also the much thinner 12.5 ~m thick
short initial cracks. single PI film, when tested in the no-precrack condition,
For the precracked specimens and the no-precrack resulted in higher Gno values (at VDNL and VD5%) than
specimens with starter films of 7.5 ~m PI or 12.5 ~m those for no-precrack tests with the 7.5 o,m PI or 12.5
PTFE, stabilization was possible when the crack ~m PTFE films. Precracked tests yielded almost identi-
lengtNhalf bend span ratio was larger than ~0.4. cal results irrespective of the material and thickness of
However, in the case of the no-precrack test with the the films used. The results for the no-precrack tests with
the 7.5 l~m PI or the 12.5 ~m PTFE film are very close
Table 4 Minimum crack length necessary to stabilize the ENF test to those for the precracked tests, though the scatter of
(top figure is crack length in mm and bottom one is a/L ratio) the former seems to be a little larger than that of the
Control method
latter. In this condition, Gik values for VD5% and Pmax
are almost identical. The Gnc values for VDNL are only
Load point Crack shear Coordinate a little smaller than these values.
displacement displacement conversion Figure 7 shows the Gilt values for the conventional
Matrix Starter control control control epoxy woven CFRP. The effects of the film material and
Brittle PI 12.5 p.m Unstable 24.6 24.6 thickness and precracking condition are not so signifi-
epoxy fracture (0.49) (0.49) cant as in the case of the UD material. Gw1~values at
PTFE t2,5 ixm Unstable 2t.0 - VDNL for the folded 25 pLmPI film are higher than those
fracture (0.42) for other film materials. The lowest Gm value was
PI 7.5 ~ m Unstable - 21,0
fracture (0,42)
obtained at VDNL for the 7.5 t~m PI film without
Precrack Unstable 20 20.0 precrack. This agrees with the results of the UD mater-
fracture (0,40) (0,40) ial, where the 7.5 ~Lm PI film resulted in the lowest GIIc
value at VDNL in the no-precrack condition. The reason
Toughened Precrack 26,2 19,5 19,5 why the 12.5 ltm PTFE film gave a higher G]k value at
epoxy (0.52) (0.39) (0.39)
VDNL in the UD and woven CFRPs may be the wavy

262 COMPOSITES Volume 26 Number 4 1995


Mode II interlaminar fracture toughness test: K. Tanaka et al.
Table 5 Gnc for various specimens with different crack starter (units = k.l m 2)

Starter No-precrack Precrack (by mode I)

VD5% VD5%
Point of calculation VDNL offset Pmax Propagation VDNL offset Pmax Propagation

(a) Material: brittle epoxy UD CFRP

Film PI 25 ixm 1.036 1.229 1,229 0,372 0.436 0,440


(28) (44) (44) (23) (14) (15)
PI 12.5 ~m 0.48 0.47 0.47 0.417 0.48 0.48
PI 7.5 tzm 0.304 0.434 0.441 0.257 0.427 0.435
(29) (19) (18) (24) (10) (10)
PTFE 7.5 gm 0.37 0.42 0,42 0.32 0,370 0.38
Average 0.310 0.430 0.435
(28) (12) (12)

(b) Material: brittle epoxy woven CFRP

Film PI 25 p~m 0,89 0.92 1,05 0,92 1.2 1.29


P[ 12.5 I~m 0.74 0.94 1.28 0.56 0,87 1.09
PI 7,5 bun 0.46 0.76 1.26 0.61 0.81 1,22
PTFE 7.5 Ixm 0.64 0.85 1.08 0.57 0.97 1.19
Average 0.628 0.918 1.182
(33) (23) (11)

(c) Material: toughened epoxy UD CFRP

Film PI 12.5 tzm 0,991 2.589 2.842 2.487 0.930 2.104 2.489 2.483
(39) (7) (12) (6) (45) (16) (12) (9)

(d) Material: APC-2 UD CFRP

Film PI 7.5 Ixm 1.477 2.034 2.014 2.582 1.530 1.922 1.959 2.318
(25) (17) (14) (10) (13) (6) (2) (6)

Note: figures in parentheses show coefficients of variation (%)

crack surfaces as well as the larger thickness of film. It


is difficult to obtain a flat crack surface with a thin PTFE
film because it has low rigidity at high cure temperatures
and cannot resist the tendency for ply waviness. The
woven CFRP has a higher Giic than that of UD CFRP
made of a similar kind of epoxy and CF materials. Gtl~
values for Pmax are significantly higher than those for
VD5%, which is different from the situation for the UD
material.
Figures 8 and 9 indicate the results for the toughened
epoxy CFRP and APC-2, respectively. In both these
figures there is a significant increase in Giic after VDNL.
The final propagation toughness values for both are
Figure 6 Influence of starter films on G~xcof brittle epoxy UD CFRP almost identical and high. The changes in G~icvalues with

Figure 7 Influence of starter films on Gi~c of brittle epoxy woven Figure 8 Influence of precracking on Gllc of toughened epoxy UD
CFRP CFRP (starter film: 12.5 ~tm PI)

COMPOSITES Volume 26 Number 4 1995 263


Mode II interlaminar fracture toughness test: K. Tanaka et al.
precracked tests are considered here. Variations of more
than 20%, and 45% at the maximum, are found in the
Gllc value at VDNL. This scatter is quite considerable
and causes us to question adoption of this value for
standardization without any further strategy for
improvement. Gnc values at VD5% or Pmax show ~ 10%
variation, which is not considered substantial in the
present situation.

Summary of RRT
The results of the RRT can be summarized as follows.

9 Careful selection of a mould release agent for the


Figure 9 Influence ofprecracking on G,,cofAPC-2 UD CFRP (starter starter film seems to be necessary especially when a
film: 7.5 ~m PI) very thin film is used.
9 A PTFE film insert between the crack surfaces is
effective in smoothing the shear movement of the
specimen and improves the stability of the test.
9 The CSD control method and the CCC method are
effective for stabilization of crack propagation. By
means of these methods, GnR curves can be obtained.
9 Starter films of 7.5 ~m thick PI and 12.5 I~m thick
PTFE give conservative test results even when tested
in the no-precrack condition.
9 For a woven CFRP and a tough matrix CFRP the
starter film preparation has less significance than for
a brittle epoxy UD CFRP.
9 Precracking did not show a significant effect on the
initiation toughness (in this case VDNL and VD5%),
even in the materials with rising R-curves.
9 Tough matrix CFRPs show a significant toughness
Figure 10 Changes in mode II fracture toughness with crack propa- increase after fracture initiation. It may be desirable
gation
to standardize a method (or methods) to analyse the
propagation toughness.
crack grouch are illustrated in Figure 10 where the R- 9 The difference in toughness values obtained by many
curves for the conventional epoxy CFRPs are also shown institutions is not significant if the 5% offset or the
for comparison. The R-curves for the conventional epoxy Pmax point is used. It becomes considerably larger
woven CFRP, the toughened epoxy CFRP and APC-2 when the starting point of non-linear behaviour is
showed a significant increase with crack propagation. used.
Their GnR values reached plateaus after 3-5 nun of prop-
agation. However, no significant change was found on
DISCUSSION
the fracture surfaces of these materials. The reason for
the increasing R-curve is not clear at this stage. Effects of equations applied and crack length adjustment
Figures 8-10 suggest that the standardized test In Table 6 Gn~ values based on the CSD measurement
method should take into account not only the initiation are summarized together with other results. Sufficient
point (whether it be the non-linear point or the 5% offset data for analysis were available only for the toughened
point) but also the propagation toughness values. GI~ at epoxy UD CFRP. Gm values are calculated by means of
the non-linear or 5% offset point is conservative and on the three different methods described in the earlier
the safe side if used as the critical value in design. section. Method 1 applies initial crack length a0, whereas
However, there seems to be no real reason why all parts method 2 used adjusted crack length ap. Both of them
and structures have to be designed based on that value. use equation (l). For calculation of G~'t~based on equa-
Depending on the safety design criteria applied and the tion (5), values of crack length are not required, but
fracture mechanics under consideration, there is the and A in equation (5) reflect the changes in the crack
possibility of utilizing the high GnR values during prop- length.
agation. Hence it is desirable to specify the methods used Also shown in Table 6 are the coefficients of varia-
to obtain this propagation toughness as well as the tion of aIlc and G~'ic. In this case the values were calcu-
c o n s e r v a t i v e Giic values. lated using all the GHcand G*,~data reported by six RRT
members. Figure 11 compares the two G~,c values (i.e.
Variation of GIlc among institutions with and without crack length adjustment) calculated at
Table 5 also shows the coefficient of variation for the various critical points. It is observed in Figure 11 that
Gllc values reported by the RRT members. Because the crack length adjustment has only a slight effect on G,I0
average values reported by the members were used for values up to the CSD5% point. However, Giic values at
calculation, Table 5 shows whether there is a difference VD5% and Pmax calculated with adjusted crack length
in the test results from the different institutions. To make are significantly higher than the values with no adjust-
the discussion simple, only the results from the ment. It may be concluded from Figure 11 that crack

264 COMPOSITES Volume 26 Number 4 1995


Mode II interlaminar fracture toughness test: K. Tanaka et al.
Table 6 Comparison of G[I~ calculated by various methods (material: toughened epoxy U D C F R P )
GI~ values (top, in kJ m ~ and coefficient of variation (bottom, in %)
Calculation
method Starter crack CSDNL VDNL CSD5% VD5% Pmax PROP

Method 1 Precrack 1.015 0.930 1.107 1.972 2.270


34 45 23 16 16
No-precrack 1,045 0,991 1.215 2.213 2.312
48 39 23 8 10
Method 2 Precrack 1.015 0.930 1.152 2.104 2.589 2.483
34 45 22 16 12 9
No-precrack 1.045 0.991 1.269 2.589 2.842 2.487
48 39 23 7 12 6
Method 3 Precrack 1.099 1.038 1.585 2.701 2.496 2.680
33 52 31 13 19 16
No-precrack 1.113 1.091 1.495 2.563 2.714
52 35 47 5 6

Figure 13 Fracture toughness values at various points in the


Figure 11 Effects of crack length calculation on Gllc for toughened load-displacement diagrams (Gnc values for toughened epoxy U D
epoxy U D C F R P C F R P obtained using calculated at)

Figure 12 Comparison of fracture toughness values of toaghened Figure 14 Coefficients of variation of fracture toughness values at
epoxy U D C F R P calculated using two equations
various points (same conditions as for Figure 13)

length adjustment is necessary in calculating GH~ for Selection of critical points


points at and after VD5%. Figures 13 and 14 compare Gnc values and the
Figure 12 is the result of a comparison of GHo and coefficients of variation of GIIc, respectively, calculated
G~lc. The test data used for this study are for the tough- for various points on the test records of toughened
ened epoxy UD CFRP, as was the case for Table 6 and epoxy UD CFRP. Here, equation (5) was used for
Figure 1I. The Gnc and G*~ values were calculated at the CSDNL and CSD5%, whereas equation (1) with ap
same point. For example, the G*~cvalue was calculated calculated by equation (3) was used for other points,
at the point of VP5% using the load and CSD data at so that reasonable matching between the measurement
that point, which is not done in ordinary cases. Figure and calculation method was obtained. From Figures 13
12 indicates that there is not a significant difference and 14 it is found that the Gtk values for CSDNL
b e t w e e n GIIc and G*
u0 values, and that both equation (1) and VDNL are fairly low (i.e. conservative) but that
and equation (5) can be applied and considered equiva- their variations are very high. Because of these high
lent to each other. variation coefficients, it is considered inappropriate to

COMPOSITES Volume 26 Number 4 1995 265


Mode II interlaminar fracture toughness test: K. Tanaka et al.

Critical point and equations. As stated earlier, design


safety and fracture mechanics considerations will decide
which of the toughness values is adopted for a given
material as the critical value. In the test the following
toughness values shall be obtained:

1. Gllc, toughness at VD5%, calculated by equation (1)


with ap adjustment by equation (3);
2. Guci, toughness at CSD5%, calculated by equation
(1) with a0 or GTici, toughness at CSD5%, calculated
by equation (5);
3. C~R, propagation toughness, calculated by equation
(1) with ap adjustment by equation (4) or G'm, prop-
Figm'e 15 Influence of unloading on GIIR of toughened epoxy UD agation toughness, calculated by equation (5).
CFRP
For calculation of 2 and 3 above, measurement of CSD
is necessary. For all the above cases, five or more tests
adopt NL points for a standard method. The fracture shall be carried out and the average value and the coef-
toughness values for the 5% offset points are lower than ficient of variation, if necessary, shall be reported.
that for the maximum load point and show smaller
variation. This suggests that they are suitable for Gnc~ and G~]ci defined above are the fracture tough-
standardization. ness values at the initial stage in the crack propagation
and are intended to represent the fracture toughness just
Effects of unloading on fracture propagation toughness after fracture initiation. It is necessary to measure CSD
Although the compliance for the calculation of prop- for calculation of these values. On the other hand, GI~r
agation toughness can be obtained from a line connect- is obtained from the load point displacement record and
ing the data point and the origin of the load versus shows the fracture toughness corresponding to clear non-
displacement (either the load point displacement or the linear deformation in the specimen. It is considered that
crack shear displacement) curve, an unloading procedure this shows the fracture toughness for the initial stage of
may become necessary in some cases, such as for speci- crack extension. Consequently, Gn~i is smaller than GIIe
mens which exhibit plastic deformation. In Figure 15 the and may be conservative when used for design safety.
effect of unloading on the GnR value is investigated for The proposed test method was adopted by the Japan
the toughened epoxy UD CFRP. The average value of Industrial Standard Committee and established in
GnR obtained in the tests with an unloading procedure March 1993 as JIS K 7086: 'Testing method for inter-
is quite similar to that obtained from the tests without laminar fracture toughness of carbon fibre reinforced
unloading. The unloading procedure did not appear to plastics'.
have any deleterious effect on the propagating toughness
in the case of the present material. However, whether the CONCLUSIONS
unloading procedure has any significant effect or not will
depend on the material tested. Hence this kind of inves- To establish a basis for standardization of the mode II
tigation always has to be made when the unloading interlaminar fracture toughness test method for CFRPs,
method is applied. two series of RRTs were run and fracture mechanics and
statistical analyses were made. The main investigation
Standard test method proposed results are as follows.
After the first and second RRTs and the analyses
shown above, the committee prepared a manuscript for 1. Precrack and a thin (7.5 ~m) PI film gave low frac-
a Japanese Industrial Standard entitled 'Testing methods ture toughness values. For the thin film, selection of
for interlaminar fracture toughness of carbon fibre rein- an appropriate release agent is necessary.
forced plastics'. This covered both mode I and mode II 2. Stabilization of crack propagation is effective in
fracture toughness test methods. In the mode II test analysis of propagation toughness.
method, the following features were included. 3. Crack length adjustment is necessary at and after the
5% offset point in the load versus load point displace-
Specimen preparation. For the starter film, PI film with ment relationship.
a thickness of 30 ~m or less is recommended. A 2-5 mm 4. No significant difference was found in the toughness
long precrack made by mode I fracture shall be used. values calculated using two different equations, one
When the CSD measurement is to be carried out, the based on the load point displacement and the other
end face where the tip of the CSD gauge contacts should on the crack shear displacement.
be smooth and square to the specimen surface. 5. Toughness values at the start point of non-linearity
in the load versus deformation records resulted in a
Test method. The three-point bend ENF test is spec- large scatter. This point, at this stage, seems to be
ified. A PTFE film insert between crack surfaces is inappropriate for standardization purposes.
recommended. Although the stabilized test is desirable, 6. It is difficult to select just one point on the
the conventional method (monotonic increase of load load-deflection diagram to represent the material.
point displacement) may be used. The test speed shall be It is reasonable to standardize multiple points as
0.5 mm rain -~ at the load point. characteristics of the material.

266 COMPOSITES Volume 26 Number 4 1995


Mode II interlaminar fracture toughness test: K. Tanaka et al.

A new standard for an interlaminar fracture toughness REFERENCES


test was established taking into account these points of 1 Research Report on Standardization of Polymer and Composite
information. Materials, Volume on Carbon Fiber Composites, Japan High
Polymer Center, Tokyo, 1992 (in Japanese)
2 Hojo, M., Kageyama, K. and Tanaka, K. Composites 1995, 26,
243
ACKNOWLEDGEMENTS 3 Russell, A.J. DREP Materials Report 82-0, Defense Research
Establishment Pacific, Victoria, BC, 1982
The authors are grateful to the Japan High Polymer 4 Carlsson, L.A., Gillespie, J.W. and Trethewey, B. J. Reirforced
Plastic Compos. 1986, 5, 170
Center for enabling them to publish the results of this 5 Carlsson, L.A. and Gillespie, J.W. in 'Application of Fracture
investigation. The members and participants of the Mechanics to Composite Materials', Vol. 6 (in Composite
research committee and the round robin tests are grate- Materials Series), Elsevier, Amsterdam, 1989
fully acknowledged. The authors are thankful to Dr H. 6 Davies, P., Kausch, H.H., Williams, J.G. and Kinloch, A.J.
6bmpos. Sci. Technol. 1992, 43, 129
Miyairi, Professor of Tokyo Medical and Dental 7 Kageyama, K., Kikuchi, M. and Yanagisawa, N. in ASTM STP
University and General Chairmen of the standardization 1110, American Society for Testing and Materials, Philadelphia,
activity of JHPC, and to Mr K. Kage and Mr H. PA, 1991, p. 210
Shinnabe of JHPC for their general support during the 8 Tanaka, K., 'Proc. Sixth Japan-US Conf. on Composite
course of the research. This research was sponsored by Materials', 1992, pp. 685-693
9 Russell, A.J. and Street, K.N. in ASTM STP 876, American
the Standard Department, Agency of Industrial Science Society for Testing and Materials, Philadelphia, PA, 1985,
and Technology. p. 349

COMPOSITES Volume 26 Number 4 1995 267

You might also like