Sie sind auf Seite 1von 10

29 3

2014 6


NEW CARBON MATERIALS

Vol. 29 No. 3

Jun. 2014

: 10078827(2014)03017610

1,2 , 1 , 3 , 1 , 1
(1. , 100191;
2. , 100020;

3. , 101300)

: ,

,
,,
,,

: ;;;;
: TQ342. + 74

: A

: ,,. Email:jennyzhaoyan@ buaa. edu. cn


:,,. Email: hao_jianwei88@ yahoo. com. cn

Experiments and finite element simulation of


interfacial properties for monofilament composites
HAO Jianwei1,2 , ZHAO Yan1 , LUO Yunfeng3 , WANG Yan1 , CHEN Da1
(1. School of Materials Science and Engineering, Beihang University, Beijing 100191, China;
2. Aviation Industry Corporation of China, Beijing 100020, China;
3. AVIC Composites Corporation, Beijing 101300, China)

Abstract: Carbon fiber / bismaleimide composites have received increasing interest, owing to their excellent properties, especially
their toughness under extreme working conditions. We established a micromechanical model for a finite element simulation of the
microdroplet test, which involves pulling a carbon fiber out of a bead of matrix using two moving knives acting on the bead as
scrapers to quantify the interfacial properties of carbon fiber reinforced bismaleimide composites. The interfacial shear strength of
carbon fiber / bismaleimide composites subjected to different hydrothermal environments was tested by microdroplet method to illus
trate the impact of moisture absorption on their interfacial properties. Hydrothermal aging caused a reduction of interfacial shear
strength, which leveled off when the immersion time in water exceeded 7 days at 71 . A numerical simulation of the debonding
process was performed based on the interface cohesive element damage model to simulate the interfacial properties of the composite
and to determine the correlation between experimental parameters and interfacial properties. The simulation successfully provided es
sential parameters for numerical analysis of the macroscopic mechanical properties of the composite. Finite element analysis of the
microdroplet test revealed that the factors that influence the interfacial shear stress distribution are the position of the knives on the
bead, thermal residual stress and hydrothermal treatment conditions.
Keywords: Microdroplet test; Carbon fibers; Hygrothermal effect; Interfacial strength; Finite element analysis
Received date: 20131226 Revised date: 20140605
Corresponding author: ZHAO Yan, Ph. D., Professor. Email: jennyzhaoyan@ buaa. edu. cn
Author introduction: HAO Jianwei, Master, Senior Engineer. Email: hao_jianwei88@ yahoo. com. cn
English edition available online ScienceDirect ( http:www. sciencedirect. comsciencejournal18725805 ) .
DOI: 10. 1016 / S18725805(14)601335

1 Introduction

Carbon fiber reinforced plastic composite


( CFRP) has been increasingly used in aerospace in

dustry during the past decades, owing to their attrac


tive properties, such as light weight, corrosion resist
ance, and easy to design. Bismaleimide (BMI) resins
are used as matrix of highperformance structural com

HAO Jianwei et al: Experiments and finite element simulation of interfacial

posites that require superior toughness and hightem


perature resistance. Epoxy blends of BMI resins can
withstand temperature as high as 245 without a con
comitant decrease in thermal stability because aromatic
polyimides are one of the most thermally stable organic
materials. Thermosetting BMI resins have played an
important role as matrix materials in hightemperature
aerospace structural applications owing to their superior
thermal stability [1] . In such applications, the compos
ites are exposed to harsh and varying environments in
cluding a wide range of temperature and hydrothermal
exposure, which accelerate declination of their me
chanical and other properties [2,3] . Thus, a variety of
correlative experimental and theoretical researches
have been done to discuss the impact of aging effect
on the performance of materials.
Different from the inplane tensile properties of
the fiber reinforced composites, the compression and
interfacial properties are dependent on the combination
of each primary component and the characteristics of
the interface between the fiber and the matrix. In con
tinuous fiberreinforced composites, the interface acts
as an intermediate bridge, which transfers the load
from matrix to fibers through the shear flow. Thus
methods for quantifying the interfacial shear strength
( IFSS) are to be established to support the study of
fiberreinforced composites. Test methods using sin
glefiber model composites, such as pullout test [4,5] ,
microdroplet test [68] , microdebond test [9,10] and
fragmentation test [11,12] are helpful in extracting the
debonding process of the interface from the overall
damage process of the composite. In particular, the
microdroplet test is a simple fiber pullout technique
to evaluate shear failure of the fibermatrix interface.
The microdroplet test involves pulling a fiber out of a
bead of matrix through a knifeedge, accompanied by
the debonding at the fiber / matrix interface, and then
it evaluates the interfacial properties based on the re
sulting force required in the pullout. This technique is
quite useful for smalldiameter fiber, such as carbon
fibers because the fiber pullout can be completed
without fiber breakage. However, the large variation
in test data caused a considerable overestimation or
underestimation of the IFSS for several reasons, such
as the bead geometry, the position of knifeedge, em
bedded length. In this case, the problem could be
solved by numerical analysis.
Recent researches have contributed to finiteele
ment modeling of the fiber pullout and pushout tests.
These models address the interfacial fracture during
the tests, using interface elements embedding the
damage zone onto the fiber / matrix interface.

177

Ananth [13] used linear spring elements to simulate an


interface, in which a stressbased failure criterion and
Coulomb friction law were applied. A remarkable as
pect of this approach was the progressive nature of the
simulation. Calibration of shear strength was carried
out using the measured pushout loads. Pochirajua [14]
conducted a stress analysis for the interface using the
socalled axisymmetric damage model based on the
Reissner s variational solution and finiteelement
modeling. Both were capable of modeling interfacial
adhesion, friction and normal debonding. Lin [15]
presented a progressive finiteelement modeling simu
lation using the cohesive zone model with a linear
tractionseparation law ( TSL ) . They captured the
load drop event at complete debonding. Chandra [16]
carried out a comparative study about the performance
of two kinds of cohesive elements with a bilinear and
an exponential TSL. Their simulation showed good a
greement with the experiment up to peak load fol
lowed by load drop.
In this study, a micromechanical modeling of the
microdroplet test was established, in order to quantify
the interfacial fracture properties of CF / BMI compos
ites. According to the IFSS results, a numerical simula
tion of debonding process is achieved on the basis of the
interface cohesive element damage model to determine
the interfacial properties of the composites. Finite ele
ment analysis of microdroplet test showed the influences
of knifeedge position, thermal residual stress and hy
grothermal environment on the IFSS distribution.

2 Experimental

2. 1 Specimen
The microbond tests were conducted using
CCF300 carbon fibers, provided by Weihai Tuozhan
Fiber Co, Ltd, and 5405 bismaleimide ( BMI) resin,
provided by Beijing Institute of Aeronautical Materi
als. One single filament was pulled straight and pas
ted at both ends of a concave metal card. A small
drop of liquidstate resin, which was carried by the
end of a metal pin, was dipped onto the fiber to form
a microdroplet, as shown in Fig. 1. The microdroplet
resin was then cured in the oven.

Fig. 1 Specimen of microdroplet test.

178

The moisture conditioning of the composites was


carried out according to ASTM 5229 standards. A se
ries of experiments was conducted in our previous
work to illustrate the performance of the composite
and mechanism of aging under the same standards to
meet the demand in industry. Based on the same con
dition, the specimens were put in an oven at 70 for
24 h to remove the moisture absorbed in air until the
weight is constant. After that, the specimens were
immersed into distilled water at 71 for moisture
conditioning. After a specified period, the samples
were moved into an 85 oven ( XMTA7000P) to
dry out until the weight is constant before next cyclic
conditioning. The period of moisture absorption were
7 and 14 d, and the time of desorption was 83 and
94 h, respectively. The symbol N m S was used to de
note a condition of cyclic hygrothermal treatment,
where N represents the period time of moisture ab
sorption, m represents the times of cycle, S repre

29

sents the wet / dry state, in which the dry state is pres
ented by D, wet state is presented by W. The condi
tions for cyclic hygrothermal treatment in this study
are 7 1 W, 7 1 D, 7 2 W, 7 2 D, 7 3 W, 7 3 D, 14 1 W,
14 1 D, 14 2 W, 14 2 D, 14 3 W and 14 3 D.
2. 2 Microdroplet test
In this study, the microdroplet test was carried
out on an interface performance evaluation device
FA620 ( Tohei Sangyo Co, Ltd, Japan) . After the
specimen is mounted, a knifeedge was adjusted to
make it contact the surface of the fiber at the bottom
of microdroplet resin, and then a steady displacement
was applied to the knifeedge to pull out of the single
fiber from the resin with a displacement rate of
0. 2 m / min as shown in Fig. 2. The test was con
ducted at room temperature under the observation with
an optical microscope, and the maximum load causing
debonding was measured by microsensors.

Fig. 2 Microdroplet test: ( a) the microdroplet test device, ( b) schematic diagram of microdroplet test,
( c) droplets before debonding and ( d) droplets after debonding.

Assuming the IFSS is approximately constant


along the entire interface, then the average IFSS can
be calculated from the following Eq. (1) :
Fd
=
(1)
2rL
Where is the average IFSS, F d is the maximum load
of interfacial failure, r is the fiber radius, and L is the
embedded length.

between the maximum load and the embedded length


in three different conditions.

3 Results and discussion

The carbon fiber / BMI composites were tested in


original condition and after water immersion for 7 or
14 d at 71 . Considering the influence of slightly
varied knifeedge positions and the embedded length
of fiber, 25 tests was taken to achieve an average re
sult in each condition. Fig. 3 shows the relationship

Fig. 3 Relationship of the maximum load and the embedded length.

HAO Jianwei et al: Experiments and finite element simulation of interfacial

As shown in Fig. 3, even if the embedded


length of the specimens was the same, the maximum
debonding load was still varied, which was related to
knifeedge positions and other factors. Data in Fig. 3
were calculated according to the formula (1), the IFSS
of single carbon fiber / BMI composite in the three differ
ent conditions was obtained and shown in Table 1.
Table 1 Average IFSS of CF / BMI composite at
different conditions at 71 .
Carbon fiber / BMI

Original

IFSS / MPa

70. 81

Water immersion Water immersion


for 7 d
for 14 d
54. 50

52. 36

The results of Table 1 show that the IFSS is re


duced after the immersion. During the moisture ab
sorption of the composites, physical changes such as
microcracks propagation and swelling, as well as
chemical changes such as hydrolysis and chemical
scission increase, lead to a degradation of the materi
als properties. Moreover, the fiber / matrix interface
can be influenced by moisture absorption [1719] because
BMI resin is extremely easy to absorb water, which
leads to volume expansion, while carbon fiber absorb
little water so the swelling of the resin generates stress
and can cause interface debonding [20,21] .
The IFSS of the composite after an immersion for
7 d was decreased by 24. 5% , while that of the com
posite immersed for 14 d was decreased by 27. 5% .
This indicated that the damage of the interface of the
composite immersed for 7 d got saturated.
To investigate the influence of cyclic hygrother
mal environment on IFSS, the specimen was im
mersed in the water for three cycles of 7 or 14 d at
71 . The results are shown in Table 2 and 3. After
the cyclic hygrothermal immersion, or extending the
cyclic time of moisture absorption, no obvious differ
ence in IFSS is found. Because size of the micro
droplet resin is around ten microns, the water mole
cules are easy to diffuse to the interface between the
carbon fiber and matrix. As the water absorption is
quickly saturated, the debonding of the interface
caused by the water immersion doesn t change appar
ently. Therefore IFSS in the two kinds of the cyclic
hygrothermal treatment are almost the same.
Table 2 Average IFSS of the CF / BMI composite after
an immersion at 71 for 7 d for different cyclic times.
Water
immersion

IFSS / MPa

Original
70. 81

71 W

71 D

72 W

72 D

73 W 73 D

54. 50 52. 2 53. 99 51. 97 52. 79 50. 66

179

Table 3 Average IFSS of the CF / BMI composite after


an immersion at 71 for 14 d for different cyclic times.
Water
immersion

IFSS / MPa

Original

14 1 W 14 1 D 14 2 W 14 2 D 14 3 W 14 3 D

70. 81

52. 36 51. 63 53. 06 50. 08 52. 28 48. 10

Based on the investigation of the microdroplet


test, the numerical analysis was established to simulate
the single fiber debonding process and to obtain the pa
rameters of interface properties for the composite.
Then, the model of droplet resin in the microdroplet
test was simulated using Carrolls theory [2224] , to ob
tain the actual profile of resin on the single fiber.

4 Finite element analysis

4. 1 Theoretical background
4. 1. 1 Linear elastic tractionseparation law [25]
The separation vector is defined as the relative
displacements u of the two contacting element faces
attached to a cohesive element as follows:
n
u n+ -u n =
+ - = u
(2)
s
u s -u s

{} {

Where the subscript n and s denote the normal and


shear component, respectively, and u + ; u denote the
displacements of an initially coincident node pair loca
ting on two contacting interface elements. The nomi
nal traction vector, T, also consists of a normal and a
shear component, T n and T s .
The nominal strains are the separations divided
by the constitutive thickness t c of the cohesive ele
ment. t c was set to unity whereas the actual thickness
was zero. With this setting, the nominal strain is
equal to :
n
n
n = = n , s = = s
(3)
tc
tc
The element stiffness becomes equal to the stiff
ness of material. The uncoupled elastic constitutive
relation is written as:
Tn
n
K nn K ns n
T=
=
=
(4)

Ts
s
K ns K ss s
4. 1. 2 Damage initiation and evolution
Damage and failure behavior of a cohesive ele
ment is controlled by the tractionseparation law
( TSL) whereas the TSL itself is defined by the initia
tion and evolution law of the damage variable. In this
work the linear maximum nominal stress criterion was
used. Damage initiation refers to the beginning of
stiffness degradation and damage is initiated when the
maximum nominal stress ratio reaches unity:
( tn ) ts tt
max o , o , o = 1
(5)
tn ts tt
Where t on t os represent the peak nominal normal and

{}

{}{}

180

29

shear stress, respectively. The symbol ( ) denotes


the Macaulay bracket signifying that a purely com
pressive load does not initiate damage.
The damage evolution law describes the form and
rate at which the effective stiffness of material de
grades after damage initiation, the whole damage
process is introduced by a scalar variable, D, it is
scaled from null ( virgin state) to unity ( full failure)
and evolves monotonically. During the displacement
based damage evolution of cohesive element, the var
iable D of the linear damage evolution as follows
Equation (6) :
0
fm( max
m - m )
D = max f 0
(6)
m ( m - m )
Where 0m and fm are the effective displacement of vir
gin state and full failure, respectively, the max
is the
m
maximum effective displacement during loading
process. The stresses are modified by D according to:
T n = ( 1-D ) T eln , T s = ( 1-D ) T els
(7)
el
el
Where T n and T s are the stress components predicted
by the linearelastic tractionseparation law ( TSL) for
current strain without damage.
4. 2 Geometric model
On the basis of Carroll s theory, the equilibrium
morphology of droplet on a fiber is shown in Fig. 4.
The morphology of droplet is based on the calculation
with the parameters indicated in the figure [22] . Ac
cording to the significance of parameters in Fig. 4,
the morphology of the resin microdroplet can be de
scribed by the following equations (8) [22,23] :

F( ,k)= 0
Besides

1k sin2
2

(11)

max = sin -1 1 1- r 2
(12)
h
k
L = 2[ arF( max ,k) +hE( max ,k) ] = f( ) (13)
Where L is the theoretical embedded length of the
droplet, L is determined by the contact angle . As
suming the theoretical embedded length is close to the
measured data l, we can calculate the as following
[26]
:
E L = l-2[ arF( max ,k) +hE( max ,k) ] =
l-f( )
(14)
The contact angle can be obtained by minimi
zation of this function, and then substituting in the
Equation ( 9 ) and ( 10 ) , a and k were calculated.
With these results and Eq. (8) the curve of the drop
let profile was established.
To solve the equations, MATLAB software was
applied, and the embedded length ( l) , droplet height
( h) and fiber radius ( r) were measured to be 42, 15
and 3. 5 m, respectively in the experiment. The con
tact angle was calculated to be 35. After plotting
the droplet profile by Eq. (8) using MATLAB, the
curve coordinates were imported into finite element
program ABAQUS to establish the geometric model of
the monofilament composite as shown in Fig. 5. It is
concluded that, compared with the profile, the micro
droplet geometric profile modeled using Carroll s the
ory is very approximate to the microdroplet profile of
real droplet resin, indicating that Carroll s theory fits
the numerical simulating well.

Fig. 5 Monofilament / microdroplet resin: ( a) photograph in the


experiment and ( b) microdroplet geometric model.
Fig. 4 Microdroplet test schematic showing the notation
used to describe the various parameters [22] .
Where, L is the embedded length, Rr is the fiber radius,
h is the bead height, is the contact angle, is the angle
from the yaxis ( referenced from a point defined by L, Rr and h) .

(1-k sin )
{yx ==h[ arF(
,k) +hE( ,k) ]
2

hcos-r
h-rcos
h2 -a2 r2
k2 =
h2
a=

(8)
(9)
(10)

4. 3 Finite element model


In consideration of the axial symmetry of the
monofilament composite, finite element analysis
( FEA ) was applied to establish an axisymmetric
model. When the embedded length is 42 m, the
droplet height is 15 m, the fiber radius is 3. 5 m,
and the knifeedge is 1. 5 m away from the interface,
the finite element model ( FEM) with interphase is set
up as shown in Fig. 6.
In the FEM, the finite element mesh simulating
the monofilament composite was created: CAX4R

HAO Jianwei et al: Experiments and finite element simulation of interfacial

( fournode plane stress element, one of the axisym


metric elements in FEM) was applied for the carbon
fiber, COHAX4 ( fournode interface cohesion ele
ment, one of the axisymmetric elements in FEM to
simulate cohesive behavior) for the interface and the
mixed of CAX4R and CAX3 ( triangular plane stress
element, one of the axisymmetric elements in FEM)
for the resin. Between the interface elements and the
fiber ( or matrix) , the transition meshing technology
(Fig. 6) was used to maintain the dense mesh of in
terface while the mesh generation of fiber and matrix
will not be too dense to affect the calculation efficien
cy. In addition, the contact algorithm was used,
where knifeedge ( treated as rigid body) contact the
resin beads in order to transmit the restraint load to the
resin beads by contacting after fixing the blades. The
loading condition is applying displacement load at the
bottom of the fiber while pulling out the fiber from
resin droplets.

Fig. 6 Finite element model of microdroplet test.

Sizing agent is the key part that affects the inter


facial connection between fibers and resin. The mate
rial properties used for numerical calculation are
shown in Table 4.
Table 4 Material properties of the fiber,
interphase and matrix.

Youngs
modulus E / GPa
CTE 10 6 /
Poissons ratio

Carbon fiber
(CCF300(J4))

Interphase

Matrix (5405)

230

3. 37

3. 45

0. 34

0. 40

0. 40

1. 5

65

65

Although the concentration and nature condition


of the sizing agent can cause variance of the surface
morphology such as roughness of carbon fiber, which
can affect the simulation results, however the simula
tion based on cohesive element model is mainly relat

181

ed to Young s modulus, coefficient of thermal ex


pansion ( CTE) and Poisson s ratio. Since the sizing
agents on carbon fiber surface will be crosslinked un
der the effect of curing agents in resin, therefore,
Zhang et al. [27] held the view that elastic modulus of
interphase is slightly lower than that of the matrix be
cause the sizing agents are always low molecular
weight additives. In this study, J4 sizing agent were
mixed with 5405 BMI resin with different propor
tions, the strength and modulus of the mixture were
found to be slowly decreased as the proportion of siz
ing agent increased. Considering the parameters from
the previous research [27,28] , the elastic modulus of in
terphase was defined as 3. 37 GPa, which is lower
than that of the matrix (5405) .

5 Simulation results

5. 1 Calculation of interphase properties


In microdroplet test, the calculation of interfa
cial shear strength was based on the assumption that
interface shear stress distribution is homogeneous a
long the interface, but actually it s not [29] . Nowa
days the stress distribution in interface can t be tested
by experiments, so the numerical calculation method
was chosen in this study. In ABAQUS FEM, cohe
sive elements are used to simulate the interphase, in
which the initial damage is defined by maximum
stress criteria and the damage evolution is defined by
energy criteria. The cohesive strength precisely re
flects the interphase strength of the composite when
the fiber / matrix debonding load is close to the maxi
mum load in experiment. According to the twodi
mensional ( axisymmetric) and threedimensional co
hesive element modeling under the definition of the
thickness direction, the fiber / matrix debonding
strength is nearly equivalent to the average maximum
load 0. 065 N ( the bead length is 42 m, measured
data of IFSS is 70. 81 MPa ) when the cohesive
strength is 77 MPa. The loaddisplacement curve is
shown in Fig. 7 ( the cohesive strength is 77 MPa, the
simulated maximum load is 0. 063 N) , and Fig. 8
shows Mises stress distribution after debonding
process when the maximum load is 0. 063 N. Based
on the typical mechanical model of bilinear forcedis
placement relationship, the maximum debonding load
determined by numerical simulation is nearly equiva
lent to the maximum load measured in test with only
3. 1% error.

182

Fig. 7 FEM formed loaddisplacement curve of debonding process.

Fig. 8 Interfacial debonding vonMises stress distribution.

29

In the process of debonding, the distribution of


axial stress and interfacial shear strength along the fi
ber axial under different loads are plotted in Fig. 9.
Fig. 9a shows a significant stress concentration where
it is close to the blades. The stress concentration in
creases with the increase of load, which may cause fi
ber tensile failure in microdroplet tests. In Fig. 9b,
the interfacial shear stress is found to be unevenly dis
tributed along the region near the blades under a small
load. With the load increasing, the interfacial shear
stress gets gradually close to the interfacial strength
while the interface debonding process begins. When
the process is completed, the interfacial shear stress
will be redistributed.
The damage can be judged by damage parameter
D, which is also defined as SEDG.
As the loads are enhanced, the debonding length
increases gradually. When the maximum debonding
load reaches 0. 063 N, the interface region is almost
damaged, indicating that the extended distance of the
crack in the interface is increased ( Fig. 10) . Moreo
ver, the axial distance, where SDEG reaches the vir
gin state, represents the location of crack tip, while
the stress reached 77 MPa ( the interface stress
strength) and the interfacial shear stress in debonding
region diminishes ( Fig. 9b) .

Fig. 9 Stress distribution under different loads: ( a) axial stress distribution of the fiber core and
( b) interfacial shear stress distribution along the interface.

5. 2 Interfacial shear stress distribution with dif


ferent knifeedge positions
Numerical method was implied to simulate the
effect of knifeedge position on the interfacial shear
stress distribution [22] . When the fiber embedded
length was 42 m and other conditions were un
changed, the knifeedge position was varied from 1. 5
to 3. 5 and 7 m away from the interface, and then
the FEM was established as shown in Fig. 11.
Fig. 10 The damage variable D impacted by different loads.

HAO Jianwei et al: Experiments and finite element simulation of interfacial

Fig. 11 The FEM of the matrix


when the knifeedge position is varied.

The axial displacement load was applied as the


knifeedge position was varied. When the strength

183

load is 0. 03 N, the axial stress distribution is shown


in Fig. 12a. When the distance between knifeedge
and the fiber rises, the stress concentration of fiber is
reduced as the stress increased, where the position is
away from the point of load. In Fig. 12b, the axial
distance is found to increase when the knifeedge po
sition is far from the fiber. On the contrary, with the
increase of the distance between the fiber and knife
edge, a decrease of the maximum debonding load in
microdroplet test is found. When distance of the
knifeedge away from the interface was 1. 5, 3. 5 and
7 m, the maximum debonding load was decreased to
0. 063, 0. 060 and 0. 055 N, respectively. This is be
cause when the knifeedge gets farer from the fiber,
the influence of the knife on the stress distribution is
declined, the stress concentration is reduced and the
load transfers more effectively between fiber and
resin.

Fig. 12 Effect of knifeedge position on the interfacial shear stress distribution: ( a) axial stress distribution of fiber core and
( b) interfacial shear stress distribution along the interface.

5. 3 Interfacial shear stress distribution with dif


ferent thermal residual stresses
The composite curing process produces residual
stresses, which affects the mechanical properties of
the composite [30] . This study only considered the
thermal residual stresses ( TRS) formed in the cooling
process after the composite was fully cured, and as
sumed that the CTE of the material components were
constant during the cooling process. The material
properties are shown in Table 4, while the CTE of
carbon fiber is negative.
The IFSS was simulated when composites were
cooled to room temperature ( 25 ) from 140 ,
100 and 65 . The distribution of TRS along inter
face is shown in Fig. 13, it can be seen that TRS of
the interface increases as the initial cooling tempera
ture increases.
As seen from Fig. 14, the maximum load de
creases when the initial temperature increases ( with
out regarding to TRS under 25 ) , this is caused by
a mismatch in CTE between fiber and matrix. The

value of CTE used in simulation refers to the CTE of


the T300 carbon fiber.

Fig. 13 Effect of cooling temperature on the thermal residual stress.

5. 4 Calculation of interphase properties in hygro


thermal environment
Assuming the stiffness of interphase of material is
similar to that of matrix in hygrothermal environment
and only the strength of interphase will be changed in
hygrothermal environment, so the maximum load as a
function of displacement was obtained by simulating

184

different cohesive strengths ( Fig. 15 ) . The maxi


mum load is 0. 052 N when the cohesive strength is
60 MPa, which is close to the experimental load under
condition of 7 1 W while IFSS is simulated to be
56. 3 MPa, which is equivalent to the IFSS experi
mental value of 54. 5 MPa with only 3. 3% error.

29

saturated. The debonding process was simulated as


the cohesive strength is 60 MPa, the axial stress of fi
ber core and interfacial shear stress distribution along
the interface are shown in Fig. 16. Although by FEM
simulation, we can not fully explain the mechanism
of IFSS change under hygrothermal environment, we
can however better understand the causation of the
distribution of IFSS by simulation in order to help ac
quire more specific and accurate data in the real ex
periment in the future.

Fig. 14 Effect of initial


cooling temperature on debonding load curves.

So the strength of interphase is 60 MPa when the


moisture absorption of composite system gets to be

Fig. 15 Effect of cohesive strength on loaddisplacement curves.

Fig. 16 Stress distribution of debonding process in hygrothermal environment: ( a) axial stress distribution of fiber core and
( b) interfacial shear stress distribution along the interface.

6 Conclusions

The essential contributions to the interface


debonding in the monofilament composite were inves
tigated based on experiments and finite element mod
eling. Measured by microdroplet test, the average
IFSS of CCF300 / 5405BMI composite is 70. 81 MPa.
Hygrothermal aging caused the reduction of IFSS,
which was saturated when moisture absorption of the
composite in water exceeding 7 d at 71 .
Based on the Carrol s theory and the cohesive
element damage model, the FEM was set up, which
successfully provides essential parameters to numerical
analysis model in macroscopic mechanical properties
of the composite including interface phase. When the
embedded length was 42 m, the droplet height was
15 m, the fiber radius was 3. 5 m, the knifeedge

was 1. 5 m, the maximum debonding load deter


mined by numerical simulation was 0. 063 N, which is
nearly equivalent to the maximum load 0. 065 N
measured in the test with only 3. 1% error.
Finite element analysis of microdroplet test re
vealed the influences of knifeedge position, thermal
residual stress and hydrothermal environment on the
interfacial shear stress distribution. When the distance
between knifeedge and the interface was increased
from 1. 5 to 7 m, the maximum debonding stress de
creased from 0. 063 to 0. 055 N. As the initial cooling
temperature increased, the TRS of the interface in
creased and the maximum load decreased. As simula
ted cohesive strength decreased in hygrothermal envi
ronment, the maximum load decreased. In addition
when the cohesive strength was 60 MPa, the simulated
result of the IFSS was 56. 3 MPa, which was close to

HAO Jianwei et al: Experiments and finite element simulation of interfacial

the experimental IFSS of 54. 5 MPa under the condi


tion of 7 1 W with only 3. 3% error.
References

[1] Li Y M, Miranda J, Sue H J. Hygrothermal diffusion behavior


in bismaleimide resin[ J] . Polymer, 2001, 42: 77917799.

[2] Purnell P, Cain J, Itterbeeck P V, et al. Service life modelling

of fiber composites: a unified approach[ J] . Composites Science


and Technology, 2008, 68(1516) : 33303336.

[3] Jedidi J, Jacquemin F, Vautrin A. Accelerated hygrothermal cy

clical tests for carbon / epoxy laminates[ J] . Composites Part A


Applied Science and Manufacturing, 2006, 37(4) : 63645.

[4] Favre J P, Merienne M C. Characterization of fibre / resin bond

ing in composites using a pullout test[ J] . International Journal


of Adhesion and Adhesives, 1981, 1: 311316.

[5] Penn L S, Lee S M. Interpretation of the force trace for Kevlar /

epoxy single filament pullout tests[ J] . Fibre Science and Tech


nology, 1982, 17(2) : 9197.

[6] Gaur U, Miller B. Microbond method for determination of the


shear strength of a fiber / resin interface: evaluation of experimen

tal parameters[ J] . Composites Science and Technology, 1989,

34(1) : 3551.

[7] Day R J, Cauich Rodriges J V. Investigation of the microme


chanics of the microbond test[ J] . Composites Science and Tech
nology, 1998, 58(6) : 907914.

[8] Wada A, Fukuda H. Evaluation of fiber / matrix interfacial shear


ing properties by means of microbond test and finite element anal
ysis[ J]. Journal of the Japan Society for Composite Materials,

2000, 26: 5864.

[9] Mandell J F, Chen J H, McGarry F J. A microdebonding test for


in situ assessment of fibre / matrix bond strength in composite ma

terials[ J ] . International Journal of Adhesion and Adhesives,

1980, 1: 4044.

[10] Ho H, Drzal L T. Evaluation of interfacial mechanical proper

ties of fiber reinforced composites using the microindentation


method[ J] . Composites Part A: Applied Science Manufactur
ing, 1996, 27(10) : 961971.

[11] HerreraFranco P J, Rao V, Drazal L T, et al. Bond strength

measurement in composites analysis of experimental techniques


[ J] . Compos Engineering, 1992, 2(1) : 3145.

[12] Wagner H D, Gallis H E, Wiesel E. Study of the interface in


Kevlar 49epoxy composites by means of microbond and frag
mentation tests: effects of materials and testing variables [ J] .
Journal of Materials Science, 1993, 28(8) : 22382244.

[13] Ananth C R, Chandra N. Elevated temperature interfacial be

havior of MMCs: a computational study[ J] . Composites Part

A: Applied Science and Manufacturing, 1996, 27 ( 9 ) : 805

811.

[14] Pochirajua K V, Tandon G P, Paganob N J. Analyses of single


fiber pushout considering interfacial friction and adhesion [ J] .

Journal of the Mechanics and Physics of Solids, 2001, 49(10) :

2307 2338.

[15] Lin G, Geubelle P H, Sottos N R. Simulation of fiber debonding


with friction in a model composite pushout test[ J]. International
Journal of Solids and Structures, 2001, 38(4647): 85478562.

[16] Chandra N, Li H, Shet C, et al. Some issues in the application


of cohesive zone models for metalceramic interfaces[ J] . Inter

185

national Journal of Solids and Structures, 2002, 39(10) : 2827

2855.

[17] Gautier L, Mortaigne B, Bellenger V. Interface damage study


of hydrothermally aged glassfiberreinforced polyester compos

ites[ J] . Composites Science and Technology, 1999, 59(16) :

23292337.

[18] Botelho E C, Pardini L C, Crezende M. Evaluation of hygro


thermal effects on the shear properties of carall composites[ J] .

Materials Science and Engineering A, 2007, 452453: 292

301.

[19] Tsenoglou C J, Pavlidou S, Papaspyrides C D. Evaluation of


interfacial relaxation due to water absorption in fiber polymer
composites [ J ] . Composites Science Technology, 2006, 66
(15) : 28552864.

[20] Wang Y, Hahn T H. AFM characterization of the interfacial


properties of carbon fiber reinforced polymer composites subjec
ted hygrothermal treatments[ J] . Composites Science and Tech
nology, 2007, 6(1)7: 92101.

[21] Guo H, Huang Y D, Meng L H, et al. Interface property of


carbon fibers / epoxy resin composite improved by hydrogen per
oxide in supercritical water [ J] . Materials Letters, 2009, 63
(17) : 15311534.

[22] Nishikawa M, Okabe T, Hemmi K, Takeda N. Micromechani

cal modeling of the microbond test to quantify the interfacial


properties of fiberreinforced composites[ J] . International Jour
nal of Solids and Structures, 2008, 45(1415) : 40984113.

[23] Song B H, Bismarck A, Tahhan R, et al. A generalized drop


lengthheight method for determination of contact angle in drop

on fiber systems [ J ] . Journal of Colloid Interface Science,

1998, 197(1) : 6877.

[24] Carroll B J. The accurate measurement of contact angle, phase

contact areas, drop volume and laplace excess pressure in drop


onfiber systems [ J ] . Journal of Colloid Interface Science,

1976, 57: 488 495.

[25] You J H, Lutz W, Gerger H, et al. Fiber pushout study of a


copper matrix composite with an engineered interface: Experi
ments and cohesive element simulation[ J] . International Jour
nal of Solids and Structures. 2009, 46(2526) : 42774286.

[26] Ash J T, Cross W M, Svalstad D, et al. Finite element evalua

tion of the microbond test: meniscus effect, interphase region,


and vise angle[ J] . Composites Science and Technology. 2003,

63: 641651.

[27] Zhang B M, Yang Z, Sun X Y, et al. A virtual experimental


approach to estimate composites mechanical properties: Model

ing with an explicit finite element method [ J] . Computational


Materials Science. 2010, 49(3) : 645651.

[28] Luo Y F, Zhao Y, Duan Y X, et al. Surface and wettability


property analysis of CCF300 carbon fibers with different sizing

or without sizing [ J] . Materials and Design. 2011, 32 ( 2 ) :

941946.

[29] Choi N S, Park J E. Fiber / matrix interfacial shear strength

measured by a quasidisk microbond specimen[ J] . Composites


Science and Technology. 2009, 69(10) : 16151622.

[30] Sun P, Zhao Y, Luo Y, et al. Effect of temperature and cyclic


hygrothermal aging on the interlaminar shear strength of carbon
fiber / bismaleimide ( BMI) composite[ J] . Materials and De
sign. 2011, 32(89) : 4341 4347.

Das könnte Ihnen auch gefallen