NII-Electronic Library Service
Analysis
of
Time-Dependent
In-Plane
Deformation
ofLaminated
Composites
Based
on
a
Homogenization
Theory
'Iletsuya
MATSUDA',
Nobutada
OHNO'"
and
Hiroki
TANAKA"
*
Department
ofMicroSystem
Engineering,
Nagoya
University,
Furo-cho,
Chikusa-ku,
Nagoya
464-8603,
Japan
"*
Department
ofMechanical
Engineering,
Nagoya
Uniyersity,
Furo-cho,Chikusa-ku,Nagoya464-8603,Japan
In
this
work,the
homogenized
elastic-yiscoplasticbehavior
oflong
fiber-reinforced
laminates
underin-plane
loading
is
preclicted
by
taking
directly
into
accountthe
micro-scopic
structure
and
stacking
sequence
of
laminae.
[[b
this
end,
a
hernogenization
theory
of nonlineartime-dependent
compositesis
appliedto
suchlaminates,
leading
to
the
macro-scopic
rate-type
constitutive
equation
of
laminates
andthe
evolution equations ofmicro-scopic and averagestresses
in
each
lamina.
The
macroscopic
constitutive
equation
has
a
stifT}iess
tensor
and a stress relaxationfunction
whichare
evaluated
explicitly
in
terms
of
the
microscopic structure and stacking sequenceof
laminae.
Tb
verify
the
present
theory,
uniaxial
tensile
tests
areperformed
on carbonfiberfepoxy
Iaminates.
It
is
thus
shown
that
the
present
theory
is
successfu1
in
predicting
the
anisotropic
yiscoplasticityin
in-plane
ten-sionof
unidirectional
and
cross-ply
laminates
and
the
negligible
viscoplasticity exhibitedby
quasi-isetropic
laminates.
Key
}Pbrtts
:Laminate,
Fiber-reinforced
composite,
In-plane
deformation,
Viscoplasticity,
Homogenization,Anisotropy
1.
Introduction
Fiber-reinforced
laminated
compositesare
nowim-portant
engineering materials.CFRP
(carbon
fiber-rein-fOrced
plastics)
laminates
be}ong
to
this
type
ofcompos-ites.
Such
composites
consist oflaminae,
each of whichis
unidirectionally
reinforced
withlong
fibers.
Macroscopic
responses
of
laminated
composites
are
usuallypredicted
by
attainingthe
overall responses of monolayers andthen
by
averaging
thern
in
accordance withthe
stackingse-quence
of
laminae,
and
appropriate
methods such asthe
Mori-Tanaka
theory(i}
and
the
cells
model
of
Aboudi{2)
can
be
employedfor
attaining
the
overall
responses
of
monolayersC])'C6}.
These
modelsare
fairly
successfu1
but
are
based
on
approximated
fields
of microscepicstress
and
strain,
resulting
in
possible
errors
especially
wheninelastic
deformation
occurs
in
constituents.
It
is,
there-fore,
worthydeveloping
a
theory
by
which
the
inelastic
behavior
of
fiber-reinforced
}aminates
can
be
simulated
accurately.
The
present
authors constructed ahomogenization
theory
for
nonlineartime-dependent
composites withperiodic
microstructures(7)'
(g).
This
theory
is
based
on unitcell
problems,
in
which
the
so-called
rperiodicity
ofperturbed
displacement
is
utilized asits
boundary
condi-tion(9Hi2).
The
theory
deyeloped
by
the
present
authorsenables
us
to
analyze
not
only
the
macroscopic
elas-tic-viscoplastic
behavior
of
composites
but
also
the
mi-croscopic
time-dependent
distributions
of
stress
and
strain
in
unit
cells.
The
present
authors
further
showed
the
fo11owing(i3):
tfthe
microscopic
distributions
of
stress
and strain are symmetric with respect
to
the
centerof
each unit ceLl,
the
field
ofperturbed velocity satisfiesthe
point
symmetry
with
respect
to
the
cell
boundary
facet
centers; as a consequence,
semiunit
cells
can
be
taken
as
the
domain
of analysis, sothat
computationtime
canbe
reduced
significantly.
Then,
the
homogenization
theory
was rebuilt using semiunit cel]s and was employed
for
computing
elastic-viscoplastic
behavior
offiber-rein-forced
unidirectional composites.The
homogenization
theory
mentioned aboveis
ex-pected
to
accuratelypredict
the
macroscopicelas-tic-viscopiastic
behavior
offiber-reinforced
laminates,
sincethe
theory
canhave
the
merit ofdirectly
bringing
the
microscopicstructure
of
laminae
into
predicting
the
macroscopic
response
of
laminates.
The
theory
canhave
another
meritof
enabling
us
to
analyze
the
microscopic
distributions
of stress and strainin
eachlamina.
These
-The Japan Society of Mechanical Engineers
TheJapanSociety of Mechanical Engineers
merits cannot
be
available
ifother
theories
areemployed.
It
is,
therefore,
of significanceto
applythe
homogeniza-tion
theory
to
simulating
the
time-dependent
nonlinearbehavioroffiber-reinforcedlaminates.
In
this
work,the
homogenization
theory
developed
by
the
present
authors
is
usedfor
deriving
a
macroscopic
constitutive
equation
applicable
to
the
in-plane
elas-tic-viscoplastic
behavior
of
long
fiber-reinfbrced
Iami-nates.
The
macroscopic
equationis
ofthe
rate-typeand
is
characterized
by
a
stiffhess
tensor
anda
stress
re]axation
function
which are evaluatedin
terms
ofthe microscopicstructure
and
stacking sequenceofIaminae.
To
verifythe
present
theory,
uniaxialtensile
tests
are
performed
oncarbon
fiberlepoxy
laminates.
It
is
thus
shown
that
the
present
theory
is
successfu1in
predicting
the
anisotropic yiscoplasticityin
in-plane
tension
of
unidirectional and cross-plylaminates
andthe
neg]igibleyiscoplasticity
ex-hibited
by
quasi-isotropic
laminates.
2.
Theory
2.1
Basicassumptions
Let
us
consider
aIaminate
in
which
long
fiber-rein-forced
laminae
are
stacked
symmetrically(Fig.
1).
Let
usassume
that
the
fibers
are
arranged
unidirectionally andperiodically
in
eachlamina,
and
that
the
fibers
deform
elastically
whilethe
matrix
exhibitelasticity
andvis-coplasticity.Let
N
andf(")
bethenumberof]aminae
andthe
voiume
fra6tion
oftheath
lamina,
respectively.It
is
then
convenient
to
employthree
kinds
ofCartesian
coordinates,i.e,,
X,
(i=1,2,3)
fbr
the
]aminate,
x,(a'
(i
=1,
2,
3)
for
the
ath
lamina,
and
yla'
(i
--
1,
2,
3)
for
the
unit cellin
the
athlamina,
Y(a),
as
shownin
Fig.
1,
The
X,
axisis
taken
in
the
stacking
direction.
The
xlf)-axisis
parallel
to
the
X,-axis
andthus
di-rected
perpendicularly
to
the
Iateral
surfaceof
the
ath
lamina.
The
x;a}-axisis
taken
in
the
fiber
direction
of
the
ath
lamina
and makesan
angle
e(a)
withthe
X,
axis,
The
yla)-axis
is
parallel
to
the
xfa)-axisbut
is
employed
solelyfor
the
unitcell
Y{cr)
.
Let
us
assume
further
that
the
laminate,
whichis
in-finitely
large
in
the
X,-and
X,
-directions,
is
subject
to
in-plane
Loading,
giying
riseto
nobending
because
ofthe
symmetry
in
stacking.
Then,
the
macroscopicstress
in
the
laminate,
X.,and
the
overall
stresses
in
laminae,
.T,(,a),
satisfy
,v
X,
-Zf("),..El:cr),
(1)
a.1
2I12 =4,=E,, =O,
e)
.xl,a)=.,xlf'=.xl.cr,)=o, cr=1,2,...,N,
(3)
where .()stands
for
the
components with respectto
the
X,
coordinatesystem.
Moreover,
themacroscopic
strain
in
the
laminate,
E,i,
and
the
oyeralistrains
in
laminae,
E,(,a)
, area]lowed
to
satisfy
E:]
=xE-(f)
Ca)
.
.E
2I
-It
is
obvious
that
Eq
the
xia)
coordinatesystem
.x{,") =.Egf}
(a}
rE2i
-.
E,,
=.Eff)
,E,,
=.EJ(]a)
,a-I,
2,...,N,
(4)
.4,E,,=Zf(a).E;:),
(5)
a=1E2i=En
==O,(6)
.Elf' =O, a=1,
2,
...,N.
(7)
s.(3)
and
(7)
also
hold
with respectto
-.2]g,a)=o, a-1,2,...,N,
(s)
.El:'=O, a=1,2,
...,
?V
.
(9)
X3
li1,
xte
(a)
Laminate
Fig.
1
!
l
!llXJ
(b)
Lamina
Structure
ofalaminate
andthree
kinds
o(c)
Unit
Cell
NII-Electronic Library Service
2.2
Homogenizationinlaminae
Let
usdenote
the
microscopicdistributions
of stress and strainin
the
unit cellY(")
ofthe
athlamina
asa;:"'(y,t)
and
aScr'(y,t),
wherey
andt
denote
y,
and
time,
respecti"ely.Then,
the
overall
stress
and
strain
in
the
athlamina,
.X,i") and .E:a}, are evaluatedby
homogenizing
crL")(y,t)and
E,ia)(y,t)
using
a
vol-ume average operhtor
<#>=yl.)
L,.,#dy`"',
ao)
where
Y(a)
indicates
the
volume ofY(").
We
assume
that
each
constituent
in
the
athtamina
has
a constitutiverelation
d},a)=cL:,)[S:,a}-13:f)],
(11)
where
(')
denotesthedifferentiation
with respecttot,
and cL? and
fl:ia)
indicate
an elastic stiffhesstensor
and
Viscoplastic
function
satisfying cL? =c:7 =c!,:) =cEZ) andlltf}=
13)1").
It
is
noted
that
cE:? andlilff)
changefrom
constituentto
constituent;
for
the
fibers,
fiEf)
yanishes.Then,
wecan
showthe
fo11owing
relations, whichsatisfy a]1
fundamental
equations
such
as
the
equilibrium equationof
stress,
the
relation
between
displacement
and strain,the
constitutiyerelation,
etc.(71
{8)i
6,1"'(y,t)=at7(y,t).,EL",(t)-i;1")(y,t),
(12)
.X',1")=<ava).E2ff'-(r,Ca'),
(l3)
whereaLZ' - cL:;
(6,,
fi,,
+z:[;a'),(i4)
r,la)=.L7(17{,a}+opf,)).
(15)
Here,
a,
denotesKronecker'sdelta,
(
),,
indicatesthe
differentiation
with respectto
y,,
and
z,1'<a)
and
ofa}
are
the
functions
to
be
determined
by
solving
boundary
value
prob]ems
L,.,cf,::zfl(,"'v;,T'dy`a'=-L,-cL:,'vfff,'dy`a',
(16)
Jl.,.,
cL:JopLVvEf'd/`a' --L,.,
qtz'af)vf.f)d}・{")
,(i7)
where vfa} signifies anyY-periodic
velocity
field
de-fined
in
Y("]
att.
2.3
In-plane
e]astic-viscoplasticconstitutive
equa-tion
of
laminates
Let
us
solve
Eq.(13)
fbr
.E'l,cr) andtransform
the
resulting
equation
in
a
matrix
form
(
,E(a))6.1 =[B(cr}]6.6
(
,iS(a))b.1+(cCa>}fi.1
.(ls)
Then,
Eqs.(8)
and(9)
allow
the
above equationto
be
reduced
to
(,E("))3.1
=[B-(a)]1.G[,.S(a)]3.i
+(C-(a)]].1,
(lg)
.Es:)
=
Bgf}..sfff'
+
Bgf)
..s;,a}
+
Blg)
..sgf)+
c;g)
,(2o)where
(
)
standsforthein-planeparts,i,e.,
(.x-'(cr)]=(.Si(,a)
..Sf,a) ..sli))T,(21)
(.E-'
`a']=(,Efa,}.Elf)
2.Ela,
}}r,(22)
and so on.
Here
(
)T
denotes
the
transpose.
Now,
let
usintroduce
further
the
in-plane
yectors of x4") and rE,(Ja)
, Le.(.S'
(a)]=(..slff).qf) ..sl:))T,
(23)
(xE=
`"')
==(yE,Cff'.Elf'
2.E:l}]T.
(24)
Then,
since
the
xf"}-
and
x;a)-axes
make
respectively
an
anglee(a)
withthe
X,-and
X,
-axes
(Fig.
1),
wehave
(.x-'
Ca)] =[p(cr)][.x=
{a)],(2s)
(,E-'
(a)]=
[e(")](.E=
(cr)],
(26)
where[PCa)]
and
[9(")]
denoterotationmatrices.
Hence,
Eqs.(25)
and
(26)
are
substituted
into
Eq.(19)
to
glve
(x.Ls'
Ca)] =[p(a}]'I[B'(a}]-
[eca}1(xEM'
{a))-
[p(a)]'i
[B-(a)]'i
{c-(a)).
(27)
Finally,
the
above equation withEq.
(4)
is
substituted
into
Eq.(1)
to
provide
a macroscopicin-plane
constitutiveequatlon
(X=]-[Ant][EL']-(Rfi],
(28)
where
-it,,
[
A']
=2
f(a)
[p(a)
]'i
[
B-(a}
]'i
[aa)
]
,
(2g)
a=1
N
[R-}
==Zf{a)
[p(al'i
[BL{a)Ii
{c'ca)]
.(3o)
a=I
Itisnotedthat
[A-]
and
{R-]
inthemacroscopicconstitutive
re1ation
(28)
depend
on
f(a),
e(a)
and
[B-(a)],which
are
evaluated
in
terrns
ofthe
stackingsequence
and
microscopic
structureoflaminae.
3.
Experimenta]Procedure
To
verifythe
theory
described
in
the
preceding
sec-tion,
uniaxialtensile
tests
at
constant
strain
rates
were
perforrned
at roomtemperature
using
coupon
specimens
cut out
fi'om
TR30t#340
carbonfiberlepoxy
plates
of300x300
mm,
manufactured
by
Mitsubishi
Rayon
Co.
Ltd,
The
plates
werelaminates
of
three
kinds,
i.e.,
unidi-rectional,
cross-ply
and
quasi-isotropic
laminates
withthe
lay-ups
of[O],,,
[O/90],,
and
[O/
±60],,,
which
had
-
619
The Japan Society of Mechanical Engineers
TheJapanSociety ofMechanical Engineers
Table
ain
gauge
Fig.2
Shapeofspecimens(unit:mm)
1
Off-axis
anglesfor
unidirectional,cross-plyand
quasi-isotropic
}am
in
ates9ooN
2
Laminate
q
(degree)
Unidirectional
Cross-pLy
Quasi-isotropic
O,
10,
20,
30,
45,
60,
90
O,
10,
20,
30,
45
O,
10,
20,
30
1.5,
2.0
and2.25
mmin
thickness,
respectively.
The
volume
fraction
offibers,
Vf
,was56
percent
in
each
ply.
The
coupon specimens with rectangularGFRP
tabs,
whichhad
the
shape shownin
Fig.
2,
wereprepared.
In
the
figure,
q
indicates
the
anglebetween
the
longitudi-na]
direction
ofspecimens
and
the
fiber
direction
of
OO
-plies.
Henceforth
op
willbe
referredto
as
the
off-axis
angle.The
values ofq
tested
in
the
present
work
are
listed
in
Table
1.
A
closed-loop servohydraulictesting
machine
with
a
Ioadlstrain
computer controllerwas
empleyed
for
the
tests.
The
longitudina]
strain
ofeach
specimen
wasdetected
by
two
straingauges
adhered
on
both
its
sides and was controlledto
increase
at acon-stant
rate.
4.
MaterialAssumptions
4.1
Fiberarrangement
The
arrangementof
carbon
fibers,
which
was
unidi-rectional
in
the
xla)direction,
was medeledto
be
hex-agonally
periodic
on
the
xl"}-x:a)
plane
in
eachlamina,
asillustrated
in
Fig.
1.
This
is
an
idealization
ofthe
ran-dom
distribution
offibers
onthat
plane
in
the
specimens
but
canbejustified
as
follows:
The
hexagonal
periodicity
of carbon
fiber
arTangementgives
riseto
the
transverse
quasi-isotropy
not onlyin
elasticitybut
alsoin
elastovis-coplasticity(7}'
(8).
Such
isotropy
matches withthe
trans-verse
isotropy
brought
about
by
the
randorn
distribution
of
fibers
on
the
xf"}-xS"}plane.
The
hexagonal
perio-dicity
therefore
does
workas
far
as
the
overall
response
ofeach
lamina
is
concerned.According
to
the
above assumption,the
unit
cell
Y{a)
was chosento
be
hexagonal
fbr
all
laminae,
as
shown
in
Fig.
3;
P';
wastaken
to
be
56
percent.
Here
it
is
notedthat
Y(a}
wastaken
to
be
2D
rather
than
3D
since eachlamina
wasassumed
to
have
no microscopic variationin
the
fiber
direction(S)'(i4}.
Since
the
hexagonal
unit
cell
has
the
point
symmetry with respectto
the
cellcenter,
it
is
suMcient
to
consider
half
of
the
unit
cell
as
the
domain
ofanalysis
for
solving
Eq.(16)
and
(17)('3).
Hence,
the
upperhalf
ofthe
unit cel] was considered andwas
divided
into
finite
elements usingfour-noded
isoparametric
elements,
as
depicted
in
Fig,
3,
4.2
Microscopic
constitutiveequations
The
carbon
fibers
were
regarded
as
a
transversely
isotropic
elastic
material.
Consequently,
the
fibers
were
assumed
to
have
five
independent
elasticconstants,
i.e,,
two
Ybung's
moduli
Ef,
and
E.t,,
two
Poisson's
ratios
vf,2 and v,f,,,
and
on'e
shear
rigidity
Gf]i,
where
the
subscripts
1,
2
and3
signifythe
yfa),
y;a)
andy:")
directions,
respectively.Table
2
showsthe
five
constants offibers
employedin
the
present
work;Ef,
waspro-vided
by
the
manufacturer,Mitsubishi
Rayon
Co.
Ltd,,
andthe
others weredetermined
by
referring
to
Kriz
and
Fig.
3Unit
cell
andfinite
element meshTable
2Material
parameters
Carbon
fiberEn
=1.55xlO`Ef3
=:2.40x
lo5
=2.47 ×1o4
G
f]I
vf,, =O.49
vf,,
=O.28
Epoxy
E.
=3.5xl03
v:. =O.35S,"
=10'5 n=35g(gP)=141.s(E-p)Ot6S+1o
NII-Electronic Library Service
150
1OOI・E,gi
50
O
O.Ol
O.02
E]3
Fig.
4
Identification
ofviscoplasticproperties
ofepoxy
matrlx
StinchcombCiS}.
The
epoxymatrix,
on
the
otherhand,
was regardedas
an
elastic-viscoplastic
materia] characterized asEL,
-
1
'E,r,m
o・,,-
Zn.'
a.a,
+iE,p
[g(a.i,)]
"
iit
,(3
1)
where
E.,
v. and n are materialconstants,
g(iP)
is
a
materiat
function
depending
on equivalentplastic
strainiP,
S,P
is
a
reference
strain rate, s,indicates
the
deyiatoric
part
of
cr,i,
and
a, =[(3f2)s,s,]"i.The
material constantsand
material
function
in
the
aboveequation
weredeterrnined
by
simulating
the
450
offLaxis
tensile
tests
ofthe
unidirectionallaminate
atE,,
==10'S,10'5
and
10'7
s':
(Fig,
4).
This
was be-causethe
effect of matrixviscoplasticity
was expectedto
be
significant
in
such effLaxistests.
The
constants
andfunction
ofthe
matrix
werethus
obtainedand
are
listed
in
Table
2.
It
is
notedthat
Eq.(31),
whichis
based
on
the
assumption of
isotropic
hardening,
can
be
yalid
if
we
consider
monotonicloading.
5.
ComparisonofExperimentsandPredictions
Uniaxial
tensile
experiments
weredone
at
Eii=
10'S
s'i
at
room
temperature
using
the
coupon
specimensof
unidirectional,
cross-ply
andquasi-isotropic
laminates.
The
experimentswere
predicted
by
meansof
the
present
theory.
The
experiments
and
the
predictions
arecompared
with each other
in
this
section.
Figure
5
deals
withthe
unidirectional
laminate
[O],,
.
As
seenfrom
the
figure,
significant
dependence
onthe
offLaxis
angle
op
wasobserved
in
the
experiments:When
q=OO,
the
tensile
behavior
was almostlinear.
However,
a]ittle
deviation
of
q
from
Oe
causedno-ticeable
nonlinearity
withdrastic
decrease
ofviscoplastic
flow
stress,
whereas
increase
ofq
beyond
450
gave
negligible
influence.
This
significant
dependence
on
rp
is
predicted
very well usingthe
present
theory,
as shownin
the
figure.
We,
therefore,
can
say
that
the
present
the-ory
is
successfu1in
describing
the
anisotropyin
long
fi-ber-reinforced
unidirectional
composites.Let
us emphasizethat
for
the
seven
tests
withdiffer-ent
values ofq
done
for
the
unidirectional
laminate,
the
predictions
have
excellent
agreement withthe
experi-ments,
as
discussed
in
the
aboye.
The
unidirectionallaminate
is
in
effbcta
larnina,
so
that
only
its
microstruc-ture
is
important.
Hence,
the
excetlent agreementbetween
predictions
and
experiments
meansthe
following:
The
hexagonal
unit cellintroduced
for
idealizing
the
micro-structure
in
each
lamina
does
accurately work asfar
as
the
overall
response
of
a
lamina
is
concerned.Then,
for
any
iaminate
consisting
of
suchlaminae,
we can expectthat
the
present
theory
is
successfu1
in
predicting
its
in-plane
deformation
behaviot
Figures
6
and7
showthe
predictions
and
experi-ments
done
for
the
cross-ply
and
quasi-isotropic
lami-nates,
[O190],,
and[O/
±60],,,
respectively,
It
is
seenthat
the
predictions
agree well withthe
experiments,
as
expected
in
the
above.
Then,
it
is
allowedto
saythat
the
homogenized
in-plane
defomiation
behavior
oflong
fi-ber-reinforced
laminates
can
be
accurately simulatedby
stacking
the
Iaminae
withtheir
microstructures
repre-sented
by
the
hexagonal
unit cells.Let
us
compare
the
stress versus strain relationsof
the
three
kinds
of
laminates.
The
cross-plylaminate
ex-hibited
nearly
the
same
anisotropy asthe
unidirectional one underin-plane
uniaxial
loading
withep
rangingfrom
OO
to
450
(Figs,5and
6).
This
suggeststhat
the
OO-
and
90e-plies
in
the
cross-plylaminate
interact
weakly witri eachother.
The
quasi-isotropic
laminate,
onthe
otherhand,
had
isotropic
and
almost
linear
behavior
r.Eu(li
4
3
2
o
Fig.
5
O.Ol
E33
Relationsofmacro-stress
4,
E,,
at constant strain rateunidirectional
laminate
O.02
and
macro-strain
E,,
=10'5s'i
for
-
621
The Japan Society of Mechanical Engineers
TheJapanSociety of Mechanical Engineers
r.E,gz
500
400
300
200
100
O
O,Ol
E33
Fig.
6
ReLations
ofmacro-stress4,
4,
at constant strain ratecross-ply
laminate
O,02
and
macro-strain
E33
=10'Ss'l
for
until
fraeture
occurred around4,=600MPa,
which
was
six
times
larger
than
the
viscoplasticflow
stress
of
the
unidirectionaland
cross-ply
laminates
of
q=45e.
The
almostlinear
behavior,
i.e.,
negligible
viscoplasticity,is
a consequence ofthe
lay-up
in
[O/
±60],,,in
whichthe
Oe-
and
±600-plies
strong]yinteract
with
one
an-other
so
as
to
induce
the
quasi-isotropy
in
in-plane
de-formation.
Let
us
repeat
that
the
present
theory
does
pre-dict
the
characteristics mentionedabove,
i.e.,
the
anisot-ropic
viscoplasticity
of unidirectional andcross-ply
laminates
and
the
negligible
yiscoplasticity ofquasi-iso-tropic
Iaminates.
6.
Conclusions
In
this
work,the
homogenization
theory
ofnonlinear
time-dependent
composites
developed
by
the
present
au-thors
was utilizedfbr
predicting
the
in-plane
elas-tic-yiscoplastic
behavior
oftong
fiber-reinfbrced
lami-nates.
We
thus
derived
a macroscopic rate-typeconstitu-tive
equation
of
laminates
as well as evolution equations of microscopic andaverage
stressesin
eachlamina,
The
macroscopic constitutiveequation
was
shown
to
have
a
stiffhess
tensor
and a stress relaxationfunction
to
be
evaluated
in
terms
ofthe
microscopicstructure
and
stacking
sequence
oflaminae.
To
verifythe
present
the-ory,
in-plane
tensile
tests
wereperformed
using
coupon
specimens
ofcarbon
fiber!epoxy
laminates
at roomtem-perature.
The
arrangement of carbonfibers
in
eachlamina
was modeledusing
a
hexagonai
unit cell.The
present
theory
then
predicted
very
accurately
the
macroscopiccharacteristics
observed suchas
the
anisotropic
vis-coplasticity
of
unidirectional and cross-plylaminates
and
the
negligible
viscoplasticity ofquasi-isotropiclaminates.
artE,gi
see
600
400
200
Fig.
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9}
(10)
(11)
(12)
C13)
(14)
{15)
O
O.Ol
E33
7
Relations
ofmacro-stressX
and ]]E]i
at
constant
strain rateEii=
quasi-isotropic
laminate
O.02
macro-straln
10-5sL'
for
References
Mori,
T.
andTanaka,
K,.
Acta
Met.,
ivbl.
21
(1973),
p.
57I-574,
Aboudi,
J.,
Mechanics
ofCompositeMaterials,
Elsevier
Science
Publishers,
{1991
).
Pindera,
M.
J.
andLin,
M.
W,,
ASME
J.
Press,
VesseL
1lechnoL
Vbl.
I11
(1989).
p.
183-190.
Tszeng,
T.
C.,
J.
Compos.
Mater.,
Vbl,
28
{1994),
p.
soo-s2e.Kawai,
M.,
Masuko,
Y.,
Kawase,
irlandNegishi,
R.,
Int,
J,
Mech.
Sci.
(to
appcar).Kawahara,
K.
andTohgo,
K.,
Preprint
ofJSME,
(in
Japanese),
No.
O03-1
(2000),
p.
I45-146,
Wu,
×.
andOhno,
N.,
Int,
J,
Solids
Struct.,
Vbl,
36
(1999),
p,
4991-5012.
Ohno,
N,,
Wu,
X.
andMatsuda,
T.,
Int.
J.
Mech.
Sci.,
Vbl.
42
(2000),
p.
1519-1536.
Bensoussan.
A..
Lions,
J,-L.
andPapanicolaou,
G,
As-ymptotic
Analysis
for
Periodic
Structures,
North
-HollandPublishingCompany,(1978).
Sanchez-Palencia
E,.
Nen-Homogeneous
Media
andVibration
1'heery,
Lecture
Notes
in
Physics
127,
Springer-Verlag,(1980).
Bakhvalov,
N.
andPanasenko,
G,
Homogenisatien:
Av-eraging