See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222917459

An enhanced mesomodel for laminates based on micromechanics

Article in Composites Science and Technology · March 2002

DOI: 10.1016/S0266-3538(01)00145-2

CITATIONS READS
158 85

2 authors:

Pierre Ladeveze Gilles Lubineau

Ecole normale supérieure de Cachan King Abdullah University of Science and Technology
592 PUBLICATIONS 15,416 CITATIONS 314 PUBLICATIONS 7,463 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Ageing of polymers and composites View project

Damage characterization and modelling of thermoplastic composites View project

All content following this page was uploaded by Gilles Lubineau on 18 October 2021.

The user has requested enhancement of the downloaded file.

 1

AN ENHANCED MESOMODEL FOR LAMINATES BASED ON

MICROMECHANICS
P. Ladevèzea , G. Lubineau a

a
LMT-Cachan (E.N.S. de Cachan / Université Paris 6 / C.N.R.S.)
61 Avenue du Président Wilson / 94235 CACHAN CEDEX
Tel : 33 - 1.47.40.24.02 Fax : 33 - 1.47.40.27.85

Abstract: An enhanced version of the damage mesomodel for laminates (D.M.L.), which has been developed
for 15 years at Cachan, is introduced in the light of the numerous works - both theoretical and experimental
- done in micromechanics. The new mesomodel is fully compatible with classical micromechanics models.

Keywords: composite, laminate, meso, micro, damage.

1. Introduction mesoconstituents. Mesodamage indicators are

linked to the stiffness variation of the mesocon-
Today, the modeling and simulation of the stituents. Aside from fiber breakage, the dam-
damage mechanisms until fracture of laminated age mechanisms taken into consideration are ma-
composite structures under complex loading is trix microcracking, fiber/matrix debonding and
still a challenge. delamination. The interface is a surface entity
Damage refers to the more or less progressive which depends of the relative orientations of the
growth of microvoids or microcracks which lead to adjacent plies; it transfers displacements and nor-
macrocracks and, finally, to fracture. Concerning mal stresses from one ply to another. (0o /0o ) in-
the modeling problem, in previous works we pro- terfaces are not considered as interfaces in our
posed a mesomodel (D.M.L.) for laminates with sense.
continuous fibers (Ladevèze [ 1]; Ladevèze [ 2]). The advantage of this procedure is that dam-
This mesomodel has been identified on numerous age mechanisms which can be very complex on
materials for aeronautics and aerospace applica- the structure’s scale can, in fact, be quite simple
tions. Today, it is being reconsidered in the light on the scale of the basic constituents. Prelimi-
of the numerous works (both theoretical and ex- nary identifications of each mesoconstituent were
perimental) done in the last twenty years on the done in previous works (Ladevèze and Le Dantec
micromechanics of laminates. The aim is to build [ 3] and Allix, Bahlouli, Cluzel and Perret [ 4] for
a bridge between micro- and mesomodels of lam- the single layer; Allix and Ladevèze [ 5] for the
inates. interface).
The main concept on which this damage me- An important point is that the state of damage
somodel is based is the use of an intermediary is assumed to be uniform within the thickness of
scale related to the scale of the laminates. On each mesoconstituent (Ladevèze [ 6]). The state
this intermediary scale, which we call mesoscale, of damage is described as constant throughout the
the material is described by means of two basic thickness of the single layer, which is a parameter;
mesoconstituents: the single layer and the inter- of course, the state of damage can vary from one
face. layer to the next in the laminate.
Damage mechanics and qualitative microanal- Thus, by using models of these two mesocon-
ysis are used to describe the degradation of the stituents, it is possible to reproduce the behavior
2

of any laminate for any loading. To get a con- 2. Damage mesomodel - behavior of the el-
sistent model, damage models with delay effects ementary ply
combined with dynamic analysis are used.
The works done on the microscale are numerous The first description of the mesomodel for lam-
and quite varied (Hashin [ 7]; Hashin [ 8]; Joffe inates can be found in (Ladevèze [ 1]). Two
and Varna [ 9]; Laws and Dvorak [ 10]; Nairn main assumptions lead to such a mesomodel. The
and Hu [ 11]; Reifsnider [ 12]; Tan and Nuismer [ first one is that the behavior of any laminated
13]). A first attempt to bridge the micromechan- structure can be reconstructed starting from that
ics and mesomechanics of laminates completely of two elementary constituents: the elementary
was done in (Ladevèze and Lubineau [ 14]); in layer (Ladevèze and Le Dantec [ 3]) and the in-
fact, two micromechanics scenarios were studied: terface (Allix, Ladevèze [ 5]).
the transverse microcracking mechanism and the
fiber/matrix debonding mechanism. The meso-
model was found to be fully compatible with these
Single layer
damage micromechanisms. Micro-meso relations
have also been derived; they introduce quantities
or relations which we call ”approximately ply- Interface
material”, which are intrinsically related to the
cracked ply’s characteristics and, therefore, inde-
pendent of the characteristics of the other plies.
In this paper, we go one step further by con-
sidering the possibility of ”diffuse” delamination Figure 1. Model of a laminate
cracks which originate at the tip of the transverse
cracks.
The central and crucial point to be discussed is
still the main homogenization hypothesis which The second major assumption is that the dam-
leads to the mesomodel: for any stacking se- age state is uniform throughout the whole thick-
quence and for any loading, the laminate can be ness of the ply (but not throughout the thickness
described through two elementary constituents: of the laminated composite) (Ladevèze [ 6]).
the single layer and the interface.
2.1. The elementary ply: damage kinemat-
We prove that this ”diffuse” delamination sce-
ics on the mesoscale
nario can be described using ”ply-material” quan-
The ply is entirely defined by the following
tities. We build intrinsic reference curves express-
strain energy:
ing relations between the micro- and mesodamage
2
indicators. Similar conclusions are presented for 1−dF hσ̄11 i+ ϕ(h−σ̄11 i2+ )
ed = 2 [ E10 + E10
the damage evolution laws.
Then, the competition between matrix micro- ν0 0
ν21 ν0 0
ν32
cracking and ”diffuse” delamination is studied. −( E120 + E20
)σ̄11 σ̄22 − ( E230 + E30
)σ̄22 σ̄33
1 2
The latter scenario occurs at a critical dimension-
2
less microcracking rate. We prove that this sat- ν0 0
ν31 1−d0 hσ̄22 i+ hσ̄33 i2+
−( E130 + E30
)σ̄11 σ̄33 ] + 2 [ E20 + E30
] (1)
uration rate of matrix microcracking is approxi- 1

mately independent of the stacking sequence. All σ̄ 2 2

σ̄13 2
σ̄23
these results are consistent with the use of a me- + 1−d
2 [ G0 +
12
G013
+ G023
]
12
somodel for laminates.
Finally, our results are discussed in the light of h−σ̄22 i2+ h−σ̄33 i2+
+ 12 [ E20
+ E30
]
the results from several experiments.
where 1 is the fiber’s direction, 2 the transverse
direction and h.i+ designates the positive parts.
 3

0.5
Consequently, if σ22 ≤ 0, microcracks are closed
and no noticeable damage occurs. 0.4
The stresses considered are, in fact, effective
0.3
stresses, including thermal residual stresses in the

d
healthy layer: 0.2

¯ = K0 "(U ) + th (2) 0.1

0
Here, K0 is the initial stiffness, "(U ) the me- 0 0.2 0.4 0.6
0.5
0.8
0.5
1 1.2 1.4

chanical loading and th the thermal residual

Y (MPa )

stress. The three damage indicators, which are

constant through the thickness, pertain to the fol-
lowing mechanisms: Figure 2. Damage evolution law on the mesoscale
• fiber breakage: dF .

• matrix microcracking and deterioration of

the fiber-matrix bonds: d,d’. Scenario 1 Scenario 2

2.2. The elementary ply: damage forces

and evolution laws
The damage evolution laws are based on the
damage forces:
Scenario 3
YF = − ∂E
∂dF , Yd
D
= − ∂E
∂d , Yd
D
0 = − ∂E D
∂d0

R H
(3)
ED = 2
−H
ed dz
2

qVery simple damage evolution laws (d0 ↔

Figure 3. Matrix microcracking, local delamina-
Y¯t0 ) and (d ↔
p
Y¯t ) have been identified for
tion and diffuse damage
a wide range of materials, where:

Y¯t = sup[Yd + bYd0 ] Y¯t0 = sup[Yd0 + b0 Yd ]

τ ≤t τ ≤t

b and b’ are coupling terms. The shape of such shows the classical damage evolution law identi-
a typical law is shown on Fig. 2. fied experimentally for this degradation scheme.
Then, transverse matrix microcracks occur
3. Micro-meso analysis: local delamina- (Nairn [ 16]). This is Scenario 1, which has been
tion induced by matrix microcracking proved to be compatible with D.M.L. (Ladevèze
and Lubineau [ 14]). This means that one can
3.1. Microphenomenology build damage indicators and damage evolution
Three basic scenarios are introduced on the mi- laws which are only ply-material quantities.
croscale. However, this scenario is unable to explain the
Our interpretation of the test performed in (La- whole degradation (d=1) of the ply. In fact, ex-
gattu, Lafarie-Frénot [ 15]) is that Scenario 3 is perimental observations (Crossmann and Wang [
the main mechanism for [45/ − 45]s composites. 17]) show that ”local delamination” spreads out
This is a diffuse degradation state which is nat- near the tips of matrix cracks. Consequently, ma-
urally consistent with the mesomodel. Fig. 2 trix microcracking leads to Scenario 2, in which
4

local delamination is assumed to occur periodi- Having solved the problem by finite element
cally near periodic matrix microcracks. calculation, we can define equivalent mesodamage
Theoretical results concerning this phe- indicators d and d’:
nomenon can be found in (Kevin and O’Brien d0 d
[ 18], Nairn [ 19]). Here, our objective is to prove a= 0 b= 0 (5)
E2 G12
that D.M.L. is fully compatible with this second
scenario (Fig. 3). D.M.L. is healthy if the main quantities (dam-
age indicators and evolution laws) depend only on
3.2. Description of the elementary cell the characteristics of the damaged ply. We aim
to emphasize these so-called ”(S’)-(S”) indepen-
dent” quantities.
(3)
ε=(H/H’) e 4. Micro-meso relations and damage indi-
cators
H’ (S") We did numerous finite element calculations for
(1) (2) different values of the stacking sequence param-
H (S)
eters. The measures of damage a, b and c were
H’ (S’) calculated. It follows that:

D Result 1: c is negligible compared to a and b.

Indeed, the typical value of c is 10−13
(J/(m3 .M pa2 )). In comparison, the magnitude
of a and b is 100 (J/(m3 .M pa2 )). D.M.L. is con-
Figure 4. The elementary cell (Ω)
sistent with this first result.
Result 2: (τ, ρ) ↔ d0 is approximately (S’)-(S”)
independent.
(S’) and (S”) are peripheral parts of the com- Fig. 5 shows (ρ, τ ) ↔ d0 for different thick-
posite. The microdamage state in (S) is defined nesses ratios ( = H H
0 ) between cracked layers
by: and homogenized parts (S’) and (S”). For a wide
1 H range of ratios, the relation (ρ, τ ) ↔ d0 is virtu-
= : dimensionless microcracking density ally unchanged. This proves that the thickness
ρ D
e effect is negligible.
τ = : delamination density The effect of peripheral orientation is shown in
H
Fig. 6. Here, again, the result is nearly inde-
If the cell is subjected to a ”plane stress” pendent of the orientation of the ply immediately
macrostate, the proof presented in (Ladevèze and adjacent to the cracked ply.
Lubineau [ 14]) remains valid. So does the follow- Thus, the micro- and mesodamage indicators
ing property: are nearly equivalent. Indicators d and d’ are
Property: the cracked ply can be homogenized viable measures of the degradation within the ply.
and its strain energy can be written as:
5. Micro-meso relations: damage evolu-
1 tion laws
ed = e0d − [ahσ̄22 i2+ + bσ̄12
2
+ cσ̄12 σ̄22 ] (4)
2
5.1. Energy release rates for Scenarios 1
where e0d is the strain energy of the un- and 2
cracked ply under the current effective Here, the two modes of degradation (matrix
stresses hσ̄22 i+ and σ̄12 . microcracking and local delamination) are com-
 5

0.8 ∂E Ω
Meso damage indicator d’

Gm = − ∂Adf |τ =cste,¯"=cste
0.6
[0/90(8)/0], τ=0.1 (6)
[0/90/0], τ=0.1
∂E Ω
[0(8)/90/0(8)], τ=0.1
[0/90(8)/0], τ=0
Gd = − ∂Add |ρ=cste,¯"=cste
0.4
[0/90/0], τ=0
[0(8)/90/0(8)], τ=0
0.2
[0/90(8)/0], τ=0.19 where EdΩ is the strain energy of the cell (see Fig.
[0/90/0],τ=0.19
[0(8)/90/0(8)], τ=0.19 4), Af the microcracked area of the matrix and
0 Ad the delaminated area.
0 0.5 1 1.5 2
Dimensionless microcracking rate: (1/ρ).
Using Eq. 6 and Eq. 4, we have:

Gm = − ∂HDe
∂Am |τ =cste,¯
d
"=cste
(7)
Figure 5. ρ ↔ d0 (τ = cste) curve for [0n /90m ] Gd = − ∂HDe
∂Ad |ρ=cste,¯
d
"=cste
T300-914 laminates. The relation (ρ, τ ) ↔ d0
appears to be nearly independent of the rela- For example, if only hσ̄22 i+ is nonzero (meaning
tive thickness of the damaged ply and peripheral that the microcracks are open and there is no
parts. effective shear stress in the central ply), we get
(since Ad = 4τ H):

H ∂a
Gm = 2 < σ̄22 >2+ 1 |τ =cste
∂ρ

H
= 2 < σ̄22 >2+ Gm | H <σ̄22 >2+ =1
1 2

0.9
(8)
H ρ ∂a
Gd = < σ̄22 >2+ 4 ∂τ |ρ=cste
Meso damage indicator d’

0.8
2
0.7

H
0.6 = 2 < σ̄22 >2+ Gd | H <σ̄22 >2+ =1
2
0.5
[45/45/0/90] s
0.4 [45/45/30/90] s
[45/45/60/90] s Degradation starts increasing as soon as Gm
0.3

0.2
and Gd reach associated critical material val-
0.1
ues. For T300-914, Gmc is assumed to be equal
0
1/ρ=cst=1.2
to 280J/m2 . For the numerical example, Gdc
0 0.05 0.1 0.15 0.2

Dimensionless delamination rate: τ .

is assumed to be constant during the delami-
nation process and equal to the mode-I critical
energy release rate identified on the DCB; there-
fore, Gdc = 180J/m2 (Allix, Lévèque and Perret
Figure 6. τ (ρ = cste) ↔ d0 curve for [45/ − [ 20]). Of course, it is possible to refine this
45/θ/90m ] T300-914 laminates. The effect of pe- value. One can define a particular law for the
ripheral orientation is negligible. evolution of Gdc during the process (Ogihara and
Takeda [ 21]). However, such refinements are
not necessary for our study since we only want
to verify whether the phenomenon is ”(S’)-(S”)
independent”. This result does not depend on
the evolution chosen for Gdc .
peting phenomena. The energy release rates
associated with matrix microcracking (Gm ) and Therefore, to know which is the dominant phe-
local delamination (Gd ) are calculated as follows: nomenon, we have to compare, for each given
6

degradation state (ρ, τ ): 0.14

t.
Gm | H <σ̄22 >2+ =1 0.12

Dimensionless delamination rate:

λ0m = 2
(9) [0(8)/90/0(8)]

[0/90/0]
Gmc 0.1 [0/90(4)/0]

and 0.08

Gd | H <σ̄22 >2+ =1 0.06

λ0d = 2
(10)
Gdc
0.04

This competitive situation is illustrated on Fig.

0.02
7, where these two functions are plotted for previ-
ous values of Gdc and Gmc . We see that damage 0
0 0.2 0.4 0.6 0.8 1
increases according to the black dashed line.
r
Dimensionless microcrack density: 1/ .

0 Figure 8. Evolution of damage mode ( ρ1 ↔ τ )

0.45
λ
m
(1/(m.Mpa ))
2

0.4

0.35

0.3

Damage evolution
0.25
m

5.2. Intrinsic properties of dimensionless

0
λ

0.2
and

0.15 0 energy release rates

λ
0.1 d
We have already proved that (ρ, τ ) ↔
d
0
λ

0.05
0.2 Gm | H <σ̄22 >2+ =1 takes the form of a ply-material
0
0.15
2
0
0.5
0.1 relation (Ladevèze and Lubineau [ 14]). This al-
Dimen
1 0.05
le
ss : t.
micro
sionle
ss
1.5
2 0
en
si
on
tio
n
ra
te
lowed us to build a governing mesodamage curve
cracki im a
ng de D in
ns
r
ity: 1\
.
de
la
m for Scenario 1.
Now we extend this conclusion to Gd :

Result 3: (ρ, τ ) ↔ Gd | H <σ̄22 >2+ =1 is a ply-

2
Figure 7. Surfaces (ρ, τ ) ↔ Gom and (ρ, τ ) ↔ God material relation.
for T300-914
Indeed, τ ↔ Gd | H <σ̄22 >2+ =1 is given on Fig. 9
2
for different stacking sequences (two values of ρ
are used: ρ = 1.04 and ρ = 0.56). The influ-
Consequently, one can plot the relation ρ ↔ τ ence of the thickness is virtually negligible even
(Fig. 8). Therefore, the relative evolution of the for thick central plies (small influence for  ≤ 4,
dimensionless microdamage indicators takes the see Fig. 4). For very thick plies ( = 8), the
form of a ply-material relation. results are not as good but are still acceptable.
One can observe that, initially, only matrix Here, again, peripheral orientation effects were
microcracking occurs. At a critical value ρc of studied by examining several [θ/−θ/ϕ/90]s lami-
ρ, diffuse delamination becomes the dominant nates. The central 90 ply is cracked and ”diffuse”
phenomenon while matrix microcracking ceases. delamination is assumed to occur at (ϕ/90) inter-
Here, a very important point is that the satura- faces. Thus, the effects of the peripheral stacking
tion density ρc takes the form of a ply-material sequence on the behavior of the central cracked
quantity. Thus, the global degradation scheme ply can be studied.
does not depend on the stacking sequence. In fact, once the layers immediately adjacent
 7

15

Gd (for H/2<σ 22 >+2 =1m.Mpa)

20
s22>=1m.Mpa)

10
15
<
H

[45/-45/0/90] s
2

10
G for (

[0/90(8)/0], r
1/ =0.56
[60/-60/0/90] s
r [30/-30/0/90] s
d

[0/90(4)/0], 1/ =0.56 5
[0/90]s
5
[0/90/0], r
1/ =0.56

[0(8)/90/0(8)], r
1/ =0.56

[0/90(8)/0], r
1/ =1.04

[0/90(4)/0], r
1/ =1.04 1/ρ=1,2
0
0 0.05 0.1 [0/90/0], 1/r=1.04
0.15
0
τ.
[0(8)/90/(8)], r
1/ =1.04 0 0.05 0.1 0.15 0.2

Dimensionless delamination rate: t. Dimensionless delamination rate: τ

Figure 9. τ ↔ Gd | H <σ̄22 >2+ =1 curve for differ- Figure 10. τ ↔ Gd | H <σ̄22 >2+ =1 curve for different
2
2
ent thickness ratios. Only the central 90 ply is [θ/ − θ/0/90]s laminates. The central 90 ply is
cracked and diffuse delamination is assumed to microcracked and ”diffuse” delamination occurs
spread along the (0/90) interfaces. at (0/90) interfaces.

15

to the cracked ply are defined, one can observe

Gd (for H/2<σ 22 >+2 =1m.Mpa)

that the remaining external plies have virtually

no influence (see Fig. 10, 11, 12). This is true 10

for every orientation considered and in agreement

with Zhang’s observations. Using a simplified an-
[45/-45/30/90]s
alytical approach based on plate theory, he found 5 [60/-60/30/90]s
[30/-30/30/90]s
effects of peripheral orientations of the same or- [30/90] s

der of magnitude as in our study (Zhang, Fan and 1/ρ=1,2

Herrmann [ 22]). 0
0 0.05 0.1 0.15 0.2
Moreover, the influence of these immediately Dimensionless delamination rate: τ
adjacent layers is quite small (approximately 8
%).

5.3. Mesodamage evolution laws Figure 11. τ ↔ Gd | H <σ̄22 >2+ =1 curve for different
2
Consequently, one can build a governing [θ/ − θ/30/90]s laminates. The central 90 ply is
mesocurve which completes the mesobehavior microcracked and ”diffuse” delamination occurs
arising from Scenarios 1 and 3 (Ladevèze and Lu- at (0/90) interfaces.
bineau [ √ 14])). A typical mesocurve is a relation
between Y¯0 and d’.
The damage force Y¯0 can easily be calculated
from previous microanalysis. Indeed, by using Moreover, one can observe that when matrix
Eq. 3 and Eq. 1, for simple cases where only microcracking is the active damage mode one has:
hσ̄22 i+ is nonzero, we have:
Gmc
< σ̄22 >2+ = H
(12)
Y¯0 = − ∂He
∂d0
d
2 G |
m H <σ̄22 >2+ =1
2
(11)
H 2 ∂a where Gmc and Gm are the critical and current
= 2 hσ̄22 i+ ∂d0
8

15 1
Gd (for H/2<σ 22 >+2 =1m.Mpa)

Diffuse delamination

Meso damage indicator: d’.

0.8

Matrix microcracking

0.6

[0(8)/90/0(8)]
[45/-45/60/90]
s
5 [60/-60/60/90] s 0.4 0(4)/90/0(4)]

[30/-30/60/90] s [0/90/0]
[60/90] s [0/90(2)/0]

[0/90(4)/0]
0.2
1/ρ=1,2 [0/90(8)/0]

0
0 0.05 0.1 0.15 0.2

Dimensionless delamination rate: τ 0

0 0.01 0.02 0.03 0.04 0.05

0.5
Meso damage force Y’

Figure 12. τ ↔ Gd | H <σ̄22 >2+ =1 curve for different

2
[θ/ − θ/60/90]s laminates. The central 90 ply is
Figure 13. Fluctuation around the mesodamage
microcracked and ”diffuse” delamination occurs
curve for T300-914. Simulations of matrix mi-
at (0/90) interfaces.
crocracking and diffuse delamination on the mi-
croscale define a cloud very close to an intrinsic
reference curve. This curve defines the mesobe-
havior of the ply.
unit energy release rates corresponding to matrix
microcracking.

In the same way, when ”diffuse” delamination

Considering the high level of damage at the on-
is the active damage mode, one has:
set of local delamination, it is legitimate to con-
Gdc sider that this phenomenon leads to total deteri-
< σ̄22 >2+ = H
(13) oration by propagation of the delamination (Ogi-
2 Gd | H 2
2 <σ̄22 >+ =1
hara and Takeda [ 23]). Their experimental re-
In the presence of matrix microcracking, using sults show very well that delamination increases
Eq.(12) and Eq. (5) in Eq. (11), one has: sharply while the density of matrix microcrack-
ing remains approximately constant during this
Gmc step. Of course, once diffuse delamination is well-
Y¯0 = (14)
E20 Gm | H <σ̄22 >2+ =1 developed, some matrix microcracking can occur
2
again, but this is not significant considering the
Similarly, in the presence of ”diffuse” delami- high level of damage within the material.
nation, one has: Therefore, one can derive from these results a
ply-material curve (Fig. 14) which fits the results
Gdc of Fig. 13. Then, delamination spreads until√ d0 =
Y¯0 = 0 (15)
E2 Gd | H <σ̄22 >2+ =1 1. A similar curve can be obtained for d ↔ Ȳ .
2

Fig. 13 shows the fluctuation

√ around a refer- 6. Comparison of model and experimental
ence mesocurve d0 ↔ Y¯0 for several [0n /90m ]s data
laminates. The first part is associated with Sce-
nario 1 and follows the curve given in the pre- So far, we proved that delamination induced by
vious companion paper. When ρ = ρc , damage matrix microcracking is consistent with D.M.L.
increases rapidly by propagation of local delami- In particular, ρc is a ply-material quantity and
nation. (ρ, τ ) ↔ Gd | H <σ̄22 >2+ =1 takes the form of a ply-
2
 9

1
of the stress respectively.
"Diffuse" Delamination

0.8
Consequently, one can calculate:
Damage indicator: d’.

0.6

Matrix Microcracking
Gd | H <σ̄22 >2+ =1 (ρc , δτ ) = ζχ (18)
2
0.4
y
Pl

with
in
Th

Diffuse
0.2 Damage 1
χ= T )2
(19)
k Ply H(σm + σ22
Thic

0
0 0.01 0.02 0.03 0.04 Therefore, in order to check whether
Meso damage force Y’0.5 Gd | H <σ̄22 >2+ =1 (ρc , δτ ) is indeed a ply-material
2
quantity, we have to look at Indicator χ, which
should be the same for any laminated structure.
Figure 14. Resulting mesodamage curve. Scenar-
ios 1, 2 and 3. Each scenario can be described on Using these experimental results, this indicator
the mesoscale by a an intrinsic reference curve. can be easily plotted for several [+25/ − 25/90n ]s
Due to the choice of the damage force Y 0 , Sce- laminates. We get the following variation of χ
nario 3 depends on the ply’s thickness, whereas with respect to the laminated structure (see Fig.
Scenarios 1 and 2 do not. 15).

0.14

material relation. 0.12

Now, let us first examine these conclusions in 0.1

Indicator χ

the light of Crossmann and Wang’s experimental 0.08

studies (Crossmann and Wang [ 17]).

0.06
For different stacking sequences, they measured
the applied strain at the initiation of matrix mi- 0.04

crocracking and at the initiation of delamination 0.02

(¯"i ). They also described the delamination shape 0

in order to determine for each specimen which 3 4 6 8

Half number of 90˚ plies : n.

was the dominant phenomenon.
At the initiation of delamination, one can write:

H
Gdc = < σ̄22 >2+ Gd | H <σ̄22 >2+ =1 (ρc , δτ ) (16) Figure 15. Evolution of Indicator χ (related to
2 2
the unit energy release rate for diffuse delamina-
Then, the effective stress < σ̄22 >+ is directly tion) with respect to the laminated structure for
related to the applied mechanical strain ¯"i and [+25/ − 25/90n ]s
T
to the residual thermal stress σ22 . Therefore, we
have:
T 2
H(σm + σ22 ) Gd | H <σ̄22 >2+ =1 (ρc , δτ ) = ζ (17) Therefore, these results are in line with our con-
2
clusion, i.e. Gd does not depend on the char-
where ζ is a constant which depends on the acteristics of (S’) and (S”). Only results with
material (but not on the structure) and σm and n ≥ 3 are represented here. Indeed, for thin-
T
σ22 are the mechanical part and the thermal part ner central plies, another phenomenon should be
10

taken into consideration. For such configurations, We have proved that the mesodamage evolu-
as mentioned by Crossmann, delamination begins tion law is nearly independent of the character-
on the edge of the laminates. Therefore, the phe- istics (orientations and thicknesses ratio) of the
nomenon studied here (diffuse induced delamina- stacking sequence.
tion) is no longer dominant. Consequently, the mesomodel is able to take
Next, let us consider the critical value ρc . Table into account the main common microdegrada-
1 gives experimental results concerning the micro- tions.
cracking rate at saturation for T300-914 (Henaff- An important issue concerning micro-meso re-
Gardin, Lafarie-Frenot, Brillaud and El Mahi [ lations is to be able to link a discontinuous
24]; Lafarie-Frenot, Henaff-Gardin and Urwald (micro-) scheme of degradation to a continuous
[ 25]; Henaff-Gardin, Goupillaud and Lafarie- one (evolution of mesodamage). This problem
Frenot [ 26]). can be solved by introducing matrix tenacity as
a material uncertainty. In this process, the per-
fectly periodic microscenario pattern is broken
and Scenario 1, which is initially discontinuous,
Laminate Saturation rate becomes nearly continuous.
[03 /90/04 ]s 1,78 Edge effects, which have not been considered
[07 /90]s 1,82 here, may nevertheless play a primary role by
[02 /90/0]s 1,78 ”initiating delamination” (Kevin and O’Brien [
[03 /90]s 1,60 18]; Crossman and Wang [ 17]). Areas with high
[03 /901,5 ]s 1,62 gradients exist in the vicinity of each edge, and it
[03 /903 ]s 1,42 would be interesting to examine the potential of
Table 1 mesomodeling in such areas.
Saturation microcracking value in 90o ply (ρc ) for As mentioned above, further studies will now
several laminates in traction take into consideration the influence of edge ef-
fects on the initiation and propagation of degra-
dation. The possibility of ply homogenization in
the presence of high gradients will be discussed.
The results of our study agree quite well with
these experimental results, which are nearly con- REFERENCES
stant. 1. P. Ladevèze, Sur la mécanique de
l’endommagement des composites, in:
7. Further comments and conclusion C.Bathias, D.Menkès (Eds.), Comptes-
rendus des JNC5, Pluralis Publication, Paris,
Here, degradation of the laminate follows three 1986, pp. 667–683.
2. P. Ladevèze, Modelling and computation un-
steps: diffuse damage, (coalescence) matrix mi- til final fracture of laminate composites, in:
crocracking and induced diffuse delamination. A. Cardon, al. (Eds.), Recent Developments
We have proved that the mesomodel is able to in Durability Analysis of Composite Systems,
model each of these degradation schemes quite Balkema, 2000, pp. 39–47.
correctly. We have proved the existence of ply- 3. P. Ladevèze, E. L. Dantec, Damage model-
ing of the elementary ply for laminated com-
material relations between the micro- and meso- posites, Composite Science and Technology
damage indicators and for the damage evolu- 43 (3) (1992) 257–267.
tion laws. The competition between degrada- 4. O. Allix, N. Bahlouli, C. Cluzel, L. Perret,
tion modes verifies similar ply-material relations Modelling and identification of temperature-
since, for example, the rate of matrix microcrack- dependent mechanical behaviour of the el-
ing at saturation (i.e. the microcracking rate at ementary ply in carbon/epoxy laminates,
Composite Science and Technology 56 (1996)
which ”diffuse” delamination becomes the domi- 883–888.
nant phenomenon) is also intrinsic. 5. O. Allix, P. Ladevèze, Interlaminar interface
 11

modelling for the prediction of laminate de- racks in laminated composites., International
lamination, Composite Structures 22 (1992) Journal of Fracture 57 (1992) 1–24.
235–242. 20. O. Allix, D. Leveque, L. Perret, Interlaminar
6. P. Ladevèze, About a damage mechanics ap- interface model identification and forecast of
proach, in: D. Baptiste (Ed.), Mechanics and delamination in composite laminates, Com-
Mechanisms of Damage in Composite and posite Science and Technology 56 (1998) 671–
Multimaterials, MEP, 1989, pp. 119–142. 678.
7. Z. Hashin, Analysis of cracked laminates: a 21. S. Ogihara, N. Takeda, Interaction between
variational approach, Mechanical of Materials transverse cracks and delamination during
4 (1985) 121–136. damage progress in cf rp cross-ply laminates,
8. Z. Hashin, Analysis of stiffness reduction International Journal of Solids and Structures
of cracked cross-ply laminates, Engineering 36 (1999) 813–846.
Fracture Mechanisms 25 (1986) 771–778. 22. J. Zhang, J. Fan, K. Herrmann, Delam-
9. R. Joffe, J. Varna, Analytical modeling of inations induced by constrained transverse
cracking in symmetric composite laminates,
stiffness reduction in symmetric and balanced International Journal of Solids and Structures
laminates due to cracks in 90o layer, Compos- 36 (1999) 813–846.
ite Science and Technology 59 (1999) 1641– 23. S. Ogihara, N. Takeda, Interaction between
1652. transverse cracks and delamination during
10. N. Laws, G. Dvorak, Progressive transverse
cracking in composite laminates, Journal of damage process in CFRP cross-ply laminates,
Composite Materials 22 (1988) 900–916. Composite Science and Technology 54 (4)
11. J. Nairn, S. Hu, Matrix microcracking, in: (1995) 395–404.
Taljera (Ed.), Damage Mechanics of Compos- 24. C. Henaff-Gardin, M. Lafarie-Frenot, J. Bril-
ite Materials, 1994, pp. 187–243. laud, A. E. Mahi, Influence of the stacking se-
12. A. Highsmith, K. Reifsnider, Stiffness reduc- quence on fatigue transverse ply cracking in
tion mechanism in composite laminates, in: cross-ply composite laminates, in: J. Masters
K. Reifsnider (Ed.), Damage in Composite (Ed.), Damage detection in composite mate-
Materials, ASTM-STP 775, 1982, pp. 103– rials, ASTM-STP 1128, 1992, pp. 236–255.
117. 25. M. Lafarie-Frenot, C. Henaff-Gardin, E. Ur-
13. S. Tan, R. Nuismer, A theory for progres- wald, Specificity of fatigue damage behaviour
sive matrix cracking in composites laminates, in carbon/epoxy cross-ply laminates, in:
Journal of Computational Materials 23 (1989) J. Baı̂lon, J. Dickson (Eds.), Proc. of Fatigue
1029–1047. 93, IFS, Montréal, Québec, Canada, 1993, pp.
14. P. Ladevèze, G. Lubineau, On a damage me- 1355–1360.
somodel for laminates: micro-meso relations, 26. C. Henaff-Gardin, I. Goupillaud, M. Lafarie-
possibilities and limits., Composite Science Frenot, Evolution of matrix cracking in cross-
and Technology To appear. ply cfrp laminates: differences between me-
15. F. Lagattu, M. Lafarie-Frénot, Variation of chanical and thermal loadings, in: A. Cardon,
PEEK matrix crystallinity in APC-2 compos- H. Fukuda, K. Reifsnider, G. Verchery, P. A.
ite subjected to large shearing deformation, Balkema (Eds.), Proc. of 4th Int. Conf. on Re-
Composite Science and Technology 60 (2000) cent Developments in Durability Analysis of
605–612. Composite Systems (Duracosys’99), Brussels,
16. J. Nairn, Matrix microcracking in compos- Belgium, 1999, pp. 69–76.
ites, in: Taljera-Manson (Ed.), Polymer Ma-
trix Composites, Ch.13, 2000, in press.
17. F. Crossman, A. Wang, The dependence of
transverse cracking and delamination on ply
thickness in graphite/epoxy laminates, in:
K. Reifsnider (Ed.), Damage in composite
materials, ASTM-STP 775, 1982, pp. 118–
139.
18. T. Kevin, O’Brien, Analysis of local delami-
nation and their influence on composite lami-
nate behavior, in: W. Johnson (Ed.), Delam-
ination and debonding of materials, ASTM-
STP 876, 1985, pp. 282–297.
19. J. Nairn, S. Hu, The initiation and growth
of delaminations induced by matrix microc-

View publication stats