Professional Documents
Culture Documents
Structural impact tests were first presented to cover typical fibre metal laminates (FMLs) subjected a low velocity projectile
impact, which produced the corresponding load-displacement traces and deformation/failure modes for the validation of numerical models. Finite element (FE) models were then developed to simulate the impact behaviour of FMLs tested. The aluminium (alloy grade 2024-0) layer was modelled as an isotropic elasto-plastic material up to the on-set of post failure stage,
followed by shear failure and tensile failure to simulate its failure mechanisms. The glass fibre laminate (woven glass-fibre reinforced composite) layer was modelled as an orthotropic material up to its on-set of damage, followed by damage initiation
and evolution using the Hashin criterion. The damage initiation was controlled by failure tensile and compressive stresses
within the lamina plane which were primarily determined by tests. The damage evolution was controlled by tensile/compressive
fracture energies combined with both fibre and matrix. The FE models developed for the 2/1, 3/2 and 4/3 FMLs plates made
with 4-ply and 8-ply glass fibre laminate cores were validated against the corresponding experimental results. Good correlation
was obtained in terms of load-displacement traces, deformation and failure modes. The validated models were ready to be used
to undertake parametric studies to cover FMLs plates made with various stack sequences and composite cores.
damage, Hashin criterion, finite element, impact, FMLs, progressive failure
PACS: 02.70.Dc, 81.05.Ni, 46.50.1a
Introduction
www.springerlink.com
1169
1170
2 Experimental work
The fibre metal laminates were based on the 2024-O aluminium alloy sheets and the woven glass fibre prepreg
EP-124-44-40. Figure 1 shows the stacking arrangement of
a 2/1 (with 4-ply composite cores) FMLs in the picture
frame mould. One 0.5 mm thick aluminium alloy sheet, four
plies of 0.1 mm thick glass fibre prepreg sheet and another
0.5 mm thick aluminium alloy sheet were placed one by one
in the 200 mm240 mm picture frame mould. The stacking
sequence of the glass fibre-based FMLs investigated is
shown in Table 1.
Configuration
2/1
1.4
2/1
1.8
3/2
2.3
3/2
3.1
4/3
3.2
4/3
4.4
1171
(1)
pl h q, pl , ,
(2)
pl total
total
, y .
E
(3)
(4)
pl
t
0
2
pl : pl dt , ,
3
(5)
Table 2 Isotropic hardening data for the aluminium alloy 2024-0 used in
this programme
Yield stress (MPa)
Figure 3 The data recording system using a high speed video camera.
Plastic strain
76
100
0.0074
113
0.0117
122
0.02
150
0.0363
176
0.07
183
0.1
186
0.2
1172
R y ,
(6)
sh
pl
pl
f
1
ii cutoff ,
3
(8)
(7)
3.2
21 E2
31 E3
1 E2
32 E3
23 E2
1 E3
1 G12
11
0
0 22
0
0 33
,
0
0 12
1 G13
0 13
1 G23 23
0
(9)
F 11 12 , 11 0.
XT
SL
t
f
(10)
Fibre compression:
2
F 11 , 11 0.
XC
c
f
(11)
Matrix tension:
2
F 22 12 , 11 0.
YT S L
t
m
(12)
Matrix compression:
2
2
2
22 YC
12
22
F
1
,
YC S L
2 ST 2 ST
c
m
(13)
Table 3 Orthotropic elasticity data of the glass fibre laminates used in this programme
E1 (GPa)
E2 (GPa)
E3 (GPa)
v12
v13
v23
G12 (GPa)
G13 (GPa)
G23 (GPa)
23
23
0.15
0.15
0.15
1 d f E1
1
Cd
1 d f 1 dm 12 E2
D
1173
1 d 1 d
f
1 d m E2
0
21
E1
,
0
1 d s GD
0
(14)
Cd ,
(15)
and where D is the overall damage variable and can be expressed as:
D 1 1 d f
1 d
m
(16)
12 21
d tf , if 11 0,
df c
d f , if 11 0,
(17)
d t , if 22 0,
d m mc
d m , if 22 0,
(18)
d s 1 1 d tf
1 d 1 d 1 d .
c
f
t
m
c
m
(19)
The damage parameters for both the fibre and matrix are
calculated based on the critical tensile and compressive
stresses in relation to the corresponding ultimate strengths.
Damage evolution is modelled by the negative slope of
the equivalent stress-displacement relation after damage
initiation is achieved. Fracture energies for fibre tension,
fibre compression, matrix tension and matrix compression
failure modes need to be specified to indicate energies disF
with values
sipated during damage, i.e. G Fft , G Ffc , GmtF and Gmc
set to 40000, 60000, 40000 and 60000 J m2 respectively in
this study. The linear damage evolution is shown in Figure
4. The typical time for the damage process and energy
change is set to 0.008 seconds, which is to ensure the complete perforation takes place. The equivalent stress and
equivalent displacement in Figure 4 are defined based on
the plain stresses and strains on fibre and matrix respectively, in conjunction with the characteristic element length.
The reason for using equivalent displacements rather than
strains is to alleviate the mesh dependency during material
softening. The value of 0eq depends on the elastic stiffness
and strength parameters of the material. The value of feq is
defined by the strain energy, which is dissipated due to failure and corresponds to the area underneath the curve in
Figure 4.
For the purpose of modelling the interaction between
various parts of pure composites and FMLs subjected to
projectile impact, the general contact interaction was defined between the two neighbouring layers and the surface-to-surface contact interaction was defined between the
projectile surface and the node set of the target centre of
each layer. Contact interaction properties for interactions
between the projectile and the aluminium alloy layer, interactions between the projectile and the glass fibre laminates
layer, interactions between the aluminium alloy layer and
the glass fibre laminated layers, and interactions between
the glass fibre laminated layer were defined respectively and
referred to by the relevant type of interaction.
Typical mesh generations, dimensions, loading and
boundary conditions for typical FMLs (2/1) are shown in
Figure 5. The FML panel is fully fixed at its four sides, i.e.
no displacement and rotation are allowed on its four boundary side faces. Two elements are generated through one
laminate. The typical element dimension ratio in the impact
region is 1.0:1.0:0.3/0.2 (thickness), which was obtained
through mesh sensitivity studies in terms of accuracy of the
modelling output and CPU time consumption. There are
totally 7200 eight-node solid elements and 2391 four-node
rigid surface elements, and the total degrees of freedom of
the model are 201492. An initial velocity is applied to the
projectile to strike the target, with the same velocity and
mass values as those in the experimental work. The size of
an individual time step was controlled automatically based
1174
Figure 5 Mesh generations, dimensions (mm), loading and boundary conditions for 2/1 FMLs.
Figure 6 Load-displacement traces for the 2/1 FMLs plate with 4-ply
composite cores subjected to a low velocity (5 m/s) impact.
Figure 7 Load-displacement traces for the 2/1 FMLs plate with 8-ply
composite cores subjected to a low velocity (5.5 m/s) impact.
Figure 8 Failure modes of the 2/1 FMLs plates made with 4-ply composite
cores subjected to a low velocity impact. (a) The test; (b) the FE simulation.
Figure 9 Failure modes of the 2/1 FMLs plates made with 8-ply composite
cores subjected to a low velocity impact. (a) The test; (b) the FE simulation.
for these two FMLs plates were quite similar, so did the
simulated ones.
Finite element models of other types of FMLs plates
subjected to a low velocity impact were also developed to
broaden the validation. Figures 10 and 11 show the numerical simulations of the experimental load-displacement traces
for the 3/2 FMLs plates made with 4-ply and 8-ply composite cores respectively subjected to an on-set perforation impact. Very good correlation was obtained between the experimental results and the numerical simulations, in terms
of the overall initial stiffness, the peak load and the perforation process.
The predicted peak loads for these two FMLs plates were
3255 and 4531 N, respectively, which are only 0.6% and
6.0% higher in comparison with the experimental results of
3235 and 4275 N, respectively. The predicted initial stiffness and the predicted displacement at the peak load were
also in reasonably good agreement with the corresponding
experimental results. The predicted perforation energies for
these two plates were 18.88 and 25.38 J, respectively.
Compared to the experimental results of 16.51 and 26.17 J,
they were 14.4% higher and 3.0% lower, respectively.
Figures 12 and 13 display the failure modes obtained
from experimental tests and numerical simulations for the
1175
Figure 11 Load-displacement traces for the 3/2 FML plate made with
8-ply composite cores subjected to a low velocity impact from the test and
the FE simulation.
Figure 12 Failure modes of the 3/2 FMLs plates made with 4-ply composite cores subjected to a low velocity impact. (a) The test; (b) the FE
simulation.
Figure 13 Failure modes of the 3/2 FMLs plates made with 8-ply composite cores subjected to a low velocity impact. (a) The test; (b) the FE
simulation.
Figure 10 Load-displacement traces for the 3/2 FML plate made with
4-ply composite cores subjected to a low velocity impact from the test and
the FE simulation.
3/2 FMLs plates made with 4-ply and 8-ply composite cores
respectively subjected to an on-set perforation impact.
Again, essential features of the experimental failure mode,
such as the cross-crack pattern and localised deformations,
were well simulated by the corresponding finite element
models.
The finite element models were also verified by test results of the FMLs plates made with the higher stacking
1176
Figure 16 The central cross-sections of the failed 4/3 FML plate made
with 4-ply composite cores subjected to a low velocity impact. (a) The test;
(b) the FE simulation.
Figure 17 The central cross-sections of the failed 4/3 FML plate made
with 8-ply composite cores subjected to a low velocity impact. (a) The test;
(b) the FE simulation,
Figure 14 Load-displacement traces for the 4/3 FML plate made with
4-ply composite cores subjected to a low velocity impact.
Figure 15 Load-displacement traces for the 4/3 FML plate made with
8-ply composite cores subjected to a low velocity impact.
Figures 18 and 19 show the comparison between the predicted peak load as well as the perforation energy and the
corresponding test results in a chart form. Clearly, very
good correlation was obtained, including the general trend.
In order to draw out the reliable relationship, more points
need to be produced by using validated computer models. In
fact, the finite element models developed are well validated
based on the reasonably good prediction of the failure
modes of the FMLs plates with the increasingly high stacking number, together with good simulations of the load-
Figure 18 Comparison of the predicted peak load and the related test
result for all FML plates studied.
4
5
6
7
Figure 19 Comparison of the predicted perforation energy and the related test results for all FML plates studied.
9
10
Conclusions
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
1
2
Vlot A. GlareHistory of the Development of a New Aircraft Material. Dordrecht: Kluwer Academic Publishers, 2001
Krishnakumar S. Fiber metal laminatesThe synthesis of metals and
32
1177