Professional Documents
Culture Documents
Modelling of Low Velocity Impact of Laminated Composite Substruct - 3
Modelling of Low Velocity Impact of Laminated Composite Substruct - 3
Scholars' Mine
1-1-2011
Fatih Dogan
Missouri University of Science and Technology, doganf@mst.edu
Ahmed M. Elmarakbi
Recommended Citation
H. Hadavinia et al., "Modelling of Low Velocity Impact of Laminated Composite Substructures,"
International Journal of Vehicle Structures and Systems, vol. 3, no. 2, pp. 96-106, MechAero Foundation
for Technical Research & Education Excellence (MAFTREE), Jan 2011.
The definitive version is available at https://doi.org/10.4273/ijvss.3.2.04
This Article - Journal is brought to you for free and open access by Scholars' Mine. It has been accepted for
inclusion in Materials Science and Engineering Faculty Research & Creative Works by an authorized administrator
of Scholars' Mine. This work is protected by U. S. Copyright Law. Unauthorized use including reproduction for
redistribution requires the permission of the copyright holder. For more information, please contact
scholarsmine@mst.edu.
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
I n te r na ti on al J our na l of
ISSN: 0975-3060 (Print), 0975-3540 (Online)
doi: 10.4273/ijvss.3.2.03
Vehicle Structures & Systems
Available online at www.ijvss.maftree.org
© 2011. MechAero Foundation for Technical Research & Education Excellence
ABSTRACT:
Impact is among the important loading scenario which is required to be addressed for fibre reinforced plastics (FRP)
materials in ground or space vehicle applications. The failure behaviour of FRP materials under impact loading is a
complex process and a detailed analysis of various mode of failure is necessary. In this paper the failure behaviour of
laminated carbon fibre reinforced plastic (CFRP) composite structures under impact loading is investigated by
conducting numerical simulations using the explicit finite element analysis ANSYS LS-DYNA software [1]. The impact
responses and failure behaviours are being investigated by performing a parametric study and sensitivity analysis in
which impact velocity, number of plies, incident angle, friction between contacted surfaces, impactor mass and the
geometry are varied in separate cases. The FE model is validated by comparing the numerical results with other
published results and good correlations are achieved.
KEYWORDS:
Composite structures; energy absorption; impact; finite element analysis; delamination
CITATION:
H. Hadavinia, F. Dogan, A. Elmarakbi and S.M.R. Khalili. 2011. Modelling of Low Velocity Impact of Laminated
Composite Substructures, Int. J. Vehicle Structures & Systems, 3(2), 96-106. doi:10.4273/ijvss.3.2.04
the structural component. These enquires have led to quasi-static contact problem as the influence of elastic
advancement in understanding of the damage micro- waves and the strain rate in the interior of the plate are
mechanisms at impact loading conditions, related to the not taken into account [18]. In other words, at low
basic properties of resin, fibres, fibre–matrix interfacing velocity impacts, the duration of the impact, i.e. the
mechanism, the architecture of the laminate and stacking interval of time elapsed from the first contact and the
sequence [4,5]. Understanding the mechanics of contact complete detachment, is much longer than the time
interaction between impactor and the target laminated required by the generated elastic waves at the point of
composite structure is also essential in order to first impact to propagate through the whole body. For
accurately predict the contact force history and to predict this reason, the impact is a highly dynamic event. When
damage evolution during the impact events [6]. In many many vibration modes of a body are excited, the
structures the damage can significantly reduce their load statically determined contact laws can be used for the
bearing capability or cause catastrophic failures. dynamic analysis as strain rate and wave propagation
In the past decades, many studies dealt with metallic effects are negligible for commonly used materials. For
structures where the mechanism of energy absorption is example, Hunter [18] showed that the energy lost by
dominated by plastic deformation. Recently, as the usage elastic waves during the impact of a sphere on an elastic
of the composite materials becomes more widespread, half-space is negligible as long as the initial velocity of
impact research is focused more on composite structures. the sphere is small compared with the phase velocity of
Many parameters have influences on the impact of compressive elastic waves in that solid [2].
composite structures. Impact velocity affects the energy According to Zukas et al. [19], impacts are
absorption of composite structures significantly and it is categorized in four different regimes. They are low
considered as the most important parameter [7]. In order velocity, high velocity, ballistic and hyper velocity
to maximizing the impact energy absorption for regimes. Zukas et al. [19] defined the velocity of these
passenger safety and comfort, it is important to limit the four ranges as <250 m/s, 250-2000 m/s, 2000-12000 m/s
maximum load transmitted to the rest of the structure. and >12000 m/s. In general low velocity impact is
This maximum load must not cause catastrophic failure defined as events which can be treated as quasi-static,
within the overall structure. the upper limit of which can vary from one to tens m/s
The heterogeneous and anisotropic nature of fibre- depending on the target stiffness, material properties and
reinforced plastic (FRP) laminates gives rise to four the mass and stiffness of the impactor [17, 20]. Cantwell
major modes of failure under impact loading (although and Morton [9] specified velocity <10 m/s as low
many others could be cited). Matrix mode failure occurs velocity while Abrate [2] assume <100 m/s as low
when cracks appear parallel to the fibres due to tension, velocity impact. Olsson [21] categorised the impact
compression or shear. Delamination mode can occur as phenomenon on composite based on the energy
the plies separate from each other under transverse associated by impact and contact time. For e.g., when the
interlaminar stresses. Fibre mode failure happens when duration of impact is the same (small impactor mass with
the tensile stress in the fibre reaches to fibre tensile high velocity) or longer (heavy mass with low velocity)
which lead to fibre breakage or in compression will lead as that of the time required for the flexural and shear
to fibre buckling, and Penetration happens when the waves to reach the target boundaries. Low velocity
impactor completely perforates the impacted surface [8]. impact is also classified according to damage [22, 23]. If
This delamination failure mode is typical of CFRPs the damage causes by delamination and matrix cracking,
laminates and is caused mainly by the interlaminar the impact can be defined as a low velocity impact.
stresses generated between plies of different fibre Otherwise, the impact is defined a high velocity impact
orientation and thus showing different flexural when penetration and fibre breakage occurs.
behaviour. This greatly contributes to decrease the In this paper, the impact responses and failure
strength of the laminate, as the different plies no longer behaviours of composite substructures are being
work together. Many studies have been done on the investigated by performing a parametric study and
behaviour of carbon fibre/epoxy laminates subjected to sensitivity analysis in which impact velocity, number of
impact loading [5, 9-15]. The first step in studying the plies, incident angle, friction between contacted surfaces,
impact behaviour of composite materials is to impactor mass and the geometry are varied in separate
characterize the type and extension of the damage cases.
induced in the structural component which could cause
fibre fracture, matrix cracking, fibre pull-out and 2. Validation of Finite Element Modelling
delamination [12-16]. The threshold energy, that is, the
In this paper, ANSYS LS-DYNA explicit finite element
impact energy below which no apparent damage is
program is used as a platform for the study of low
induced within the laminate, as well as the damage
velocity impact of various substructures made from
extent for a given impact energy above this threshold,
laminated CFRP composites. The explicit solution
are of great practical interest.
method used by LS-DYNA provides solutions for short-
The transition between different regimes of impacts
time, large deformation dynamics, quasi-static problems
based on the impactor velocity is not strictly defined and
with large deformations and multiple non-linearity, and
there are disagreements on the definition [8]. Sburlati’s
complex contact/impact problems. The models can be
[17] definition is based on the time during which the
transferred between ANSYS and LS-DYNA to perform
impactor and the plate remain in contact. If the contact
sequential implicit-explicit/explicit-implicit analyses,
time is very long in comparison to the lowest period of
such as those required for drop test, spring back and
free vibrations of the plate, the impact is counted as a
97
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
other applications [24]. The composite substructures the laminate occurred. In the FEA model, the CFRP
which will be discussed later are impacted at right angle plate is clamped at lateral end tabs (ux=uy=uz=0) and
and the effect of incident angle on the energy absorption simply supported along the longitudinal edges (uy=0).
of these structures has also been investigated. No rotational constraints are imposed on the edges. The
In order to be assured of reliability of the modelling results of experimental and FEA impact energy time-
results, the FEA results need to be compared with history for unstressed and pre-stressed plates are
experimental tests. For this study, the low velocity compared in Fig. 2(a). The beginning of the plateau of
impact tests on laminated plates carried out by Heimbs et the curve coincides with the loss of contact between the
al. [25] is chosen for verification of the modelling impactor and the plate and this is the plate absorbed
methodology (see Fig. 1). In their study, they used energy. The plateau energy is made of kinetic and elastic
carbon fibre-reinforced epoxy laminate with a energies, and absorbed energy due to laminate damage.
symmetric, quasi-isotropic lay-up made of 24 plies with It is commonly assumed that in FRP composite materials
stacking sequence of [-45/0/45/90]3s. The Cytec© the first two energies are much lower than the third one,
unidirectional prepreg material is made of the 12 K HTS so the total absorbed energy practically equivalent to the
carbon fibres and 977-2 epoxy matrix. The laminated energy dissipated by damage in the laminate.
plate were made by stacking plies according to the During the experiment the contact force between the
required fibre orientation angles and were cured in an impactor and the plate was also recorded. Fig. 2(b)
autoclave at 180 C and 7 bar. The resulting average compares the time history of experimental contact force
cured plate thickness was 2.7 mm with free specimen and FE results for both unstressed and pre-stressed
size of 300mm150mm. The specimen length was plates. The experimental and FEA results are in good
400mm and the specimen was bonded with a 50mm agreement with high degree of accuracy for both
width end tabs on both ends, see Fig. 1. The reported unstressed and pre-stressed plate. The contact force
material properties of CFRP are =1.46 g/cm3 E11= 153 shows high oscillation during the impact period due to
GPa, E22= 10.3 GPa, G12= 5.2 GPa, Xt=2540 MPa, elements failure.
Xc=1550 MPa, Yt=82 MPa, Yc=236 MPa, SC=90 MPa
50
and ν12= 0.3 [25]. Unstressed- FEA
Unstressed- Experiment
Pre-stressed- FEA
40 Pre-stressed- Experiment
Energy (kJ)
30
20
10
(a)
0
0 2 4 6 8 10
Time (ms)
(a) Energy
10
Unstressed- FEA
Unstressed- Experiment
Fig. 1: Dimensions of experimental impacted specimen [25] and Pre-stressed- FEA
LS-DYNA model of laminated composite plate with the rigid 8 Pre-stressed- Experiment
spherical impactor
6
using thin shell element SHELL163 with 24 sub-layers
through the thickness. For modelling damage evolution
during the impact of the plate, material MAT54, 4
MAT_ENHANCED_COMPOSITE_DAMAGE is used.
This material model is based on Chang-Chang failure
2
criteria explained in detail in section 3. The impactor is
modelled as a spherical rigid body using a rigid material
model MAT20. Note that the strength of the elements 0
(b)
surface. Tiebreak contact *CONTACT_TIEBREAK_ When failure in all through the thickness composite
SURFACE_TO_SURFACE is one of the recommended layers has occurred, the element is deleted. The
methods for modelling adhesives in LS-DYNA [27]. parameter can be used to scale the shear stress
Tiebreak contacts have been placed between surfaces 2-6 interaction in the fibre tensile failure criterion. Hashin
and 3-7. Tiebreak contact functions the same way as criterion will be resulted if =1 which overestimates the
common contacts under compressive loading. Under shear stress interaction as stated by Schweizerhof et al.
tensile load, tiebreak allows the separation of the tied [33]. For =0 simple maximum stress criterion without
surface under the following failure criterion: interaction will be resulted which compare better with
2 2 experiments. In this study the value of β=0 has been
n
s
1 (1) chosen.
NFLS SFLS MAT20 is used to model the rigid cylindrical
Where NFLS is normal failure strength and SFLS is impactor. Impactor mass is normally limited in real tests
depending to the testing capability. Cylindrical impactor
shear failure strength for adhesive and n and s are the
mass is approximately 62 kg. The mass density of
applied normal and shear stresses, respectively. In this
cylindrical impactor is taken as 15 times the mass
study, the adhesive strength in normal and shear modes
density of steel, ρ=7826 kg/m³, to reduce the
are assumed to be NFLS= 56 MPa and SFLS= 44 MPa
computation time. The Young’s modulus and Poisson’s
which are the common properties of structural epoxy
ratio have been selected the same as steel, i.e. E=207
adhesive.
GPa and =0.28. In reality, a box beam can be
manufactured using welding at corners or using
appropriate adhesives. In modelling, welding can be
defined by using spot-weld option in LS-DYNA. In this
study, welding will be used for simple box and stiffened
box beam. In both of these models, mid-planes of the
vertical plates are offset from the horizontal plates’
edges. Spot welds are inserted to connect the edge nodes
of vertical and horizontal plates. Spot welds are created
using *CONSTRAINED_SPOTWELD command (see
Figure 4). More details about spot welds can be found in
LS-DYNA Keyword User’s Manual [27]. These spot
Fig. 4: Definition of contact between different surfaces welds modelled as massless and rigid therefore they do
not have any effect on the mass of the beams.
The orthotropic, linear elastic material law MAT54
The wall thickness is t=4mm and distance between
based on Chang-Chang failure criteria [29-32] control
failure in longitudinal (fibre) and transverse (normal to nodes (element edge size) is d=10mm. Due to stress
fibre) directions in tensile, compressive and shear concentration at the corner, the transverse tensile
loading of the fibre and the matrix. These failure modes strengths and in-plane shear strength were reduced by
in the Chang-Chang failure criterion for composite shell 50% giving:
elements are: 1 1
S n Yt t d , S s Sc t d (6)
Tensile fibre mode (aa > 0) 2 2
2
0 failed (2) Where Yt is the transverse tensile strength and Sc is in-
e2f aa ab 1
t
X c
S 0 elastic plane shear strength given in Table 1 [34]. Sn and Ss are
normal and shear strength of the spot weld. Putting
If failed then Ea = Eb = Gba = vab = 0. values given in Table 1 in Eq. (6), Sn=960 N and Ss =
Compressive fibre mode (aa < 0) 1380 N. During the analysis, spot welds will fail when
2
0 failed (3) the following condition is met:
ec2 aa 1
Xc 0 elastic N 2
f f
n s (7)
If failed then Ea = vba = vab = 0. 1
Sn Ss
Tensile matrix mode (bb > 0)
2 2
0 failed (4) where fn and fs are normal and shear forces in the spot-
2
em bb ab
S
1
Yt c 0 elastic weld element at the end of each increment of the
loading. N and M are the failure criteria exponents which
If failed then Eb = vbs = 0 .Gba = 0. are set to 1.0 in the current work.
Compressive matrix mode (bb < 0)
2 Y
2 4. Parametric Studies
e d2 bb c 1 bb
2 S c Yc (5)
2S c Before analysis of the composite substructures a
2
0 failed parametric studies were carried out to find the effects of
ab 1
various parameters. Many parameters have significant
Sc 0 elastic
influences on the behaviour of laminated composites
If failed then Eb = vba =vab = 0 .Gab. Xc =2Yc for during low velocity impacts. For example, impactor
50% fiber volume. velocity is one of the most important parameter.
100
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
Different impact velocities cause various levels of The deflections of the beam are shown in Fig. 6(b).
damage and energy absorptions. Other important The deflection is increased at higher impact velocity as
parameters are incident angle, the impactor mass, an impact at a higher initial velocity has higher kinetic
number of plies, frictions between contacted surfaces, energy. When the velocity is 3 m/s, cylinder bounced
and structures geometry. A rectangular composite box back. The energy absorbed by the box at various
beam is chosen as the base model for sensitivity analysis. velocities is shown in Fig. 6(c).
The laminates in the base model made of five plies
14
(IP=5) impacted at a velocity of 4 m/s with an impactor V= 3 m/s
mass of 62 kg with static coefficient of friction 0.1 and 12
V=
V=
4
5
m/s
m/s
dynamic coefficient of friction 0.0225. In the sensitivity V= 6 m/s
both of the vertical and top horizontal plates. When (b) Deflection
impactor velocity is 3 m/s no buckling (damage) 1000
occurred on vertical plates. Top horizontal plate resists
to the impactor load itself.
800
V = 4 m/s V= 3 m/s
V= 4 m/s
600
V= 5 m/s
Energy (kJ)
V= 6 m/s
400
V = 5 m/s 200
(c)
0
0 10 20 30 40 50 60 70
(c) Energy
Fig. 6: Time history for box beam at different impactor velocities
500
m = 31 kg
400
Energy (kJ)
=30 =75
200
100 m = 124 kg
0
0 10 20 30 40 50
Time (s)
m = 248 kg
(a) Energy
5
=0 =45
=15 =60
=30 =75
4 Fig. 8: Damage of composite box beam for different impactor mass
at t=50 ms
Impactor Velocity (m/s)
40
3
m= 31 kg
m= 62 kg
30 m= 124 kg
m= 248 kg
2
20
Reaction Force (kN)
1
10
0
0
0 10 20 30 40 50
Time (s)
-10
(b) Impactor velocity
(a)
Fig. 7: Time history at different incident angles for box beam -20
0 10 20 30 40 50 60
(plies) are set to 2, 5 and 8. The base model has five impact velocity and impactor mass have significant
integration points. When number of plies is 2, top effects on the behaviour of composite structures during
horizontal plate is not damaged by impactor. By low velocity impacts, the static and dynamic coefficients
increasing the number of plies the damage on composite of friction and number of plies have a negligible effect
box is increased slightly. The peak reaction force value on the energy absorption and reaction force.
and peak time remained the same for all cases Fig. 10(a). 30
The deflection of box beam is smaller with increasing
the layers while keeping the thickness unchanged, Fig. 25
s=0.1
10(b). Hence using more integration points makes the
s=0.2
beam stiffer. 20
s=0.4
25 10
NIP=2
NIP=5
20 5
NIP=8
0
Reaction Force (kN)
15
10 -5
(a)
5 -10
0 10 20 30 40 50 60
0 Time (ms)
Time (ms) 8
=0.1
s
=0.4
s
10 6
8
NIP=2 4
NIP=5
NIP=8
Average Deflection (mm)
6
2
(b)
4
0
0 10 20 30 40 50 60
Time (ms)
2
(b) Deflection
(b) Fig. 11: Time history for box beam for various static coefficients of
0
0 10 20 30 40 50
friction
30
Time (ms)
25
(b) Deflection
=0
d
Fig. 10: time history for box beam for different number of plies 20
=0.0225
d
=0.05
4.5. Static and dynamic friction coefficients d
Reaction Force (kN)
15
these peaks are not changed as shown in Fig. 11. Also, Time (ms)
there are no significant changes in reaction force and
displacement when coefficient of dynamic friction is Fig. 12(a): Reaction force history for box beam for various
dynamic coefficients of friction
changed as shown in Fig. 12. In summary though the
103
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
d
=0.05
6
d beam structural integrity remains nearly intact while the
simple plate and stiffened plate broken to pieces.
5
4
2
3
under impact velocity of 4 m/s and other base load Time (ms)
35
scenario defined in Section 4 are studied. The results of
deflections, energy, impactor velocity and contact forces 30
for the considered four substructures are compared in
Figs. 13(a-d). From Fig. 13(a), the deflection of box 25
Plate
Stiffened Plate
Contact Force (kN)
beam and stiffened box beam bounces back to zero while Box Beam
Cylindrical Tube
plain plate, stiffened plate and cylindrical tube are 20 Stiffened Box Beam
Time (ms)
30
Average Deflection (mm)
Fig. 13: (c) Impactor velocity (above) and (d) Contact force
(below) histories at impact velocity of 4 m/s
20
Plate
Stiffened Plate
Box Beam (a) Simple plate
10 Cylindrical Tube
Stiffened Box Beam
(a)
-10
0 10 20 30 40 50 60 70 80 (b) Stiffened plate
Time (ms)
500
400
Plate
300 Stiffened Plate
Box Beam
Cylindrical Tube
Stiffened Box Beam
200
Time (ms)
Fig. 13: (a) Deflection (above) and (b) Energy (below) histories at Fig. 14: Damage of composite substructures at t=30 ms
impact velocity of 4 m/s
104
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
105
H. Hadavinia et al. 2011. Int. J. Vehicle Structures & Systems, 3(2), 96-106
[17] R. Sburlati. 2002. The contact behavior between a foam [28] S.H. Benabdallah. 2007. Static friction coefficient of some
core sandwich plate and a rigid indentor, Compos.: Part plastics against steel and aluminium under different
B, 33 (4), 325-332. doi:10.1016/S1359-8368(02)00014-8 contact conditions, Tribology Int., 40, 64-73. doi:10.1016
[18] S.C. Hunter. 1957. Energy absorbed by elastic waves /j.triboint.2006.02.031
during impact, J. Mech. Phys. Solids, 5, 162-171. doi:10. [29] F.K. Chang and K.Y. Chang. 1987. A progressive damage
1016/0022-5096(57)90002-9 model for laminated composites containing stress
[19] J.A. Zukas, T. Nicholas, H.I. Swift, L.B. Greszczuk, and concentrations, J. Compos. Mater., 21, 834-855. doi:10.11
D.R. Curran. 1982. Impact Dynamics, John Wiley & Sons. 77/002199838702100904
[20] A.E.H. Love. 1862. A Treatise on the Mathematical [30] H.Y. Choi, H.Y.T. Wu, and F.K. Chang. 1991. A new
Theory of Elasticity, Cambridge University Press. approach toward understanding damage mechanisms and
mechanics of laminated composites due to low-velocity
[21] R. Olsson. 2000. Mass criterion for wave controlled impact: Part II, J. Compos. Mater., 25, 1012-1038.
impact response of composite plates, Composites: Part A,
31, 879-887. doi:10.1016/S1359-835X(00)00020-8 [31] H.Y. Choi and F.Y. Chang. 1990. Impact damage
threshold of laminated composites. Failure Criteria and
[22] D. Liu and L.E. Malvem. 1987. Matrix cracking in Analysis in Dynamic Response, AMD Vol. 107, ASME
impacted glass/epoxy plates, J. Compos. Mater., 21, 594- Applied Mechanics Division, Dallas, TX, 31-35.
609. doi:10.1177/002199838702100701
[32] H.Y. Choi, H.S. Wang, and F.K. Chang. 1992. Effect of
[23] A. Faggiani and B.G. Falzon. 2010. Predicting low- laminate configuration and impactor’s mass on the initial
velocity impact damage on a stiffened composite panel, impact damage of graphite/epoxy composite plates due to
Composites: Part A, 41, 737-749. doi:10.1016/j.composit line loading impact, J. Compos. Mater., 26(6), 804-827.
esa.2010.02.005 doi:10.1177/002199839202600603
[24] ANSYS V11.0, USER HELP MENU
[33] K. Schweizerhof, M. Maier, A. Matzenmiller, and W.
[25] S. Heimbs, S. Heller, P. Middendorf, F. Hahnel, and J. Rust. 1993. Energy absorption with composite crash
Weiße. 2009. Low velocity impact on CFRP plates with elements in frontal crash – an analysis with LS-DYNA3D,
compressive preload: Test and modelling, Int. J. Impact 11th CAD-FEM user’s meeting, 315-325.
Engg., 36, 1182-1193.doi:10.1016/j.ijimpeng.2009.04.006
[34] J.R. Smith, L.C. Bank, and M.E. Plesha. 2000.
[26] M. Langseth, O.S. Hopperstad, and T. Berstad. 1999. Preliminary study of the behavior of composite material
Crashworthiness of aluminium extrusion: validation of box beams subjected to impact, 6th Int. LS-DYNA Users
numerical simulation, effect of mass ratio and impact Conf., Vol. 11, Dearborn, 1–15.
velocity, Int. J. Impact Engg., 22, 829-854. doi:10.1016/ [35] C.Y. Tu and C.C. Chao. 1999. Three-dimensional contact
S0734-743X(98)00070-0
dynamics of laminated plates: Part 2 -Oblique impact with
[27] LS-DYNA Keyword’s User Manual, Livermore Software friction, Composites: Part B, 30, 23–41. doi:10.1016/S135
Technology Corporation, 2003, Version 970. 9-8368(98)00039-0
106