Professional Documents
Culture Documents
Composites: Part B
journal homepage: www.elsevier.com/locate/compositesb
a r t i c l e i n f o a b s t r a c t
Article history: The weight reduction has become a key driver in the automotive design. The study and the simulations of
Received 10 September 2013 non-conventional materials has become very important. This work is focused on the numerical
Received in revised form 23 November 2013 crashworthiness design of carbon fibres reinforced plastics (CFRP) structures. After a literature research
Accepted 9 January 2014
on the currently numerical models specifically used for CFRP, two different numerical models were
Available online 16 January 2014
developed in order to reproduce experimental crush test on CFRP tube previously made. The influence
of several different parameters on the final results was analysed. The work put in evidence the advanta-
Keywords:
ges of crushing zone modelling compared to a stacked shell one.
A. Carbon fibre
C. Finite element analysis (FEA)
Ó 2014 Elsevier Ltd. All rights reserved.
C. Computational modelling
B. Impact behaviour
1. Introduction the design phase. For metallic materials the available numerical
codes are capable to provide good levels of predictability. The
Nowadays the use of composite materials is becoming a com- energy dissipation by plastic deformation in ductile metallic struc-
mon practice in several industry fields such as the automotive tures is well understood and may be modelled by elastic–plastic
and the aerospace ones [1–3]. The much higher specific strength material models in FEA. For modelling CFRP it is vital to reproduce
and stiffness showed by this kind of material, when compared to precisely the complex physics behind the composite’s crushing
the traditional employed metallic alloys, opened a large horizon behaviour and therefore to apply more complex modelling.
of possibilities concerning weight saving but also crashworthiness There are approaches using a very fine discretization of the
[4–6]. In the automotive industry, carbon fibres reinforced plastics composite structure, modelling details of individual tows.
(CFRP) have been successfully used making possible the design of Currently these approaches consume too much computational
vehicles with higher performance, lower fuel consumption and, resources to be applied to analysis [10] of large automotive
as a consequence, lower CO2 emissions. structures.
Compared to traditional metallic materials, composite materials It is necessary to develop models that are simple enough to be
offer increased possibilities to tailor the performance. A part from employed in practical analysis situations but at the same time
the matrix and reinforcements selected, material orientation and capable to provide results with a suitable level of accuracy. At
stacking as well as processing parameters influence the structural the same time the approach shall be numerically robust and prac-
behaviour. tical in model build phase.
Concerning crashworthiness, CFRP structures fail through a The goal of this work is to study the overall capability of the ex-
combination of several different fracture mechanisms [7–9]. These plicit solver in modelling the crushing behaviour of a CFRP tube.
complex failure modes, together with friction effects, contribute to The most promising modelling techniques were adopted. After a
the overall energy absorption. This number of effects makes the de- deep analysis of the state of art of the subject, two different finite
sign of energy absorbing structures much more difficult than met- element modelling approaches were developed and the influence
als, which are extensively used for this kind of applications. of the most important model parameters was examined. The re-
In crashworthiness design, the finite element simulations play a sults analysis, where the numerical data were compared to exper-
very important role thanks to its possibility to predict behaviour in imental reference values, allowed defining the best practice for the
early design phases, thus enabling cost reduction and shortening of simulation of this kind of problems.
⇑ Corresponding author. Tel.: +39 011 0906937; fax: +39 011 0906999. 2. Experimental methods
E-mail addresses: pedrombussadori@gmail.com (B.P. Bussadori), karsten.
schuffenhauer@lamborghini.com (K. Schuffenhauer), alessandro.scattina@polito.it Numerous tests have been proposed in literature to determine
(A. Scattina). the mechanical properties relevant in crashworthiness analysis.
1359-8368/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compositesb.2014.01.020
726 B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735
The proposed tests range from large scale to small scale specimen, The average crushing force F is evaluated after the initial peak of
from high speed to low speed quasi-static tests. force, during the stable crush zone.
The crush test consists of a constant axial compression of the The CE is defined as the ratio between the peak force and the
tested specimen. It is one of the most important tests to study average crushing one:
the crashworthiness of a structure [11] since it allows to measure
F peak
the required properties and at the same time allows good observa- CE ¼ ð3Þ
tion of the behaviour at low cost. The most important result exam- F
ined is the force versus displacement curve. This curve (Fig. 1) If the crushing efficiency is equal to one the behaviour is called
usually has three well distinguished stages [6]. In the first one as perfect plastic and the load versus displacement chart would
the compressive load reaches its maximum value, just before the have a rectangular shape. The absolute values of the average and
beginning of the trigger collapse. The trigger has the function to peak forces are of great relevance for crashworthiness design, be-
initiate the collapse of the structure in a well-controlled way and cause they are strictly related to the accelerations transmitted to
also in a well-defined location. The value of the peak load and the vehicle’s occupants in a real situation of crash. Not only a cer-
the drop from it to the average crushing load, are strongly tain amount of energy has to be absorbed but also the values of the
dependent on the trigger geometry [6]. The second crushing stage peak and average loads have to remain below a properly threshold
is the transition one, in which the compressive load drops from its [12].
peak value and after some oscillation reaches the average crushing Starting from experimental results discussed in the following
load. The transition phase is usually very short. In the third final subsection 2.1, a theoretical reference crushing curve was defined.
stage the crushing proceeds through the specimen at a constant This reference curve was used to compare and to evaluate the
average load. This phase usually has a serrated behaviour, with numerical simulations results.
an oscillation of the crushing load around an average value. It is
during this phase that the major part of the energy is absorbed
2.1. Experimental results
by the specimen.
From this curve some relevant quantities can be calculated.
The specimens considered for this work were manufactured
These include: the specific energy absorption (SEA), the average
using a carbon-epoxy pre-impregnated fabric material. Table 1
crushing stress, the peak load, the average crushing load and the
contains a summary of the main physical and mechanical proper-
crushing efficiency (CE). The SEA is the energy absorbed per unit
ties of the used material.
of mass of crushed materials:
There is no standardised crush test for composite materials, a
E large number of different specimen geometries have been used
SEA ¼ ð1Þ
qAd for previously crashworthiness studies [6]. For this work a thin-
walled square tube was chosen. This kind of geometry profile is
where E is the energy absorbed obtained integrating the load over
widely used by the automotive industry, especially when concern-
the total displacement d, q is the material density, and A the cross
ing crashworthiness structural components [11,12]. The specimens
section area. The SEA is not an intrinsic material property. It de-
have a length of 150 mm with an inner side of 50 mm. The wall
pends not only on the material properties but also on several other
thickness is 2.08 mm. For the production of the specimens, during
parameters, especially the specimen geometry. Two specimens
the lamination phase, a female mandrel made of sand was used.
made of the same material but with different geometries can col-
The lamination lay-up is shown in Fig. 2, where there is a top view
lapse in very distinct ways, and as a consequence can achieve very
of the specimen. The bottom part of the tubes was attached to a re-
different values of SEA. The SEA allows comparing the energy
sin plate to avoid sliding on the test machine during the crush. The
absorption capabilities of two different materials and geometries
tests were performed in quasi-static conditions using an INSTRON
when the weight is an important parameter of project.
5900 250 kN electric machine, with a constant forward velocity set
The average crushing stress r is the average crushing force di-
to 20.0 mm/min throughout the test. The final vertical displace-
vided by the specimen cross sectional area:
ment was set to 100 mm (2/3 of the overall specimen’s length)
F while the load and the displacement of the machine cross beam
r¼ ð2Þ were recorded at a frequency of 10 Hz. The test was replied for
A
three different specimens. Fig. 3 contains the average experimental
curves obtained for the tested specimens, while the Table 2 sum-
Force marises all the relevant average parameters concerning the energy
absorption of the tests. Fig. 4 shows the specimen before and after
the crush test.
Peak load
Fpeak
Table 1
F: average Carbon-epoxy material properties.
crush load
Property Value
Nominal density (g/cm3) 1.528
Nominal cured ply thickness (mm) 0.227
Transition Stable crush
Resin strain energy release rate (J/m2) 250
zone zone
Filament count/tow 3000
Weave topology Twill 2 2
Dry weave areal weight (g/cm2) 95
Fibre volume fraction (%) 43
Tensile strength (MPa) 800
Tensile modulus (GPa) 65
Compressive strength (MPa) 800
Displacement Compressive modulus (GPa) 62
In-plane shear strength (MPa) 89
Fig. 1. Typical crushing load vs. displacement curve.
B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735 727
The load curve has a serrated pattern, with very small oscilla-
tions of the force around its average value during the stable crush
phase, very similar to the typical curve showed in Fig. 1. In this test,
the predominant collapse mode showed by the tube was the lam-
ina bending, or splaying one [13–15], clearly identified in Fig. 4. It
is possible to see that large amounts of delamination occurred be-
tween two main groups of plies which remained attached to each
other along the test. The outer plies (first group), moved to the out-
side, leading to the formation of large material fronds as also the
tearing of the corners. These plies were largely bent causing the
micro-fracture of the fibres and matrix which compose them. The
second group of fibres (inners one), remained stuck in the interior
of the tube, being completely crushed, leading to high amounts of
fibre and matrix fracture.
With these results the reference curve was made (Fig. 1). In this
curve the stable crush zone, after the first peak load and the de-
crease of the load, is a straight line. This means constant load
increasing the displacement. A SEA equal to 70 kJ/kg was defined
Fig. 2. Lamination lay-up scheme, top view of the specimen.
as a reference value for this theoretical curve. For the peak force
a value 20% higher with respect to the average one was used.
Starting from the expression described before, the values (Table 3)
of the main relevant parameters for the reference curve were
60 4.8
obtained.
50 4
3. Numerical simulations
Absorbed energy (kJ)
Table 3
0 0 Reference crushing values.
0 20 40 60 80 100
Reference parameters Value
Displacement (mm)
SEA (kJ/kg) 70
F (kN) 47.80
Fig. 3. Average experimental load and energy vs. displacement curves of the CFRP
Fpeak (kN) 57.36
specimens.
r (MPa) 105.84
Table 2
Main parameters obtained from experimental tests on carbon-epoxy specimen.
Total absorbed energy (kJ) Collapsed mass (kg) Average crushing force (kN) Average crushing stress (MPa) SEA (kJ/kg)
4.32 0.067 42.41 95.32 64.48
Fig. 4. Composite tube before (a) and after (b) the test.
728 B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735
In literature, large variety of models can be found. In Fig. 5, a In the stacked shell models the composite material is modelled
classification of these models is proposed. It is possible to distin- with more layers of shell elements joined together by specific ele-
guish two major groups. ments (cohesive elements, springs). These joining elements work
The first group is the micro-mechanical one [10,16–20]. In this according to a specific model, which has the goal of reproducing
group the finite element models try to simulate the composite the composite’s matrix phase properties. The global stiffness ma-
crushing phenomenon through a detailed modelling of its micro- trix of this type of model is the assembly of the stiffness matrix
mechanical behaviour. A very fine solid mesh is required to model of the shell elements and of the joining elements. The stacked shell
accurately the micro-mechanics matrix crack propagation phe- model allows a compromise between accuracy and efficiency
nomenon. The computational effort demanded by this kind of depending on the utilised number of layers. This kind of model is
model is very high, what makes it unpractical for engineering crash capable to provide a better physical representation of the stratified
analysis. This approach is used mainly to perform simulations con- composite structure but at the same time it keeps the simplicity
cerning the delamination phenomenon, in which the growth inherent of the macro-mechanical approach. These two features
behaviour of a single crack is studied in a very detailed way. are the main reasons behind its wide use for this kind of
The second group is the macro-mechanical one [21–30]. This simulation.
type of models provides a macro-mechanical description of the In this work the two different typologies of the macro-mechan-
material collapse. It is much more computational effective, and ical models have been taken into consideration, developing a
consequently it is a suitable choice for engineering crash analysis. stacked shell model on one hand and a single layer shell model
However, it is not capable to model precisely all the main collapse on the other hand. The two numerical models are described in
modes that occur simultaneously during a crush event. details in the following sections.
The macro-mechanical group, in turn, can be divided in two
main different types of models: the single shell layer models 3.1. The stacked shell model
[17,21,22,24] and the stacked shell models [25,29,30].
The single shell layer models use a single layer of shell elements The stacked shell numerical model developed, shown in Fig. 6,
to model the specimen. This kind of model is not capable to model can be divided in four components: the machine plate, the trigger,
the inter-laminar collapse modes showed by composites under the composite tube structure and the interlaminar elements, called
crushing in an accurately way. However it is useful if a not detailed also as tied elements in the following.
representation of the failure physics but only load and energy level The machine plate was modelled using shell elements as a flat
predictions are required. The main advantages are its simplicity rigid body located few millimetres just above the tube. The com-
and computational effectiveness, so for that point they are highly posite tube is modelled using different layers of shell elements
suitable for practical engineering crash analysis. On the other hand, joined together with cohesive elements. The number of layers used
they have a notable lower robustness due to large amount of in the specimen modelling is a key feature of this kind of model,
parameters calibration required to obtain acceptable global results, having a great influence on the results and on the demanded com-
for a given test configuration. putational time. In the following, the influence of this parameter is
SOFT parameter
calibration
Standard models
Global
approach Post-damage stress
calibration
Single shell layer
Global behaviour CZONE Model
models
(Abaqus)
Tied contact
Fracture mechanics
Small solid elements
techniques
Rigid Plane
Tied Element
Trigger zone
studied. To model the trigger, the first row of nodes at the tube end condition. Only the relevant distinct features presented in this
was translated some millimetres towards the tube’s exterior and model compared to the stacked shell one described in the previous
the thickness of these elements was reduced (Fig. 6). For the tube section, are exposed here.
and trigger shell elements the material type 131 was used while In the crushing zone model the specimen and the trigger are
the ply properties were modelled through the fabric composite modelled by using a single layer of shell elements. The trigger
global ply model, type 7, using five integration points through elements have a different material model formulation. This formu-
the thickness and four Gaussian point on each element. The fabric lation is progressively activated for the elements located at the
reinforced composite ply is modelled as a homogeneous orthotro- crush front (the region of the tube where the contact with the
pic elastic damaging material, whose properties are degraded on machine plate occurs) when they reach a certain value of a pre-
loading, by micro-cracking prior to ultimate failure. The inter-lam- defined parameter defined by the user. The parameter which
inar material, which connects the shell layers to each other, is controls the activation of the crushing formulation for the subse-
modelled by using one dimensional cohesive element, which ties quent rows of elements is the area ratio of the crushed elements
one shell layer to each other according to a mathematical model (actual area ratio initial one). When the crushing formulation is
which reproduces the inter-laminar material micro-mechanics. In activated, the mechanical behaviour of the element becomes elas-
particular the cohesive interface was modelled by using the mate- tic-perfectly plastic in the axial crushing direction. When the area
rial type 303. A penalty formulation is used to connect the slave ratio reaches a minimum allowed value, this element is eroded
node from one shell layer to the master segment from the other from the model. To avoid numerical problems due to large
shell layer [31]. The failure of the inter-laminar material is based amounts of deformation, the strains are calculated using the coor-
on the elastic energy stored in the crack proposed by Pickett dinates with respect to the moment when the option was triggered
et al. [32] and depends on the elastic energy stored by the crack and not the initial one.
in modes I and II of propagation. The elastic energy limit for both In the model developed the crushing formulation of the trigger
modes is directly computed by the maximum allowed stress de- element is already activated before the initial contact, in order to
fined as an input material property. Two different sliding interface initiate the crushing at the right location. For the other elements
formulations (coulomb friction) are used in order to model the not located at the crush front, the same law described for the pre-
interaction between the model parts that are not permanently con- vious model is still valid.
nected by standard finite element connectivity conditions. The first
sliding interface defines the contact between the tube structure
and the machine plate. The second sliding interface models the 4. Stacked shell model simulations and results
possible self-contact between the different parts of the tube struc-
ture during the crushing process by using the self-contact treat- In this section the simulations performed by using the stacked
ment. Fully-integrated elements are used in order to avoid the shell model and their results are exposed and analysed in details.
problem of hourglass deformation. The nodes located at the base The section was divided in several subsections, each one corre-
of the tube were constrained in all their six degrees of freedom. sponding to a specific parameter analysed.
To the machine plate a constant velocity along the tube axis was
applied. In order to reduce the computational time, using the expli-
4.1. Influence of friction coefficient and element dimension
cit time integration and being the material model used not strain
rate sensitive, the velocity used in the numerical model was in-
The influence of the friction coefficient between the tube and
creased respect to the experimental one to 0.5 m/s.
the machine plate (l1) and of the element dimension was studied.
The tube was modelled first by using three shell layers with a
3.2. The single shell layer model 7.0 mm size mesh. The mesh sizes used in this work are those typ-
ically adopted for simulations of full vehicle impact. Three different
With the single layer models the deformation modes possible values of friction l1 (0.1, 0.2 and 0.3) were examined, while the
are limited, several proposal for enhancements are made in litera- friction for the tube self-contact (l2) was kept constant and equal
ture and are implemented into commercial solvers. to 0.65. These values were chosen basing on coefficients find in
Here the crushing zone formulation was chosen for the single literature [33,34]. The same simulations with the three different
shell layer model. With this formulation the mechanical behaviour l1 values were replied using a finer mesh (element size equal to
changes when an element is assumed to go into crushing 4 mm).
730 B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735
Table 4
7.0 and 4.0 mm model energy results.
Simulation set-up SEA (kJ/kg) Total energy (kJ) Contact energy (kJ) Contact energy (%) Internal energy (kJ) Internal energy (%)
7.0 mm, l1 = 0.1 38.12 1.27 0.23 18.50 1.04 81.50
7.0 mm, l1 = 0.2 43.02 1.43 0.34 23.43 1.09 76.57
7.0 mm, l1 = 0.3 45.29 1.51 0.41 27.09 1.10 72.91
4.0 mm, l1 = 0.1 35.94 1.20 0.20 16.30 1.00 83.70
4.0 mm, l1 = 0.2 47.73 1.59 0.31 19.41 1.28 80.59
4.0 mm, l1 = 0.3 42.25 1.41 0.41 28.85 1.00 71.15
Fig. 7. 7.0 (top) and 4.0 (bottom) mm model with l1 0.1, 0.2 and 0.3 (from the left to the right) at the crush end.
B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735 731
Fig. 10. Load and energy vs. displacement curves, influence of layers.
1.57 times more energy than the one present in the 4 and 5-layers
models. Apparently this is an unexpected result because the 4 and
5-layers tubes have a greater number of cohesive interfaces and
also a larger inter-laminar potential energy. For all the three mor-
phologies basically there was no delamination between the con-
nected shell layers; they remained tied to each other in a firmly
way with no relative detachment. All the initial tied connections
are still present in the model at the end of the crush, and the in-
ter-laminar material did not reach the full failure, the delamination
did not occur. Increasing the numbers of layers the deformations of
tied connections reduced: the smallest relative displacement be-
tween the layers was found for the 5-layers model. This explains
why the energy absorbed by each cohesive interface is much smal-
ler in the 5-layers model if compared to the 3-layers one.
Fig. 9. Models with 3, 4 and 5 layers (from the left to the right) at the crush end.
732 B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735
Table 6
Comparison of energy results for models with different layers.
Model morphology SEA (kJ/kg) Total energy (kJ) Inter-laminar energy (kJ) Inter-laminar energy (%)
3-Layers 46.49 1.55 0.096 6.36
4-Layers 48.43 1.61 0.062 3.85
5-Layers 40.28 1.34 0.061 4.55
until now. Where the lower value of GIu and GIIu (attempt 3 and 4 in
Table 7 Table 7) were used, larger amounts of delamination occurred. The
Inter-laminar fracture energies values examined.
variation of this parameter from its physical value up to one or two
Attempt 1 Attempt 2 Attempt 3 Attempt 4 orders of magnitude higher did not promote a significant variation
GIu 2
(J/m ) 1e5 1e4 1e3 250 of the SEA. An unrealistic amount of delamination occurred be-
GIIu (J/m2) 1e5 1e4 1e3 500 tween the material layers, only for the two higher values this phe-
nomenon did not occur. Using values close to the real ones, the
cohesive interface shows an unrealistically brittle behaviour. The
crushed specimen showed a realistic post-crushing state only
investigated the same phenomena but for woven carbon-epoxy when the strength of the cohesive interface was virtually increased
material. They found that GIIu is 1.4 times higher than GIu . In this by large amounts.
work a value of GIIu two times greater than GIu as did by Greve
and Andrieux in their work [39] is adopted. The other three tested
configurations are an order of magnitude higher than the previous 4.4. Discussion
one.
The other simulation parameters are element dimension of The highest value of SEA obtained in the different simulation
7 mm, and friction coefficient l1 and l2 equal to 0.3 and 0.65 with the stacked shell models (48.43 kJ/kg), is considerably lower
respectively. than the reference one (70 kJ/kg). There is a significant underesti-
Fig. 11 shows for each morphology (3, 4 and 5 layers), the differ- mation of the energy of about 30% due to the fact that some energy
ent tested configurations at the end of the simulation, while the absorption mechanisms present in the real crush event are not ta-
force and energy results are presented in Figs. 12–14 and in ken into account by this kind of model. The overall dissipated fric-
Table 8. tion energy is under estimated. The highest value found for this
The variation of the critical energy release rate did not promote energy component was around 28%. In literature [15,40,41] it has
a different main mode of collapse with respect to the one observed been found that friction have a higher effect on crush test results
Fig. 11. Model comparison at the crush end: from left to the right different energy rates values, from top to bottom different shell layers adopted.
B.P. Bussadori et al. / Composites: Part B 60 (2014) 725–735 733
and it is responsible for more than 50% of the total energy absorbed
during crushing. Analysing the simulations, it is possible to observe
that the stacked shell models developed are capable to take into
account the friction between the fronds and machine plate and
the friction concerning the tube self-contact, but is not capable
to reproduce the debris edge which is formed in the crush zone.
Moreover the relative sliding of two adjacent layers does not pro-
mote any frictional energy dissipation because the shell layers are
not in direct contact, but they make only inter-laminar energy
absorption, due to the deformation of the tied elements.
This kind of model has very sensitivity behaviour: a small
modification in the parameters can completely change the way
in which the tube collapses, and consequently the absorbed
energy.
The underestimation of energy is the limited representativeness
of this kind of model concerning the reproduction of the complex
way in which the real tube collapses.
Fig. 12. 3-Layers model inter-laminar force (F) and energy (E) vs. displacement
curves. 5. Crushing zone model simulations and results
Table 8
Comparison of energy results for models with different inter-laminar fracture energies values considered.
No of layer GIu and GIIu values (J/m2) SEA (kJ/kg) Total energy (kJ) Inter-laminar energy (kJ) Inter-laminar energy (%)
6. Conclusions
Acknowledgements [18] Greco F, Luciano R. A theoretical and numerical stability analysis for composite
micro-structures by using homogenization theory. Compos Part B: Eng
2011;42(3):382–401.
The research was supported by Automobili Lamborghini S.p.A. [19] Sun H, Di S, Zhang N, Pan N, Wu C. Micromechanics of braided composites via
The authors wish to express their appreciation to Luciano De Oto, multivariable FEM. Comput Struct 2003;81(20):2021–7.
[20] Tabiei A, Aminjikarai BS. A strain-rate dependent micro-mechanical model
Attilio Masini and Stefan Klaus Nothdurfter of Lamborghini’s
with progressive post-failure behavior for predicting impact response of
Advanced Composites Research Center for the support and advice. unidirectional composite laminates. Compos Struct 2009;88(1):65–82.
Luigi De Sario and Carlo Vitiello of Lamborghini’s Advanced [21] Crashworthiness Working Group of the CMH-17. Simulation of the quasi-static
crushing of a fabric composite plate. Abaqus Technology Brief; 2011.
Composites Research Center are kindly acknowledged for their
[22] Feindler N, Drechsler K, Doll J. Test method to analyse the energy absorption of
help during production and testing of specimen in the testing composite material using flat coupon testing. In: Proceedings of the 5th
phase. international conference on composites testing and model identification,
Lausanne, 14–16 February, 2011.
[23] Borovkov A, Palmov V, Banichuk N, Saurin V, Barthold F, Stein E. Macro-failure
criterion for the theory of laminated composite structures with free edge
delaminations. Comput Struct 2000;76:195–204.
References [24] Johnson A, Pickett A. Impact and crash modelling of composite structures: a
challenge for damage mechanics. In: Proceedings of the 9th user conference
[1] Smith GF. Design and production of composites in the automotive industry. EURO-PAM, Darmstadt, 7–8 Octoberm, 1999.
Compos Manuf 1990;1(2):112–6. [25] Greve L, Andrieux F. Deformation and failure modelling of high strength
[2] McCarthy RFJ, Haines GH, Newley RA. Polymer composite applications to adhesives for crash simulation. Int J Fract 2007;143(2):143–60.
aerospace equipment. Compos Manuf 1994;11(6):83–9. [26] Tang CY, Tsui CP, Lin W, Uskokovic PS, Wang ZW. Multi-level finite element
[3] Davies G. Material for automobile bodies. Oxford: Elsevier; 2000. analysis for progressive damage behavior of HA/PEEK composite porous
[4] Feraboli P, Masini A. Development of carbon/epoxy structural components for structure. Compos Part B: Eng 2013;55:22–30.
a high performance vehicle. Compos Part B: Eng 2004;35:323–30. [27] Barut A, Madenci E, Tessler A, Starnes Jr JH. A new stiffened shell element for
[5] Kang K-T, Chun H-J, Park J-C, Na W-J, Hong H-T, Hwang I-H. Design of a geometrically nonlinear analysis of composite laminates. Comput Struct
composite roll bar for the improvement of bus rollover crashworthiness. 2000;77:11–40.
Compos Part B: Eng 2012;43(4):1705–13. [28] Mohite PM, Upadhyay CS. Region-by-region modeling of laminated composite
[6] Garner DM, Adams DO. Test methods for composites crashworthiness: a plates. Comput Struct 2007;85:1808–27.
review. J Adv Mater 2008;40(4):5–26. [29] Huang J, Wang X. Numerical and experimental investigations on the axial
[7] Guoxing L, Tongxi Y. Energy absorption of structures and crushing response of composite tubes. Compos Struct 2009;91(2):222–8.
materials. Cambridge: Woodhead Publishing; 2003. [30] Joosten MW, Dutton S, Kelly D, Thomson R. Experimental and numerical
[8] Guimard J-M, Allix O, Pechnik N, Thevenet P. Energetic analysis of investigation of the crushing response of an open section composite energy
fragmentation mechanisms and dynamic delamination modelling in CFRP absorbing element. Compos Struct 2011;93(2):682–9.
composites. Comput Struct 2009;87:1022–32. [31] PAM-CRASH FE code. Engineering systems international, Rungis Cedex.
[9] Jumahat A, Soutis C, Jones FR, Hodzic A. Fracture mechanisms and failure [32] Pickett AK, Johnson AF, Rozicky P. Computational methods for predicting
analysis of carbon fibre/toughened epoxy composites subjected to impact damage in composite structures. Compos Sci Technol
compressive loading. Compos Struct 2010;92(2):295–305. 2001;61(15):2183–92.
[10] Chatiri M, Gull T, Matzenmiller A. An assessment of the new LS-DYNA layered [33] Dhieb H, Buijnsters J, Eddoumy F, Celis J. Surface damage of unidirectional
solid element: basics, patch simulation and its potential for thick composite carbon fiber reinforced epoxy composites under reciprocating sliding in
structures analysis. In: Proceedings of the 7th European LS-DYNA conference, ambient air. Compos Sci Technol 2011;71(15):1769–76.
Salzburg, 14–15 May, 2009. [34] Schön J. Coefficient of friction and wear of a carbon fiber epoxy matrix
[11] Peroni L, Avalle M, Belingardi G. Comparison of the energy absorption composite. Amsterdam: Wear; 2004. p. 395–407.
capability of crash boxes assembled by spot-weld and continuous joining [35] Hull D. A unified approach to progressive crushing of fibre-reinforced
techniques. Int J Impact Eng 2008;5(7):498–511. composite tubes. Compos Sci Technol 1991;40(4):377–421.
[12] Morello L, Rosti Rossini L, Pia G, Tonoli A, editors. The automotive body, vols. I [36] Li JCM, editor. Microstructure and properties of materials. Singapore: World
and II. London: Springer; 2011. Scientific; 2000.
[13] Farley GL, Jones RM. Crushing characteristics of continuous fiber-reinforced [37] Jurf R, Pipes R. Interlaminar fracture of composite materials. J Compos Mater
composite tubes. J Compos Mater 1992;26(1):37–50. 1982;16(5):386–94.
[14] Eshkoor RA, Oshkovr SA, Sulong AB, Zulkifli R, Ariffin AK, Azhari CH. Effect of [38] Nikbakht M, Choupani N. Experimental investigation of mixed-mode fracture
trigger configuration on the crashworthiness characteristics of natural silk behaviour of woven laminated composite. J Mater Sci 2009;44(13):3428–37.
epoxy composite tubes. Compos Part B: Eng 2013;55:5–10. [39] Greve L, Andrieux F. Deformation and failure modelling of high strength
[15] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. The static and adhesives for crash simulation. Int J Fract 2007;143(2):143–60.
dynamic axial crumbling of thin-walled fibreglass composite square tubes. [40] Fairfull A, Hull D. Energy absorption of polymer matrix composite structures:
Compos Part B: Eng 1997;28(4):439–51. frictional effects in structural failure. In: Wierzbicki T, Jones N, editors.
[16] Wagner W. FE – modeling of fiber reinforced polymer structures. In: Structural failure. New York: John Wiley & Sons, Inc.; 1989.
Proceedings of the 5th world congress on computational mechanics, Vienna, [41] Carruthers J, Kettle A, Robinson A. Energy absorption capability and
7–12 July, 2002. crashworthiness of composite material structures: a review. Appl Mech Rev
[17] Feraboli P. Development of a corrugated test specimen for composite material 1998;51(10):635–49.
energy absorption. J Compos Mater 2008;42(3):229–56.