Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Composite Structures
journal homepage: www.elsevier.com/locate/compstruct
Department of Mechanics (DIMEC), Politecnico di Torino, Duca degli Abruzzi, 24, 10129 Torino, Italy
School of Science and Technology, University of Camerino, Madonna delle Carceri, 9, 62032 Camerino, Italy
a r t i c l e
i n f o
Article history:
Available online 2 September 2011
Keywords:
Lightweight design
CFRP tubes
Impact attenuator
Crashworthiness
a b s t r a c t
Carbon bre composites have shown to be able to perform extremely well in the case of a crash and are
being used to manufacture dedicated energy-absorbing components, both in the motor sport world and
in constructions of aerospace engineering. While in metallic structures the energy absorption is achieved
by plastic deformation, in composite ones it relies on the material diffuse fracture. The design of composite parts should provide stable, regular and controlled dissipation of kinetic energy in order to keep the
deceleration level as least as possible. That is possible only after detailed analytical, experimental and
numerical analysis of the structural crashworthiness.
This paper is presenting the steps to follow in order to design specic lightweight impact attenuators.
Only after having characterised the composite material to use, it is possible to model and realise simple
CFRP tubular structures through mathematical formulation and explicit FE code LS-DYNA. Also, experimental dynamic tests are performed by use of a drop weight test machine.
Achieving a good agreement of the results in previously mentioned analyses, follows to the design of
impact attenuator with a more complex geometry, as a composite nose cone of the Formula SAE racing
car. In particular, the quasi-static test is performed and reported together with numerical simulation of
dynamic stroke. In order to initialize the collapse in a stable way, the design of the composite impact
attenuator has been completed with a trigger which is consisted of a very simple smoothing (progressive
reduction) of the wall thickness. Initial requirements were set in accordance with the 2008 Formula SAE
rules and they were satised with the nal conguration both in experimental and numerical crash
analysis.
2011 Elsevier Ltd. All rights reserved.
1. Introduction
In order to ensure the drivers safety in case of high-speed
crashes, special impact structures are designed to absorb the race
cars kinetic energy and limit the deceleration acting on the human
body. In current automotive development, in order to improve
their crashworthiness and increase stiffness to weight ratio, composite material is introduced with the scope of optimisation of
car body components. In fact, composites have a greater capacity
to absorb energy compared to metals, mainly due to the different
modes of failure that govern energy absorption.
Crash investigations on composite structures reported in the literature are mainly based on experimental test analysis of small
plates submitted to bending impact and on simple bars, of circular
Corresponding author.
E-mail addresses: jovan.obradovic@polito.it (J. Obradovic), simonetta.boria@unicam.it (S. Boria), giovanni.belingardi@polito.it (G. Belingardi).
URLs: http://www.dimec.polito.it (J. Obradovic), http://www.sst.unicam.it (S.
Boria), http://www.dimec.polito.it (G. Belingardi).
0263-8223/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2011.08.005
or rectangular cross section, of prismatic or tapered shape, submitted to axial impact [17]. Also, a couple of analytical models have
been proposed to predict the energy absorption characteristics of
thin-walled tubular structures [1114]. Furthermore, some studies
can be found in the literature concerning composite crash-boxes
for automotive applications, but they are still few and do not cover
all aspects of composite structure modelling [810,1619].
An important aspect of crashworthiness research is the validation of analytical and numerical models for accurate simulation
of structural response to crash impacts. Indeed, they constitute
the necessary tools for the designer to study the response of the
specic structures to dynamic crash loads, to predict global
response to impact, to estimate probability of injury and to evaluate numerous crash scenarios, not economically feasible with full
scale crash testing.
This study covers the steps to follow during the design of a specic impact attenuator. After the mechanical characterisation of
the CFRP material, it is possible to calibrate the numerical material
model, to properly design and to perform experimental tests on
thin-walled tubular structures. The good correlation between
424
Nomenclature
gc
Faverage
Fmax
raverage
SEA
t
r0
rm
RB
H
t1
t2
LC
ls1
crash load efciency (dened as the ratio between average and peak load)
average load
peak load
axial average strength
specic energy absorption
wall thickness
ultimate stress in uni-axial tension
matrix shear strength
mean radius in B
axial length of tube
thickness of the plies bending outside the shell radius
thickness of the plies bending inside the shell radius
length of central crack
side length of the wedge inscribed to the external bent
frond
ls2
/
a1
a2
l1
l2
Wb
Wh
Wm
Wf
WT
s
Table 1
Material properties for used prepreg.
Property
Composite CFS003/LTM25
carbon-epoxy fabric prepreg
Density
Young modulus in bre longitudinal direction
Young modulus in transverse direction
Poissons ratio
Shear modulus12
Shear modulus23
Shear modulus31
Longitudinal tensile strength
Transverse tensile strength
Longitudinal compressive strength
Transverse compressive strength
Compressive strength in direction 12
In plane shear strength
425
Wb
p
2
p
r0 t21 a1 RB ls1 sin / r0 t22 a2 RB ls2 sin /
2
First zone length = 30 mm, laminate thickness = 0.8 mm simulated with four plies.
426
ls1
W m 2prm ls1 sin a1 t1 RB
sin / 2prm ls2
2
ls2
sin a2 t 2 RB
sin /
2
WT
W f s l1
r0 ls1 sin a1 l2 cos a1
s
r0 ls2 sin a2 l2 cos a2 l2 r0 ls1 ls2
For the impact attenuator has been followed the same procedure that was used for cylindrical tubes. It is implemented the
rigid wall planar moving forces card with a mass of 300 kg,
which is presenting the obstacle that is impacting the attenuator
with an initial velocity of 7 m/s. The support of the attenuator is
modelled as rigid wall planar, with prescribed friction coefcient
equal to 0.4.
Particular attention was given to the material denition of composites. In particular, for the LS-DYNA library the material types 54
and 55, that implement the Chang-Chang and Tsay-Wu criteria,
were used for modelling of composite tubes and impact attenuator,
thanks to their ability to give a numerical behaviour near to the
experimental ones. To avoid ductile behaviour with folding, it is
important to change the element strength at some point of the collapse evolution. This is obtained thanks to a time-step failure
parameter (TFAIL), that takes a value of 0.8.
5. Experimental quasi-static and dynamic tests
Table 2
Experimental crash-tests results.
Diameter
(mm)
Thickness
(mm)
Impact velocity
(m/s)
Peak
deceleration (g)
Average
deceleration (g)
Absorbed
energy (kJ)
Residual height
(mm)
gc (Faverage/
raverage
Fmax)
(MPa)
SEA (kJ/
kg)
80
80
80
50
50
50
2.5
2
1.5
2.5
2
1.5
3.7
3.6
4
3.9
4.2
4.2
23.10
18.81
11.01
16.68
10.62
7.01
16.18
14.51
7.55
12.59
8.46
5.14
2.02
1.96
2.35
2.24
2.59
2.59
165
156
115
146
120
73
0.70
0.77
0.68
0.75
0.79
0.73
73.41
82.83
57.80
89.84
76.06
62.19
65
65
54
76
76
63
427
The tested CFRP tubes have absorbed impact energy by a progressive crushing process, through various combinations of failure
mechanisms (Fig. 6). In particular the specimens have shown two
different crushing modes: splaying with axial splitting and fragmentation with debris compacted inside the tube and preventing
further crushing. However, in all congurations, the residual part
of tubes results perfectly integer.
5.2. Experimental quasi-static tests on impact attenuator
The two competing nose cone designs, as previously discussed,
are tested in quasi-static axial compression. The tests are con-
Fig. 5. Velocity and stroke vs. time (a), measured and theoretical absorbed energy vs. stroke (b).
Fig. 6. Failure modes of the cylindrical tubes tested: (a) diameter 80 mm, (b) diameter 50 mm.
Fig. 7. Quasi-static test: Initial position of attenuator (left) and after the performed test (right).
428
ducted with an electromechanical Zwick Z100 machine at a crosshead speed of 0.5 mm/s. The compression is performed with upper
moving plate and the load is measured by the load cell.
The thinner nose cone (version 1) is tested up to the displacement in axial direction of 100 mm. The structure with thicker wall
(version 2) is tested up to the displacement of 160 mm. The thinner
impact attenuator, at two subsequent stages of the test i.e. before
quasi-static test and after the displacement of 100 mm, is shown
in Fig. 7.
The comparison of force vs. displacement diagram of two different impact attenuator design versions is shown in Fig. 8. It is evident that the trend is the same and the desired smooth curve
behaviour is obtained for both cases. After the initial peak, it is visible the second force peak at the displacement of 30 mm and the
third one at the displacement of about 100 mm. This is because
of the wall thickness change, which is shown in Fig. 2. The wall
thickness change is very sensitive parameter and in order to
decrease force peaks, the difference in ply thickness between different impact attenuator zones should be as less as possible.
6. Comparison between analytical, numerical and experimental
results
6.1. Cylindrical tubes
After having performed a mesh sensitivity analysis, a uniform
mesh with elements of about 2 mm per side has been chosen. A
sequence of the deformed shapes of a fabric tube (diameter
50 mm and thickness 2 mm) at different simulation times, is
shown in Fig. 9.
The numerical and experimental deceleration vs. shortening
curves for the tube with diameter 50 mm and thickness 2 mm
are compared in Fig. 10, while some crash parameters obtained
from experimental tests are reported together with the analytical
and numerical results in Table 3 for all specimens. As the experi-
mental data, also the numerical ones are ltered with the same lter CFC60.
From the comparison of the results it is evident that the analytical model and FE model are able to simulate the brittle composite
material behaviour with material separation from the specimens
and consequently approximate the absorbed energy and the actual
crushing with good accuracy.
In all cases the difference between analytical, numerical and
experimental results is lower than 12%. The discrepancy between
analytical results and experimental ones is mainly due to the difculty to accurately model the phenomenon of energy absorption
for composite materials. Instead the difference between numerical
results and experimental ones is mainly due to the complexity to
simulate the initialization of the fragmentation phase and defects
in the construction of the tubes.
6.2. Impact attenuator
In the diagram of the Fig. 11, it is shown the evolution of decelerations during the time, calculated by numerical analysis. The rst
peak of the impact attenuator is due to the initialization of structural collapse. The decrease of the initial peak value can be
obtained by a proper modication of impact attenuator structure
that includes crush initialisation triggers. After the initial peak, a
nearly at diagram of the deceleration is obtained as expected.
Also, it is recommended in SAE rules that average deceleration
of the vehicle must not exceed 20 g. This constrain is completely
satised in numerical circumstances.
The force is directly dependent on acceleration, and thats why
the trend of forcedisplacement curves is almost the same. In
Fig. 12 the comparison of force vs. displacement curves is presented, obtained by experimental quasi static tests and crash
dynamic numerical simulations carried out by LS-DYNA code. It
is well visible that the rst critical peak is mostly a numerical problem that does not appear in the experiment. During the quasi static
test, the rst maximum peak of the force is 52 kN. A nearly at diagram of the impact force is obtained, which means a nearly constant value of the deceleration.
Fig. 9. Sequence of deformation and Von Mises stress for tube with 50 mm diameter and 2 mm thickness.
429
Analytical model
FE model
Diameter
thickness (mm)
Absorbed
energy (kJ)
Crushing
(mm)
Absorbed
energy (kJ)
802.5
802
801.5
502.5
502
501.5
2.05
2.05
2.35
2.22
2.61
2.58
31
39
87
49
73
140
2.01
2.01
2.01
2.01
2.47
2.01
Experimental test
Error% on crushing
Crushing
(mm)
Absorbed
energy (kJ)
Crushing
(mm)
FE vs. experimental
difference%
38
45
87
53
82
132
2.02
1.96
2.35
2.24
2.59
2.59
35
44
85
54
80
127
11.4
11.4
2.3
9.2
8.7
10.2
8.5
2.2
2.3
1.8
2.5
3.9
7. Conclusions
Fig. 12. Comparison of force vs. displacement curves obtained by LS-DYNA code
and quasi-static experiment.
Fig. 13. Flexion of the attenuators wall subjected to quasi-static compression test
after the displacement of 100 mm-comparison of experimental and numerical
results.
[1] Thornton PH, Edwards PJ. Energy absorption in composite tubes. J Compos
Mater 1982;16:52145.
430
[12] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. Energy
absorption capability of berglass composite square frusta subjected to
static and dynamic axial collapse. Thin-Wall Struct 1996;25(4):26995.
[13] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. Analytical
modeling of the static and dynamic axial collapse of thin-walled berglass
composite conical shells. Int J Impact Eng 1997;56:47792.
[14] Gupta NK, Velmurugan R. Analysis of polyester and epoxy composite shells
subjected to axial crushing. Int J Crashworthiness 1999;5(3):33344.
[15] Formula SAE rules; 2008.
[16] Heimbs S, Strobl F, Middendorf P, Gardner S, Eddington B, Key J. Crash
simulation of an F1 racing car front impact structure. In: 7th European Ls-Dyna
conference; 2009.
[17] Belingardi G, Obradovic J. Design of the impact attenuator for a formula
student racing car: numerical simulation of the impact crash test. J Serbian Soc
Comput Mech 2010;4(1):5265. ISSN: 1820-6530.
[18] Bisagni C, Di Pietro G, Fraschini L, Terletti D. Progressive crushing of berreinforced composite structural components of a formula one racing car.
Compos Struct 2005;68:491503.
[19] Belingardi G, Obradovic J. Crash analysis of composite sacricial structure for
racing car. Mobility Vehicles Mech 2011;37:15. ISSN: 1450-5304.
[20] Belingardi G, Chiandussi G. Vehicle crashworthiness design general
principles and potentialities of composite material structures. In: Abrate S,
editor. Impact engineering of composite structures. Wien: Springer; 2011. p.
193264.