You are on page 1of 10

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

net/publication/245574572

Behaviour of an Impact Attenuator for Formula SAE Car under Dynamic


Loading

Article  in  International Journal of Vehicle Structures and Systems · July 2010


DOI: 10.4273/ijvss.2.2.01

CITATIONS READS

16 5,436

1 author:

Simonetta Boria
University of Camerino
71 PUBLICATIONS   967 CITATIONS   

SEE PROFILE

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

Two-wheel drive electric motorcycle modelling and control View project

AI and fault detection in data of mechanical industry View project

All content following this page was uploaded by Simonetta Boria on 05 June 2014.

The user has requested enhancement of the downloaded file.


S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53
ISSN: 0975-3060 (Print), 0975-3540 (Online)
In tern at iona l Jou rnal o f
doi: 10.4273/ijvss.2.2.01 Vehicle Structures & Systems
© 2010. MechAero Foundation for Technical Research & Education Excellence Available online at www.ijvss.maftree.org

Behaviour of an Impact Attenuator for Formula SAE Car under Dynamic


Loading
Simonetta Boriaa
Department of Mechanical, Nuclear and Production Engineering,
University of Pisa, Pisa, Diotisalvi, 2.
Email: simonetta.boria@aliceposta.it
a
Presently - Department of Mathematics and Informatics,
University of Camerino, Camerino,Madonna delle Carceri, 9.

ABSTRACT:
A Formula SAE car must have structural devices that able to absorb most of the kinetic energy by means of progressive
crushing therefore minimizing the forces and decelerations transferred to the occupant during frontal collisions. This
paper presents the prediction of dynamic behaviour of an impact attenuator for a Formula SAE car under frontal
impact conditions by using numerical and experimental approaches. The impact attenuator construction is a
combination of sandwich panels and aluminium inner sheets. Firstly, the analysis of sandwich structures is carried out
to better understand their behaviour and model them properly in the numerical simulation using LS-DYNA. In order to
obtain the best configuration for the impact attenuator in terms of maximum absorbed energy, minimum deceleration
and weight saving, its length and the number of inner sheets are optimised. Finally, a crash-test is performed on the
impact attenuator to compare the experimental results with the numerical ones. The results obtained show that the
impact attenuator is able to absorb the total impact energy with progressive and plastic deformation to contain the
average deceleration as below 20g.

KEYWORDS:
Formula SAE car, Impact, Finite element simulation, Sandwich structure, Impact attenuator

CITATION:
S. Boria. 2010. Behaviour of an impact attenuator for formula SAE car under dynamic loading, Int. J. Vehicle
Structures & Systems, 2(2), 45-53.

of the system must be of 300kg. The impact velocity


1. Introduction against a rigid barrier must be of 7m/s. During the test,
One of the most important aspects during the design of a the average deceleration of the trolley must not exceed
racing car is to take into account of its behaviour in case 20g with peaks under 40g. The final deformation must
of an impact [1]. Therefore, it is necessary to provide the be limited to the impact attenuator only.
car with structural devices (called impact attenuators or A truncated pyramidal shape, as shown in Fig. 1, has
crash-boxes) that are able to absorb most of the kinetic been chosen for the impact attenuator. This choice is
energy and protect the driver from serious injuries during dictated by two main reasons: the first is due to the
impact. In order to meet the requirements of Formula common shape in the actual race prototypes; the second
SAE competition [2], the impact attenuator has to be is that increasing the cross sectional area prevents the
designed with acceptable dimensions and to meet Eulerian instability during deformation. The angle
specific performances such as the average deceleration between the loading axis and the plates induces the
and deformation during impact. In particular, the design bending leading to the formation of local plastic hinges
and assembly of the impact attenuator has to meet the that are effective to absorb the impact energy in a
following requirements [2]: controlled manner.
• It must be at least 200mm long (along the main
axis), 200mm wide and 100mm high;
• It must be installed at the bulkhead front either by
welding or using at least 4 bolts (M8, Grade 8.8);
• It must not penetrate the bulkhead during impact;
• It must guarantee the occupant safety during off-
axis and off-centre impacts.
The impact attenuator can be installed on the racing
car only after having passed a specific crash-test. During
this dynamic test, the impact attenuator and the front part
of the chassis are attached to a trolley. The total weight Fig. 1: The impact attenuator shape as a hollow beam
45
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

During the deformation process, the lateral panels (4 Several Finite Element (FE) models of the impact
off) are characterized by an alternation of hinges and attenuator have been developed using ANSYS [11] as
straight zones. With the increase of plastic hinges, the pre/post-processor and LS-DYNA [12] as a solver.
crushing of the straight zones come in contact with one The material characteristics are defined by combining
another. This has lead us to use aluminium sandwich the experimental tests on the sandwich structures with
structures [3] with hexagonal core for the impact simple numerical models. Afterwards, the global system
attenuator design. This material is available from E- comprising an impact attenuator, striking mass and
Team at the University of Pisa. Sandwich panels [4] barrier is analysed. In order to obtain the best
store energy in bending and in compression along the configuration, the impact attenuator geometry is
axis of cells more effectively than simple sheets. For the optimised for the required average deceleration and
required performances, the energy to absorb is set as stroke efficiency. Finally, the FE analysis results are
relatively low. Thanks to previous simulations [5-10] of compared with the crash-test data.
impact attenuators made of honeycomb materials. A
well-designed attenuator with two sandwich panels that 2. Material Characterisation
are placed on the lateral sides seems to satisfy the quoted
The sandwich panels are made of aluminium alloy
requirements.
AA5052, while the inner sheets are of aluminium alloy
Few aluminium sheets have been assembled using
AA5005. The material properties are given in Table 1.
rivets between the two panels making the sandwich
structure to work effectively in the ways mentioned as Table 1: Aluminium properties
above. These membranes create higher stiffness areas Type of aluminium AA5052 AA5005
and consequently trigger the instability and folding of Density (kg/m3) 2700 2700
the sandwich structure. After the first impact, the face Elasticity modulus (GPa) 70 70
sheets have no structural task and work as an instability- Poisson’s ratio 0.3 0.3
starter only. A general shape of the chosen impact Yield stress (MPa) 130 41
attenuator is shown in Fig. 2. To avoid the structure Ultimate stress (MPa) 210 124
under non-frontal impact and to form a hinged Elongation at break (%) 9 7
parallelogram, an additional sheet is used at the top of
the impact attenuator. The top sheet is attached directly The analysis of sandwich structures is carried out to
to the lateral sandwich panels and transfers the impact better understand their behaviour and model them
load to both the panels in case of an off-axis collision. properly in the numerical simulation. Out-of-plane
The steps for the assembly of the impact attenuator compressive tests have been performed on the sandwich
are as follows: specimen with quasi-static deformation (0.12 mm/s). The
• Folding two strips of each plate, to create the specimen geometry is described in Table 2, with
surface to apply the rivets; reference to Figs. 4 and 5.
• Drilling holes in the inner sheets and sandwich
panels and then apply rivets between them;
• Bolted joining of the attenuator with the front
bulkhead using four ribbed L-shaped angles as
shown in Fig. 3.

Fig. 4: Sandwich panel geometry

Fig. 2: The impact attenuator using sandwich panels

Fig. 5: Honeycomb core geometry


Table 2: Sandwich specimen dimensions

Dimension Value
Panel geometry axbxh (mm) 70x70x13
Core thickness hc (mm) 12
Face skin thickness tf (mm) 0.5
Cell geometry Sxdxtc (mm) 6.35x3.67x0.0381
Fig. 3: Bolted joining of the attenuator and front bulkhead Cell geometry α (°) 120
46
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

The stress-strain response from the experimental


results of sandwich specimen is shown in Fig. 6. The
initial peak indicates that the sandwich panels are not
pre-crushed. After a large displacement, the cell walls
start touching each other and the core reaches its total
compaction condition. This situation causes a huge
increase in stiffness leading the core to behave nearly as
a solid aluminium block. The energy absorption
capability of the honeycomb structure [13] is determined
by the average stress in the plateau area of the stress-
strain curve.
Fig. 8: Force vs. Displacement of honeycomb for various velocities

3. FE Modelling of Attenuator
3.1. Materials Modelling
Material type 24 (*MAT_PIECEWISE_LINEAR_
PLASTICITY) of the LS-DYNA library is used for the
aluminium sheets and external skins. The material
property is defined using a bilinear true stress-true strain
curve. The ultimate true stress is calculated from the
Fig. 6: Stress vs. Strain under out-of-plane compression
ultimate engineering stress (σri ) using:

The crush strength of aluminium honeycomb (


σ rv = σ ri 1 + ε ri / 100 ) (1)
increases linearly [14] with the impact velocity. This is Where ν is the Poisson’s ratio. The ultimate true strain is
due to three possible causes: calculated from the ultimate engineering strain (εri)
• Compression and temperature increase due to using:
presence of air in the cells;
• Effects due to inertia; (
ε rv = ln 1 + ε ri / 100 ) (2)
• Strain rate effects of aluminium. The tangent modulus of the true stress-true strain curve
Compression of air within the cells is negligible [6] is calculated using:
and is not solely affect the measured response. The ( )(
ETAN = σ rv − σ y / ε rv − ε y ) (3)
inertia effects appear to be sensitive [5] to the crush
strength, but their nature is not yet completely Where εy is the strain at yield stress (σy) and is calculated
understood. Therefore, the variation of crush strength of from the elastic modulus (E) as follows:
the sandwich panel due to inertia is ignored in this study. εy =σy /E (4)
Aluminium alloys exhibit the strain rate effects only at
extremely high impact velocity. As the impact velocity is The final failure of inner sheets and face skins during the
7 m/s, it is hard to achieve such high strain rates into a impact loading is characterised by a failure criterion.
large proportion of the sandwich panels. However, to Failure means a loss of stiffness and is detected when the
validate this assumption a simple FE model representing elongation of elements in any direction reaches the
a portion of honeycomb cell as shown in Fig. 7(a) is allowable strain.
analysed [6] for compression along its main axis. The Different approaches [9] exist for the FE modelling
stages of crushing of honeycomb cell for various impact of honeycomb structures. They depend on the
velocities from the FE simulation are shown in Figs. idealisation, loading, computational cost, and accuracy
7(b). As expected, the crushing behaviour exhibits of the results. A detailed representation of the hexagonal
multiple crushing of the cell walls along its main axis. cells can predict the deformation of cell wall reasonably
Fig. 8 shows the impact force response of the well, but it is unsuitable for large-scale FE models. The
honeycomb cell for the considered impact velocities. The core can also be simplified as an orthotropic continuum
force is higher for 36m/s velocity. The overall response based on its effective material properties [7, 15]. The in-
shows a poor sensitivity with the strain rate. plane properties of the honeycomb core are much lower
than the out-of-plane ones. Experimental tests such as
shear, three-point bending, and buckling under in-plane
compression are undertaken to characterise the in-plane
orthotropic properties. The honeycomb core behaviour
under compression is shown in Fig. 9. This
simplification is commonly used for the behaviour of a
sandwich panel, where there are neither fluctuations nor
initial peak in the stress-strain curve exist. This
simplification also reduces the calculation time and
implies that every element in the FE model deforms
Fig. 7: Honeycomb: a) Portion of the cell analysed, b) Deformation
simultaneously.
47
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

Gauss integration is used for the convergence. This rule


allows a faster convergence than the trapezoidal rule and
is accurate enough for this study. The striking mass and
honeycomb core are idealised using 8-noded SOLID164
type elements. These elements have been used as fully
integrated to avoid the any hourglass phenomena during
numerical solution.
As the failure of the bond between the core and the
outer face skins precedes the failure of honeycomb core
in traction, the tensile rigidity between the core and face
skins has to be correctly represented in the FE model.
Fig. 9: Stress-Strain curve for the honeycomb under compression
Hence, the honeycomb is considered as a collection of
hexagonal prisms. Each prism is idealised using
When the honeycomb material model 26 (*MAT_ LINK167 type elements whose area represents the
HONEYCOMB) of LS-DYNA material library is used, a transverse section of the cell. LINK167 element is an
lack of stiffness along the cells axis is observed for the explicit tension-only connector that behaves like a cable
tension loading. In order to correctly model the with no compressive or bending stiffness. It should be
honeycomb tensile rigidity, LINK type elements are used noted that the LINK167 element is not useful to
between the core and the face skins. The present analysis represent a stretched honeycomb core.
is limited for predicting the global deformation of the In the FE model, there are as many LINK167 type
sandwich structure and not to the local deformation of elements as the number of nodes on the face skins. An
honeycomb. equivalent area is assigned to the LINK167 element such
The car body and the barrier are idealised as rigid. that the total stiffness of the link elements equals the
During the crash-test, the impact attenuator hits a rigid total stiffness of prisms using:
wall. Hence, the assumption of striking mass being rigid
Aeq = AHEX N HEX / N N (5)
leads to a conservative simulation and allows the
attenuator’s main function as to absorb all the kinetic Where NN is the number of nodes on the skin, AHEX is
energy by cancelling any time-delay during loading. the area of the transverse section of the prism and is
calculated from the honeycomb core geometry S and tc
3.2. Geometry Idealisation
(see Table 2 and Fig. 5) using:
The total system consists of an impact attenuator, a
striking mass and a barrier wall. During the assembly of AHEX = 4St c / 3 (6)
the impact attenuator, care is taken to ensure that the NHEX is the number of hexagonal prisms filling the area
sandwich panels are not pre-crushed. The upper sheets of the skin (AS) obtained from:
orthogonal to the inner sheets and the sandwich panels
are not idealised. Rivets are not idealised in the FE (
N HEX ≈ 2 AS S 2 3 ) (7)
model due to a computational overload. Also, the failure
A failure criterion for the link elements is defined such
of face skins of the sandwich panel will precede the
that they break when their tensile stress reaches the
failure of the rivets. Therefore, the elements representing
ultimate stress of the adhesive bond (15.5MPa) between
the inner sheets are directly connected to the skin nodes
the core and face skins.
at this interface. In order to reduce computation time,
only a quarter of the total system (see Fig. 10) is 3.3. Frictional Contacts Modelling
simulated using appropriate boundary conditions. There exist two contacts namely inner sheets to inner
sheets and inner sheets to the wall during crash. The
material used for the wall is not known as a priori.
Striking mass
Typical friction coefficient between two aluminium
Attenuator surfaces is in the range of 1.2-1.7 [16, 17]. The explicit
Barrier
code - LSDYNA does not accept a coefficient of friction
which is greater than 1. Hence, a coefficient of friction
of 0.9 is used in the FE simulation. If the wall is made of
steel, the coefficient of friction will be 0.47 and 0.6 for
the kinematic and static conditions respectively.
Simulations are carried out with these coefficients of
friction and 0.9 as the coefficient of friction for both
kinematic and static conditions. A comparison of force
Fig. 10: FE model of total system for dynamic analysis
vs. time response for these two analyses is shown in Fig.
11. The differences in Force vs. Time response due to
The sandwich face skins, inner sheets, and the wall this change in the coefficient of friction is not noticeably
are idealised using 4-noded SHELL163 type elements. great (peak force is within 3%). The friction between the
These elements have both the bending and membrane wall and the attenuator increases due to the heat during
deformation capabilities. They have been used with a impact. Hence, all the coefficients of friction have been
fully-integrated Belytschko-Tsay element formulation assumed to be 0.9 for the worst impact conditions.
with 5 integration points through the thickness. The
48
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

Fig. 11: Sensitivity of Force vs. Time to coefficient of friction


Fig. 12: Sensitivity of deceleration to length and inner sheets
3.4. Loads and Boundary Conditions
The following degrees of freedom of the nodes at the
symmetry planes have been constrained:
• The displacement along the orthogonal directions
to the symmetry plane, and
• The rotations around all the directions in the
symmetry plane.
The nodes at the interface of upper sheet to the
sandwich panels are constrained such that the onset of
plastic hinges is not allowed in bending. Stiffness and
non-failure of bolted joint conditions are plausible
during impact. Therefore, the base of attenuator (close to
the barrier) is fully constrained such that the nodes of the
sandwich panels at the base can move together with the
motion of the vehicle. The barrier is constrained in all
degrees of freedom. The displacement of striking mass
along the forward direction is permitted only with a
constant initial impact velocity of 7m/s. Fig. 13: Sensitivity of stroke efficiency to length and inner sheets
In order to run the explicit dynamic simulation using
LS-DYNA, it is necessary to fix some control
parameters such as termination time (0.08s), time-step,
and output requests.

4. FE Simulation Results
In order to obtain the best configuration for the impact
attenuator, its length and the number of inner sheets are
optimised [10]. The number of inner sheets is varied
from 4 to 7 for varied impact attenuator lengths of 300,
350 and 400mm. Figs. 12, 13, and 14 respectively show
the sensitivity of deceleration, stroke efficiency (ratio of
final shortening over initial length), and impact force to
the variations in the length and the number of inner
sheets of the attenuator. The attenuators of 300mm long
with 4 or 5 inner sheets show high peaks in the
deceleration response. The same trend is observed for
the attenuators of 350 and 400mm long with 4 inner
sheets. The impact attenuators of 350 and 400mm long
with 5 inner sheets show good performances. The impact Fig. 14: Sensitivity of impact force to length and inner sheets
attenuators with 6 inner sheets seem to be the best choice The attenuator of 300mm long with one inner sheet
based on the deceleration and force vs. displacement is analysed and the results of the configuration with 6
(measure of energy absorption capacity) responses. inner sheets is compared in Table 3. When the number of

49
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

inner sheets is increased, an increase in the absorbed


energy and a reduction in the deceleration are observed.
The impact attenuator of 300mm long with 6 inner
sheets is regarded as the optimum configuration in terms
of the required average deceleration (less than 20g) and
stroke efficiency (less than 1). This configuration has the
least volume and weighs about 3kg. The final assembly
and optimum geometry of the impact attenuator are
shown in Figs. 15 and 16 respectively. The spacing
between the inner sheets is 42mm.
Table 3: Sensitivity of the number of inner sheets on crash
parameters for 300mm length attenuator

Characteristics 1 sheet 6 sheets


Max. shortening (m) 0.296 0.25
Max. deceleration (g) 340 18.7
Max. force (N) 280968 14093
Absorbed energy (J) 1397.7 1813.1

Fig. 16: Attenuator geometry for the optimum configuration

The sequence of attenuator deformation for various


time steps of the FE simulation under impact loading is
shown in Fig. 17. As expected, the crushing of the wall
and the contact between the inner sheets during the
stages of simulation have shown considerable crash
Fig. 15: Attenuator assembly with 6 inner sheets performance.

Fig. 17: Simulated crash performance of the impact attenuator

7m/s. A rapid unfastening system, a piston driven by


5. Crash-Test Results compressed air, is used to drop the striking mass. Two
The experimental setup to study the behaviour of impact steel ropes are used to guide the striking mass. The
attenuator under dynamic loading is schematically accelerometers are positioned at the top of the striking
shown in Fig. 18. A striking mass of 300kg is dropped mass as shown in Fig. 19. The base of the impact
from a height of 2.5m, to generate an impact velocity of attenuator is fastened to a steel plate on the concrete by
using four L-angles as shown in Fig. 20.
50
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

Fig. 21: Impact attenuator after the crash-test

The acceleration response with a sampling


frequency of 5kHz is shown in Fig. 22. It is evident that
the impact between 2.19s and 2.26s. The deceleration
peaks greater than 40g have been filtered with the
Channel Filter Class (CFC) 60 in accordance with the B-
3.21.2 rule of the Formula SAE car competition [2]. The
filtered data, as shown in Fig. 23, has a deceleration peak
of 21.2g. The average deceleration of the raw data
during the time interval of 2.194 – 2.266s is 9.81g.
Hence, the attenuator design satisfies the required
deceleration levels (less than 40g).

Fig. 18: Experimental setup of the crash-test

Fig. 19: Location of the accelerometers on the striking mass


Fig. 22: Un-filtered acceleration response at 5kHz

Fig. 20: Impact attenuator before the crash-test

The crash-test is performed in compliance with the


Formula SAE car rules. During the impact loading, the
inner sheets and wall of the attenuator contact one
another and then lead to a final deformation as shown in Fig. 23: Comparison of raw and filtered acceleration response
Fig. 21 where the inner sheets are compacted together.
51
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

The experimental impact velocity vb can be checked


using the average acceleration a and the total impact
time using:
vb = a(t e − t b ) ≈ 6.9m/s (8)
Where tb and te are the beginning and end time of the
impact respectively. The length (lf) of the impact
attenuator after the impact can be estimated as:
l f = li − vb (t e − t b ) + 0.5a(t e − t b ) 2 ≈ 0.06m (9)
Where li is the initial length of the attenuator. From the
deceleration response, as shown in Fig. 23, is integrated
by two steps to obtain the velocity (Fig. 24) and the
displacement (Fig. 25) responses. The velocity and
displacement responses from the crash-test data have
shown good agreement with the FE simulation results.

Fig. 26: Impact force response – FE simulation vs. Crash-test

Table 4 shows a comparison of the crash parameters


from the experimental and numerical results. Despite the
simplifications adopted in the FE model, the average
deceleration from simulation is higher (~23%) than the
experimental one. The max. force and the absorbed
energy from the FE analysis correlate well with the
experimental ones. The FE analysis is able to reproduce
the progressive crushing of the impact attenuator. The
crash parameters from the FE simulation correlate
reasonably well (~17%) with the crash-test results.
Table 4: Crash characteristics – FE simulation vs. Crash-test

Fig. 24: Velocity response – FE simulation vs. Crash-test Diff.


Characteristics Simulation Test
(%)
Average deceleration (g) 12.1 9.81 23.3
Residual length (m) 0.05 0.06 16.6
Final shortening (m) 0.25 0.24 4.2
Stroke efficiency 0.83 0.8 3.7
Max. force (N) 56372* 62388 9.6
Absorbed energy (J) 7252.4* 7141.5 1.5
*As only a quarter of the total system is idealised in the FE
model, the peak force and the absorbed energy for the total
system have been factored by 4.

6. Conclusions
The behaviour of an impact attenuator for Formula SAE
car has been described using detailed FE simulation and
a crash-test. The number of inner sheets and the
attenuator length are optimised for good containment
Fig. 25: Displacement response – FE simulation vs. Crash-test and higher energy absorption during the impact. The
The response of the force transmitted from the crush behaviour and the predicted crash parameters from
striking mass to the impact attenuator during impact is the FE simulation have shown a reasonable agreement
obtained by multiplying the acceleration response with with the crash-test results. Based on these results, it is
the striking mass. Fig. 26 shows force-displacement demonstrated that the proposed impact attenuator design
response from the FE simulation experimental results. has satisfied the requirements of Formula SAE rules.
The energy absorption capacity is characterised by the
area under the force vs. displacement curve. The ACKNOWLEDGEMENTS:
variation in the FE simulation data is due to the simple The author would like to thank the Prof. Ing. Giuseppe
idealisation of the impact attenuator and the maturity of Forasassi, Technicians at the Scalbatraio Laboratory, and
the contact characteristics. The first peak in the the E-Team Squadra Corse at the University of Pisa for
experimental data is dominated by the buckling of the their collaboration in this work.
top plate. This peak is not observed in the numerical
results as the top plate is not idealised in the FE model.
52
S. Boria. 2010. Int. J. Vehicle Structures & Systems, 2(2), 45-53

REFERENCES: [9] L. Aktay, A.F. Johnson, and B.-H. Kröplin. 2008.


[1] A.N. Mellor. 2002. Impact testing in formula one, Int. J. Numerical modelling of honeycomb core crush behaviour,
Crashworthiness, 7(4), 475-486. Engg. Fracture Mechanics, 75, 2616-2630.
[2] Formula SAE Rules. 2009. SAE International. [10] S. Boria and G. Forasassi. 2009. Numerical simulation of
crash-test for a Formula SAE car, Proc. 21st Int. Tech.
[3] D. Zenkert. 1997. The Handbook of Sandwich Conf. Enhanced Safety of Vehicles, Stuttgart, Germany.
Construction, EMAS Publishing.
[11] LS-DYNA Keyword Users Manual. 2007. Version 9.71.
[4] W. Goldsmith and J. Sackman. 1991. Energy absorption
by sandwich plates: A topic in crashworthiness, ASME- [12] ANSYS Users Manual. 2007. Version 10.
Crashworthiness and Occupant Protection in [13] L.J. Gibson and K. Ashby. 1999. Cellular Solids:
Transportation Systems, AMD-126/BED-19. Structure and Properties, Cambridge University Press.
[5] H. Zhao and G. Gary. 1998. Crushing behaviour of [14] E. Wu and W-S. Jiang. 1997. Axial crush of metallic
aluminium honeycomb under impact loading, Int. J. honeycombs, Int. J. Impact Engg., 19, 439-456.
Impact Engg., 21, 827-836. [15] S. Heimbs, T. Mehrens, P. Middendorf, M. Maier, and A.
[6] M. Yamashita and M. Gotoh. 2005. Impact behavior of Schumacher. 2007. Numerical determination of the
honeycomb structures with various cell specifications – nonlinear effective mechanical properties of folded core
numerical simulation and experiment, Int. J. Impact structures for aircraft sandwich panels, Proc. 6th European
Engg., 32, 618-630 LS-DYNA Users’ Conf., Gothenburg, Sweden.
[7] S. Heimbs, P. Middendorf, and M. Maier. 2006. [16] R.T. Barrett. 1990. Fastener Design Manual. NASA
Honeycomb sandwich material modelling for dynamic Reference Publication, 1228.
simulations of aircraft interior components, Proc. 9th Int. [17] L.S. Lerner. 1996. Physics for Scientists and Engineers,
LS-DYNA Users’ Conf., Dearborn, USA. Jones and Bartlett Publishers.
[8] S.T. Hong, J. Pan, T. Tyan, and P. Prasad. 2008. Dynamic
crush behaviours of aluminium honeycomb under
compression dominant inclined loads, Int. J. Plasticity, 24,
89-117.

53

View publication stats

You might also like