You are on page 1of 11

Materials and Design 30 (2009) 2393–2403

Contents lists available at ScienceDirect

Materials and Design


journal homepage: www.elsevier.com/locate/matdes

Crushing response of foam-filled conical tubes under quasi-static axial loading


Z. Ahmad *, D.P. Thambiratnam
Faculty of Built Environment and Engineering, Queensland University of Technology, GPO Box 2434, 2 George Street, Brisbane, QLD 4001, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Foam-filled thin-walled tubes are considered to be desirable energy absorbers under axial loading due to
Received 15 September 2008 their higher energy absorption compared with empty tubes. This paper treats the axial crushing and
Accepted 18 October 2008 energy absorption response of foam-filled conical tubes under quasi-static axial loading, using non-linear
Available online 30 October 2008
finite element models. Influence of important parameters such as wall thickness, semi-apical angle and
density of foam filler was investigated and the results highlight the advantages of using foam-filled con-
Keywords: ical tubes as energy absorber. Results also indicate that the crush and energy absorption performances of
Conical tube
conical tubes are significantly enhanced by foam filling. The primary outcome of the study is new
Foam-filled
Axial crushing
research information and development of empirical relations which will facilitate the design of foam-
Energy absorption filled conical tubes as energy absorbers in impact applications.
Finite element Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction section and thus a conical tube has greater benefit than other ta-
pered tubes. In regard to the study of foam-filled thin-walled tubes,
Adverse effects of impact can be mitigated by absorbing the im- much of the focus has been on foam-filled thin-walled circular and
pact energy in a controlled manner in order to protect the structure square tubes under axial loading. Since energy absorption capacity
under consideration. Thin-walled tubes of various geometrical of a foam-filled tube is dependent on foam density, high-density
shapes such as circular, square, rectangular and hat-sections have foam needs to be used if high-energy absorbing capacity is desir-
been known to be excellent impact energy absorbers because of able. Nevertheless filling the circular tubes with a high-foam den-
their progressive axial folding. Tapered tubes also prove advanta- sity of 0.7 g/cm3 leads to global Euler buckling which is an
geous for impact and energy absorption application due to their undesirable buckling mode as reported in [5].
desirable response and energy absorption capacity [1]. Due to Relatively few studies have been reported on the energy
high-energy absorption efficiency under compressive deformation, absorption performance of foam-filled tapered tubes. Some useful
light weight foam material has also become increasingly attractive experimental and numerical investigations have been carried out
in impact applications. A combination of thin-walled tube and light to establish the load–deflection and collapse mode responses of
weight foam is therefore anticipated to enhance the energy absorp- foam-filled tapered tubes [4,9,10]. Research on the crush response
tion capacity, as evidenced by initial studies [2–5]. Furthermore, of a foam-filled rectangular tapered tube was pioneered by Reid
there are also other benefits in addition to the primary one of and Reddy [9]. More recently, Mirfendereski et al. [10] have
increasing the energy absorbing capacity. It is evident that the numerically carried out a parametric study on foam-filled tapered
crushing loads of foam-filled tubes are higher than the sum of rectangular tubes under static and dynamic loadings. Those studies
the separate crushing loads of foam and tube due the interaction showed that the crush load of an empty rectangular tube can be in-
effect between the foam filler and tube wall and it is a function creased by foam filling. Also Gupta and Velmurugan [4] experi-
of foam density [2,6]. mentally found that the energy absorption capacity of a foam-
Considering empty tubes, a tapered thin-walled tube is prefer- filled conical composite tube is significantly greater than that of
able to a straight tube as it has a stable load–deflection response an empty conical tube. As a matter of fact, this is the only research
and minimizes the chances of collapse by buckling in the Euler conducted on the energy absorption of foam-filled conical tube
mode as found in previous studies of tapered rectangular and con- using composite material in which a different axial collapse re-
ical tubes at a semi-apical angle range of 5°–15° [1,7]. As pointed sponse and failure mode were observed. Above all, it is envisaged
out by Mamalis et al. [8], a circular cross section is more effective that by filling a metallic conical tube with foam, the energy absorp-
in progressive crushing compared to a square or rectangular cross tion and crushing resistance of the conical tube can be substan-
tially improved.
Despite its potential to be an efficient energy absorber, very
* Corresponding author. Tel.: +61 7 3138 1824; fax: +61 7 3138 1515.
E-mail address: z.ahmad@qut.edu.au (Z. Ahmad). limited studies have been conducted on energy absorption

0261-3069/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2008.10.017
2394 Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403

capacity of foam-filled conical tubes and no attempt has been The moving mass is described as a rigid body free to translate
made to understand their crush response. In response to this, a only along the z axis, and the supporting mass was fully con-
comprehensive study has been undertaken to investigate the axial strained as shown in Fig. 1b. Symmetrical boundary conditions
crushing response and energy absorption of foam-filled metallic were applied along all free edges. Although explicit codes are com-
conical tubes, using finite element (FE) techniques. Parametric monly used for simulating high-speed impact events, a general
and factorial studies have been performed to evaluate the tube re- quasi-static modeling problem could also be undertaken effec-
sponse under axial quasi-static loading and to quantify their en- tively with reasonable computing time and accuracy as addressed
ergy absorption capacity with respect to the wall thickness, previously in the literature [7,12,13]. The deformation rates were
semi-apical angle and foam density parameters. The primary out- relatively small for an explicit solver; therefore, mass scaling was
come of the study is new research information and development
of empirical relations which will facilitate the design of foam-filled
conical tubes as energy absorbers in impact applications.

2. Finite element modelling


V
2.1. Description of finite element model

Velocity (mm/ms)
The finite element models of empty and foam-filled conical
tubes in this study were developed using the explicit non-linear fi-
nite element code LS-DYNA [11]. Fig. 1a shows the mesh used to
model the empty and foam-filled conical tubes. Half models were
used to simulate both the empty and foam-filled conical tubes
resulting in reduction in computational time as also observed from
a previous study on foam-filled circular tube [3].
In the present study, the Belytschko–Lin–Tsay shell elements
with five integration points through the thickness were employed
to model the conical tube profile. For the foam-filled conical tube
model, the foam was modeled using 8 noded solid elements with tR tT
the reduced integration techniques in combination with hourglass Time (ms)
control. Stiffness-based hourglass control was employed in order
Fig. 2. Load curve used in quasi-static simulation.
to avoid spurious zero energy deformation modes and the use of
reduced integration was to avoid volumetric locking [11]. In order
to decide the size of elements, a mesh convergence study was car-
ried out to minimize mesh refinement effect on the accuracy of the 600
numerical results. It was found that mesh sizes of 2.5 mm elements
500
and 4.0 mm elements along the volume, for the conical tube and
foam, respectively, were able to give converged solutions within 400
a reasonable amount of time and hence this mesh size was used
throughout the study. The supporting and moving mass were mod- 300
eled using 8 noded brick (solid) elements and were treated as rigid
bodies. Self-contact interaction was simulated using an ‘‘automatic 200
single surface” to the part of conical tubes in order to avoid inter-
100
penetration of tube wall. To account for contact between the rigid
bodies, foam and tube, ‘‘node to surface” and ‘‘automatic surface to 0
surface” contacts were defined with static and dynamic coeffi- 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
cients of friction of 0.3 and 0.2, respectively.
Fig. 3. True stress versus plastic strain distribution for carbon steel material.

z
Rigid body
translates in z
direction y

θ Symmetry boundary
conditions applied along
unloaded tube edges
z Fully fixed
rigid body

x y Base of tube fully


fixed to rigid body
Fig. 1. (a) Finite element mesh of half model of empty and foam-filled tubes and (b) loading and boundary condition arrangement.
Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403 2395

Table 1 LS-DYNA. The conical tube material is carbon steel with the follow-
Material parameters for the Alulight aluminium foam with different foam densities. ing mechanical properties: initial yield stress of ry = 401.4 MPa,
qf (g/cm3) qp (N/mm2) a a2 (N/mm2) b c (N/mm2) eD Young’s modulus, E = 200 GPa, Poisson ratio, m = 0.3 and density,
0.22 2.14 2.12 169 2.942 2.45 2.5074 qo = 7809 kg/m3. To accurately define the hardening characteristic
0.534 12.56 2.12 1544 3.680 1.00 1.6206 in the FE model, the engineering stress–strain curve of the carbon
0.71 22.18 2.12 4295 4.718 6.438 1.3357 steel was obtained using a standard tensile test. Fig. 3 shows the
true stress versus plastic strain relationship that was incorporated
into LS-DYNA for the analyses. The model selected to represent the
applied in order to achieve sufficiently small time steps and speed material behaviour of aluminum foam filler was Deshpande-Fleck
up the solution while still maintaining a quasi-static deformation foam (material model 154) which has been implemented as a user
condition. In the present study, the quasi-static loading was simu- subroutine in LS-DYNA. Material model for the foam filler was
lated by performing mass scaling and prescribing a ramp velocity- developed by using experimental data of three different Alulight
time loading history. The velocity was initially ramped up within aluminum foam (AlMg1Si) with densities: 0.22 g/cm3, 0.534 g/
ramping time, tR = 50 ms (which provided acceptable results) and cm3 and 0.71 g/cm3 provided by Alulight International GmbH,
then followed by constant velocity, V of 2 m/s throughout the total Germany.
loading duration, tT as shown in Fig. 2. A similar approach has been In order to determine the material parameters of Alulight alu-
employed previously in established studies [14,15]. The ratio of the minum foam, the yield stress, ry based on the strain hardening rule
total kinetic energy to the total internal energy was found to be given by Eq. (1) was adopted. In this equation, a2, b, and c are
less than 5% over the period of the axial crushing, ensuring the qua- material parameters, and ^ e is the equivalent strain. Herein, rp
si-static strain rates. and eD are the plateau stress and densification strain, respectively
The conical tube is modeled with a piecewise linear elastic– 0 1
plastic behaviour with strain hardening (material model 24) in ^e B 1 C
ry ¼ rp þ c þ a2 ln @  b A ð1Þ
eD 1 ^e
eD
Table 2
Comparison of quasi-static simulation results for empty conical tube.
As the above strain hardening rule was calibrated to a uni-axial
Geometry h (mm) h (deg) Initial peak load (kN) Mean load (kN) compression test carried out by Alulight International GmbH, the
Mamalis FE model Mamalis FE model densification strain, eD can be expressed by the equations
et al. [7] et al. [7]  
9 þ a2 qf
E1 2.41 5 121.0 125.0 86.5 86.4 eD ¼  ln ð2Þ
E2 1.19 5 50.2 54.4 21.4 22.9 3a2 qfo
E3 0.51 5 20.3 21.1 6.9 7.2 9 ð1  2mp Þ
E4 0.28 5 13.1 8.3 2.8 2.6 a2 ¼ ð3Þ
2 ð1 þ mp Þ
E5 0.28 10 4.1 5.2 2.4 2.2

In Eq. (2) qf is a foam density and qfo is the density of base material
Table 3 which is often assumed as 2.7 g/cm3. The parameter a in Eq. (3) de-
Details of foam-filled conical tube specimens used in the axial quasi-static loading. fines the shape of the yield surface and is a function of the plastic
No. of specimen h (deg) L (mm) Db (mm) Dt (mm) h (mm) qf (kg/m3) coefficient of contraction, mP. For aluminum foam, mP is assumed
equal to zero [3,16]. Hence material parameters for each foam den-
5 5 150.4 79.9 54.1 2.0 220
sity were determined using the provided engineering stress–strain

Dt

h
t

θ L

Db
Fig. 4. Quasi-static test set-up and configuration of tube specimen.
2396 Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403

60
100
FE Mamalis et al.[7]
50 90
80

Mean load (kN)


Load (kN)

40 70
60
30 50
40
20
30
10 20
10 FE Mamalis et al.[7]
0 0
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Deflection (mm) Deflection (mm)
Fig. 5. (a) Quasi-static load–deflection curves (h = 5°, h = 1.19 mm) and (b) mean load–deflection curves (h = 5°, h = 2.41 mm) for empty conical tubes.

data and Eqs. (1)–(3) coupled with least square method and simple results [8] as depicted in Figs. 6 and 7 show a comparison of the
programming. Material parameters of foam filler used in the quasi- experimental and numerical load–deflection and mean load–
static simulation are tabulated in Table 1. deflection responses together with typical deformation modes for
the foam-filled conical tubes under quasi-static loading. In this fig-
2.2. Validation of finite element model ure, results for all three test repetitions of the experiments are
shown with the following identification system: Fi where ‘‘F”
To ensure that the FE model is sufficiently accurate, it was val- stands for the filled tube, and i is the specimen number. It is evi-
idated using experimental and numerical results which predict the dent that there is good correlation between the present FE and
mean loads, initial peak loads and collapse modes for the axial experimental results. A typical deformation mode of the filled tube
crushing of empty and foam-filled conical tubes under quasi-static is also well described by the present simulation with similar col-
loading. The FE model for the empty conical tube was validated by lapse formation and number of created lobes as in the experiments.
comparing the load–deflection response, mean load, initial peak Overall, it is evident that there is good correlation of the present
load and collapse modes with the existing results [7,8]. The valida- numerical results with those from experiments and existing re-
tion of the empty tube was carried out for an average length of sults. As such, the FE model is capable of simulating the physical
127 mm and for different combinations of wall thickness and response of both the empty and filled tubes under quasi-static ax-
semi-apical angles as described in Table 2. ial loading. The validated models therefore will provide sufficiently
There are no existing experimental data, theoretical relations or accurate predictions of the response of empty and foam-filled con-
numerical results in the literature for the response of foam-filled ical tubes under quasi-static axial loading.
conical tubes. The FE model for the foam-filled conical tube was
therefore validated using results of experiments carried out by
the authors, by comparing the load–deflection response, mean 3. Parametric study
load–deflection response and mode of deformation. The details of
the foam-filled conical tubes investigated are shown in Table 3. 3.1. Effect of wall thickness, h, semi-apical angle, h and foam density,
Fig. 4 shows the test set-up and configuration of tube specimen qf on the tube response
for axial quasi-static testing. To reduce the number of experiments
and ensure repeatability and reduction in experimental errors, five The axial crushing and energy absorption responses of empty
specimens of foam-filled conical tubes, all with the same semi-api- and foam-filled conical tubes are treated in this section and the
cal angle of 5°, were tested. Foam-filled tubes with other dimen- influence of geometry and material parameters namely the wall
sions could be later analysed using the validated finite element thickness, h and semi-apical angle, h and foam density, qf is
model in a parametric study. Only three representative test results investigated.
for the foam-filled conical tubes are presented in this paper. A Crush response is described with commonly used response
screw-type Tinius Olsen universal testing machine was used for parameters, namely the mean crush load, the initial peak load, load
all quasi-static tests at a crosshead speed of 10 mm/min. ratio and the energy absorption capacity. To allow meaningful
Fig. 5 shows the quasi-static load–deflection curves (h = 5°, comparison of the results, the dimensions of the empty and
h = 1.19 mm) and mean load–deflection curves (h = 5°, foam-filled conical tube were kept same with the height of all
h = 2.41 mm) for the empty conical tube. Results in this figure tubes 200 mm. The semi-apical angle was increased from 5° to
show a reasonable agreement between the existing results and 15° by increasing the bottom end diameter and keeping the top
present FE predictions, as also found for other wall thicknesses of end diameter constant at 40 mm. Table 4 shows the details of
empty conical tubes. Comparison of additional results from the the tubes in the quasi-static parametric study.
present quasi-static simulations with those from existing numeri- The effects of foam density and wall thickness on the mean load
cal model [7] for the empty conical tube is tabulated in Table 2, and are shown in Fig. 8a for a constant semi-apical of 10°. Clearly,
indicates good agreement between the two sets of results. Strictly introducing foam filler leads to a significant increase in the mean
speaking, the mean collapse load of a conical tube slightly in- load, thus enhancing the energy absorption capacity particularly
creases throughout the deformation process as found in [8]. How- when the tubes are filled with high-density foam. Also it can be
ever, in this study, the mean collapse load is defined as the seen that an increase in wall thickness causes an increase in the
equivalent constant load with a corresponding amount of displace- mean load up to a given deflection for both the empty and filled
ment. The deformation modes of empty conical tubes predicted by tubes as was also found by Gupta and Venkatesh [17] for the empty
the FE model also compared well with those from the experimental conical tube. Moreover, for a given wall thickness, the mean load
Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403 2397

apical increases as shown in Fig. 8b. From Fig. 8, the conical tube
filled with a foam density of 0.71 g/cm3 has the highest mean load
for a given wall thickness and semi-apical angle. It is also evident
that the foam density has more influence on the mean load as
the wall thickness and semi-apical angle increase.
All the above phenomena is due to the presence of foam filler
where the load will increase slowly as the foam cells deform plas-
tically and increase dramatically approaching densification region
as the cell edges progressively touch each other. This effect will be
most pronounced when using high-density foam as such foam re-
sists the tube wall to buckle inward leading to increase mean
loads as deflection increases. Nevertheless, density of foam is
not the sole parameter as tubes filled with foam densities above
a certain range (depending on tube thickness and semi-apical an-
gle) may tend to lose their crashworthiness efficiency [19]. Thus
other response parameters were also investigated in the next
section.
Fig. 9a and b shows the influence of wall thickness and foam
density, and semi-apical angle and foam density, respectively, on
the load ratio, rP defined as the ratio of mean load to initial peak
load. From these figures, it is evident that the tubes filled with
higher values of foam density (considered in this paper) have sig-
nificantly larger load ratios. Wall thickness seems to have only a
small influence on the load ratio as seen in Fig. 9a. The load ratios
of tubes with higher density foams, however, increase with semi-
apical angle as seen from Fig. 9b due to progressive increase in
the crush load with deflection. This feature may not be desirable
in the design for efficient energy absorption.
Finally, Fig. 10a shows the effect of foam density and wall thick-
ness on the initial peak load. It is evident that wall thickness has a
significant effect on the initial peak loads of all the tubes. Since
there is increasing amount of material available for plastic defor-
mation in a foam-filled conical tube, the initial peak load increases
moderately due to the foam filling for each wall thickness as can be
observed in Fig. 10a. However, there was no significant variation in
peak loads among the filled tubes. Thereby, it may be anticipated
that no significant increase in initial peak load will be observed
by using foam-filled conical tubes. This effect is practically desir-
able as filling the tube with foam results high-energy absorption
capacity and at the same time, without significant increase in the
initial peak load.
The effect of the semi-apical angle on the initial peak load for
each of the tube is depicted in Fig. 10b. From this figure, it is obvi-
ous that the initial peak load of empty conical tube decreases sig-
nificantly with semi-apical angle. Similar phenomenon has been
observed in previous study on the buckling of thin conical frusta
[20]. However, the initial peak load of filled tubes slightly increases
when introducing the semi-apical angle of 10° and followed by ra-
pid decrease when increasing semi-apical angle to 15°. Thus it ap-
pears that the response of the filled tube displays certain
dependency on the semi-apical angle. But no theoretical represen-
tation of the relationship between the initial peak load and semi-
apical angle for foam-filled conical tube was found in the literature
Fig. 6. Comparison of collapse modes of empty conical tube from the numerical and
experimental [8] results. (a) h = 2.41 mm, (b) h = 1.19 mm, and (c) h = 0.28 mm. for comparison. Obviously the initial peak load of filled tube is
slightly higher for each semi-apical angle compared to empty tube
due to the presence of foam filler and greater amount of material
across the section of the tube, which effectively increases the axial
increases with increase in foam density, as was also found for other stiffness of the tube and hence the higher load required to initiate
semi-apical angles. buckling.
When varying the semi-apical angle from 5° to 15°, empty tube
is less influenced by semi-apical angle as the mean load remains 3.2. Factorial study for energy absorbing characteristics under quasi-
reasonably constant but slightly decreases. These results for empty static loading
conical tube are similar to what was pointed out in Mamalis and
Johnson [18] in which the mean load decreased with semi-apical The enhancement of mean load and energy absorption capacity
angle. On the other hand, metallic foam-filled conical tube with are indeed due to the interaction effect between foam filler and
qf = 0.22–0.71 g/cm3 exhibits an increase in mean load as semi- conical tube as addressed in [2,21]. For the present study, a facto-
2398 Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403

180 120
160
100
140

Mean load (kN)


120 80
Load (kN)

100
60
80
60 40
40
20
20
F1 F2 F3 FE F1 F2 F3 FE
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70
Deflection (mm) Deflection (mm)

Fig. 7. (a) Quasi-static load–deflection curves, (b) mean load–deflection curves for foam-filled conical tubes (h = 5°, h = 2.0 mm, and qf = 220 kg/m3), and (c) Typical
deformation modes of foam-filled conical tube.

Table 4 mates main and interaction effects between the input and re-
Model details used in the quasi-static parametric study. sponse parameters. The ‘effect’ of a factor refers to the change in
Conical h (°) L (mm) Db (mm) Dt (mm) h (mm) qf (g/cm3) response of the system due to the factor as that factor moves from
tube a low to a high value. The study can also detect ‘interaction effects’,
Empty 5 200 75.0 40 1.5, 2.0, 2.5 0
in which two or more input parameters have a combined effect on
10 200 110.5 40 1.5, 2.0, 2.5 0 the system’s response. For a detailed explanation of factorial design
15 200 147.2 40 1.5, 2.0, 2.5 0 and the procedure used to obtain the results presented in this sec-
Foam- 5 200 75.0 40 1.5, 2.0, 2.5 0.22, 0.534 0.71 tion, the reader is referred to [22]. The values for each parameter at
filled 10 200 110.5 40 1.5, 2.0, 2.5 0.22, 0.534 0.71 the low and high levels used in the factorial study are shown in Ta-
15 200 147.2 40 1.5, 2.0, 2.5 0.22, 0.534 0.71 ble 5. The validated FE model is used for the study. Instances where
factorial design has been used for analysing energy absorbers is
also provided in [23,24].
Fig. 11a shows the relative effect of each parameter on the spe-
rial study was used to determine the effect of controlling parame- cific energy absorption (SEA). SEA is defined as energy absorption
ters such as wall thickness, semi-apical angle and foam density on capacity per unit mass which is a common measure to specify
the tube response namely the specific energy absorption (SEA) and the mass efficiency of an energy absorbing structure. This parame-
energy absorbed per unit crush length, Ecl. A factorial study esti- ter provides a comparison between different tube geometries to

250 290

200 240
Mean load (kN)
Mean load (kN)

150 190

100 140

50 90

0 40
1.5 2 2.5 5 10 15
Wall thickness (mm) Semi-apical angle (deg)

f = 0g/cm3 f = 0.22g/cm3 f = 0.534g/cm3 f = 0.71g/cm3

Fig. 8. Effect of foam density, wall thickness and semi-apical angle on mean load (a) constant h = 10° and (b) constant h = 2.0 mm.
Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403 2399

1.6 3.0

1.4 2.5

1.2 2.0
rP

rP
1.0 1.5

0.8 1.0

0.6 0.5

0.4 0.0
1.5 2.0 2.5 5 10 15
Wall thickness (mm) Semi-apical angle (deg)

f = 0g/cm3 f = 0.22g/cm3 f = 0.534g/cm3 f = 0.71g/cm3

Fig. 9. Effect of foam density, wall thickness and semi-apical angle on the load ratio (a) constant h = 5° and (b) constant h = 1.5 mm.

155 85
145 84
135
83
125
Pp(kN)

Pp(kN)
82
115
81
105
95 80

85 79
75 78
1.5 2 2.5 5 10 15
Wall thickness (mm) Semi-apical angle (deg)

f = 0g/cm3 f = 0.22g/cm3 f = 0.534g/cm3 f = 0.71g/cm3

Fig. 10. Initial peak load vs. wall thickness and semi-apical angle (a) constant h = 5° and (b) constant h = 1.5 mm.

Table 5 determine which is more efficient in terms of the amount of energy


Range of parameters used in the factorial study. they absorb for a given tube mass, and is a useful parameter when
Level h (mm) h (deg) weight reduction is important. The results show that the SEA in-
creases (i.e. a positive effect) with increasing wall thickness, yet
Low 1.5 5
High 2.5 15 decreases (i.e. a negative effect) with increasing semi-apical angle
due to the associated increase in mass. In other words, each tube

8 120
6 100
4
Average SEA

80
Average Ecl

2
θ 60
0
h 40
-2
h-θ 20
-4
-6 0
-8
h θ h-θ
-20
Geometry parameter Geometry parameter

f = 0g/cm3 f = 0.22g/cm3 f = 0.534g/cm3 f = 0.71g/cm3

Fig. 11. Relative effect of wall thickness, semi-apical angle and foam density on: (a) specific energy absorption, SEA and (b) energy absorbed per unit crush length, Ecl.
2400 Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403

Empty Foam-filled Fig. 12 shows the effect of foam filling on the number of lobes
and approximated fold length, df. Obviously, the number of lobes
significantly increases when introducing the foam filler for the
δf δf tube with semi-apical angle of 5° as also observed in a previous
study on the numerical simulation of foam-filled circular tube
undergoing axial collapse [5,6]. Fundamentally, as more collapse
lobes were created, higher strain energy will be dissipated in the
tube wall. Nevertheless, for foam-filled tube with semi-apical angle
of 15°, the number of lobes and approximated fold length were al-
most similar to those for empty conical tube. This result shows that
the increase in mean load of such tube is due to the densification of
θ = 5o
foam core rather than collapsing of tube wall. In general, the num-
ber of lobes in a filled tube will depend on the foam density, wall
thickness and semi-apical angle.
Fig. 13 shows deformation modes and load deflection profiles of
empty and filled tubes at different semi-apical angles. It is note-
worthy that the presence of foam filler substantially changes the
deformation mode of empty conical tube from diamond mode to
concertina mode as shown in this figure. Similar results were ob-
served in previous studies on the axial crushing of foam-filled alu-
θ = 15o minum tubes [2,25]. For each tube, collapse starts at the small end
and progresses down the length of the tube. Empty conical tube of-
Fig. 12. Effect of foam filling on the created lobes and fold length (here h = 2.0 mm). ten fails initially in concertina mode followed by diamond mode up
to maximum crush distance particularly for higher semi-apical an-
gle. For instance, empty tube with semi-apical angle of 5° appears
can absorb more energy for a given mass as the wall thickness in- collapsing in concertina mode and followed by three lobes dia-
creases. It can be seen that the SEA in general is more influenced by mond mode. While for higher semi-apical angle, the collapse be-
the semi-apical angle than by wall thickness as the foam density gins by yielding with a concertina mode, and thereafter it
increases. Finally, the h–h interaction shows that the wall thickness progresses by multi-lobe diamond fold mechanism. Due to foam
has more influence on the SEA for empty tubes but has less influ- filling, the filled tube with semi-apical angle of 10° collapses en-
ence on the SEA for filled tubes as the semi-apical angle increases. tirely under concertina mode, whereas for the filled tube of semi-
Overall, it can be concluded that there will be an increased mass of apical 5° and 15°, collapse is initiated progressively and continu-
the absorber as the foam density is increased, despite the associ- ously by the formation of concertina mode from its smaller end
ated increase in absorbed energy. Consequently, in order to practi- and followed by the progressive collapse of the shell by the forma-
cally optimize the SEA, semi-apical angle needs to be controlled in tion of three-lobe diamond when about to reach maximum crush
using metallic foam-filled conical tubes as energy absorbers. distance. Tube inversion is expected to occur for the empty tube
The relative effect of each geometry parameter and foam den- beyond a critical semi-apical angle and this could be delayed by
sity on the energy absorbed per unit crush length, Ecl is shown in filling the tube with foam filler. In accordance with Gupta et al.
Fig. 11b. It can be seen that the Ecl can be increased by either [26], the reverse bending (tube inversion) seems to occur at a
increasing the wall thickness or the semi-apical angle. Clearly, semi-apical angle higher than 40°. This feature is not studied as
increasing the wall thickness significantly increases the energy ab- the expected critical semi-apical angle is out of range of semi-api-
sorbed up to a given deflection, and this is obviously due to the in- cal angles considered in the present study.
creased amount of material available for plastic deformation and From a design viewpoint, progressive collapse of concertina
subsequent energy absorption. Fig. 11b also shows that for all mode is desirable in designing energy absorber as this type of col-
the tubes the wall thickness and semi-apical angle have a positive lapse tends to absorb high energy. Hence, filled tube appears to
influence on the energy absorbed per unit crush length, which also collapse in such deformation mode resulting in higher energy
depends on the foam density. This highlights the significant effect absorption capacity. It is also evident from Fig. 13 that for filled
which the foam density may have on the response. The h–h inter- tubes the load gradually increases with deflection and that this
action shows that the wall thickness in general has more influence is more prominent with increase in semi-apical angle. In addition,
on the Ecl as the semi-apical angle increases, except for foam use of foam filler can resist the radial inward movement of created
density of 0.534 g/cm3. Practically, the wall thickness therefore is lobes for higher semi-apical angle in order to retain the increase in
more effective for controlling the energy absorbed per unit crush energy absorption capacity. Furthermore, since the number of
length. lobes is generally higher for foam-filled tube, the load–deflection
response oscillates steadily about mean load resulting in an effec-
3.3. Deformation profiles of empty and foam-filled conical tube under tive crush response for energy absorption application.
quasi-static loading

In order to allow tubular structures to absorb sufficient 4. Developing an empirical relation for load ratio of foam-filled
energy for impact mitigation, they must collapse in a progres- conical tube
sive manner. Euler buckling mode (global buckling) is an unde-
sirable collapse mode as it might significantly reduce the A simple empirical relation involving dimensionless parameters
energy absorption capacity. Deformation mode is sensitive to can be a useful tool for initial assessment of the crush and energy
the wall thickness, semi-apical angle and foam density. This absorption responses of foam-filled conical tubes for design pur-
can be verified by observing the number of lobe created and pose. With this preliminary design information, a suitable geome-
the fold length for each tube due to varying geometry and foam try and foam material could then be chosen and further examined
density. under simulated or experimental impact loading condition [23].
Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403 2401

Isometric view Top view Side view

120
100
80

P (kN)
60
40
20
0
0 20 40 60 80 100

θ =5 o
δ (mm)
140
120

P (kN)
100
80
60
40
20
0
0 20 40 60 80 100

θ = 10o δ (mm)

160
140
120

P (kN)
100
80
60
40
20
0
0 20 40 60 80 100

θ = 15o δ (mm)

200

150
P (kN)

100

50

0
0 20 40 60 80 100
θ=5 o
δ (mm)

300
250
P (kN)

200
150
100
50
0
0 20 40 60 80 100

θ = 10 o
δ (mm)
350
300
250
P (kN)

200
150
100
50
0
0 20 40 60 80 100
δ (mm)
θ = 15 o

Fig. 13. Effect of the semi-apical angle on the deformation mode and load deflection profiles of (a) empty (h = 2.0 mm) and (b) foam-filled conical tube (here h = 2.0 mm,
q = 0.534 g/cm3).

The results of the validated numerical model were used to develop tween these dimensionless groups and energy absorption response
a simple empirical formula which could be used to predict the qua- which is represented by the load ratio, rP over a range of values for
si-static axial crushing and energy absorption performance of each dimensionless group. Evaluation of load ratio is an elegant way
empty and foam-filled conical tubes. of assessing the performance of an energy absorber. The proposed
The geometry and material parameters of foam-filled conical empirical relation for the load ratio, rP is as follows where h/Dm
tube namely the wall thickness, h, mean diameter, Dm and foam and qf/qo are slenderness ratio and relative density, respectively
density, qf can be taken into account to form dimensionless groups  ao   bo
for each semi-apical angle, h of 5°, 10° and 15°. The purpose of the h qf
rp ¼ C 1þ ð4Þ
empirical model is to represent a semi-empirical relationship be- Dm qfo
2402 Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403

 0:11  
1.6 h q 3:34
rp ¼ 0:433 1þ f h ¼ 5 ð5Þ
1.4
Dm qfo
 0:32  
h q 4:25
1.2 rp ¼ 0:206 1þ f h ¼ 10 ð6Þ
D
 m 0:27 
qfo 
rP

1.0 p h q 5:44
r ¼ 0:230 1þ f h ¼ 15 ð7Þ
0.8 Dm qfo
0.6 From the above equations, it can be seen that power coefficient,
ao is relatively small compared with the corresponding value for
0.4
0 0.05 0.1 0.15 0.2 0.25 0.3 power coefficient, bo. This shows that the slenderness ratio has
only a small influence on the load ratio. Obviously increase in rel-
Relative density ( f / fo)
ative foam density results in an increase of load ratio for a given
slenderness ratio. Also the power coefficient, bo becomes larger
h/Dm = 0.026 h/Dm = 0.035 h/Dm = 0.044
as the semi-apical angle increases from 5° to 15°. This confirms
what was evident from Fig 13 which shows the influence of
Fig. 14. Variation of the load ratio with relative foam density and slenderness ratio
semi-apical angle on the load deflection profiles. Eventually, the re-
for the foam-filled conical tube (h = 5°).
sults from all these equations can be used to obtain the ‘‘best fit”
curve to the finite element results.
Herein C, ao and bo are constant and power coefficients, respec-
In order to evaluate the accuracy of the empirical model, the
tively, and mean diameter, Dm was used as the conical tube has a
load ratio of the foam-filled conical tubes with various slenderness
varying diameter. A similar slenderness ratio has been used exten-
ratios and foam densities, including one (qf = 0.35 g/cm3) which
sively in studying effect of varying slenderness ratio on mean
was not used in parametric study, were calculated using Eqs.
crushing load of empty conical tubes [18,20]. Since energy absorp-
(5)–(7) and then compared with the results from the validated FE
tion response of foam-filled tube strongly depends on the density
models for semi-apical angle 5°–15°. The comparison of rP is sum-
of foam filler, relative density is included in this relation. The sec-
marised in Table 6. As can be seen, reasonable correlation is ob-
ond term in Eq. (4) was so chosen in order to avoid zero value of
tained with less than 7% discrepancy.
load ratio when zero density is substituted to represent the empty
Thus, a useful empirical formulation has been developed which
conical tube. Furthermore, aspect ratio was not considered here as
can provide the designer with an initial estimate of the load ratio
the load ratio of tapered tube is relatively independent of aspect
(ratio of mean load to initial peak load) for the empty or foam-filled
ratio as pointed out by Nagel and Thambiratnam [23].
conical tubes of given dimensions. Moreover, the relationship pro-
Fig. 14 shows the variation of load ratio, rP with the relative
vides a guide as to which parameters have most influence on the
foam density and slenderness ratio for the foam-filled conical tube
load ratio and energy absorption response. Practically, the load ra-
with a constant semi-apical angle of 5°. In calculating load ratio,
tio can be used to measure a uniformity of crush load and energy
the mean and initial peak loads were obtained using the validated
absorber performance. In addition, the range of foam density pro-
finite element model. It has been reported in literature that very
viding desirable energy absorption performance may be deter-
high value of foam density may have an adverse effect on the crush
mined for each semi-apical angle in order to acquire the required
response (i.e. global buckling) due to significant increase in crush
load ratio of a filled tube.
load at the densification region [5,19]. To evaluate the range of
Above all, it should be noted that the empirical relation was
foam density for satisfactory performance and set an upper limit
developed for the foam-filled conical tube of 200 mm length with-
on the foam density, the parameter load ratio, rP is used. Practi-
in a certain range of semi-apical angles, wall thicknesses and foam
cally, an ideal energy absorber collapses progressively with initial
densities, namely; 5°–15°, 1.5–2.5 mm and 0–0.71 g/cm3, respec-
peak load similar to mean load. Thus, if the load ratio becomes
tively. Thus, these equations are most accurate for values of each
greater than unity, undesirable energy absorber performance
parameter within these ranges. The chosen range of wall thickness
may be obtained. Clearly, from Fig. 14, the load ratio increases as
and tube length is commonly used in the real world applications
the relative density increases due to the presence of foam filler
for designing energy absorbers. Furthermore, tube wall tends to
which increases the mean load over the maximum crush distance.
fail under rupture mechanism if the thickness is less than
Furthermore, the load ratio initially increases with slenderness ra-
1.5 mm, whereas foam filling is not practical and weight effective
tio but the trends reverse as the relative foam density increases.
anymore if the wall thickness and semi-apical angle are more than
Using the least squares theorem and simple programming, the
2.5 mm and 15°, respectively. Similar empirical relations may be
values of the constants in Eq. (4) were then determined, giving
developed for other tube lengths using relevant results from finite
the following relationships between the load ratio, rP and geometry
element analyses to enhance the design information across a
parameters for each semi-apical angle
broader range of parameters.

5. Conclusion
Table 6
Prediction of the load ratio using Eqs. (5)–(7) and the finite element model.
This paper has investigated and described the crush and energy
h (deg) h (mm) Dm qf qf/qfo h/Dm Empirical FE
absorption responses of empty and metallic foam-filled conical
(mm) (g/cm3) (g/cm3) (mm)
tubes under quasi-static axial loading using validated FE models,
5 1.75 57.5 0 0 0.030 0.64 0.62
for variations in wall thickness, semi-apical angle and foam den-
5 1.75 57.5 0.22 0.081 0.030 0.83 0.80
5 2.0 57.5 0.35 0.130 0.035 0.94 0.99
sity. Overall, it was found that foam-filled conical tube is superior
10 2.25 75.25 0.534 0.198 0.030 1.36 1.30 to empty conical tube in energy absorption performance based on
10 1.5 75.25 0.35 0.130 0.020 1.21 1.28 the energy absorption capacity, load ratio, interaction effect and
15 2.25 93.6 0.22 0.081 0.024 0.96 0.91 stability of crush response. At present, there are no design aids
15 2.5 93.6 0.35 0.130 0.027 1.19 1.21
for using foam-filled conical tubes as energy absorbers, despite
Z. Ahmad, D.P. Thambiratnam / Materials and Design 30 (2009) 2393–2403 2403

their potential to be effective energy absorbers. This paper gener- References


ated design information regarding the relative effect of each geom-
etry and material parameter on the tube response and [1] Nagel GM, Thambiratnam DP. A numerical study on the impact response and
energy absorption of tapered thin-walled tubes. Int J Mech Sci
demonstrated the feasibility and superior performance of foam- 2004;46(2):201–16.
filled conical tubes. Moreover, a simple empirical model for calcu- [2] Yamada Y, Banno T, Xie ZK, Wen C. Energy absorption and crushing behaviour
lating the load ratio of energy absorber has been developed which of foam-filled aluminium tubes. Mater Trans 2005;46(12):2633–6.
[3] Borvik T, Hopperstad OS, Reyes A, Langseth M, Solomos G, Dyngeland T. Empty
can be then effectively used as an initial guide in designing foam- and foam-filled circular aluminium tubes subjected to axial and oblique quasi
filled conical tubes as energy absorbers. It is noteworthy that the static loading. Int J Crashworthiness 2003;8(5):481–94.
use of foam-filled conical tubes as energy absorbers contributes [4] Gupta NK, Velmurugan R. Axial compression of empty and foam filled
composite conical shells. J Compos Mater 1999;33(6):567–91.
new knowledge in structural crashworthiness and impact applica-
[5] Seitzberger M, Rammerstorfer FG, Degischer HP, Gradinger R. Crushing of
tions. The main conclusions and design information from the study axially compressed steel tubes filled with aluminium foam. Acta Mech
are summarised below. 1997;125(1–4):93–105.
[6] Hanssen AG, Langseth M, Hopperstad OS. Static and dynamic crushing of
circular aluminium extrusions with aluminium foam filler. Int J Impact Eng
(1) The mean load and energy absorption response of foam- 2000;24(5):475–507.
filled conical tubes subjected to axial quasi-static loading [7] Mamalis AG, Manolakos DE, Ioannidis NB, Kostazos PK. Numerical simulation
is more influenced by the foam density and wall thickness of thin-walled metallic circular frusta subjected to axial loading. Int J
Crashworthiness 2005;10(5):505–13.
than the semi-apical angle. Practically, the foam density [8] Mamalis AG, Johnson W, Viegelahn GL. The crumpling of steel thin-walled
and wall thickness are more effective for controlling the tubes and frusta under axial compression at elevated strain rates: some
energy absorbing capacity. experimental results. Int J Mech Sci 1984;26:537–48.
[9] Reid SR, Reddy TY. Axial crushing of foam-filled tapered sheet metal tubes. Int J
(2) Initial peak load of foam-filled conical tube slightly increases Mech Sci 1986;28(10):643–56.
due to the presence of foam filler. This increase does not vary [10] Mirfendereski L, Salimi M, Ziaei-Rad S. Parametric study and numerical
significantly for the range of foam densities treated herein. analysis of empty and foam-filled thin-walled tubes under static and dynamic
loadings. Int J Mech Sci 2008;50(6):1042–57.
However, of greater significance is the increase in energy [11] Hallquist JO. LS-DYNA 3D: theoretical manual. Livermore: Livermore Software
absorption capacity of filled tubes. Technology Corporation; 2006.
(3) Conical tube filled with metallic foam appears to collapse in [12] Mathisen KM, Hopperstad OS, Okstad KM, Berstad T. Error estimation and
adaptivity in explicit nonlinear finite element simulation of quasi-static
steady manner and thus reduces the chances of collapse by
problems. Comput Struct 1999;72(4–5):627–44.
global buckling mode. Hence foam-filled conical tube is pref- [13] Santosa SP, Wierzbicki T, Hanssen AG, Langseth M. Experimental and
erable compared to empty or foam-filled straight circular numerical studies of foam-filled sections. Int J Impact Eng 2000;24(5):509–34.
tube for use as energy absorbers. [14] Tarigopula V, Langseth M, Hopperstad OS, Clausen AH, Axial crushing of thin-
walled high-strength steel sections. In: International symposium on the
(4) The ratio of mean load to initial peak load (load ratio) crashworthiness of light-weight automotive structures. Int J Impact Eng
increases with increasing semi-apical angle and foam den- 2006;32(5):847–82.
sity. Furthermore, as the foam density increases, the influ- [15] Seitzberger M, Rammerstorfer FG, Gradinger R, Degischer HP, Blaimschein M,
Walch C. Experimental studies on the quasi-static axial crushing of steel
ence of wall thickness on the load ratio generally decreases columns filled with aluminium foam. Int J Solids Struct 2000;37(30):4125–47.
while the influence of semi-apical angle increases. As a [16] Reyes A, Hopperstad OS, Berstad T, Hanssen AG, Langseth M. Constitutive
design guide, it has to be noted that load ratio is more sen- modeling of aluminum foam including fracture and statistical variation of
density. Eur J Mech A/Solids 2003;22(6):815–35.
sitive to varying the foam density as the semi-apical angle [17] Gupta NK, Venkatesh. Experimental and numerical studies of impact axial
increases. compression of thin-walled conical shells. Int J Impact Eng 2007;34(4):
(5) If weight reduction is a main goal, the combination of tube 708–20.
[18] Mamalis AG, Johnson W. The quasi-static crumpling of thin-walled circular
geometry and foam density must be carefully chosen as cylinders and frusta under axial compression. Int J Mech Sci 1983;25(9–
the SEA of filled tubes tends to reduce due to the increase 10):713–32.
of mass. [19] Zarei H, Kroger M. Optimum honeycomb filled crash absorber design. Mater
Des 2008;29(1):193–204.
(6) The crush load (and the load ratio) increases with deflection
[20] Gupta NK, Sheriff NM, Velmurugan R. A study on buckling of thin conical frusta
for higher density foams (treated herein) and this feature under axial loads. Thin-Wall Struct 2006;44(9):986–96.
might adversely affect the energy absorption performance. [21] Lu G, Yu T. Energy absorption of structures and materials. Cambridge:
The proper choice of foam density is therefore important Woodhead; 2003.
[22] Box GEP, Hunter WG, Hunter JS. Statistics for experimenters. New York: Wiley;
in design. 1978.
(7) The energy absorbed within a given crush distance for a [23] Nagel GM, Thambiratnam DP. Computer simulation and energy absorption of
foam-filled conical tube can be maximized by: tapered thin-walled rectangular tubes. Thin-Wall Struct 2005;43(8):1225–42.
[24] Reyes A, Langseth M, Hopperstad OS. Crashworthiness of aluminum extrusions
(i) increasing the wall thickness, and/ or, subjected to oblique loading: experiments and numerical analyses. Int J Mech
(ii) increasing the semi-apical angle, and/or, Sci 2002;44(9):1965–84.
(iii) increasing the density of foam filler. [25] Kavi H, Tasoy AK, Guden M. Predicting energy absorption in a foam-filled thin-
walled aluminum tube based on experimentally determined strengthening
(8) Empirical relations have been developed for calculating the coefficient. Mater Des 2006;27(4):263–9.
load ratio of empty and metallic foam-filled conical tubes [26] Gupta NK, Easwara Prasad GL, Gupta SK. Plastic collapse of metallic conical
with semi-apical angle of 5°–15° in terms of non-dimen- frusta of large semi-apical angles. Int J Crashworthiness 1997;2:349–66.
sional controlling parameters. This empirical model can be
used for preliminary design evaluation particularly when
initial estimate of energy absorption performance is desired
for empty and foam-filled conical tubes.

Acknowledgements

The authors acknowledge the support of Alulight International


GmbH, Germany represented by Mr. Philipp Zach for providing
the compression test data of Alulight foam (AlMg1Si).

You might also like