Composite Structures 132 (2015) 393405

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

On functionally graded composite structures for crashworthiness

Yong zhang a, Minghao Lu a, Guangyong Sun b,, Guangyao Li b, Qing Li c
a
College of Mechanical Engineering and Automation, Huaqiao University, Xiamen 361021, China
b
State Key Laboratory of Advanced Design and Manufacture for Vehicle Body, Hunan University, Changsha 410082, China
c
School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney, Sydney, NSW 2006, Australia

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

Article history: The foam-lled thin-walled composite structures have proven an ideal energy absorber in automotive
Available online 27 May 2015 engineering for its extraordinary energy absorption ability and lightweight features. Unlike existing uni-
form foam and thickness (UFT) structure, this paper introduces functionally graded foam (FGF) to ll into
Keywords: functionally graded thickness (FGT) thin-walled structure, named as double functionally graded (DFG)
Crashworthiness tube, where different congurations of foam and wall thickness gradients are taken into account. To sys-
Composite structures tematically explore the crashworthiness of DFG structures, rst, experimental results were performed to
Metallic foam
validate nite element (FE) models. Second, a comparison of crashworthiness was carried out for (1) four
Energy absorption
Functionally-graded structures
different DFG structures, (2) four single functionally-graded (SFG) structures and (3) one traditional UFT
structure. The results showed that the DFG structures have better energy absorption capacity than the
SFG and UFT structures, especially with a convex gradient conguration. In addition, the specic energy
absorption (SEA) values of these four DFG structures are fairly close to each other, while their loading
responses highly depend on the combination of gradients. Of these DFG structures, Ascending
Ascending conguration exhibits best overall crashworthiness characteristics. Finally, parametric studies
were performed and the results indicated that widening the ranges of foam density and tube wall thick-
ness can improve the energy absorption of the AscendingAscending DFG structures without increasing
the initial peak load. Therefore, the DFG structure of AscendingAscending gradient is recommended for a
potential absorber.
2015 Elsevier Ltd. All rights reserved.

1. Introduction Recently, such cellular materials as metallic foams have

Energy absorber plays a critical role in protecting occupant and
goods safety when vehicular collision occurs. As a class of typical
energy absorbing devices, thin-walled structures are considered
superior and one of most common energy absorbers for crashing
protection attributable to its well-controlled deformation pattern
of progressive folding. Different types of thin-walled structures,
such as circular [1,2], square [2], top-hat and double-hat tubes
[3,4], have received extensive attention. The previous studies
demonstrated that cross-sectional congurations have critical
inuences in energy absorption capacity of thin-walled columns.
Specically, circular prole was found to be more effective in pro-
gressive folding than other cross sectional proles [1,2]. For this
reason, the study on crashworthiness of circular thin-walled tubes
is of considerable interests.
Corresponding author. Tel.: +86 731 8881 1445; fax: +86 731 8882 2051.
E-mail address: sgy800@126.com (G. Sun).
received extensive attention because they can undergo large defor-
mation at nearly constant plateau load. Therefore, metallic foams
have led to increased applications as ller materials in
thin-walled structures to increase energy absorption without
sacricing too much weight. For example, Seitzberger et al. [5]
performed quasi-static experiments and identied that foam ller
greatly improved weight efciency with respect to energy absorp-
tion. Hanssen and his coworkers [6,7] carried out axial crushing
tests and derived some close-form formulas for prediction of aver-
age force, maximum force and effective crushing distance of
foam-lled tubes. They also identied the strong interaction
between foam and tube wall, which explained the reason why
the structure absorbs more energy than the sum of empty tube
and foam themselves [6,8]. Furthermore, Reddy and Wall [9] found
that low density foam prevents irregular buckling of thin-walled
tubes and leads to more symmetric buckling deformation, thus
improving stability of crushing process. Nevertheless, Seitzberger
et al. [10] pointed out that lling the circular tube with high den-
sity foam could result in global buckling and somehow reduce SEA
of foam-lled structures. For this reason, design optimization is

http://dx.doi.org/10.1016/j.compstruct.2015.05.034
0263-8223/ 2015 Elsevier Ltd. All rights reserved.
394 Y. zhang et al. / Composite Structures 132 (2015) 393405
required to seek a best possible combination of foam material
parameters and geometric conguration for foam-lled structures.
In this regard, Hou et al. [11] exploited various surrogate modeling
techniques to seek optimal design of uniform foam-lled thin-wall
structures for crashworthiness. Zarei and Kroger [12,13],
Nariman-Zadeh [14] and Zhang et al. [15] also adopted optimiza-
tion techniques to maximize energy absorption and minimize the
weight of uniformly foam-lled aluminum tubes.
Although the UFT columns have improved crashworthiness to a
considerable extent, such conguration may not exert their maxi-
mum capacities in material usage. In other words, the uniform
structures may not make best use of material for meeting the
requirements of vehicular lightweight [16]. There is an urgent need
to develop new structural congurations with non-uniform distri-
butions of material and/or thickness for maximizing crashworthi-
ness and material usage. Recently, functionally graded materials
(FGMs), where microstructural details are spatially varied through
a non-uniform pattern, are drawing increasing attention attributa-
ble to their tailored multifunctional behaviors. These advanced
materials with engineered gradients of composition, structure or
specic properties in the preferred direction/orientation are supe-
rior to homogeneous materials made of the same constituents and
uniform geometry. For FGMs, the resultant mechanical properties
such as Youngs modulus, Poissons ratio, shear modulus and mate-
rial density can vary in preferred directions [17]. In nature, FGM
structures widely exist, from bamboos (Fig 1(a)) [18] and cuttle-
bone consisting of horizontal lamellae separated by vertical pillars
[19] (Fig. 1(b)), to the spongy trabecular bone [20]. In the biomi-
metic context, the gradient concept has been taken into account
in a wide range of engineering applications, where optimal use of
materials is essential, e.g. functionally graded piezoelectric materi-
als, dental implants [21], heat exchanger [22], and novel concrete
structures [23], etc.
To further improve crashworthiness of foam-lled thin-wall
structures, graded foam material and tube wall thickness have
shown certain benets. In this respect, functionally graded foam
(FGF) materials [24,25], where foam density varies continuously
in a predened form, have been attempted to replace uniform foam
llers. Several studies were conducted to investigate the energy
absorption performance of FGF congurations. Cui et al. [26] pro-
posed a FGF model, and showed that FGFs have superior
energy-absorbing capability than uniform foam. Sun et al. [27]
explored the crashworthiness of FGF lled square tube with axial
gradient, and they found that FGF is superior to its uniform coun-
terparts in overall crash behaviors. Yin et al. [28] studied the axial
crushing behavior of FGF lled tapered tube. Fang et al. [29]
explored the crashworthiness of the FGF column under lateral
impact. Yin et al. [30] and Attia et al. [31] subsequently extended
the foam ller from axial density gradient to lateral density
gradient.
Furthermore, to improve the utilization of wall materials, func-
tionally graded thickness (FGT) was proposed, in which the
Tailored Rolling Blanks (TRB) technology has been used to produce
sheet metal for continuously changing wall thickness [32]. In this
regard, Sun et al. [16] rst explored the crushing characteristics
of FGT square tubes and found that FGT tube is superior to its
Fig. 1. Examples of FGMs in nature and engineering (a) bamboo (b) cuttlebone and SEM image of the transverse cross section of cuttlebone.
uniform counterpart in overall crashing behaviors. More recently,
Li et al. [33] compared the novel FGT tube with the conventional
tapered tube for withstanding oblique impacting, and found that
FGT tube is more benecial within a given spatial constraint.
These studies showed that the column with a graded wall thick-
ness is more preferred than with a uniform thickness for its stable
load-deformation responses and reduced risk of global buckling.
Nevertheless, these previous studies on functionally graded
structures largely focused on single gradient for either column wall
thickness or foam ller. To the authors best knowledge there has
been no report available for taking into account the dual gradients
for foam ller and wall thickness. This paper presents a novel
composite structure, named as double functionally graded (DFG)
conguration. To explore crashworthiness of DFG structures, a
comparative study on the four different FGF-FGT circular columns,
four corresponding SFG structures and one UFT structure with the
same weight, is conducted here, in which SEA and Initial Peak Load
(IPL) are selected as the indicators to evaluating crashworthiness.
2. DFG structure and nite element modeling
2.1. Single functionally-graded structures
Previously explored functionally graded structures have been
largely restricted to the SFG structures, where only one compo-
nent, either outer tube wall or inner foam ller, uses
functionally-graded conguration. Fig. 2 illustrates the SFG struc-
tures proposed in literature, namely longitudinally graded wall
thickness (Fig. 2(a) [16,33]) and foam density (Fig. 2(b)
[27,28,31]); as well as laterally-graded wall thickness (Fig. 2(d))
and foam density (Fig 2(c) [2931]). This paper proposes a novel
DFG conguration, combining graded wall thickness and graded
foam density in longitudinal direction.
2.2. Double functionally-graded structures
To adapt the functional requirements for crashworthiness,
novel processing methods such as friction stir processing [34,35],
aluminum melt direct foaming route [36] have made it possible
to fabricate foam in a desired gradient with continuous varying
density along a certain direction [27]. To model the axially graded
foam ller, the continuous gradient conguration was discretized
with a limited number of layers, and the foam density of each layer
was regarded uniform [16,27]. Fig. 3(a) is a schematic of such a FGF
conguration, and foam density q is determined by the following
power-law function.
8  10m
<q qmax  qmin xL
min ascending pattern
qx; m 1
:q x10m
max  qmax  qmin L descending pattern
where q represents foam density of each layer, qmin and qmax are the
minimum and maximum densities, respectively; m denotes the gra-
dient parameter that governs the variation pattern of foam density;
x and L are the distance from the xed end (rigid wall) and axial
length of the tube shown in Fig. 3(a), respectively. The direction
 Y. zhang et al. / Composite Structures 132 (2015) 393405 395

Fig. 2. Single functionally-graded structures reported in literature.

Fig. 3. Scheme of DFG tube.

of density gradient follows the direction of axial impact. Here, if than 0, seen in Fig. 4(a). For the descending pattern, on the con-
foam density increases from rigid wall to the impactor, such a dis- trary, an opposite tendency is observed, shown in Fig. 4(b). It
tribution is named as ascending pattern; descending pattern is should be noted that, for a given gradient exponent, ascending or
dened otherwise. descending pattern leads to opposite trend of concavity or convex-
The column with functionally graded thickness can be manufac- ity of distribution.
tured with FGT sheet, whose thickness gradient was dened as
follows: 2.3. Finite element models
8 x10n
<t tmax  tmin ascending pattern
min L Based on the schematic shown in Fig. 3(c), a nite element (FE)
tx; n x10n 2
:t  t max  tmin descending pattern model of DFG circular tube (200 mm long and 60 mm in diameter)
max L
subjected to axial impact loading is constructed in Fig. 5, in which
where t min and t max are the minimum and maximum wall thick- the bottom (distal) surface of the foam-lled column is xed to the
nesses, respectively; n denotes the gradient parameter that governs rigid ground. In order to simulate the front crash of a passenger car,
the variation in thickness; x and L are the distances shown in and a rigid block with a mass of 600 kg impacts onto the top (inci-
Fig. 3(b). The denitions of ascending and descending patterns for dental) surface at an initial velocity of v = 10 m/s. To approximate a
thickness are similar to those in the graded foam. continuous change of foam density and wall thickness, the struc-
Finally, double functionally-graded (DFG) structure with the ture is evenly divided into 40 discrete layers. The depth of each
FGF lled the FGT circular tube is shown in the schematic in layer is 5 mm, which was considered small enough for modeling
Fig. 3(c). the gradients [31].
For the ascending pattern, gradient exponent changes from con- In literature, indentation triggers are often introduced
vexity to concavity when n or m varies from less than 0 to greater [6,3740] in the outer tubes to stabilize the folding process and
396 Y. zhang et al. / Composite Structures 132 (2015) 393405
Fig. 4. Distribution of grading patterns.
Fig. 5. Finite element modeling of double gradient tubal structure.
peak force. In this study, a trigger is introduced with a depth of
0.3 mm at about 7.5 mm below the top surface of the circular tube,
in which the distance of 7.5 mm is approximately equal to the
half-wavelength of the rst fold.The nonlinear explicit FE code
LS-DYNA 971 is used to simulate the crashing behavior and energy
absorbing capability of the FGF-lled FGT circular tube. The outer
FGT tube is meshed with BelytschkoTsay four-node shell ele-
ments with ve integration points across the thickness and one
integration point in the plane [29,33,41,50]. The FGF ller is mod-
eled with eight-node solid elements with one reduced integration
point to avoid volumetric locking [30,31,41]. Convergence test
shows that the shell elements with dimension of 2.5 mm  2.5 mm
[42,43] for the external tube and the solid elements with
dimension of 2.5 mm  2.5 mm  2.5 mm for the foam ller are
sufcient to predict the crushing behavior of FGF-lled FGT tube.
An automatic surface-to-surface contact is chosen to simulate the
contact between the specimen and rigid wall, as well as the contact
between foam and inner wall of tube. An automatic single-surface
contact is adopted to avoid penetration during tubal folding. In
these contacts, the static and dynamic friction coefcients are both
0.20 [30].
To carry out comparative study on different graded types of
foam lled thin-wall structures, nine different congurations,
including four DFG (i.e. FGT lled with FGF), four SFG (i.e. FGT lled
with uniform foam, or uniform thickness tubes lled with FGF) and
one UFT (uniform thickness tube lled with uniform foam) pat-
terns, were considered here. These nine Ascending, Descending
and Uniform patterns are represented by A, D and U for wall thick-
ness and foam ller, respectively in a sequence from outer wall
thickness to inner foam ller. For example, AD denotes the
FGF-lled FGT tube with Ascending tube thickness and
Descending foam density. The schematics of these nine models
with different gradient patterns are depicted in Fig. 6.
To allow meaningful comparison, the mass of column wall and
foam-ller of uniform tube should be equal to that of
functionally-graded counterparts. Thus, the corresponding foam
density and thickness of uniform tube under the same mass of
DFG tube are calculated as:
1 X 1 L
qUF q V FGF i 3
V UF i1 FGF i
L
1 X 2
tUF t FGT i AFGT i 4
AUF i1
where qFGF i and V FGF i are the density and volume of the ith layer of
FGFs, respectively; qUF and V UF represent the corresponding average
density and average volume of UF. L1 is the number of layers for
FGFs. Similarly, tFGT i and AFGT i are the thickness and surface area
of the ith layer of the FGTs; t UF and AUF denote the thickness and
surface area of the UTs. L2 is the number of layers for FGT tubes.
2.4. Material properties
The tube material is aluminum alloy AA6060 T4, whose
mechanical properties are as follows: density q 2:7
103 kg=m3 , Youngs modulus E = 70 GPa, initial yield stress
ry 70 Mpa, ultimate stress ru 173 Mpa and Poissons ratio
v = 0.3 [44]. The constitutive model for aluminum tube is
established based on an elasticplastic material model with von
Mises isotropic plasticity algorithm. Plastic hardening is dened
by several piecewise lines in this material model. The effect of
strain rate is neglected due to the strain rate insensitivity of alu-
minum alloy materials [45,46].
The constitutive model of foam ller is based on the isotropic
uniform material model developed by Reyes et al. [35]. The yield
criterion of this foam material is dened as:
 Y. zhang et al. / Composite Structures 132 (2015) 393405 397

Fig. 6. Schematic diagram of nine different congurations of FE models.

U re  Y 6 0 5 where qf and qf 0 denote the foam density and the base material
where U denotes the yield surface. Y is the yield strength and re is density, respectively. And C 0 , C 1 and q are the material constants,
an equivalent stress, which is given as: which are summarized in Table 1. It can be found from Eq. (9) that
Ep is also a function of foam density.
1  2  Fracture is often modeled in nite element (FE) analyses by
r2e 2 2
a2 rv a rm 6
removing (eroding) elements if their strain reaches a critical
1 3
(threshold) value. In this study, two different criteria are
where rv and rm represent the von Mises effective stress and the used. First, the plastic volumetric strain em exx eyy ezz
mean stress, respectively. a is a parameter that determines the e1 e2 e3 , measured as positive in tension and negative in com-
shape of the yield surface, written as pression, is chosen as an indicator for determining whether or not
an element should be eroded. When a critical, positive value of the
91  2v p
a2 7 volumetric strain is reached, the element will be removed [44]. The
21 v P erosion of element is expressed as: em P ecr , where ecr denotes the
where v p is the plastic Poissons ratio. For the aluminum foam, critical volumetric strain and is dened as 0.1 for aluminum alloy
v p 0 in most cases [34], then a 2:12. The strain hardening rule in the present study [28]. Second, a stress failure criterion was
is adopted in this material model as [44]: evaluated. Note that when the volumetric strain is used in the frac-
ture criterion, only hydrostatic deformation is included. In the
" #
stress criterion, the element will be eroded if r1 P rcr , i.e. the rst
e 1
ry rp c e a2 ln 8 (maximum) principal stress reaches a critical level, where rcr is the
eD 1  ee  eD b critical stress. [44].

where ee is the equivalent strain. rp ; a2 ; c; b and eD are material

Table 1
parameters, which can be expressed as an exponential function of Material constants for aluminum foam [43].
the ratio of foam density to base material density, as
8    ap (Mpa) a2 (Mpa) 1=b c (MPa) Ep (MPa)
> q
< rp ; a2 ; c; 1b ; Ep C 0 C 1 q f C 0 (MPa) 0 0 0.22 0 0
f0
  9 C 1 (MPa) 720 140 320 42 0:33  106
>
: eD  9a22 ln qqf
3a f0
q 2.33 0.45 4.66 1.42 2.45
398 Y. zhang et al. / Composite Structures 132 (2015) 393405

2.5. Validation of FE model

As a novel structure, the FE model of FGF-FGT circular tube has

not been reported in literature yet. Theoretically, traditional uni-
form foam-lled thin-walled (UFT) tube can be considered a speci-
c case of FGF-lled FGT tube. Therefore, validation of the FE model
was performed against the UFT tube against quasi-static crushing
experiment as illustrated in Fig 7(a) and (b). The uniform wall
thickness of tube is 1 mm and the density of uniform foam ller
is 0.28 g/cm3. The specimen was compressed at a constant speed
of 2 mm/min and the characteristic deformation is accounted for
75% of original length (Fig. 7(c)). Note that in the explicit FE simu-
lation a much higher velocity was used [47], in order to balance
computational time and accuracy. Fig. 7(d) and (e) compare the
deformed shapes between FE predictions and experimental results
for the same UFT tube. The overall deformation of the FE model in
terms of collapse mode and number of folds is in good agreement
with that of the experiments. Fig. 8 compares the loaddisplace-
ment curves for the FE modeling and experimental test, in which
the reasonable agreement can be observed. Therefore, the FE
model developed in this study is considered effective for the com- Fig. 8. Comparison of loaddisplacement curve between the FE simulation and
experiment test.
parative studies on the foam-lled FGT tube.

3. Crashworthiness analysis In the automotive engineering, initial peak force is often consid-
ered critical to survival rate of occupants when impact occurs. High
3.1. Crashworthiness indicators initial peak load may lead to severe injury or even death of
occupants [26,50]. A typical forcedisplacement curve of cellular
To evaluate the performance of energy-absorption devices, materials features three key stages: rst, the axial crushing force
there have been a number of crashworthiness indicators proposed rumps up quickly; then it stays at relative stable level; and nally
[48]. Of which specic energy absorption, dened as the energy rises up to a high level when the foam is densied [51]. The
absorbed per unit mass, has been extensively employed, as [49]: optimum loaddisplacement curve should be a ramp function of
EAd positive slope in order to reduce the aggressiveness of the vehicle
SEAd 10
M in low speed impact [52]. For a FGF-FGT structure, the overall force
where d is the axial crushing distance, M is the total mass of the may gradually increase at a relatively proportional manner.
structure and EA is total energy absorbed during crushing, calcu- Moreover, when impact occurs, initial peak force corresponds to
lated as: the maximum outward radial motion in the formation of the rst
Z d fold, then load starts to decrease as the new fold is moved inwards
EAd Fxdx 11 [53]. Thus, for the graded structures, such as tapered tube, the
0 initial peak force (IPF) is often adopted as a key indicator to
where F(x) represents the instantaneous crushing force. evaluation of loading responses [50,53].

Fig. 7. (a) and (b) test specimen, (c) test setup and comparison of deformation patterns for axial crushing loading (d) experimental result (e) simulation result.
 Y. zhang et al. / Composite Structures 132 (2015) 393405
3.2. Crashworthiness of DFG, SFG and UFT structures
In order to compare the crashworthiness of DFG structures with
the SFG and UFT counterparts, full factor design is adopted here to
sample the gradient parameters m and n in 5 different levels. As
such, nine different structures generate totally 225 sample points
for the FE analyses. Table 2 summarizes the full factor design for
the DFG structures with average tube thickness and foam density
corresponding to each combination. Furthermore, to compare the
crashworthiness of different graded structures with the uniform
structures that have the same mass, each DFG model is grouped
with two corresponding SFG and one uniform foam and thickness
(UFT) counterparts. Hence, four groups of the models with different
graded patterns are: AA/AU/UA/UFT (Group 1), AD/AU/UD/UFT
(Group 2), DA/DU/UA/UFT (Group 3) and DD/DU/UD/UFT
(Group 4), respectively. The maximum and minimum values of
foam density and thickness of column are consistent for all
functionally-graded structures: qmax 0:6 g=cm3 , qmin 0:2
g=cm3 and tmin 0:8 mm, respectively. The crushing distance is
considered as 130 mm for all these models, accounting for 65% of
the original length, covering the majority of plateau region.
3.2.1. Comparison of specic energy absorption
With the 225 design responses (SEA, IPF) obtained from the full
factor sampling FE analyses, SEA of the four groups of models are
illustrated with the 3D bar charts in Fig. 9 through different com-
binations of n and m. Taking Group 1 as an example (Fig. 9(a)), SEA
has different changing trends for m and n. For the DFG structures
considered here, SEA monotonously declines as m and n changing
from convex to concave; whereas for the SFG and UFT structures,
SEA decreases as m changing from convex to concave, but increases
as n changing from convex to concave. Generally speaking, the DFG
structures exhibited superior energy absorption capacity, espe-
cially at the convex regions. The similar results can also be
observed for Groups 2, 3 and 4 in Fig. 9(b)(d), respectively.
To compare the energy absorption capacity of the DFG struc-
tures with the SFG and UFT counterparts, the SEA values of the
DFG structure are subtracted from those of SFG or UFT structures,
thus the surfaces of discrepancy are constructed by tting the sub-
tracted values. The reference plane of zero value is included to dis-
tinguish the positive and negative discrepancies here. Finally, the
discrepancy of SEA for these 4 groups is plotted in Fig. 10.
First, the DFG structures are compared with conventional
uniform structures, as we are mostly concerned with the perfor-
mance of DFG structures over such UFT counterparts. Fig. 10(a)
shows that SEA of AA is higher than UFT except for the right
triangular concave gradient region. Similar observations can be
made in Fig. 10(d), (g) and (j), indicating that convex gradient
parameters strengthen the energy absorption capacity of DFG
structures.
Second, the DFG structures are compared with the two corre-
sponding SFG structures. For comparing the SFG structures lled
with uniform foam density, Fig. 10(b) shows that SEA of AA
conguration is higher than the AU counterpart besides the
same concave gradient region with a shrunk area. Consistent
ndings are shown in Fig 10(e), (h) and (k). Therefore, ascending
foam pattern and descending foam pattern can absorb energy
more efciently than uniform foam except for the less concave
gradient regions. For comparing the DFG structures with the
SFG structures that have uniform thickness,
Fig. 10(c), (f), (i) and (j) shows that at the concave tube thick-
ness gradient regions, DFG has lower SEA than the SFG struc-
tures with uniform tube thickness. From the above ndings, it
is known that the advantage of DFG will be more prominent
when choosing a convex gradient.
399
Overall, capacity of energy absorption of DFG structures is supe-
rior to SFG and UFT structures. Nevertheless, the gradient expo-
nents govern the performance of DFG structures: convex
gradients are a desirable choice for DFG structures, while concave
gradients should be avoided. Concave foam gradient causes foam
ller to undergo quick densication, absorbing only a small
amount of energy [26]. The convex foam gradient for both ascending
and descending patterns makes the ller denser in a progressive
fashion, thereby being more efcient to absorb impact energy.
Similarly, concave thickness gradient makes tube wall relative thin
overall, which reduces the stability in progressive folding; whereas
convex thickness gradient enhances the stability.
3.2.2. Comparison of Initial Peak Load
Initial Peak Load (IPL) reects the initial peak decelerations dur-
ing impact. The peak deceleration transmitted to the impacted
body may be detrimental to the structure under consideration
[53], or fetal to the passengers as it may cause internal dam-
age/bleeding. Therefore, it is of great importance to comparing
the IPL response of DFG structures with the SFG and UFT struc-
tures. Here we collected the force transmitted to the rigid wall,
which symbolizes the protected structures during impact.
Fig. 11 compares the IPL results, in which the effect of gradient
parameters is represented with average tube thickness and average
foam density due to the opposite trend of convexity and concavity
under ascending and descending patterns, respectively.
Firstly, comparison was made to identify the IPL responses for
different foam gradients. Fig. 11(a) compares IPL of AA/AU/AD pat-
terns with ascending wall thickness. It can be seen that IPL of
descending foam pattern is higher than that of uniform foam;
whereas the ascending foam pattern keeps IPL at fairly low level.
Therefore, the ascending foam pattern is considered favorable for
energy absorber as it also retains IPL and deceleration at a rela-
tively low level. For the uniform and descending foam pattern,
IPL is higher than that for the ascending foam pattern. Therefore,
descending or uniform foam conguration is considered unfavor-
able in reducing the impact loading. Similar phenomenon is also
observed in Fig. 11(b) and (c), where uniform thickness pattern is
favorable to descending thickness.
Secondly, comparison was made to investigate loading
responses for different wall thickness patterns.
Fig. 11(d), (e), and (f) compare the IPL of different graded thick-
nesses under the same foam gradient. For example, Fig. 11(d) rep-
resents the IPL of the three graded wall thicknesses (ascending,
descending and uniform) which have the same ascending foam l-
ler. It can be seen from these gures that the descending wall
thickness causes higher IPL than uniform wall thickness does. On
the contrary, ascending wall thickness has the lowest IPL. As a
result, it is concluded that the ascending thickness pattern is favor-
able of these three graded wall thicknesses. Besides, it can be seen
that the difference of IPL between the descending thickness and
ascending thickness (Fig. 11(d)(f)) is smaller than that between
the descending foam and ascending foam (Fig. 11(a)(c)).
Therefore, foam gradient has larger effect than thickness gradient
for the loading responses in the current setting.
On the whole, it can be stated that, descending foam and
descending thickness congurations generate a higher IPL than
the other patterns. Moreover, IPL of uniform and descending foam
pattern exhibits great sensitivity to the average foam density. On
the contrary, wall thickness pattern does not display much sensi-
tivity to average wall thickness.
3.2.3. Crashworthiness analysis of different DFG structures
The previous discussion revealed that in general, the DFG struc-
tures are advantageous over the SFG and UFT structures from
400 Y. zhang et al. / Composite Structures 132 (2015) 393405

Table 2
Average tube thickness and foam density for given combinations of gradient parameters (ascending and descending pattern).

m n
1 ascend 0.5 ascend 0 ascend 0.5 ascend 1 ascend
1 descend 0.5 descend 0 descend 0.5 descend 1 descend
1 ascend taver 1:422 taver 1:326 t aver 1:15 t aver 0:973 t aver 0:871
1 descend qaver 556:6 qaver 556:6 qaver 556:6 qaver 556:6 qaver 556:6
0.5 ascend taver 1:422 taver 1:326 t aver 1:15 t aver 0:973 t aver 0:871
0.5 descend qaver 500:5 qaver 500:5 qaver 500:5 qaver 500:5 qaver 500:5
0 ascend taver 1:422 taver 1:326 t aver 1:15 t aver 0:973 t aver 0:871
0 descend qaver 400 qaver 400 qaver 400 qaver 400 qaver 400
0.5 ascend taver 1:422 taver 1:326 t aver 1:15 t aver 0:973 t aver 0:871
0.5 descend qaver 298:9 qaver 298:9 qaver 298:9 qaver 298:9 qaver 298:9
1 ascend taver 1:422 taver 1:326 t aver 1:15 t aver 0:973 t aver 0:871
1 descend qaver 240:6 qaver 240:6 qaver 240:6 qaver 240:6 qaver 240:6
The unit for tube thickness and foam density are mm and kg=m3 , respectively.

Fig. 9. SEA of 4 Groups.

energy absorption perspective. As for loading response, however, evaluate the overall crashworthiness characteristics of DFG struc-
the performance of DFG structures highly depends on the gradient tures, comparison of AA/AD/DA/DD is further conducted here.
pattern. Ascending pattern has quite different loading response From Fig 12(a), it is seen that the SEAs of these four DFG
with descending pattern. Therefore, to more comprehensively structures exhibit fairly similar patterns. All of these four DFG
Y. zhang et al. / Composite Structures 132 (2015) 393405 401

Fig. 10. SEA Discrepancy surfaces for Groups 14.

402 Y. zhang et al. / Composite Structures 132 (2015) 393405

Fig. 11. Comparison of Initial Peak Load in different gradient congurations.

structures absorb energy at a rather close level for the given com-
bination of gradient parameters. In other words, grading direction
has marginal effects on energy absorption capacity (SEA) of the
DFG structures [30]. To further explore this phenomenon, the
deformation processes of sectional views of AA/AD/DA/DD cong-
urations are displayed in Fig. 13 at 5 ms, 10 ms and 15 ms, respec-
tively. It is noted that the deformation lengths and the numbers of
folds of all the four gradient congurations are very close to each
other at the given time step, explaining their relatively close level
of energy absorption.
Fig. 12(b) compares the IPL of these four DFG structures. It can
be seen that AA and DA congurations have a relatively low IPL,
whereas AD and DD patterns have a higher IPL. Among which AA
tube has the lowest Initial Peak Load overall. Furthermore, the

Fig. 12. SEA and Initial Peak Load of AA/AD/DA/DD congurations.

 Y. zhang et al. / Composite Structures 132 (2015) 393405 403

Fig. 13. Deformation process of AA/AD/DA/DD congurations.

Fig. 14. SEA and Initial Peak Load of AA conguration with varying tube thickness range.

deformation pattern exhibited in Fig. 13 can also be utilized to
explain this observation. It can be noted that the progressive fold-
ing of tube wall and compression of foam ller rst occur simulta-
neously at their low stiffness regions independently, then
propagates to the rest of the tube structure. Since the low stiffness
regions are placed next to rigid wall (which stands for protected
structure) for AA-Type, initial kinetic energy is quickly absorbed
due to easier deformation, thus the force transmitted to protected
structure is relatively low.
details of functionally-graded elements are: gradient parameters,
range of the graded parameters, and the pattern of gradient param-
eters. Since the effects of gradient parameters have been discussed
in the previous section; the following section explores the effects
of tube thickness range and foam density range, respectively, aim-
ing to provide a guideline for further improving the performance of
DFG structures.
4.1. Wall thickness range

4. Parametric study of AA-type graded structure
The previous section indicates that the AA tube is the most
preferable DFG structures, it is thus to further perform a paramet-
ric study on the main factors that affect the crashworthiness of AA
structure. The three factors that determine the congurational
Tube thickness is an important factor for crashworthiness as it
largely inuences deformation pattern [54]. Thus, we extended
the upper thickness bound of current model to tmax 2 mm (tmin
remains the same as 0.8 mm) in this study, while the foam ller
is kept unchanged. The simulation results under the same impact
conditions are presented in Fig. 14. It is shown that a wider
404 Y. zhang et al. / Composite Structures 132 (2015) 393405
Fig. 15. SEA and Initial Peak Load of AA conguration with varying foam density range.
thickness range could improve SEA overall, as seen in Fig 14(a).
Nevertheless, Fig. 14(b) indicates that the structure with wider
thickness range generates a higher IPL. Interestingly, it is noted
that the increase in the magnitude is moderate for SEA and IPL
by widening the thickness range. Therefore, it is believed that tube
thickness does not seem to be a major factor that alters the crash-
worthiness characteristics greatly in the present graded composite
structures.
4.2. Foam density range
The previous studies demonstrated that FGF can be efciently
used to absorb impact energy and reduce IPL, upon being properly
designed. The variation in the foam density range
(Dq qmax  qmin ) affects the performance of FGF considerably
[25]. Nevertheless, the upper bound of foam density in the current
model already falls in a high density range. Therefore, to investi-
gate the effect of foam density range in the composite structure,
we lowered the upper bound of foam density to qmax = 0.4 g/cm3
for comparison. The simulation results under the same impact con-
dition are presented in Fig. 15. It can be seen from Fig. 15(b) that
the IPL with a lower foam density range does not reduced dramat-
ically. However, SEA of the tube with a smaller density range falls
at drastically at the convex foam gradient region, seen in
Fig. 15(a). It is thus noted that a greater foam density range is of
more benet to improve energy absorption efciency of the AA
tube [25].
5. Conclusions
The crushing behaviors of DFG/SFG and conventional uniform
foam-lled thin-walled circular tubes have been studied in this
paper. The parametric study demonstrated that the AA congura-
tion of DFG structures has desirable overall crashworthiness in
terms of energy absorption and peak load response. For the AA
conguration, further parametric studies were performed. The
main ndings of this study are summarized as follows:
(1) Functional graded structures exhibit superior energy
absorption capacity over corresponding uniform counter-
part. DFG structures absorb more impact energy per unit
mass than corresponding SFG structures and UFT structures
overall.
(2) Energy absorption capacity of DFG is greatly inuenced by
the combined gradient parameters. The combination of con-
vex gradients improves the SEA of DFG structures, whereas
combination of concave gradients downgrades SEA. From
the examples, gradient of foam density is more inuential
than gradient of tube thickness.
(3) The effect of gradient direction on energy absorption was
found marginal under the given conditions. However, the
gradient direction is critical for generating initial peak load.
Ascending pattern, where less stiff part was attached to the
rigid wall, and stiffest part subjected to the impactor, is
shown to be an efcient mechanism for reducing initial peak
force. Therefore, AA tube was found to be the best of these
possible candidates for designing future energy absorption
devices.
(4) The crashworthiness of AA conguration can be improved if
the ranges of foam density and tube thickness are increased,
in which increasing foam density range was shown to be
more effective.
Acknowledgments
This work is supported by The National Natural Science
Foundation of China (51205141, 51305143, 11202072), Fujian
Province Natural Science Foundation of China (2015J01204),
Promotion Program for Young and Middle-aged Teacher in
Science and Technology Research of Huaqiao University
(ZQN-PY202), China Postdoctoral Science Foundation funded pro-
ject (2014T70613).
References
[1] Abramowicz W, Jones N. Dynamic axial crushing of circular tubes. Int J Impact
Eng 1984;2(3):26381.
[2] Abramowicz W, Jones N. Dynamic progressive buckling of circular and square
tubes. Int J Impact Eng 1986;4(4):24370.
[3] White MD, Jones N. Experimental study into the energy absorbing
characteristics of top-hat and double-hat sections subjected to dynamic
axial loading. Proc Inst Mech Eng, Part D: J Automob Eng 1999;213(3):
25978.
[4] White MD, Jones N, Abramowicz W. A theoretical analysis for the quasi-static
axial crushing of top-hat and double-hat thin-walled sections. Int J Mech Sci
1999;41(2):20933.
[5] Seitzberger M, Rammerstorfer FG, Gardinger R, Degischer HP, Blaimschein M,
Walch C. Experimental studies on the quasi-static axial crushing of steel
columns lled with aluminium foam. Int J Solids Struct 2000;37:412547.
 Y. zhang et al. / Composite Structures 132 (2015) 393405
[6] Hanssen AG, Langseth M, Hopperstad OS. Static and dynamic crushing of
circular aluminium extrusions with aluminium foam ller. Int J Impact Eng
2000;24:475507.
[7] Hanssen AG, Langseth M, Hopperstad OS. Static and dynamic crushing of
square aluminium extrusions with aluminium foam ller. Int J Impact Eng
2000;24(3):47383.
[8] Reid SR, Reddy TY, Gray MD. Static and dynamic axial crushing of foam-lled
sheet metal tubes. Int J Mech Sci 1986;28:295322.
[9] Reddy TY, Wall RJ. Axial compression of foam-lled thin-walled circular tubes.
Int J Impact Eng 1988;7(2):15166.
[10] Seitzberger M, Rammerstorfer FG, Degischer HP, Gradinger R. Crushing of
axially compressed steel tubes lled with aluminium foam. Acta Mech
1997;125(14):93105.
[11] Hou SJ, Li Q, Long SY, Yang XJ, Li W. Crashworthiness design for foam lled
thin-wall structures. Mater Des 2009;30(6):202432.
[12] Zarei HR, Kroger M. Crashworthiness optimization of empty and lled
aluminum crash boxes. Int J Crashworthiness 2007;12(3):25564.
[13] Zarei HR, Kroger M. Optimization of the foam-lled aluminum tubes for crush
box application. Thin-Walled Struct 2008;46(2):21421.
[14] Nariman-Zadeh N, Darvizeh A, Jamali A. Pareto optimization of energy
absorption of square aluminium columns using multi-objective genetic
algorithms. Proc Inst Mech Eng Part B-J Eng Manuf 2006;220(2):21324.
[15] Zhang Y, Sun GY, Li GY, Luo Z, Li Q. Optimization of foam-lled bitubal
structures for crashworthiness criteria. Mater Des 2012;38:99109.
[16] Sun GY, Xu FX, Li GY, Li Q. Crashing analysis and multiobjective optimization
for thin-walled structures with functionally graded thickness. Int J Impact Eng
2014;64:6274.
[17] Liew KM, Lei ZX, Zhang LW. Mechanical analysis of functionally graded carbon
nanotube reinforced composites: a review. Compos Struct 2015;120:907.
[18] Habibi MK, Lu Y. Crack propagation in bamboos hierarchical cellular structure.
Sci Rep 2014;4:5598.
[19] Cadman J, Chang C, Chen JN, Chen YH, Zhou SW, Li W, et al. Bioinspired
lightweight cellular materials understanding effects of natural variation on
mechanical properties. Mater Sci Eng, C 2013;33:314652.
[20] Jiang YB, Zhao J, Liao EY, Dai RC, Wu XP, Genant HK. Application of micro-CT
assessment of 3-D bone microstructure in preclinical and clinical studies. J
Bone Miner Metab 2005;23:12231.
[21] Lin D, Li Q, Li W, Swain MV. Bone remodeling induced by dental implants of
functionally-graded materials. J Biomed Mater Res: Part B Appl Biomater
2010;92:4308.
[22] Bohidar SK, Sharma R, Mishra PR. Functionally graded materials: a critical
review. Int J Res (IJR) 2014;1(7):289301.
[23] Evangelista F, Roesler J, Paulino G. Numerical simulations of fracture resistance
of functionally graded concrete materials. Transp Res Rec 2009;2113:12231.
[24] Brothers AH, Dunand DC. Mechanical properties of a density-graded replicated
aluminum foam. Mater Sci Eng, A 2008;489(12):43943.
[25] Matsumoto Y, Brothers AH, Stock SR, Dunand DC. Uniform and graded
chemical milling of aluminum foams. Mater Sci Eng, A 2007;447(12):1507.
[26] Cui L, Kiernan S, Gilchrist MD. Designing the energy absorption capacity of
functionally graded foam materials. Mater Sci Eng, A 2009;507(12):21525.
[27] Sun GY, Li GY, Hou SJ, Zhou SW, Li W, Li Q. Crashworthiness design for
functionally graded foam-lled thin-walled structures. Mater Sci Eng, A
2010;527:19119.
[28] Yin HF, Wen GL, Fang HB, Qing QX, Kong XZ, Xiao JR, et al. Multiobjective
crashworthiness optimization design of functionally graded foam-lled
tapered tube based on dynamic ensemble metamodel. Mater Des
2014;55:74757.
[29] Fang JG, Gao YK, Sun GY, Zhang YT, Li Q. Parametric analysis and
multiobjective optimization for functionally graded foam-lled thin-wall
tube under lateral impact. Comput Mater Sci 2014;90:26575.
[30] Yin HF, Wen GL, Hou SJ, Qing QX. Multiobjective crashworthiness optimization
of functionally lateral graded foam-lled tubes. Mater Des 2013;44:41428.
405
[31] Attia MS, Meguid SA, Nouraei H. Nonlinear nite element analysis of the crush
behaviour of functionally graded foam-lled columns. Finite Elem Anal Des
2012;61:509.
[32] Yang R, Fu Y, Li G. Application of tailor rolled blank in vehicle front end for
frontal impact. SAE Technical Paper; 2007. 2007-01-0675.
[33] Li GY, Xu FX, Sun GY, Li Q. A comparative study on thin-walled structures with
functionally graded thickness (FGT) and tapered tubes withstanding oblique
impact loading. Int J Impact Eng 2015;77:6883.
[34] Yoshihiko H, Kazuya T, Takao U, Soichiro K, Osamu K, Nobuhiro Y. Fabrication
of functionally graded aluminum foam using aluminum alloy die castings by
friction stir processing. Mater Sci Eng, A 2012;534:7169.
[35] Yoshihiko H, Kazuya T, Ryo Y, Takao U, Soichiro K, Osamu K, et al.
Nondestructive observation of pore structure deformation behavior of
functionally graded aluminum foam by X-ray computed tomography. Mater
Sci Eng, A 2012;556:67884.
[36] He SY, Zhang Y, Dai G, Jiang JQ. Preparation of density-graded aluminum foam.
Mater Sci Eng, A 2014;618:4969.
[37] Zhang X, Zhang H, Wen ZZ. Axial crushing of tapered circular tubes with
graded thickness. Int J Mech Sci 2015;92:1223.
[38] Langseth M, Hopperstad OS, Hanssen AG. Crash behaviour of thin-walled
aluminium members. Thin-Walled Struct 1998;32(13):12750.
[39] Zhang Xiong, Zhang Hui. Energy absorption of multi-cell stub columns under
axial compression. Thin-Walled Struct 2013;68:15663.
[40] Ali N, Masoud RR. Mechanics of axial plastic collapse in multi-cell, multi-
corner crush tubes. Thin-Walled Struct 2011;49:112.
[41] Li GY, Zhang ZSH, Sun GY, Xu FX, Huang XD. Crushing analysis and
multiobjective optimization for functionally graded foam-lled tubes under
multiple load cases. Int J Mech Sci 2014;89:43952.
[42] Ahmad Z, Thambiratnam DP. Crushing response of foam-lled conical tubes
under quasi-static axial loading. Mater Des 2009;30:2393403.
[43] Zhang X, Cheng GD. A comparative study of energy absorption characteristics
of foam-lled and multi-cell square columns. Int J Impact Eng 2007;34:
173952.
[44] Hanssen AG, Hopperstad OS, Langseth M, IIstad H. Validation of constitutive
models applicable to aluminium foams. Int J Mech Sci 2002;44(2):359406.
[45] Hou SJ, Han X, Sun GY, Long SY, Li W, Yang XJ, et al. Multiobjective
optimization for tapered circular tubes. Thin-Walled Struct 2011;49:85563.
[46] Zhang ZH, Liu ST, Tang ZL. Comparisons of honeycomb sandwich and foam-
lled cylindrical columns under axial crushing loads. Thin-Walled Struct 2011;
49(9):10719.
[47] Mamalis AG, Manolakos DE, Ioannidis MB, Papapostolou DP. The static and
dynamic axial collapse of CFRP square tubes: nite element modelling.
Compos Struct 2006;74:21325.
[48] Reyes A, Hopperstad OS, Berstad T, et al. Constitutive modeling of aluminum
foam including fracture and statistical variation of density. Eur J Mech A/Solids
2003;22(6):81535.
[49] Kim HS. New extruded multi-cell aluminum prole for maximum crash energy
absorption and weight efciency. Thin-Walled Struct 2002;40(4):31127.
[50] Ahmad Z, Thambiratnam DP. Dynamic computer simulation and energy
absorption of foam-lled conical tubes under axial impact loading. Comput
Struct 2009;87(34):18697.
[51] Gibson LJ, Ashby MF. Cellular solids: structures and properties. 2nd
ed. Cambridge, UK: University Press; 1997.
[52] Daneshi GH, Hosseinipour SJ. Grooves effect on crashworthiness
characteristics of thin-walled tubes under axial compression. Mater Des
2003;23:6117.
[53] Ghamarian A, Zarei HR, Farsi MA, Ariaeifar N. Experimental and numerical
crashworthiness investigation of the empty and foam-lled conical tube with
shallow spherical caps. Strain 2013;49(3):199211.
[54] Yamashita M, Gotoh M, Sawairi Y. Axial crush of hollow cylindrical structures
with various polygonal cross-sections numerical simulation and experiment. J
Mater Process Technol 2003;140:5964.