Thin-Walled Structures 82 (2014) 183195

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Theoretical prediction and crashworthiness optimization

of multi-cell triangular tubes
TrongNhan Tran a,c, Shujuan Hou a,b, Xu Han a,b,n, Wei Tan a,b, NhatTan Nguyen a,d
a

State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha 410082, Hunan, PR China
College of Mechanical and Vehicle Engineering, Hunan University, Changsha 410082, Hunan, PR China
c
Faculty of Mechanical Engineering, Industrial University of Ho Chi Minh City, Go Vap District, HCM City, Vietnam
d
Center for Mechanical Engineering, Hanoi University of Industry, Tu Liem District, Ha Noi, Vietnam
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 2 January 2014
Received in revised form
13 March 2014
Accepted 26 March 2014
Available online 17 May 2014

The triangular tubes with multi-cell were rst studied on the aspects of theoretical prediction and
crashworthiness optimization design under the impact loading. The tubes' proles were divided into
2-, 3-, T-shapes, 4-, and 6-panel angle elements. The Simplied Super Folding Element theory was
utilized to estimate the energy dissipation of angle elements. Based on the estimation, theoretical
expressions of the mean crushing force were developed for three types of tubes under dynamic loading.
When taking the inertia effects into account, the dynamic enhancement coefcient was also considered.
In the process of multiobjective crashworthiness optimization, Deb and Gupta method was utilized to
nd out the knee points from the Pareto solutions space. Finally, the theoretical prediction showed an
excellent coincidence with the numerical optimal results, and also validated the efciency of the
crashworthiness optimization design method based on surrogate models.
Crown Copyright & 2014 Published by Elsevier Ltd. All rights reserved.

Keywords:
Crashworthiness
Multiobjective optimization
Triangular tube
Multi-cell
Energy absorption
Impact loading

1. Introduction
Thin-walled extrusions have been extensively applied in vehicle
crashworthiness components to absorb impact energy in the past
three decades. The tests and the theoretical expressions of square
and circular tubes under axial quasi-static and dynamic loading cases
were rst described by Wierzbicki and Abramowicz [1] and Abramowicz and Jones [2]. From then on, DiPaolo et al. [3,4], Guillow et al.
[5], Ullah [6] Zhang and Zhang [7], Alavi Nia and Parsapour [8] also
did many researches on these aspects. Beside square and circular
tubes, several other proles were also studied on their quasi-static or
dynamic responses, such as triangular tubes [912], hexagonal tubes
[13], etc. The structural collapse modes of triangular and square tubes
are different from those of circular tubes. Nevertheless, the crushing
curves of forcedisplacement of triangular and square tubes are
similar to those of circular tubes. The crushing curves of force
displacement of all the proles show that the crushing force rst
reaches an initial peak, then drops down and then uctuates around
a value of the mean crushing force. The extensional deformation has

n
Corresponding author at: College of Mechanical and Vehicle Engineering,
Hunan University, Changsha, Hunan 410082, PR China.
E-mail addresses: shujuanhou@hnu.edu.cn (S. Hou),
hanxu@hnu.edu.cn (X. Han).

http://dx.doi.org/10.1016/j.tws.2014.03.019
0263-8231/Crown Copyright & 2014 Published by Elsevier Ltd. All rights reserved.

more dominant effect on the crushing responses while the quasiinextensional mode occurs normally [14].
According to studies by Wierzbicki and Abramowicz [1], the
number of angle elements on cross-section of tube decided,
to a certain extent, the effectiveness of energy absorption. As
a matter of fact, it is necessary to design thin-walled multi-cell
tubes for weight-efcient energy absorption. Chen and Wierzbicki
[15] examined the axial crushing resistance of single-cell, doublecell and triple-cell hollow tubes, and the respective foam-lled
tubes under the quasi-static axial loading. The Simplied Super
Folding Element (SSFE) theory was applied to simplify SFE theory,
and three extensional triangular elements and three stationary
hinge lines were comprised instead of the kinematically admissible
model of SFE [1]. The average folding wavelength and the theoretical expression of the mean crushing force were deduced by
dividing the cross-sectional tube into distinct panel section and
angle element, assuming that the roles of each panel and of angle
element were at the same level. The work of Chen and Wierzbicki
[15] showed that the multi-cell tube could increase the specic
energy absorption SEA by approximately 15%, compared to the
respective hollow tube. Kim [16] used Chen and Wierzbicki's model
[15] to study multi-cell tubes with four square elements at the
corner. The SEA of new multi-cell tube was reported to increase by
190%, compared to conventional square tube. Zhang et al. [17] also
applied SSFE theory to derive a theoretical expression of the mean

184

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

crush force of multi-cell square tubes under the dynamic impact

loading. In Zhang's work, the cross-section of tube was divided into
three basic angle elements, and the study also came to the
contribution that plastic energy of each element type was dissipated through membrane action. It was assumed from the
theoretical expression that the average wavelength for the dissimilar folds developed at corners. Thereafter, the SSFE theory was also
adopted by Zhang et al. [18] to predict the mean crushing force of 3panel angle element. At the same time, the SFE theory was
extended by Naja and Rais-Rohani [19] to explore the crushing
characteristics of multi-cell tubes with two different types of threepanel elements. A closed form expression of mean crushing force
was also put forward by Naja and Rais-Rohani [19].
Dynamic progressive buckling of thin-walled multi-cell tubes
under axial impact loadings was studied by Jensen et al. [20] and
Karagiozova and Jone [21]. Then, the structural dynamic progressive buckling under the axial loading was summarized by Karagiozova and Alves [22] from a phenomenological point of view.
Consequently, the desirable energy-dissipating mechanism was
a stable and progressive folding deformation pattern for the
structural deformation. On the other hand, the global bending on
a structure was an undesirable energy-dissipating mechanism
mode. At the beginning, multi-cell tubes were mostly employed
from the aspects of theoretical researches, such as by Kim [16] and
Naja and Rais-Rohani [19], Nowadays, either FE solutions [23] or
surrogate models [13,2426] developed appeared in the search
eld of multi-cell tubes under the impact loading. However, there
is seldom a combination study of theory, numeric and optimal
method for thin-walled multi-cell tubes.
Above all, the axial crushing of tube types I, II, III was studied
on both theoretical prediction and numerical optimization design
in this paper. Based on the SSFE theory, theoretical expressions of
the mean crushing force for the three types were derived. All the
proles studies in this paper were divided into 2-, 3-, T-shape,
4 and 6-panel angle elements. In order to obtain the optimal
proles under the crashworthiness criterion, dynamic nite
element analysis code ANSYS/LS-DYNA was executed to simulate
tubes and to obtain the numerical results at the design sampling
points. The multiobjective optimization design was utilized to
obtain the optimal congurations. Finally, the theoretical expressions are employed to validate the numerical optimal solutions.

wavelength 2H for different lobes was ignored. To analyze energy

dissipation over the collapse of a fold, the triangular multi-cell
thin-walled tubes were divided into several basic elements: the 2-,
3-, 4- and 6-panel angle element as shown in Fig. 1.
Based on the principle of global equilibrium for shells, the
internal and external energy dissipations are of equal rate E_ ext
E_ int . The external energy work for a complete single fold is equal to
the sum of dissipated bending and membrane energy. That is
1
P m 2H Eb Em

where Pm, 2H, Eb and Em respectively denote the mean crushing

force, the length of the fold, the bending energy and the membrane
energy, and is the effective crushing distance coefcient. The panel
of folding element after deformation is not completely attened as
shown in Fig. 2. Hence, the available crushing displacement is
smaller than 2H. In this study, the value of was taken as 0.7 since it
was found between 0.7 and 0.8 [1].
2.1.1. The bending energy of tube
The SSFE theory was applied to calculate the dissipated energy
in bending of each panel. Only three extensional and compressional triangular elements and three stationary hinge lines were
used in SSFE theory, which was different from SFE theory. In SFE
theory, a model was built with trapezoidal, toroidal, conical and
cylindrical surfaces of moving hinge lines.
In the work of Chen and Wierzbicki [15], the energy dissipation
for bending of each panel Eb was estimated by summing up the
energy dissipation at three stationary hinge lines. Then
4

Efb M 0 i b
i1

where M 0 s0 t 2 =4 is the fully plastic bending moment of the

panel, b is the sectional breadth and denotes the rotation angle
at the stationary hinge lines.
In this case, the panel was supposed to completely atten after
the deformation of the wavelength 2H. Consequently, the four
rotation angles at three stationary hinge lines were /2 one by
one (as shown in Fig. 2). By applying Eq. (2), the bending energy at

2. Theoretics
2.1. Theoretical prediction of multi-cell triangular tube
The SSFE theory was applied to solve the axial collapse of
triangular multi-cell thin-walled tubes. In the SSFE theory, the wall
thickness was assumed to be constant and the variation of

Fig. 2. Bending hinge lines and rotation angles on basic folding.

Fig. 1. Cross-sectional geometry of triangular multi-cell tube and typical angle element. (a) Tube type I (b) Tube type II and (c) Tube type III.

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

stationary hinge lines of panel is obtained as

Efb

2 M 0 b

Since the role of structural each panel is similar and the multicell tube is constituted by m panels (as shown in Fig. 1), the energy
dissipation for bending of multi-cell tube is inferred as
2 M 0 mb 2 M 0 B
Etube
b

The energy dissipation in membrane of right corner element

for the symmetric mode is determined as

2.1.2. The membrane energy of angle element

2.1.2.1. The membrane energy of 2-panel, 3-panel and T-shape angle
element. The basic folding element (BFE) was created by using the
triangular elements and the stationary hinge lines (Fig. 3) so as to
calculate the membrane energy of right-corner through asymmetric
or symmetric deformation mode in the SSFE theory. Two possible
collapse modes of the asymmetric and symmetric deformation were
supposed in the establishment of BFE. The symmetric and asymmetric
modes came from the extensional and quasi-inextensional modes
respectively. As for the asymmetric mode, the three triangular
elements were developed for each web after the deformation.
However, the symmetric mode had two triangular elements for each
panel after the deformation. Thus for the asymmetric mode during
one wavelength crushing, the energy dissipation in membrane Em of
each panel was evaluated by integrating the area of triangular
elements (shaded areas in Fig. 3a). Then
Z
1
H2
5
Easym
s0 tds s0 tH 2 2M 0
mf
2
t
s
Each panel was assumed to have the similar contribution. For the
asymmetric mode, the dissipated membrane energy of the right
corner element is double of that in one single panel. Then
H2
t

For the symmetric mode during one wavelength crushing, the

dissipated energy in membrane of each panel was estimated by
integrating the area of triangular elements (shaded areas in Fig. 3b).
Then
Z
H2
Esym
7

s0 tds s0 tH2 4M0

mf
t
s

where B is the sum of side and internal web lengths.

f
Easym
m_r  c 2E m 4M 0

185

sym
Esym
m_r  c 2E m_f 8M 0

H2
t

Concerning hollow tubes such as triangular, square, pentagonal

and hexagonal, the cross-section proles were formed by 2-panel
angle elements. The crushing forces of hollow tubes were strengthened in turn from triangular tube to hexagonal tube. When the
central angle varies from 301 to 1201, the crushing force of 2-panel
angle elements has an appropriate increase [27]. Eq. (1) shows that
the mean crushing force is a function of bending and membrane
energy. Because the role of each structural panel is assumed to be
similar, the bending energy of angle elements with the same
numbers of panel will not change. Then, the mean crushing force
will vary with respect to the membrane energy. Thus, the membrane
energy of right corner element is bigger than that of 2-panel angle
element with central angle of 601. From Fig. 4a, the membrane
energy of a 2-panel angle, during one wavelength crushing, is
calculated as
 panel
Eind:2
Easym
m
m_r  c cos 4M 0

H2
cos
t

The energy dissipation in membrane of 3-panel angle element

was calculated by Zhang and Zhang [18]. Then, the membrane

Fig. 3. Basic folding element: (a) asymmetric mode [15] and (b) symmetric (extensional) mode.

Fig. 4. (a) Relationship between the membrane energy of right corner and of 2-panel angle element and (b) 3-panel angle element.

186

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

energy of 3-panel angle element, during one wavelength crushing,

is
E3m panel 4M 0

H2
1 2 tan =2
t

10

T-shape element was created by one right-corner element and

one additional panel as shown in Fig. 5. For this reason, the
membrane energy of T-shape element was calculated by a sum
between membrane energy of right-corner element for the symmetric mode and membrane energy of one additional panel.
Because the contribution of each panel is assumed to be similar,
the energy dissipated in membrane of T-shape element, during
one wavelength crushing, is more than triple each panel's membrane energy. Then
ETm shape 3Esym
12M 0
mf

H2
t

membrane energy of 4-panel angle element was estimated by

summing up the membrane energy of 2-panel angle element for
symmetric mode and the membrane energy of two additional
panels.
It is not easy to give a precise calculation for membrane energy
of two additional panels. In this case, it is too complicated for the
SFE theory to be applied here. So, a simplied deformation model
of two additional panels was proposed and the SSFE theory was
utilized to solve this problem. Fig. 7b shows the membrane
triangular elements developed at the corner line. Accordingly,
the membrane energy Em of 4-panel angle element, during one
wavelength crushing, is evaluated by integrating the triangular
areas as


H2
1
1
12
E4m panel 8M 0
t
cos

11

2.1.2.2. The membrane energy of 4-panel angle element. The 4-panel

angle element is a symmetric structure and is constituted by one
2-panel angle element and two additional panels. Therefore, the
energy absorption in membrane of a 4-panel angle element was
calculated by a sum of membrane energy of 2-panel angle element
and membrane energy of two additional panels. Due to similar
deformation mode shown in Fig. 6, it is assumed that 2-panel
angle element and the corresponding 2-panel angle element
in the 4-panel angle element are of equal crushing resistance.
Simultaneously, the deformation mode of 2-panel angle element is
a symmetric mode in this case. For this reason, the dissipated

2.1.2.3. The membrane energy of 6-panel angle element. As

a symmetric structure and being composed of six panels, the
dissipated energy in membrane of a 6-panel angle element was
calculated by the sum of membrane energy absorbed by all
6 panels. Because of the symmetric structure and the similar
contribution of each angle element, the 6-panel angle element was
formed by two 4-panel angle elements (Fig. 8a). As a result, the
membrane energy of 6-panel angle element, during one
wavelength crushing, was estimated by the sum of membrane
energy absorbed by two 4-panel angle elements. This is


H2
1
E6m panel 2E4panel
1

16M
13
0
m
t
cos

Fig. 5. (a) Collapse mode of T-shape element and (b) extensional elements.

Fig. 6. (a) 2-panel angle element and (b) 4-panel angle element.

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

187

Fig. 7. (a) Collapse mode of 4-panel angle element and (b) extensional elements.

Fig. 8. (a) Collapse mode of 6-panel angle element and (b) extensional elements.

2.1.3. The mean crushing force in quasi-static case

To build a quantitative prediction of crushing resistance of
triangular multi-cell tubes and to nd out how angle element
affects the structure of tube, the theoretical solution of the mean
crushing force of multi-cell tube was introduced. Fig. 1a shows the
three independent 2-panel angle elements and three 4-panel
angle elements form prole of tube type I. Substituting the terms
of Eqs. (4), (9) and (12) into Eq. (1), the general theoretical
equation of the mean crushing force of tube type I was obtained.
This is

Rigid wall
Lumped Mass

v = 10 m/s

L0 = 250 mm

Fig. 9. Schematic of the computational model.

P m  I 2H

 panel
Etube
3Eind:2
E4m panel
b
m

2 M 0 B 2M 0




H2
1
6 cos 12 1
:
t
cos

14

Substituting Eq. (16) back into Eq. (15), the mean crushing force in
quasi-static loading of tube type I is obtained as
p

Transforming Eq. (14), we obtain




Pm  I B H
1
B H

6 cos 12 1
F ;

H t
H t
M0
cos

Pm  I
15

B F ;
H

)H

s
Bt
F ;

17

where

The half-wavelength was obtained under the stationary condition of the mean crushing force P m =H 0, then
0 

F ;
M0 B M0 H

F ; 0:5 s0 t 1:5 B0:5

H
t
2

16


F ; 6 cos 12 1


1
:
cos

As presented in Fig. 1b, the prole of tube type II was

constituted by three 3-panel angle elements, three T-shape angle

188

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

elements and one 6-panel angle element. Substituting the terms

of Eqs. (4), (10), (11) and (13) into Eq. (1), the general theoretical
expression of the mean crushing force of tube type II is obtained
as
P m  II 2H Etube
3E3m panel ETm shape E6m panel
b
2 M 0 B 2M 0



H2
8
32 12 tan =2
t
cos

18

An alternative form of Eq. (18) is



P m  II B H
8
B H

32 12 tan =2
G;

H t
H t
M0
cos

19

By using the stationary condition of the mean crushing force,

the half-wavelength can be obtained as P m =H 0. Then
s
B G;
Bt
)H
0  2
20
t
G;
H

Fig. 10. Deformation process of three tubes.

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

Substituting Eq. (20) back into Eq. (19), the equation of the mean
crushing force for tube type II under the quasi-static loading is
obtained as
p
M0 B M0 H
G;
P m  II

G; 0:5 s0 t 1:5 B0:5

21
2
H
t
where
G; 32 12 tan =2

8
:
cos

The structure of the tube type III (Fig. 1c) was formed by three
independent 2-panel angle elements, six 4-panel angle elements,

189

and one 6-panel angle element. Substituting the terms of Eqs. (4),
(9), (12) and (13) into Eq. (1), the general theoretical solution to
the mean crushing force of tube type III is obtained as
P m  III 2H Etube
3E2m panel 6E4m panel E6m panel
b


H2
32
32 6 cos
2 M 0 B 2M 0
t
cos

22

Transforming Eq. (22), we obtain



P m  III B H
32
B H

32 6 cos
Q ;

H t
H t
M0
cos

23

By using the stationary condition of the mean crushing force,

the half-wavelength is expressed as P m =H 0. Then
s
B Q ;
Bt
)H
24
0  2
t
Q ;
H
Substituting Eq. (24) into Eq. (23), the mean crushing force of tube
type III under quasi-static loading is
p
Q ;
M0 B M0 H
P m  III

Q ; 0:5 s0 t 1:5 B0:5

25
H
t
2
where
Q ; 32 6 cos

32
:
cos

2.2. Optimization design methodology

Among all the indicators of crashworthiness optimization
design, the vital analytical objective was the energy-absorption.
Hence, in order to estimate the energy absorption of structural
unit mass m, specic energy absorption (SEA) was formulated as
SEA

EA
m

26

In fact, a higher SEA indicates a better capability of energy

absorption. In Eq. (26), the total strain energy during crushing is
Table 1
Design matrix of three types of tube for crashworthiness.
n

Fig. 11. The crushing foredisplacement curve of (a) tube I, (b) tube II and
(c) tube III.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

t (mm) a (mm) Tube type I

1
1.3
1.6
1.9
2.2
1
1.3
1.6
1.9
2.2
1
1.3
1.6
1.9
2.2
1
1.3
1.6
1.9
2.2
1
1.3
1.6
1.9
2.2

80
80
80
80
80
85
85
85
85
85
90
90
90
90
90
95
95
95
95
95
100
100
100
100
100

Tube type II

Tube type III

SEA
PCF (kN) SEA
(kJ/kg)
(kJ/kg)

PCF (kN) SEA

(kJ/kg)

PCF (kN)

18.262 39.362
20.226 52.274
22.774
65.287
23.854 77.82
24.732 88.936
17.635 42.151
19.491 55.941
21.079 69.78
22.231 82.904
23.891 94.376
16.689 48.183
18.63
63.406
19.883 80.385
21.425 97.058
22.916 111.899
16.092 50.984
17.793 67.207
19.279 85.237
20.865 102.754
22.276 117.81
15.569 50.155
17.083 66.443
19.26
83.029
20.628 98.954
21.053 113.035

49.572
65.965
82.467
98.125
111.611
53.133
70.384
88.192
104.633
118.899
60.979
69.095
99.84
120.552
138.568
65.978
83.194
105.95
128.907
148.295
63.148
83.857
104.85
124.628
142.093

52.592
69.599
85.801
101.135
114.541
56.038
74.202
92.462
109.979
125.572
64.544
85.478
107.641
128.512
146.48
68.196
90.665
114.112
135.856
153.825
66.649
87.852
110.016
130.951
157.552

17.356
19.876
21.86
22.003
23.107
16.842
19.272
20.924
21.602
22.693
16.559
18.9
20.273
20.679
21.524
15.253
16.954
18.229
19.146
19.71
14.398
16.139
17.748
18.55
18.715

23.339
25.227
28.199
29.092
29.644
22.753
24.273
26.388
27.564
28.107
21.338
24.05
25.536
26.272
26.42
20.932
22.371
24.579
25.306
25.948
19.337
20.011
22.979
23.92
24.152

190

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

Fig. 12. The response surface of (a) peak crushing force and (b) SEA.

Fig. 13. (a) SEA vs structural weight and (b) Pm vs structural weight.

estimated as
Z d
Pxdx
EA
0

27

where P(x) is the instantaneous crushing force. In addition, the

initial peak crushing force (PCF) of multi-cell thin-walled tube was
used for estimating the impact characteristics. Another crashworthiness indicator is the mean crushing force Pm which is
computed by
Z
E
1 d
Pxdx
28
Pm A
d d 0
where d is the crushing displacement at a specic time.
2.2.1. Response Surface Method (RSM)
A typical surrogate modelling technique was considered appropriate in the multivariate optimization process involving material,
geometrical nonlinearities and contact-impact loading nonlinearities. The primary concept of RSM was applied to the construction
of regression functions for crashworthiness indicators by using the
function values at the design sampling points. The mathematical

Fig. 14. Pareto spaces for multi-objective optimization: (a) tube type I, (b) tube
type II and (c) tube type III.

expression of RSM is expressed as

m

~ x
yx  yx
i i
i1

29

~ and y(x) are respectively the surrogate surface approxwhere yx

imation and the numerical solution denoting y(x). m represents
the total number of basic functions i(x), and i is the unknown

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

coefcient. Taking n dimensional problem for example, the full

linear polynomial basis function is
1; x1 ; x2 ; ; xn

30

and the full quartic polynomial basis function is expressed as

1; x1 ; x2 ; :::; xn ; x21 ; x1 x2 ; :::; x1 xn ; :::; x2n ; x31 ; x21 x2 ; :::; x21 xn ; x1 x22 ; :::;
x1 x2n ; :::; x3n ; x41 ; x31 x2 ; :::; x31 xn ; x21 x22 ; :::; x21 x2n ; :::; x1 x32 ; :::; x1 x3n ; :::; x4n

191

mathematically given as
Maximize x; xL ; xR L  R

33

where L arctanf 2 xL  f 2 x=f 1 x  f 1 xL and R arctanf 2 x 

f 2 xR =f 1 xR  f 1 x are the left and right bend-angles of x.

31

The full quartic polynomial basis function was proved to be

a better choice for the regression analysis [13,2426]. The quartic
response surface models were consequently adopted in this study.
2.2.2. Multi-objective optimization
With two objectives of SEA and PCF, the multiobjective optimization problem for minimizing PCF and maximizing SEA was
dened by the linear weighted average methods (LWAM) [24].
Then, the mathematical denition for the crashworthiness optimization in terms of the LWAM is given as
8
SEAn
>
Minimize Ft; a wPCFt;a
>
PCFn 1  wSEAt;a
>
>
<
s:t w A 0; 1
32
>
1
r t r 2:2 mm
>
>
>
:
80 r a r 100 mm
where SEAn and PCFn are the given normalizing values for each
cross-sectional prole.
2.2.3. Knee point
In some certain cases, the designer must choose the most
preferred solution (termed as knee point) from optimal solutions
to meet their requirement. Several methods were proposed to
determine a knee point from Pareto set such as Turevsky and
Suresh [28] and Sun et al. [29]. However, if there is a great deviation
among the orders of magnitude of different objectives, these
methods [29] seem to be less effective. A modied multi-objective
evolutionary algorithm suggested by Branke et al. [30] was utilized to
seek the knee regions. Deb and Gupta [31] have recently suggested a
solution to nd a knee point with maximum bend-angle, which is

3. Numerical simulation and crashworthiness optimization

3.1. Numeric simulation
In this section, the FE model was carried out by ANSYS/LS-DYNA
to simulate the triangular multi-cell thin-walled tubes subjected to
axial dynamic loading with 4, 6 and 9 cells (as shown in Fig. 1). The
side-length a of the cross-sections and the thickness t were chosen
to be design variables, and the design interval is given in Eq. (32).
The total length L0 of all the tubes is 250 mm.
In this study, the thin-walled tubes were modelled with the
BelytschkoTsay four-node shell element with the optimal mesh
density of 2.5  2.5 mm. The material AA6060 T4 was modelled
with material model #24 (Mat_Piecewise_Linear_Plasticity) with
mechanical properties: Young's modulus E 68,200 MPa, initial
yield stress sy 80 MPa, ultimate stress su 173 MPa, Poisson's
ration 0.3, and power law exponent n 0.23 [32]. Since the
aluminum was insensitive to the strain rate effect, this effect was
neglected in the nite element analysis. An automatic node to
surface contact between thin-walled tube and rigid wall was
dened to simulate the real contact. Alternatively, an automatic
single surface contact algorithm was utilized for the self-contact
among the shell elements to avoid interpenetration of folding
generated during the axial collapse. In the contact denition,
a friction coefcient of 0.3 among all surfaces was employed. To
generate enough kinetic energy, one end of tube was attached
with a lumped mass of 500 kg whereas another end impacted onto
a rigid wall with an initial velocity of 10 m/s. The schematic of the
computational model is shown in Fig. 9.
All of the tubes were axial symmetric structures. Despite the
same length, same side-length and same thickness, the three tubes

Table 2
Difference of numeric result and theoretical prediction for three tubes.
n

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

Tube type I

Tube type II

Tube type III

Num. Pm (kN)

Theo. Pm (kN)

Diff. (%)

Num. Pm (kN)

Theo. Pm (kN)

Diff. (%)

Num. Pm (kN)

Theo. Pm (kN)

Diff. (%)

22.23
34.107
46.497
61.379
77.295
23.173
34.142
47.275
62.465
79.196
23.787
35.095
48.408
64.341
81.044
24.151
35.645
49.676
66.14
82.19
24.819
36.812
51.437
67.423
83.731

23.016
34.028
46.344
59.817
74.337
23.736
35.098
47.809
61.717
76.710
24.435
36.137
49.230
63.561
79.013
25.115
37.146
50.611
65.352
81.250
25.776
38.129
51.956
67.096
83.427

3.54
 0.23
 0.33
 2.54
 3.83
2.43
2.80
1.13
 1.20
 3.14
2.73
2.97
1.70
 1.21
 2.51
3.99
4.21
1.88
 1.19
 1.14
3.86
3.58
1.01
 0.49
 0.36

25.894
41.1
55.331
72.1
91.64
26.75
42.39
58.3
74.137
94.125
27.482
43.184
59.954
76.129
95.039
28.264
43.859
60.278
77.219
95.897
29.109
44.984
60.994
79.107
96.597

27.148
40.143
54.680
70.589
87.738
27.996
41.404
56.406
72.827
90.533
28.820
42.627
58.080
74.998
93.244
29.620
43.816
59.707
77.108
95.879
30.400
44.974
61.291
79.162
98.443

4.84
 2.33
 1.18
 2.10
 4.26
4.66
 2.33
 3.25
 1.77
 3.82
4.87
 1.29
 3.13
 1.49
 1.89
4.80
 0.10
 0.95
 0.14
 0.02
4.44
 0.02
0.49
0.07
1.91

39.973
60.35
85.345
111.642
139.519
41.129
61.05
87.9
113.899
144.17
42.716
62.715
89.541
117.161
148.092
43.862
63.768
91.482
120.81
151.03
45.268
65.507
93.192
122.495
153.514

41.221
60.954
83.028
107.183
133.223
42.510
62.868
85.648
110.581
137.467
43.760
64.726
88.190
113.878
141.583
44.976
66.531
90.660
117.082
145.584
46.160
68.290
93.066
120.201
149.477

3.12
1.00
 2.72
 3.99
 4.51
3.36
2.98
 2.56
 2.91
 4.65
2.44
3.21
 1.51
 2.80
 4.39
2.54
4.33
 0.90
 3.09
 3.61
1.97
4.25
 0.14
 1.87
 2.63

192

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

were different in weight. Tube I is the lightest one while tube III is
the heaviest. The axial crushing of multi-cell tubes was presented
with a displacement equal to about 70% of the initial length. Fig. 10
shows the deformation process of three tubes at different times.
Sometimes the exact value of the effective crushing distance on
the crushing forcedisplacement curve was not unique. The
corresponding crushing forcedisplacement curves of three tubes
are also shown in Fig. 11. After reaching the initial peak and before
rising steeply whenever the deformation capacity is exhausted at
the effective crushing distance, the crushing force fell sharply and
then uctuated periodically and around the values of the mean
crushing force in correspondence with the formation, and nally
completed the collapse of folds one by one.

3.2. Crashworthiness optimization

For obtaining the response functions of SEA and PCF, a series of
25 design sampling points (based on a and t) were selected in the
design space to provide sampling design values for FEA and
regression analysis of three types of tubes (Table 1) so as to obtain
the response surface of the SEA and PCF. Fig. 12 shows that the
SEA's and PCF's RS of tube types I, II and III cases behave
monotonically over the design domain. In addition, the curves in
Fig. 13 illustrate the variation of SEA and Pm with changes in
weight. Meanwhile, energy absorptions of the tube types I and III
were better than those of tube type II.
The Pareto sets for these three cross-sectional proles were
obtained by changing the weight coefcient w in Eq. (32), and the
Pareto frontiers are plotted in Fig. 13. In fact, any point on the
Pareto frontier can be an optimum. As a result, some methods
were proposed to determine the best solution (knee point) which
has a large trade-off value in comparison with other Paretooptimal points. In this case, methods of [31,32] were utilized to
determine the knee region and the knee point, respectively. The
results of expression (33) showed that Pareto solutions (Knee
points) for tube types I, II and III were 0.7924, 0.7818 and 0.7773,
respectively. The relative errors (REs) of FE simulation value and
RS approximate value are summarized in Table 3. Therefore the FE
simulation value and RS approximate value at the Knee points
were exactly close to each other. According to the relationship
among the weighted average method and those of [31,32], these
optimal results are plotted in Fig. 14.

4. Theoretical validation and discussion

The theoretical expressions of the mean crushing force of tube
types I, II and III are proposed in Section 2.1.3. Nevertheless, these
expressions were applied in axial quasi-static loading case in
which the effect of dynamic crushing was not considered. In the
dynamic loading case, dynamic amplication effects, consisting of
inertia and strain rate effects, were considered in the theoretical
equations above. In reality, the aluminum alloy is insensitive to the
strain rate effect that can be neglected. A dynamic enhancing
coefcient was suggested to take the inertia effect into account.
It is not easy to nd the accurate value for dynamic enhancing
coefcient, and this coefcient is even a variable for different
geometric parameters as described by Langseth and Hopperstad
[33] and Hanssen et al. [34]. According to these studies, the
coefcient was proposed in the range of 1.31.6 for AA6060 T4
extruded tubes. These coefcients () were simply set to 1.41,
1.3 and 1.45 for tube types I, II and III, respectively. Thus, the
theoretical solution to tube type I is applied as follows:
0:5
P dym:
s0 t 1:5 B0:5
m  I I P m  I I

p
F ;
2

where


F ; 6 cos 12 1

1
cos

for tube type II

0:5
s0 t 1:5 B0:5
P dym:
m  II II P m  II II

p
G;
2

where
Fig. 15. Comparison between numerical prediction and theoretical prediction:
(a) tube type I, (b) tube type II and (c) tube type III.

G; 32 12 tan =2

34

8
:
cos

35

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

In Eqs. (34)(36), s0 is a ow stress of material with power law

hardening, which is calculated [35] as

For tube type III

p
Q ;
0:5
1:5 0:5

s
t
B
P dym:
III
m

III
III
0
m  III
2

36

s0

where
Q ; 32 6 cos

32
:
cos

Table 3
Optimal results by using the method of Deb and Gupta (Knee point).
Type of
cross-section

Terms

Optimal design
variables (mm)

SEA
(kJ/kN)

PCF
(kN)

Type I

Approximate value
FE analysis value
RE
Approximate value
FE analysis value
RE
Approximate value
FE analysis value
RE

t 1.23, a 80

19.897
19.728
0.856
25.282
25.717
 1.691
25.100
24.800
1.210

49.315
49.088
0.462
63.436
63.056
0.603
63.967
64.267
 0.467

Type II

Type III

t 1.25, a 80

t 1.21, a 80

193

sy su

1n

0:106 GPa

37

where sy and su denotes the yield strength and the ultimate

strength of the material, respectively, and n is the strain hardening
exponent.
Hence, the mean crushing forces were calculated by expressions (34)(36) and these values were used to compare with the
value of mean crushing force of tubes at a displacement of 60% in
each different case. Remarkably, this mean crushing force is
dened as an equivalent constant force with a corresponding
amount of displacement. The deviations among theoretical equations above and numerical predictions for all cases of three types
of tubes are listed in Table 2. For tube types I and II, the differences
of Eqs. (34) and (35) and numeric results were, respectively,
ranging from  3.83% to 4.21% and from  4.26% to 4.84%. With
regard to tube type III, the dissimilarities between Eq. (36) and
numeric results varied in the range of  4.65% to 4.33%. Obviously,
these differences belong to available range (Table 2). Continuously,
Fig. 15 reveals the very close agreement between the theoretical
solutions and the numerical predictions for all cases.

Fig. 16. (a) Deformation result and (b) crushing forcedisplacement curve of tube I.

Fig. 17. (a) Deformation result and (b) crushing forcedisplacement curve of tube II.

194

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

Fig. 18. (a) Deformation result and (b) crushing forcedisplacement curve of tube III.

Deriving from the optimal results of the method of Deb and

Gupta (Table 3), the optimal triangular sections of tube types I, II
and III were considered in this analysis. The deformation results
and crushing forcedisplacement curves of three tubes are presented in Figs. 1618. For the optimal tube type I with 4 cells in
Table 3, the side-length and the wall thickness were of 80 mm and
1.23 mm, respectively (as shown in Fig. 16). The mean crushing
force obtained from FE analysis is 30.43 kN. Visibly, with this
prole, the sum of side-length and of internal web length B is of
352.62 mm. Substituting items into Eq. (34), the theoretical prediction of mean crushing force is
0:5
P dym:
 0:106  1:231:5
m  I 1:41 

352:620:5

6:46
31:335 kN
2  0:7

38

With 6 cells, the optimal tube type II in Table 3 has the sidelength of 80 mm and the wall thickness of 1.25 mm (Fig. 17). The
mean crushing force received from FE analysis was 37.01 kN.
Simultaneously, for this optimal tube, the sum of side-length and
internal web length B is 439.06 mm. To substitute items into
Eq. (35), the theoretical prediction of mean crushing force is
0:5
P dym:
 0:106  1:421:5
m  II 1:3 

439:060:5

7:41
37:864 kN
2  0:7

39

As listed in Table 3, optimal tube type III has 9 cells (Fig. 18).
The width and the wall thickness of this cross-section are
respectively of 80 mm and 1.21 mm. Then, the mean crushing
force in FE analysis was 53.67 kN. As a matter of course, the
parameter of prole of tube III is B 439.376 mm. Replacing items
into Eq. (36), the theoretical prediction of mean crushing force is
0:5
P dym:
 0:106  1:361:5
m  III 1:45 

439:3760:5

10:089
54:903 kN
2  0:7

40

From the results above, Eq. (36) were adopted to calculate the
mean crushing force for three optimal tubes. Subsequently, the
differences between numerical predictions and theoretical solutions for optimal tube types I, II and III were respectively of 2.97%,
2.3% and 2.05%. These differences show that the proposed equations
are appropriate to the numerical predictions. In addition, the stable
and progressive folding deformation patterns that are developed in
all the three types of tube are the desirable energy-dissipating
mechanism.

5. Conclusions
The proles of three types of tubes were divided into the basic
elements: 2-, 3-, T-shape, 4- and 6-panel angle element. Based on
the Simplied Super Folding Element theory, theoretical expressions of the mean crushing force were proposed for the three
types of triangular multi-cell thin-walled tubes under the axial
crushing loading. Numerical simulations of tubes under the axial
dynamic impact loading were also carried out, and a dynamic
enhancement coefcient was introduced to account for the inertia
effects of aluminum alloy AA6060 T4. Numerical results showed
that tube types I and III were better than tube type II in the aspect
of energy absorption. Simultaneously, the stable and progressive
folding deformation patterns appeared for all the three types
of tubes.
The two RS models of PCF and SEA for each tube were constructed. Pareto sets were obtained by using the linear weighted
average methods (LWAM). In this paper, the Pareto solutions of three
types of tubes were identied to seek out the knee points. The
relative errors between RS approximate value and FE analysis value
at the Knee points were obtained and those were also acceptable.
Finally, the theoretical expressions excellently agreed with the
numerical results, and simultaneously validated the efciency of
the crashworthiness optimization design method based on the
surrogate models and the numerical analysis techniques.

Acknowledgments
The nancial supports from National Natural Science Foundation of China (Nos. 11232004 and 51175160), New Century Excellent Talents Program in University (NCET-12-0168) and Hunan
Provincial Natural Science Foundation (12JJ7001) are gratefully
acknowledged. Moreover, Joint Center for Intelligent New Energy
Vehicle is also gratefully acknowledged.
References
[1] Wierzbicki T, Abramowicz W. On the crushing mechanics of thin-walled
structures. J Appl Mech 1983;50:72734.
[2] Abramowicz W, Jones N. Dynamic axial crushing of square tubes. Int J Impact
Eng 1984;2:179208.
[3] DiPaolo BP, Monteiro PJM, Gronsky R. Quasi-static axial crush response of a
thin-wall, stainless steel box component. Int J Solids Struct 2004;41:370733.
[4] DiPaolo BP, Tom JG. Effects of ambient temperature on a quasi-static axialcrush conguration response of thin-wall, steel box components. Thin-Walled
Struct 2009;47:98497.

T. Tran et al. / Thin-Walled Structures 82 (2014) 183195

[5] Guillow SR, Lu G, Grzebieta RH. Quasi-static axial compression of thin-walled

circular aluminum tubes. Int J Mech Sci 2001;43:210323.
[6] Ullah H. Buckling of thin-walled cylindrical shells under axial compression. Int
J Numer Methods Eng 2009;79:133253.
[7] Zhang X, Zhang H. Energy absorption of multi-cell stub columns under axial
compression. Thin-Walled Struct 2013;68:15663.
[8] Alavi Nia A, Parsapour M. An investigation on the energy absorption characteristics of multi-cell square tubes. Thin-Walled Struct 2013;68:2634.
[9] Krolak M, Kowal-Michalska K, Mania R, Swiniarski J. Experimental tests of
stability and load carrying capacity of compressed thin-walled multi-cell
columns of triangular cross-section. Thin-Walled Struct 2007;45:8837.
[10] Liu Y. Collapse behaviour and simplied modeling of triangular crosssection
columns. Indian J Eng Mater Sci 2009;16:718.
[11] Hong W, Jin F, Zhou J, Xia Z, Xu Y, Yang L, et al. Quasi-static axial compression
of triangular steel tubes. Thin-Walled Struct 2013;62:107.
[12] Hong W, Fan H, Xia Z, Jin F, Zhou Q, Fang D. Axial crushing behaviors of multicell tubes with triangular lattices. Int J Impact Eng 2014;63:10617.
[13] Hou S, Li Q, Long S, Yang X, Li W. Design optimization of regular hexagonal
thin-walled columns with crashworthiness criteria. Finite Elem Anal Des
2007;43:55565.
[14] Abramowicz W. Thin-walled structures as impact energy absorbers. ThinWalled Struct 2003;41:91107.
[15] Chen W, Wierzbicki T. Relative merits of single-cell, multi-cell and foam-lled
thin-walled structures in energy absorption. Thin-Walled Struct
2001;39:287306.
[16] Kim H-S. New extruded multi-cell aluminum prole for maximum crash
energy absorption and weight efciency. Thin-Walled Struct 2002;40:31127.
[17] Zhang X, Cheng G, Zhang H. Theoretical prediction and numerical simulation
of
multi-cell
square
thin-walled
structures.
Thin-Walled
Struct
2006;44:118591.
[18] Zhang X, Zhang H. Numerical and theoretical studies on energy absorption of
three-panel angle elements. Int J Impact Eng 2012;46:2340.
[19] Naja, A, Rais-Rohani, M. Inuence of cross-sectional geometry on crush
characteristics of multi-cell prismatic columns. In: Proceedings of the 49th
AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics and materials
conference (2008).

195

[20] Jensen , Langseth M, Hopperstad OS. Experimental investigations on the

behaviour of short to long square aluminum tubes subjected to axial loading.
Int J Impact Eng 2004;30:9731003.
[21] Karagiozova D, Jones N. Dynamic buckling of elasticplastic square tubes
under axial impactII: structural response. Int J Impact Eng 2004;30:16792.
[22] Karagiozova D, Alves M. Dynamic elasticplastic buckling of structural
elements: a review. Appl Mech Rev 2008;61:040803.
[23] Tang Z, Liu S, Zhang Z. Analysis of energy absorption characteristics of
cylindrical multi-cell columns. Thin-Walled Struct 2013;62:7584.
[24] Hou S, Li Q, Long S, Yang X, Li W. Multiobjective optimization of multi-cell
sections for the crashworthiness design. Int J Impact Eng 2008;35:135567.
[25] Hou S, Li Q, Long S, Yang X, Li W. Crashworthiness design for foam lled thinwall structures. Mater Des 2009;30:202432.
[26] Hou S, Han X, Sun G, Long S, Li W, Yang X, et al. Multiobjective optimization
for tapered circular tubes. Thin-Walled Struct 2011;49:85563.
[27] Zhang X, Huh H. Crushing analysis of polygonal columns and angle elements.
Int J Impact Eng 2010;37:44151.
[28] Turevsky I, Suresh K. Efcient generation of pareto-optimal topologies for
compliance optimization. Int J Numer Methods Eng 2011;87:120728.
[29] Sun G, Li G, Zhou S, Li H, Hou S, Li Q. Crashworthiness design of vehicle by
using multiobjective robust optimization. Struct Multidiscip Optim
2011;44:99110.
[30] Branke J, Deb K, Dierolf H, Osswald M. Finding knees in multi-objective
optimization. In: Yao X, Burke E, Lozano J, Smith J, Merelo-Guervs J, Bullinaria J,
Rowe J, Tio P, Kabn A, Schwefel H-P, editors. Parallel problem solving from
naturePPSN VIII. Berlin, Heidelberg: Springer; 2004. p. 72231.
[31] Deb K, Gupta S. Understanding knee points in bicriteria problems and their
implications as preferred solution principles. Eng Optim 2011;43:1175204.
[32] Santosa SP, Wierzbicki T, Hanssen AG, Langseth M. Experimental and
numerical studies of foam-lled sections. Int J Impact Eng 2000;24:50934.
[33] Langseth M, Hopperstad OS. Static and dynamic axial crushing of square thinwalled aluminum extrusions. Int J Impact Eng 1996;18:94968.
[34] Hanssen AG, Langseth M, Hopperstad OS. Static and dynamic crushing of
circular aluminum extrusions with aluminum foam ller. Int J Impact Eng
2000;24:475507.
[35] Tam LL, Calladine CR. Inertia and strain-rate effects in a simple plate-structure
under impact loading. Int J Impact Eng 1991;11:34977.