Professional Documents
Culture Documents
State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University, Changsha, Hunan 410082, PR China
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:
Received 15 February 2013
Received in revised form
25 September 2013
Accepted 12 October 2013
Available online 19 October 2013
Thin-walled structures have exhibited signicant advantages in light weight and energy absorption and
been widely applied in automotive, aerospace, transportation and defense industries. Unlike existing
thin-walled structures with uniform thickness, this paper introduces functionally graded structures with
changing wall thickness along the longitudinal direction in a certain gradient (namely, functionally
graded thickness e FGT). Its crashing behaviors are the key topics of the present study. We examine the
crashing characteristics of functionally graded thin-walled structures and evaluate the effect of different
thickness gradient patterns on crashing behaviors. It is shown that the gradient exponent parameter n
that controls the variation of thickness has signicant effect on crashworthiness. To optimize crashworthiness of the FGT tubes, the Non-dominated Sorting Genetic Algorithm (NSGA-II) is used to seek for
an optimal gradient, where a surrogate modeling method, specically response surface method (RSM), is
adopted to formulate the specic energy absorption (SEA) and peak crashing force functions. The results
yielded from the optimization indicate that the FGT tube is superior to its uniform thickness counterparts
in overall crashing behaviors. Therefore, FGT thin-walled structures are recommended as a potential
absorber of crashing energy.
Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved.
Keywords:
Multiobjective optimization
Tailor rolled blank (TRB)
Crashworthiness
Non-uniform thickness sheet
Functionally graded thickness
1. Introduction
Over the past years, research interests in crashworthiness have
resulted in a series of systematic investigations into crash responses of various thin-walled structures via analytical, experimental and numerical approaches [1e8]. Two signicant aspects,
namely weight and crashworthiness, have drawn primary attention
in these studies. As an effective structure, thin-walled components
have showed signicant advantages over other solid elements and
are capable of carrying substantial loads with desired deformation,
which could be appreciably higher than the corresponding ultimate
or bulking loads [9,10]. In reality, thin-walled structural members
play a critical role on enhancing the capability of energy absorption
in impact engineering.
The automobile body in white (BIW) is mainly composed of
thin-walled structural parts, which are made by stamping or
forming process of traditional metal sheets with uniform thickness
[11e19]. It is of great interests in investigating the crashworthiness
of thin-walled structures for improving the vehicle safety and light
* Corresponding author. Tel.: 86 731 8882 1717; fax: 86 731 8882 2051.
E-mail addresses: sgy800@126.com (G. Sun), gyli@hnu.edu.cn (G. Li).
weight. In this regard, Zhang et al. [12] evaluated the energy absorption characteristics of regular polygonal and rhombic columns
under quasi-static axial compression. Song et al. [13] introduced
origami patterns into thin-walled structures and minimized the
initial peak force and subsequent uctuations. Tang et al. [20]
presented a cylindrical multi-cell column to improve energy absorption. Naja and Rais-Rohani [21] proposed a sequentially
coupled nonlinear nite element analysis (FEA) technique to
investigate the effects of sheet-forming process and design parameters on energy absorption of thin-walled tubes made of
magnesium alloy. Acar et al. [22] studied the crashing performances
of tapered tubes using multiobjective optimization. Although such
thin-walled structures have been extensively used as energy absorbers for their high energy absorption capacity, light weight and
low cost [23], all these thin walled structures were based upon the
uniform material and/or the same wall thickness. The inherent
shortcoming resides on that such structures may not exert their
maximum capacities in crashworthiness, and furthermore, a uniform wall thickness does not necessarily make best use of material
for meeting the requirements of vehicular light weight [24e26]. So
there is an urgent need to develop new structural conguration
with different material and/or thickness combinations for maximizing crashworthiness and material usage.
0734-743X/$ e see front matter Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijimpeng.2013.10.004
According to Yang et al. [27], a metal sheet with varying thickness could be a more desirable structure because it not only uses
material more efciently, but also increases design exibility
considerably. It has been demonstrated that with an optimal choice
of different materials grades (e.g. via tailored welded blanks (TWB),
or hybrid blanks) and/or thicknesses (e.g. via tailor rolling blanks),
crashing performance of the combined components can be
improved to a higher extent. Indeed, design of specic thin-walled
components with desired materials/thicknesses in a more efcient
manner could represent new potential for further reducing weight
and enhancing performance of the products. Of these components
with variable material/thickness, the TWB structures, which consists of laser-welded sheet metals of different thicknesses and
materials, provide a exible combination of component materials
and thicknesses, which has been adopted in vehicular oor
component [28], B-pillar [29], front-end structure [30], and door
inner panels etc. [31,32]. The main shortcoming of those blanks lies
in that it consists of discrete thickness sections and may lead to
stress concentration in the interfaces. To overcome such defects of
TWB, a relatively new rolling process, named tailor rolled blank
(TRB), has been developed. In the newly developed TRB process, the
rolling gap can be varied, which leads to a continuous thickness
variation in the workpiece. Applications of such a rolling process
allow reducing more weight compared with traditional stamping
or forming processes. As such, varying sheet thickness can better
meet more and more demanding design requirements, thereby
enhancing utilization of material and/or thickness comparing with
traditional stamping uniform sheets.
There have been some reports on TRB in sheet metal forming.
For example, Zhang et al. [33] investigated the effects of transition
zone length, blank thickness variation, friction coefcient and die
clearance on the springback of TRB component. Meyer et al. [34]
used TRB to increase the maximum drawing depth compared to
the blanks with constant thickness. Urban et al. developed a design
tool by combining numerical simulation and optimization algorithm to improve the formability of TRB [35]. To the authors best
knowledge, however, very limited studies on crashworthiness
design of thin-walled TRB structures have been performed to date.
To make use of TRB thin-walled structures with functionally
graded thicknesses (FGT) for impact engineering, it is essential to
understand the energy absorption characteristics in comparison
with those well-studied uniform thickness (UT) thin-walled
structures. More importantly, it is of particular importance to
seeking the best possible thickness gradient for crashing performance with different measures. Thus the objective of this paper
resides in quantifying and improving crashing behaviors of thinwalled structures with functionally graded wall thickness. For this
reason, two critical issues need to be addressed in this paper: (1) a
direct problem that quanties the crashing characteristic of functionally graded thin-walled structures with variable wall thickness
and evaluates the effects of the different thickness pattern on both
specic energy absorption (SEA) and peak impact forces; (2) an
inverse problem that seeks optimal gradient for maximizing the
specic absorption energy (SAE) and minimizing peak crashing
force (Fmax).
As for the functionally graded thickness structure, the thickness
of thin-walled varies throughout the depth in an ascending or
descending gradient. It is expected that the gradient exponential
parameter (n) has a signicant effect on crashworthiness. To
represent such complex crashworthiness objective functions with
respect to gradient parameter, which has not been explored in
literature before, a surrogate model technique, namely specically
response surface method (RSM), will be attempted here. To maximize the energy absorption and minimize the peak crashing force,
the multiobjective optimizations for the FGT structures are
63
s0
2
X
i1
!
q
_
Qi 1 expCi
1
_ 0
(1)
Zd
EA
Fddd
(2)
SEA
EA
m
(3)
64
Unit: mm
t=1.2
R=3
a=60
Seam weld
L2=100 mm
Impacting
mass=600 kg
L1=310 mm
b=60
(a)
(b)
Fig. 1. Schematic of (a) experimental set-up for dynamic tests and (b) geometry description of square tube [39].
Favg
EA
(4)
From Zarei and Kroger [41], the crash force efciency (CFE) and
specic energy absorption can increase simultaneously. The CFE
indicates the uniformity of forceedisplacement curve, meaning
that the higher the CFE, the more efcient the structure. Thus, CFE
of a structure is formulated as other key indicator,
CFE
Favg
100%
Fmax
(5)
1200
1000
800
600
400
200
0
0.00
0.05
0.10
0.15
Plastic Strain
0.20
0.25
Fig. 2. True-stress versus true-plastic strain curves at different strain rates for DP800
high-strength steel [39].
tf x
xn
n
tmax tmax tmin xL
(6)
where x is the distance from the origin (top) of the column, and n is
the grading exponent and assumed to vary between 0 and 10, L
denotes the total length of column. When the total layer Ns is
approaching to innite, the wall thickness will be graded continuously. Thus, the wall thickness increases along the length with a
gradient function changing from convexity to concavity when the n
value varies from less than 1 to greater than 1, as shown in Fig. 5a.
On the contrary, the wall thickness decreases along length, an
opposite tendency is observed as shown in Fig. 5b. Herein, the
ascending pattern is considered to perform the crashworthiness
design with the FGT structures.
3. Numerical modeling
3.1. Finite element (FE) modeling
The FE model used to simulate the crashing process was based
upon the experimental tests conducted by Tarigopula et al. [39]. An
explicit FE code, LS-DYNA, was used to implement the parallel
computing. Fig. 6 shows the FE mesh using the 4-node shell elements with 6 degrees of freedom at each node. In this model, the
bottom end is xed and a rigid mass-block of 600 kg was linked to
the loading end through a master node. In the crash scenario, the
mass block is assigned an initial velocity of 10 m/s through the
master node. The contact between the mass block surface and the
FGT tubes was modeled with a friction coefcient of 0.2, while for
the self-contact of the tube, frictional effects were neglected [43].
To nd the optimum mesh size for the numerical simulation, a
convergence test with ve different mesh sizes was carried out. In
comparison, a UT column having the same mass as its FGT counterpart was also employed to perform the mesh convergence.
Therefore, element size 3 mm 3 mm is adopted in the nite
element model. The energy absorption characteristics predicted by
different element sizes (exponent n 0) are summarized in Fig. 7,
which shows that there is very small difference between element
sizes 2 mm 2 mm and 2.5 mm 2.5 mm. It can be seen that the
mesh size adopted for all models of 2.0 2.0 mm is sufcient.
As for FGT, it is assumed that the depth of each layer is the same and
dened as Le, an equivalent thickness of UT column that has the same
volume of high-strength steel material can be calculated as follows,
tavg
NS
X
i1
ti Le =NS Le
NS
X
i1
ti =NS
(7)
0.33 7850
495
200
233
_ 0 (1/s)
76 10 0.0116 0.001
65
1000
3.2. Validation of the numerical models
900
Fmax
800
In order to validate the developed FE models, the modeling results of the UT square tubes under axial dynamic loading are rst
compared with the theoretical solutions available in the literature
[44], i.e.
Force F (kN)
700
600
500
1=3
(8)
400
where s0 is the characteristic stress of tube material, bw and t are
the width and wall thickness of tube, respectively. The comparison
of the crashing force versus displacement of the UT square tubes
(thickness t 1.2 mm) under an axial dynamic loading rate of 10 m/
s is shown in Fig. 9. It is seen that the present FEA results are in good
agreement with the theoretical solutions for the UT square tubes.
Favg
300
200
Ea
100
0
50
100
350
400
1.0
0.8
n=0.1
n=0.2
0.6
n=0.5
n=1
0.4
n=2
n=5
0.2
n=10
0.0
0.0
0.2
0.4
0.6
Normalized distance x/L
0.8
1.0
(a)
1.0
n=10
0.8
Crash speed v0
1
2
3
4
5
6
7
8
9
10
11
...
Crash speed v0
Top end (ttop)
NS-11
NS-10
NS-9
NS-8
NS-7
NS-6
NS-5
NS-4
NS-3
NS-2
NS-1
NS
n=5
n=2
0.6
n=1
0.4
0.2
n=0.5
n=0.2
n=0.1
0.0
0.0
0.2
0.4
0.6
0.8
Normalized distance x/L
1.0
(b)
Fig. 5. Variation of thickness vs normalized distance. (a) ascending gradient pattern,
and (b) descending gradient pattern.
66
Fig. 6. Finite element model for dynamic impact simulation. (a) Experimental test; (b) Finite element model (2D); (c) 3D model.
140
120
67
100
80
60
40
20
0
in terms of SEA and Fmax in the later stage of impact (e.g. Fig. 15).
Therefore, the FGT component has signicant potential and seems
to be superior and ideal to the crashworthiness design. While
knowing considerable effect of the exponent of thickness gradient
function (Eq. (6)) on crashworthiness of FGT column, it remains a
question how to seek for best possible n by optimizing the crashing
characteristics of corresponding FGT columns, which forms another
goal of this study below.
40
60
80 100 120
Displacment (mm)
140
160
180
MaxSEAn; Fmax n
s:t: nLower n nUpper
(9)
120
O riginP ro
E valuation
O riginP ro
21 layers
31 layers
E v a l u a52
t i o layers
n
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
160
140
120
E valuation
100
80
60
40
20
0
50
100
150
200
Displacement (mm)
(a)
250
300
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
E valuation
O riginP ro
21 layers
31 layers
52 layers
E valuation
40
O riginP ro
E valuation
O riginP ro
E valuation
20
O riginP ro
E valuation
O riginP ro
E valuation
100
180
20
Fig. 9. Crashing force and displacement curves for square tubes under axial dynamic
loading.
80
60
50
100
150
200
Displacement (mm)
(b)
250
300
68
Pm
2
yj b
yj
R 1
2
Pm
j1 yj y
j1
Pm
b
j1 yj y j
RAAE P
m
j1 yj y
RMAE
maxfjy1 b
y 1 j; :::; jym b
y m jg
Pm
b
j1 yj y =m
(10)
(11)
(12)
where b
y j and y are the corresponding predicted (or surrogate) and
mean values for the FEA value yj at each checking point j, respectively, m represents the number of these checking points. In general, the larger the R2 values, the more accurate the surrogate
model. The smaller the RAAE and RMAE, the better the metamodel.
4.1.3. Multiobjective genetic algorithm
Genetic algorithm (GA) is a popular global optimization tools
that was originated from mechanisms of natural evolution and
genetic principles, which is superior to many traditional optimization algorithms because of capability of avoiding trapping in local
optima for searching an optimum [51]. To deal with multiobjective
optimization, the non-dominated sorting GA (NSGA) algorithm has
proven effective by ranking the solutions with non-dominated
sorting and assigning the tness with the ranking. As an
improved version of NSGA, NSGA-II has been demonstrated to be
one of the most efcient algorithms for multiobjective optimization
in a number of benchmarking problems [52]. The NSGA-II algorithm was used in this study and the relevant NSGA-II parameters
are listed in Table 3.
8
23:29696 2:80979n 0 n 1
>
< SEAG n
21:14585 0:56497n 1 < n 10
222:37182 114:94364n 0 n 1
*>
:FG
max n
141:69533 3:31388n 1 < n 10
Linear : 8
22:58031 4:98897n 0 n 1
>
< SEAU n
16:71194
0:42712n 1 < n 10
304:14318 121:91000n 0 n 1
>
:FU
max n
181:59733 6:94079n 1 < n 10
(13)
23:487984:08330n1:27351n2 0 n 1
SEAG n
21:539870:76198n0:01791n2 1 < n 10
2
286:80818544:51939n429:57576n 0 n 1
>
G
*>
: Fmax
n
2
148:558676:74555n0:31197n 1 < n 10
Quadratic : 8
23:223369:27598n4:28701n2 0 n 1
>
U n
>
SEA
<
19:005771:57404n0:10427n2 1 < n 10
>
328:04650281:26548n159:35548n2 0 n 1
>
U
: Fmax
n
211:4640021:87412n1:35758n2 1 < n 10
(14)
23:26355 0:51731n 8:07786n2 6:23425n3 0 n 1
SEAG n
2
3
19:43504 1:10489n 0:38697n 0:02453n 1 < n 10
316:18392 1011:26717n 1653:56469n2 815:99262n3 0 n 1
>
G
*>
: Fmax
n
153:91233 11:49396n 1:34152n2 0:06240n3 1 < n 10
Cubic :
8
23:30785 10:61836n 7:80724n2 2:34682n3 0 n 1
>
U n
>
SEA
<
20:01888 2:47261n 0:29909n2 0:01181n3 1 < n 10
>
335:79308 404:34992n 482:12937n2 215:18260n3 0 n 1
>
U
: Fmax
n
231:33800 39:50129n 5:17950n2 0:23163n3 1 < n 10
(15)
8
>
>
<
8
>
>
<
69
23:06291 6:44927n 42:91078n2 61:96692n3 27:86634n4 0 n 1
SEAG n
22:29852 2:56623n 0:95643n2 0:15902n3 0:00834n4 1 < n 10
>
330:47713
1507:55942n 4135:02593n2 4786:33061n3 1985:16900n4 0 n 1
G
*>
: Fmax
n
2
3
4
199:33333 69:72601n 22:64916n 2:97400n 0:13235n 1 < n 10
Quartic :
8
23:33132 11:43353n 11:88308n2 8:86816n3 3:26067n4 0 n 1
>
U n
>
SEA
<
21:74548 4:68621n 1:10907n2 0:12249n3 0:00503n4 1 < n 10
>
335:39238 390:43675n 412:56352n2 103:87723n3 55:65268n4 0 n 1
>
U
: Fmax
n
245:57500 57:75386n 11:85828n2 1:14426n3 0:04148n4 1 < n 10
8
>
>
<
160
Experiment (ds04)
Simulation
140
Force (kN)
120
Table 2
Change range of SEA and Fmax at different thickness range (Unit: %).
80
60
40
20
20
40
60
Dt 1.0 mm
DSEA
DSEA
DFmax
DSEA
DFmax
13.54
6.11
7.65
13.45
14.26
22.44
33.86
19.40
1.08
20.71
45.72
39.36
38.09
30.47
23.21
29.59
11.64
13.51
14.18
8.52
4.56
3.11
12.00
8.68
13.31
10.15
21.07
16.22
14.85
16.52
41.14
30.55
24.29
15.58
14.93
15.2
11.91
9.52
8.03
6.88
19.10
10.89
6.44
7.84
11.75
1.09
11.67
8.67
6.56
3.34
16.10
13.64
12.63
10.11
7.98
8.38
7.05
5.31
4.60
3.90
1.3 0
Fig. 10. Comparison of force versus displacement curves between experiments [39]
and FE simulations.
10
12
1.0
1.2
1.2
0<n<1
1.1
Weight (kg)
1.0
0.9
0.8
0.7
Dt 0.7 mm
Fmaxb
Dt 1.4 mm
0.2
0.4
0.6
0.8
1
2
4
6
8
10
100
(16)
1<n<10
0.6
0.5
0.0
(a)
0.2
0.4
0.6
0.8
Graded exponent n (0<n<1)
(b)
Fig. 12. (a) Schematic of FGT thin-walled square column (31 layers); (b) Mass variation
with different gradient exponent n.
Fig. 11. Variation of SEA and Fmax for various values of Dt and n.
70
18
350
n=0
UT
FGT
16
SEA (kJ/kg)
14
n=0.2
n=0.4
n=0.6
n=0.8
12
10
n=1.0
n=2.0
n=4.0
n=8.0
n=0
UT
FGT
300
n=0.2
250
n=0.4
n=0.6
n=0.8
n=1.0
200
n=2.0
n=8.0
150
n=4.0
100
n=10.0 n=6.0
6 n=10.0 n=6.0
50
0.5
0.6
0.7
1.1
1.2
0.5
1.3
0.6
0.7
(a) SEA
1.1
1.2
1.3
(b) F max
Fig. 13. Comparisons of SEA and Fmax of FGT and UT columns at the timeframe of 15 ms.
18
350
UT
FGT
n=0.4
14
n=1.0
12
n=0.6
n=0.8
n=4.0
10 n=8.0
n=2.0
8 n=10.0 n=6.0
0.5
0.6
n=0.2
250
n=0.4
n=0.6
n=0.8
n=1.0
200
n=2.0
n=8.0 n=4.0
150
100
0.7
1.1
1.2
n=0
UT
FGT
300
n=0.2
SEA (kJ/kg)
16
n=0
n=10.0 n=6.0
1.3
0.5
0.6
(a) SEA
0.7
1.1
1.2
1.3
(b) F max
Fig. 14. Comparisons of SEA and Fmax of FGT and UT columns at the timeframe of 20 ms.
SEA 21.7712 kJ, Fmax 121.8934 kN), the 2nd order polynomial
function may yield better optimal results. It is thus concluded that
the higher metamodeling accuracy of RSM does not necessarily
yield better nal optimal results.
In order to further discuss the results of the two single objective
optimizations, SEA and Fmax predicted by the different RS models
are listed in Tables 6 and 7, respectively. These single objective
optima given in both tables correspond to the special (or idealized)
points in the Pareto space, which lie at each end of the Pareto
curves as in Fig. 18. From both tables, the optimum exponents of the
FGT tube for the same objective function generally differ from the
different RS models. To minimize Fmax, for instance, the optimal
24
350
n=0
UT
FGT
22
14
n=8.0 n=6.0
n=10.0
SEA (kJ/kg)
18
16
300
n=0.2
n=0.4
n=0.6
n=0.8
n=1.0
20
UT
FGT
n=2.0
n=4.0
12
10
0.4
0.6
0.8
1.0
Mass (kg)
(a) SEA
1.2
n=0
1.4
n=0.2
250
n=0.4
n=0.6
n=0.8
n=1.0
200
n=2.0
150
n=4.0
n=6.0
n=10.0
100
n=8.0
0.4
0.5
0.6
0.7
1.1
(b) F max
Fig. 15. Comparisons of SEA and Fmax of FGT and UT columns at the timeframe of 25 ms.
1.2
1.3
1.4
Fig. 16. Deformation modes of FGT and UT columns with different graded parameters n at different time steps.
71
72
Value
Population size
Number of generations
Crossover probability
Crossover distribution index
Mutation distribution index
20
50
0.9
10.0
20.0
Yes
Satisfy?
No
5. Conclusions
This paper proposed to characterize and optimize the crashing
performance of thin-walled structures with functionally graded
thickness (FGT). The crashworthiness of the FGT structures was
examined and the effects of the different thickness gradients on
both specic absorption energy (SEA) and peak force (Fmax) levels
24
Table 4
Accuracy assessment of the RSM metamodels (0 n 1).
SAE
FG
max
SAEU
FU
max
Order
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
R2
0.95296
0.95335
0.95394
0.95496
0.88159
0.89639
0.91112
0.95557
0.99374
0.99478
0.99512
0.99513
0.99473
0.99929
0.99976
0.99976
23
Fitting indicators
RAAE
RMAE
0.15160
0.14830
0.14912
0.14650
0.27700
0.23092
0.27446
0.16865
0.06528
0.06732
0.06412
0.06376
0.06879
0.02005
0.01266
0.01285
0.75046
0.75292
0.74572
0.74061
0.66889
0.67663
0.53360
0.45332
0.15181
0.11677
0.11813
0.12022
0.10292
0.05267
0.02571
0.02571
22
SEA (kJ/kg)
Objective
21
20
FGT from Eq. (13)
FGT from Eq. (14)
FGT from Eq. (15)
FGT from Eq. (16)
19
18
17
-360
-320
21
SAEG
FG
max
SAEU
FU
max
Order
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
0.98745
0.99113
0.99373
0.99525
0.98875
0.98928
0.98943
0.99017
0.92491
0.92566
0.93214
0.93311
0.98995
0.99912
0.99939
0.99984
19
Fitting indicators
RAAE
RMAE
0.07209
0.06297
0.05478
0.05980
0.07963
0.08803
0.08366
0.07934
0.16616
0.16722
0.20066
0.16312
0.09155
0.02413
0.02299
0.01013
0.29503
0.25297
0.18699
0.19098
0.21789
0.19839
0.21692
0.24795
0.70909
0.72494
0.65886
0.66710
0.13862
0.06688
0.04259
0.02590
SEA (kJ/kg)
Objective
-240 -200
-Fmax (kN)
-160
-120
-80
20
Table 5
Accuracy assessment of the RSM metamodel (1 < n 10).
-280
18
17
16
15
14
13
12
-200
-180
-160
-140
-Fmax (kN)
-120
-100
Fig. 18. Pareto fronts of FGT thin-walled columns and corresponding UT columns.
RS order
Single objective
Grade n
SEA (kJ/kg)
Fmax (kN)
UT
Linear
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
2.25e-08
1.0000
1.92e-10
0.8826
6.34e-09
1.0000
5.19e-11
1.0000
7.24e-10
1.0000
7.51e-10
0.6341
8.23e-09
0.4686
0.0924
0.8239
22.580
17.591
23.223
18.376
23.308
18.150
23.331
18.173
23.297
20.487
23.488
21.411
23.264
21.889
23.339
21.064
304.143
182.233
328.047
203.937
335.793
198.390
335.392
197.989
222.372
107.428
286.808
114.253
316.184
121.440
222.855
133.173
Quadratic
Cubic
Quartic
FGT
Linear
Quadratic
Cubic
Quartic
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Table 7
Ideal optimums of the two single objective functions for the FGT and UT columns
(1 < n 10).
Columns
RS order
Single objective
Grade n
SEA (kJ/kg)
Fmax (kN)
UT
Linear
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
Ideal
1.0000
10.0000
1.0000
8.0568
1.0000
10.0000
1.0000
9.4953
1.0000
10.0000
1.0000
10.0000
1.7073
10.0000
1.0000
8.9055
16.285
12.441
17.536
13.092
17.834
13.394
18.051
13.277
20.581
15.496
20.796
15.711
20.316
16.329
20.538
15.461
174.657
112.189
190.947
123.352
196.785
122.643
198.577
123.946
138.381
108.557
142.125
112.300
137.889
110.728
149.415
106.600
Quadratic
Cubic
Quartic
FGT
Linear
Quadratic
Cubic
Quartic
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
Max SEA
Min Fmax
73
74
[39] Tarigopula V, Langseth M, Hopperstad OS, Clausen AH. Axial crushing of thinwalled high-strength steel sections. Int J Impact Eng 2006;32:847e82.
[40] Paik JK, Kim BJ, Park DK, Jang BS. On quasi-static crushing of thin-walled steel
structures in cold temperature: experimental and numerical studies. Int J
Impact Eng 2011;38:13e28.
[41] Zarei HR, Krger M. Crashworthiness optimization of empty and lled
aluminum crash boxes. Int J Crashworthiness 2007;12:255e64.
[42] Zhang X, Zhang H. Optimal design of functionally graded foam material under
impact loading. Int J Mech Sci 2013;68:199e211.
[43] Kazanc Z, Bathe KJ. Crushing and crashing of tubes with implicit time integration. Int J Impact Eng 2012;42:80e8.
[44] Abramowicz W, Jones N. Dynamic progressive buckling of circular and square
tubes. Int J Impact Eng 1986;4:243e70.
[45] 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;527:1911e9.
[46] Hou SJ, Han X, Sun GY, Long SY, Li W, Yang XJ, et al. Multiobjective optimization for tapered circular tubes. Thin-Walled Struct 2011;49:855e63.
[47] Kurtaran H, Eskandarian A, Marzougui D, Bedewi NE. Crashworthiness design
optimization using successive response surface approximations. Comput
Mech 2002;29:409e21.
[48] Hamad H, Al-Smadi A. Space partitioning in engineering design via metamodel acceptance score distribution. Eng Comput 2007;23:175e85.
[49] Sun GY, Li GY, Gong ZH, He GQ, Li Q. Radial basis functional model for multiobjective sheet metal forming optimization. Eng Optim 2011;43:1351e66.
[50] Jin R, Chen W, Simpson TW. Comparative studies of metamodelling techniques under multiple modelling criteria. Struct Multidiscipl Optim 2001;23:
1e13.
[51] Franulovic M, Basan R, Prebil I. Genetic algorithm in material model parameters identication for low-cycle fatigue. Comput Mater Sci 2009;45:505e10.
[52] Murugan P, Kannan S, Baskar S. NSGA-II algorithm for multi-objective generation expansion planning problem. Electric Power Syst Res 2009;79:622e8.
[53] Nagel GM, Thambiratnam DP. A numerical study on the impact response and
energy absorption of tapered thin walled tubes. Int J Mech Sci 2004;46:201e16.