Mechanics of Advanced Materials and Structures

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/umcm20

Crashworthiness of nested corrugation square

energy-absorbing tubes with circumferentially
cosine profile

Honghao Zhang, Dongtao Yu, Danqi Wang, Tao Li, Yong Peng, Lin Hou &
Zhixiang Li

To cite this article: Honghao Zhang, Dongtao Yu, Danqi Wang, Tao Li, Yong Peng, Lin Hou &
Zhixiang Li (27 Sep 2023): Crashworthiness of nested corrugation square energy-absorbing
tubes with circumferentially cosine profile, Mechanics of Advanced Materials and Structures,
DOI: 10.1080/15376494.2023.2257694

To link to this article: https://doi.org/10.1080/15376494.2023.2257694

View supplementary material

Published online: 27 Sep 2023.

Submit your article to this journal

Article views: 214

View related articles

View Crossmark data

Citing articles: 7 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=umcm20
MECHANICS OF ADVANCED MATERIALS AND STRUCTURES
https://doi.org/10.1080/15376494.2023.2257694

ORIGINAL ARTICLE

Crashworthiness of nested corrugation square energy-absorbing tubes with

circumferentially cosine profile
Honghao Zhanga, Dongtao Yua, Danqi Wangb, Tao Lic,d, Yong Pengc,d, Lin Houc,d , and Zhixiang Lic,d
a
Key Laboratory of High Efficiency and Clean Mechanical Manufacture (Ministry of Education), School of Mechanical Engineering, Shandong
University, Jinan, China; bCollege of Automobile and Mechanical Engineering, Changsha University of Science and Technology, Changsha,
China; cKey Laboratory of Traffic Safety on Track of Ministry of Education, School of Traffic and Transportation Engineering, Central South
University, Changsha, China; dJoint International Research Laboratory of Key Technology for Rail Traffic Safety, Central South University,
Changsha, China

ABSTRACT ARTICLE HISTORY

With the development of railway technology, high-speed trains have been invested in practical Received 1 July 2023
operation increasingly. The traffic accident that train runs at high speed will cause casualties and Accepted 7 September 2023
property losses. The energy absorption tube can effectively alleviate the damage of the vehicle in
KEYWORDS
the collision process and protect passengers in the buffering process. In this paper, nested corru­
Nested tubes; cosine
gation square energy-absorbing tubes (NCSTs) with circumferentially cosine profile and different corrugation; crashworthi­
inner structures are analyzed by adopting finite element (FE) simulation. The quasi-static compres­ ness; numerical simulation;
sion tests and proposed theoretical model based on Simplified Super Folding Element (SSFE) the­ theoretical prediction
ory are employed to validate the effectiveness of numerical models. The simulation results
indicates that the crushing modes possess certain sensitiveness to cross-sectional conformation
and the NCSTs present more stable and orderly deformation modes compared with non-nested
tubes. The specific energy absorption (SEA) and crushing force efficiency (CFE) of NCST can be
increased by 52.49 and 19.07% compared to ordinary corrugation tube, respectively. In addition,
the tube with two-layers octagonal wall (SO1D) is selected as optimal structure because it is
ranked first under cases of 40% by adopting the technique for order preference by similarity to
ideal solution (TOPSIS) method. Finally, the sensitivity analyses based on SO1D are conducted. The
results show that the parametric variation of trib and tout fairly effect on the deformation pattern.
And increases in wall thickness t and section size of the middle wall Dmid will all result to increases
in initial peak crushing force (IPCF) and significant variation on other crashworthiness criteria. This
paper offers certain reference significance for the study of energy-absorbing structures.

1. Introduction lead to more reasonable multi-corner unit distribution and

more regular and progressive folding mode of folds [8,
Thin-willed tubes are widely used as energy absorber in
19, 20].
fields of vehicle, rail transportation, aerospace and other
During the collision process, the energy absorbers would
industries owing to the stiffness weight ratio, specific energy be transferred tremendous energy which are mainly dissi­
absorption and corrosion resistance [1, 2]. When impacted, pated in corner units [21, 22]. Therefore, the corrugation
the thin-walled tubes can effectively reduce the impact on configuration is proposed and various types have been
passengers. In recent years, many scholars have been com­ derived, such as trigonometric corrugation [23], Fourier cor­
mitted to improving the crushing capacity of thin-walled rugation [24], triangular corrugation [25] and biomimetic
tubes and have made satisfactory progress [3–5]. Multiple corrugation [26, 27]. Sadighi et al. [28, 29]proposed a new
methods have been applied to ameliorate the crashworthi­ axially half-corrugated thin-walled tube by introducing four
ness performance, including changing the cross-section uniformly distributed grooves along the tube wall. The
shape [6, 7], changing material strength [8, 9], adding results show that the inversion deformation mode can
inducements [10], filling structures [11–13] and optimizing increase the friction force area and improve the crashworthi­
parametric configuration [14–16]. Current studies have ness of the structure. In the studies of Jamal-Omidi et al.
showed that the design for structural cross-section is a most [30] and Deng et al. [31], a new lateral corrugation config­
effective and direct approach to enhance the crushing cap­ uration was introduced into thin-walled circle tubes, and the
acity [17, 18]. The design of cross-section is prior to para­ results show that the number and amplitude of sinusoidal
metric optimization, the reasonable configuration would corrugation greatly affected the crashworthiness. Liu et al.

CONTACT Lin Hou [email protected]; Zhixiang Li [email protected] Key Laboratory of Traffic Safety on Track of Ministry of Education, School of
Traffic and Transportation Engineering, Central South University, Changsha, China.
Supplemental data for this article can be accessed online at https://doi.org/10.1080/15376494.2023.2257694.
� 2023 Taylor & Francis Group, LLC
2 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

[32] studied the longitudinal corrugation tube and found nearest point of the corrugation outer wall perform steadier
that dynamic asymptotic buckling can achieve optimal collapse modes and better crashworthiness. Therefore, it can
energy dissipation effect and peak force can be significantly be confirmed that nested configurations have good potential
reduced. Furthermore, cosine corrugation-reinforced tube for improving the crashworthiness of structures.
has been proved that to be a very reasonable configuration, As mentioned above, current researches mainly focus on
and more stable and orderly crushing pattern can occur by the energy absorption tubes with single feature cross-section
combining multi-cell and multi-corner configuration while the combining the nested structure with the inner rib
[15, 33]. is of great help for energy absorption. It has been proved
However, the corrugation configuration can only change that corrugation square corrugated tube has obvious advan­
the deformation mode of the structure to a certain extent, tages in energy absorption compared with circular tubes
and the crushing form of the outer layer will be fairly [33]. Therefore, in this paper, a novel nested corrugation
affected by the internal structure. Studies found that the square energy-absorbing tubes (NCSTs) with circumferen­
reinforced ribs inside the tube can guide deformation and tially cosine profile and different inner structures are pro­
improve the crashworthiness [34, 35]. Li et al. [33] devel­ posed. Distinct with the existing research, this work is
oped circumferentially corrugated square tubes (CCST) with mainly reflected in the following three aspects: 1) designing
three types of cross inner ribs to induce the deformation of a series of nested corrugation square energy-absorbing tubes
CCSTs and found the tubes deformed progressively due to (NCSTs) and conducting crashworthiness numerical analysis
the existing of crisscross inner rib. The single tube can be under axial compression; 2) considering multiple crash­
divided into several units when the inner ribs were intro­ worthiness criteria and determining the optimal structure by
duced. Compared with the mono-cell tube, the multi-cell TOPSIS method; 3) proposing a theoretical prediction model
tube has a significantly stronger energy dissipation capacity based on Simplified Super Folding Element (SSFE) theory
[36, 37]. Alavi Nia et al. [37] introduced ribs configuration and verify effectiveness.
into various shapes of thin-walled tubes and the results indi­ In Section 2, the FE models of NCSTs are established
cate that the multi-cell sections perform greater energy- and validated by experimental data. In Section 3, the numer­
absorbing capacity compared with that of simple sections. ical simulations of NCSTs are compared to discuss the
Zhang et al. [38] developed a multi-cell hybrid cross-section energy absorption characteristic under axial loading and the
tube by raising the number of ribs and concluded that this technique for order preference by similarity to ideal solution
multi-cell tube is more stable. In addition, Fu et al. [39] and (TOPSIS) method is selected to determine the ideal struc­
Qin et al. [40], respectively conducted research on biomim­ ture. In Section 4, the theoretical prediction model is pro­
etic thin-walled structures, the experiments and simulation posed. In Section 5, sensitivity analysis of parameters is
results all show that rib configuration make a great contri­ applied for ideal structure. Ultimately, some conclusions are
bution to improving the crashworthiness of structures. drawn in Section 6.
In addition to the guiding effect of the ribs, the inter­
action between the inner and outer walls may also improve
2. Numerical model and verification
the crushing performance of the structure, such as nested
tubes and sandwich tubes. A kind of novel sandwich sinus­ This section introduces geometric design and numerical
oidal lateral corrugated tubes were proposed in the research models of 22 thin-walled tubes. In addition, the results of
of Deng et al [41]. The interaction among outer circular simulation are compared with experiment data. It is found
tubes, middle lateral corrugation tubes and inner circular that numerical models have considerable reliability so that
tubes will make the sandwich structure have higher energy the structures can be developed by numerical simulation in
dissipation capacity that the value of energy absorption (EA) the follow.
raised 36%. Chen et al. [42] investigated a new sandwich
design structures including three types of shape and found
2.1. Description of geometric characteristic
the hexagonal corrugated sandwich column performed the
par excellence energy-absorbing capacity. Yang et al. [25] In this paper, 22 corrugation square energy-absorbing tubes
designed a new type hierarchical truncated conical shells with circumferentially cosine profile are examined. NCSTs
(HTCS) with corrugated cores and observed that the layer are formed by adding one or two internal structures to the
number of corrugated cores HS and corrugations along cir­ outer wall of the corrugated square tube, including circular,
cumference N both are conducive to HTCS achieving octagonal, square and corrugated square tubes. Cross-section
enhanced overall buckling resistance and enhanced energy shapes of structures of 22 thin-walled structures are shown
absorption. Tran et al. [43] investigated the collapse mode in Figure 1.
of multi-stage nested aluminum alloy tubular tubes with dif­ For example, the tube SO2D is the structure is nested
ferent number of layers under dynamic axial loading. The twice; the first letter “S” represents outer configuration
decision-making results show that excessive number of “square” tube, the second letter “O” represents the inner
layers can lead to a decrease in comprehensive performance, “octagonal” structure, the last letter “D” represents the struc­
and the most reasonable configuration is three layers. In ture is nested twice and the number “2” indicates the nest­
addition, Albak [44] developed a series of nested thin-walled ing modality of the structure. Figure 2 shows cross-section
tubes. The investigation shows that the ribs connected to the of circumferentially corrugated square outer wall and its
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 3

Figure 1. Cross-section diagrams of 22 structures. The Baseline is regarded as original structures which consists of corrugation square outer wall. Sribs is formed by
introducing star-shaped ribs into the foundation outer wall. For SC, SO, SS and SCOS tubes, the structures are formed by nesting circle tubes, octagon tubes, square
tubes and square tubes with cosine contours once or twice on the base of the first two elements, respectively.

Figure 2. The thin-walled tube and its cross street schematic diagram (Taking SO1 and SO1D as examples): (a) the energy-absorbing tube; (b) cross section of tubes
with once and twice nested structures. Purple dotted line is expressed by the cosine function with an interval of −20 to 20 and the other three edges can be
obtained by circular array.

cosine expression is defined as bellow: to the origin. In order to convenience of analysis, all of the
� � � � structures have the equal normal lengths L0, height h and
5 L0 L0 L0
y ¼ A cos px þ ; x2 , (1) thickness t, that L0 ¼ 40 mm, h ¼ 100 mm and t ¼ 1 mm.
L0 2 2 2
Table 1 shows the detailed levels of geometric parameters
where A is the amplitude and L0 is the nominal side length. and mass of 22 thin-walled structures and the number of
In this study, amplitude of the configuration A is deter­ nested internal structure layers are represented by N.
mined as 3 mm. The other three edges are obtained by pass­ The design of internal structure size refers to the research
ing the magenta edge line through the circular array relative of Albak [44]. The external corrugated square is inscribed
4 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

pffiffi
with a circle with a diameter of 40 2 mm. For once nested The nonlinear explicit FE analysis program LS_DYNA is
structures, inner square and p octagon
ffiffi are inscribed with a performed for FE numerical simulation. The FE model con­
circle with a diameter pffiffi of 18.4 2 mm, and the diameter of sists of three components, namely moving plate, fixed plate
the circle are 18.4 2 mm. For twice nested structures, mid­ and test tube, which is shown in Figure 3.
dle square andpoctagon are inscribed with a circle with a The test tube is clamped between the moving plate and the
ffiffi
diameter of 22 2 mmpand fixed plate. The fixed plate is fixed at the bottom and the mov­
ffiffi the inner are tangent to a circle
with a diameter of 10 2 mm, and the diameters of the cir­ ing moved uniformly in the vertical direction. The test tubes
cle are respectively 22 and 10 mm. The corrugation square is are modeled with the Belytschko-Tsay shell element with one
designed with a nominal length of 18.4 mm for tubes with a center integration point and five through-thickness integration
single internal structure, and a nominal length of 10 and points. The global element size is 1 � 1mm. The loading vel­
22 mm for tubes with two internal structures. The inner ocity is defined 3 m/s which is able to improve computing effi­
structures nested in the tube are connected to the corrugated ciency and the crushing distance is 60 mm. The contact
outer wall through the star inner ribs. between moving plate and test tubes is defined using the
“AUTOMATIC_NODE_TO_SURFACE” contact algorithm
and the “AUTOMATIC_SINGLE_SURFACE” contact algo­
2.2. Establishment of FE model rithm is used to define self-contact of test tubes. The coeffi­
cients of friction of dynamic and static are defined as 0.3.
FE numerical simulation is applied to investigate crash­
Both of the plates are modeled with MAT_RIGID (Mat_
worthiness characteristic of 22 structures under axial load.
20) considering the rigid wall and the test tubes are modeled
with MAT_PIECEWISE_LINEAR_PLASTICITY (Mat_24).
Table 1. The detailed parameters of 22 thin-walled tubes.
The material of test tubes is aluminum alloy AA6061O and
Name N MASS (kg) L0 (mm) h (mm) t (mm)
the material properties are shown in Table 2. The true
Baseline 0 0.06 40 100 1
Sribs 0 0.11 40 100 1 stress-strain feature of AA6061O is shown in Figure 4 [33].
SC1 1 0.10 40 100 1
SC1D 2 0.13 40 100 1
SO1 1 0.11 40 100 1 2.3. Validation of FE model
SO1D 2 0.13 40 100 1
SO2 1 0.10 40 100 1 The FE simulation result of Baseline and Sribs are used to
SO2D 2 0.13 40 100 1
SO3D 2 0.13 40 100 1 compare with the experiment results of previous studies [33].
SO4D 2 0.13 40 100 1 The test machine MTS Landmark was used to conduct quasi-
SS1 1 0.11 40 100 1 static compression test with the loading velocity of 2 mm/min.
SS1D 2 0.13 40 100 1
SS2 1 0.11 40 100 1
SS2D 2 0.13 40 100 1 Table 2. Material properties of AA6061O.
SS3D 2 0.13 40 100 1 Properties Values
SS4D 2 0.13 40 100 1
SCOS1 1 0.11 40 100 1 Density q ¼ 2700 kg/m3
SCOS1D 2 0.14 40 100 1 Young’s modulus E ¼ 69.79 GPa
SCOS2 1 0.11 40 100 1 Yield stress ry ¼ 54 MPa
SCOS2D 2 0.14 40 100 1 Ultimate stress ru ¼ 163.46 MPa
SCOS3D 2 0.14 40 100 1 fracture strain ef ¼ 0.18
SCOS4D 2 0.14 40 100 1 Poisson’s ratio t ¼ 0.33

Figure 3. Establishment of finite element model of energy-absorbing tubes: (a) the test aluminum tube is fixed on the testing machine; (b) schematic diagram of
finite element model: moving plate is shown as the red part, which push along z-axis with the constant speed of 3 m/s. Test tube compressed is the grey part which
is fastened on the Fixed plate which is constrained in six degrees of freedom.
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 5

The materials and parameters of the experimental samples are to be simulated in this paper. In addition, there are many
the same as those of the simulation. The deformation form uncontrollable factors during the experiment, such as manu­
and force-displacement of Baseline and Sribs are shown in facturing error and operation error. The FE simulation has
Figure 5. higher stability relatively that result to more ideal outcomes.
As shown in Figure 5(c), the deformation patterns of In addition, according to Figure 5, the force-displacement of
Baseline are generally same under experiment and simula­ FE simulation generally coincide with experiment result.
tion results, which all present global bending mode. When Moreover, some significant crashworthiness indicators are
Sribs is examined, the folds of experiment display bigger also compared in the Table 3.
size and fewer number compared with simulation results, From Table 3, it can be seen that all errors are less than
which is shown in Figure 5(d). It can be explained that 8.50% for crashworthiness criteria. The reasons for errors
small size element is selected due to the complex structures can be explained by the complexity of structure and instabil­
ity of experiment. In summary, the result shows that accur­
acy of the constructed FE model is validated and it can be
applied for further investigation of nested structures.

3. Comparison of the nested structures under axial

loading
In this section, The NCSTs are investigated by comparing
deformation modes, force-displacement curves and scores of

Table 3. Comparison of energy absorption index values between experiment

and FE simulation.
Tube IPCF (kN) MCF (kN) EA (kJ) SEA (kJ/kg)
Baseline Test 18.34 5.70 0.34 6.14
FE 17.95 5.60 0.34 6.08
Error (%) 2.17 1.78 1.19 0.98
Sribs Test 37.31 25.73 1.54 13.93
FE 40.35 27.11 1.63 14.67
Figure 4. Ture stress-strain curve of AA6061O [33]. Error (%) 8.14 5.36 5.84 5.31

Figure 5. Comparison of the result between experiment and FE simulation: (a) the bedstand for crushing experiment; (b) aluminum tubes used for testing: Baseline
and Sribs; (c) is the force-displacement curves of baseline in test and FE model and the deformation patterns when the tubes is compressed to 40% of the high; the
force-displacement curves and the deformation patterns compressed by 60% of Sribs are shown in (d).
6 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

Figure 6. The deformation modes of 22 thin-walled structures under axial load. All of the tubes are crushed to 40% of the original length.

crashworthiness criteria. Furthermore, a multi-criteria deci­ The MCF is described as follow:

sion-making (MCDM) approach is developed to evaluate Ð Smax
EA f ðxÞdx
and determine the ideal structures. IPCF ¼ ¼ 0 (4)
Smax Smax
IPCF and MCF are important parameters. Usually, two
3.1. Description of energy absorption capacity
indicators are required to evaluate the energy absorption
Some crashworthiness indicators are defined in order to efficiency. Therefore, the crushing force efficiency is defined:
quantify the energy absorption capability [32, 45]. In this Fmean
paper, five indictors, include total energy absorption (EA), CFE ¼ (5)
Fpeak
specific energy absorption (SEA), Initial peak force (IPCF),
mean crushing force (MCF) and crushing force efficiency When MCF is closer to IPCF, it indicates that the structure
(CFE), are selected to evaluate the crushing performance of can absorb the load more effectively, and the structure is more
structures. stable. In other words, the closer the CFE value is to 1, the bet­
The IPCF is very important for studying the impact load ter the energy absorption efficiency of the structure is.
on the crew, which can be observed visibly on the force-dis­
placement curve. It represents the maximum load that the
3.2. Analyzation of the nested structures under axial
structure is subjected, which ordinarily appears in the initial
loading
stage of the crash.
EA represents the amount energy dissipated during the In this study, the proposed NCSTs are analyzed under axial
whole process from the initiate of compressing to the ter­ loading by using FE simulation method to compare energy
minate of loading: absorption capacity of various nested structures. To investi­
ð Smax gate the advantages of NCSTs, Baseline and Sribs are used
EA ¼ f ðxÞdx (2) to contrast. The deformation mode, force-displacement and
0
crashworthiness indictors of all structures are displayed
where Smax is the max crushing length under load, and f(x) respectively in Figures 6–9.
is the impact force function.
The SEA presents the energy absorbed per unit mass and
is described as follow: 3.2.1. Comparison of NCSTs for crushing modes
Under axial impact, global bending occurs in the thin-walled
EA tube named ‘Baseline’. The reason for this mode is that
SEA ¼ (3)
m thin-walled structure with a sufficiently long initial vertical
where m is the mass of energy absorbing tube. length can occur this bending [46, 47]. From the Figure
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 7

Figure 7. The folding pattern of constituent elements in SC1D.

Figure 8. Force-displacement curves of 22 thin-walled structures between 0 mm to 60 mm: (a) Baseline-SC1D; (b) SO1-SO4D; (c) SS1-SS4D; (d) SCOS1-SCO4D.

9(b), Baseline reports the worst EA value. Usually, global mode and unstable mode. This is caused by the connection
bending mode will seriously undermine the energy absorp­ of the star shaped ribs to the outer walls [33]. From the
tion capacity [23]. Actually, most of the tubes reported overall perspective, tubes nested with octagonal structures,
mixed mode which is situated between ordered progressive such as SO1D, SO2D and SO4D have more ordered
8 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

Figure 9. The scores of four crashworthiness criteria of 22 thin-walled structures: (a) IPCF, (b) EA, (c) SEA, (d) CFE.

deformation modes but corrugation square structures uni­ 3.2.2. Comparison of NCSTs for force-displacement response
versally present more unstable. This is because that the tree- The force-displacement curves of 22 nested structures with
shaped elements are generated when the ribs intersect corru­ different inner structures are plotted in Figure 8. At the
gation walls, which will induce stronger local stiffness. The beginning of loading, the force of all nested structures will
best deformation mode occurs to SO2D, the worst deform­ rise significantly, and then drop after reaching the peak
ation mode is presented in SCOS1D. Interestingly, the value, until it is stable and floats in a small range, forming a
SO1D is observed has the largest SEA value CFE value platform. It is found that the crushing energy of is mainly
instead of SO2D. consumed by the way of bending deformation and mem­
In order to clearly observe the deformation patterns of brane deformation. The bending energy mainly depends on
the corner components and typical folding units of NCST, the cross-section length, and the membrane deformation
the constituent elements of SC1D are separated and exam­ usually depends on the corner element. [42, 49]. Compared
ined individually, which are shown in Figure 7. with tubes nested once, tubes nested twice have more cross-
According to Figure 7, the constituent elements of SC1D can section lengths and corner elements. All of the tubes nested
be characterized as five types, including Y-shape element I, T- twice have the plateau fluctuations that are more intensive
shape element I, shell corner element, convex tree-shaped elem­ than tubes nested once. This is because that during the
ent and Y-shape element II. It can be seen that five elements crushing process, the tubes with two internal structures pre­
exhibit diverse deformation shapes depending on the crushing sent more folds that are conducive to energy absorption. At
mode of the folds. As shown in fi, both sides of the shell corner the same time, these structures nested twice appear more
element deform toward the same direction, ultimately forming progressive deformation modes than the structures nested
an typical in-extensional mode [33, 48]. For the T-shape element once, which is shown in the Figure 8. Corner elements will
I which shown in ›, the vertical blade sandwiched in the mid­ result in high local stiffness and engender tubes have higher
dle also folds in the same direction on both sides. For the Y- peak force and make tubes absorb more energy, which cor­
shape element I and II, all of the arc-shaped blades collapse responds to Figure 9 where SCOS4D has the highest peak
toward the end that deviates from the center of the circle. But force with 65.92 kJ and the highest EA values with 3.05 kJ.
distinct with Y-shape element I, the folds on the blades of II are In addition, the fluctuation of SCOS is more unstable com­
more orderly and progressive, which means that the deform­ pared with the tubes which have same inner structure quan­
ation mode on the inner side of the structure is better than on tity. The result can be explained by the fact that SCOS has
the outer side. less ordered deformation mode.
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 9

The force-displacement curves of 22 nested structures preference by similarity to ideal solution (TOPSIS) is
with different inner structures are plotted in Figure 8. At employed to evaluate the crashworthiness of nested corruga­
the beginning of loading, the force of all nested structures tion square energy-absorbing tubes with circumferentially
will rise significantly, and then drop after reaching the peak cosine profile. In this approach, the similarity with positive
value, until it is stable and floats in a small range, forming a ideal solution and negative ideal solution is determined by
platform. It is found that the crushing energy of is mainly calculate Euclidean distance [50, 51]. The TOPSIS method is
consumed by the way of bending deformation and mem­ now widely used in the research of energy absorbing struc­
brane deformation. The bending energy mainly depends on ture owing to its applicability and simplicity [52–57]. The
the cross-section length, and the membrane deformation main procedure of TOPSIS method is depicted as bellow:
usually depends on the corner element. [42, 49]. Compared Step 1: Establish the initial decision matrix X.
with tubes nested once, tubes nested twice have more cross- 2 3
section lengths and corner elements. All of the tubes nested x11 x12 ::: x1n
twice have the plateau fluctuations that are more intensive 6 x21 x22 ::: x1n 7
6 7
than tubes nested once. This is because that during the X ¼ ½xij �mn ¼6
6 ... .. .. .. 7 (6)
4 . . . 7
5
crushing process, the tubes with two internal structures pre­ ..
sent more folds that are conducive to energy absorption. At xm1 xm1 . xmn
the same time, these structures nested twice appear more
progressive deformation modes than the structures nested where xij denotes the numerical values of the i-th alterna­
once, which is shown in the Figure 8. Corner elements will tive relative to the j-th design indicator, m and n represent
result in high local stiffness and engender tubes have higher the quantity of alternatives and indictor. In this study, four
peak force and make tubes absorb more energy, which cor­ selected crashworthiness values of 20 nested structures
responds to Figure 9 where SCOS4D has the highest peak shown in Table 4 are utilized to establish the initial decision
force with 65.92 kJ and the highest EA values with 3.05 kJ. matrix.
In addition, the fluctuation of SCOS is more unstable com­ Step 2: Convert to normalized decision matrix R.
pared with the tubes which have same inner structure quan­ In the four evaluation criteria selected, IPCF and MASS
tity. The result can be explained by the fact that SCOS has are cost criteria and SEA and CFE are benefit criteria. The
less ordered deformation mode. former means the bigger the better, while the latter means
the opposite. It is necessary to compute the dimensionless
normalized decision matrix so that the criteria with different
3.2.3. Comparison of NCSTs for crashworthiness criteria dimensions and units can be compared. The normalized
Figure 9 exhibits the performances of crashworthiness crite­ decision matrix R can be derived as follow:
ria of 22 thin-walled nested structures. The SO1D has the 8
most prominent SEA value which is 267.93% higher than > maxxij − xij
>
> i
the value of Baseline and 52.49% higher than the value of >
> , i ¼ 1, 2, :::, k; j 2 j1 :
< maxxij − maxxij
Sribs. In addition, SO1D and SO3D have the maximum CFE R ¼ ½rij �mn ¼ i i
(7)
> xij − maxxij
values of 0.80, which is 158.06% higher than Baseline. When >
> i
, i ¼ 1, 2, :::, k; j 2 j2 :
>
>
the EA values are examined, the maximum value is at : maxxij − maxxij
i i
SCOS4D with 50.86 kJ, followed by SCOS1D and SCOS2D
with 50.32 and 50.07 kJ. Meanwhile, they have the high peak
force with 65.92, 65.85 and 65.80 kJ. It can be seen that
SCOS4D has the most prominent EA and the highest IPCF
value. However, it is difficult to distinguish the thin-walled Table 4. The numerical results to generate decision matrix.
structures only by qualitative analysis, so it is necessary to No. Tube IPCF (kN) SEA (kJ/kg) MASS (kg) CFE
conduct decision-making to systematically determine the 1 SC1 48.12 19.48 0.06 0.71
2 SC1D 61.03 21.10 0.11 0.77
optimal structure. 3 SO1 48.91 19.87 0.10 0.72
4 SO1D 62.28 22.37 0.13 0.80
5 SO2 47.64 19.39 0.11 0.71
3.3. Selection of the ideal structure 6 SO2D 60.49 20.97 0.13 0.77
7 SO3D 61.47 22.09 0.10 0.80
In this section, NCSTs are evaluated by using MCDM 8 SO4D 60.95 21.08 0.13 0.77
9 SS1 48.53 18.65 0.13 0.68
method to select the ideal structure. To comprehensively 10 SS1D 60.77 20.67 0.13 0.75
evaluate the crashworthiness of the structure, three crash­ 11 SS2 48.90 18.84 0.11 0.69
worthiness indicators, IPCF, SEA, CFE, are chose as the 12 SS2D 61.39 20.92 0.13 0.75
13 SS3D 60.08 20.23 0.11 0.74
decision criteria. Considering the lightweight, the mass of 14 SS4D 60.79 20.49 0.13 0.74
structure, MASS, is also introduced. 15 SCOS1 50.86 19.61 0.13 0.71
16 SCOS1D 65.85 21.46 0.13 0.76
17 SCOS2 50.92 19.81 0.11 0.72
18 SCOS2D 65.80 21.36 0.14 0.76
3.3.1. Main processes of TOPSIS method 19 SCOS3D 65.65 21.09 0.11 0.75
In order to sort reasonably the options with conflicting cri­ 20 SCOS4D 65.92 21.69 0.14 0.77
teria, and find the optimal scheme, the technique for order
10 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where j1 is the set of benefit criterion and j2 is the set of uX
u n � �2
cost criteria. Dþ
i ¼t vij − vjþ , i ¼ 1, 2, :::, m; j ¼ 1, 2, :::, n: (11)
Step 3: Determine the weighted normalized decision j¼1
matrix V.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V ¼ ½vij �mn ¼ xj � rij , i ¼ 1, 2, :::, m; j ¼ 1, 2, :::, n: (8) uX
u n
D−i ¼t −
ðvij − vj Þ2 , i ¼ 1, 2, :::, m; j ¼ 1, 2, :::, n: (12)
where xj is the weighed factor of design criteria which rep­ j¼1
resents relative importance of each criterion to weight the
normalized decision. The ten groups of weight factor are Step 6: Obtain the similarities with ideal solutions and
P
listed in Table 5. For each group of weight, nj¼1 xj ¼ 1: rank results according to preference.
Step 4: Obtain the positive ideal solution Iþ and negative The similarity represents the proximity degree of alterna­
ideal solution I−. tives to the ideal scheme. It is expressed by the Equation
� (13):
I þ ¼ max vij , i ¼ 1, 2, :::, mÞ ¼ ðv1þ , vþ þ
2 , :::, vn Þ (9)
i
D−i
Ciþ ¼ , i ¼ 1, 2, :::, m: (13)
� Dþ
i þ D−i
I − ¼ min vij , i ¼ 1, 2, :::, mÞ ¼ ðv−1 , v−2 , :::, v−n Þ (10)
i
where the upper the Ciþ is, the more ideal the option is.
Step 5: Calculate the distances between alternatives and
positive or negative ideal solutions.
3.3.2. Results
The distance between alternatives and positive ideal solu­
In different actual working conditions, the same crash­
tion Dþ and distance between alternatives and negative ideal
worthiness indicator may have different levels of impor­
solution D− can be obtained by using Euclidean distance:
tance, resulting in different relative weights [58]. To
guarantee the rationality of assessment result, ten groups of
Table 5. Ten cases of weight combinations about four criteria. weight combinations about four crashworthiness indicators
Value of weight are used to comprehensively select the ideal structure and
Condition IPCF SEA MASS CFE listed in Table 5.
Case 1 0.25 0.25 0.25 0.25 As shown in Table 5, a criterion has different value in
Case 2 0.7 0.1 0.1 0.1 different weight combine cases. For example, in case 1, the
Case 3 0.1 0.7 0.1 0.1
Case 4 0.1 0.1 0.7 0.1 four criteria have the same weight value with 0.25; in case 2,
Case 5 0.1 0.1 0.1 0.7 the weight of IPCF achieves the highest value with 0.7 and
Case 6 0.5 0.5 0 0
Case 7 0 0.5 0.5 0
the other have the same values with 0.1. The rank results
Case 8 0 0 0.5 0.5 and relative closeness coefficients of 20 nested structures in
Case 9 0.3 0.3 0.3 0.1 ten cases of weight combinations are shown in the
Case 10 0.1 0.3 0.3 0.3
Supplementary Table A1. The Figure 10 shows the radar
response results of decision-making.

Figure 10. The scores of the structures evaluated by the TOPSIS method under the 10 working conditions are displayed in the radar chart. Where, the closer to the
outside of the radar chart means that the structure is more desirable under this condition.
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 11

From Figure 10, it is found that the SO1D ranked first in classified into various constituent elements to predict the
40% of the cases and SC1 ranked first in the 30% of the membrane deformation [59–61]. In this section, the mem­
cases, which can be concluded that the SO1D and SC1 have brane energy of solitary element is calculated, then the total
the advantage in aspect of energy absorption. However, membrane energy is obtained by summing. The thin-walled
when deformation mode and force-displacement curve are structures are segmented and classified as: shell corner elem­
considered, the SO1D emerges more stable crushing mode ent, T-shaped element I and II, Y-shaped element I and II,
and more ideal force-displacement. Meanwhile, the SO1D tree-shaped element, 3-plane element and criss-cross elem­
has the EA value of 50% higher than SC1. This phenomenon ent, which are shown in Figure 11 (The nested structures
shows that SO1D perform the best considerable crashworthi­ with the same internal structure are composed of the same
ness and is select ideal structure and used in the subsequent constituent elements, therefore only typical structures are
parameter investigate. analyzed here).
The membrane energy is extremely dissipated in areas of
multiple planes interactions and corner [48, 62]. The values
4. Theoretical validation of the nested structures of membrane energy quite depend on the included angle of
Theoretical prediction is likewise an extremely indispensable adjacent planes for constitute element. However, the
approach to analyze directly the energy-absorbing perform­ included angles of planes are difficult to determine in some
ance of thin-walled structures. In this section, Simplified element because the shell has certain radian, for instance T-
Super Folding Element (SSFE) approach which is widely shaped element I and convex tree-shaped element. In order
used and confirmed a simply and credible way is employed to achieve simplification and certain prediction accuracy, the
to predict the theoretical values of the MCF of endergonic arcuate shells in the intersection area are approximated as
structures [59, 60]. tangent planes, thus equivalent to plane elements [59, 60].
According to SSFE theory, bending and membrane deform­
ation are main forms to consume external work when 4.2.1. The membrane of Sribs
impacted. Meanwhile, the wavelengths of folds are assumed all For the Sribs, there are four types elements consist cross-
equal 2H and the wall thickness t is supposed to be uniform. section, including shell corner element, T-shaped element I,
The energy balance equation can be described as follow: Y-shaped element I and 8-plane element. According to the
Pm 2Hk ¼ Eb þ Em (14) simplified method, the shell corner element can be equiva­
lent to plane corner element with central angles of 180�
where Pm denotes the mean crashworthiness force; Eb and Em
which is shown in Figure 11(a), and according to Li and
represent the bending energy and membrane energy dissipated
Zhang, the plane corner element generates the deformation
under crushing, respectively. Take account of the actual situ­
of extensional with the membrane energy can be expressed
ation, the crushing length is smaller than 2H. Therefore, the
as [62]:
effective compression factor k is introduced into Equation (14).
In this paper, k was determined as 0.7 [61]. 4M0 H 2 tanðh=2Þ
Ecorner
m
element I
¼
t tan ðh=2Þ þ 0:05=tanðh=2ÞÞ=1:1
4.1. The prediction of bending energy (18)
�
where the h ¼ 180 : On the base of SSFE theory, the deform­
The bending energy can be obtained by calculating the sum
ation mode of 3-plane element is divided three patterns: type I,
of the absorbed energy at stationary hinge lines which was
type II and mixed type, which is shown in Figure 12.
defined as follow [59]:
When the angle a < 90� , 3-plane element perfect to be
Eb ¼ 2pM0 Lc (15) capsuled as type I; when the angle a > 120� , 3-plane element
where Lc is the total length of the section profile, M0 is the perfect to be capsuled as type II; when 90� �a � 120� , 3-
full plastic bending moment per unit length, which can be plane element would be capsuled as mixed type which per­
expressed as: haps generate both types [56, 62]. For Sribs, the T-shaped
element I is equivalent to the a with the angle of 30� [44],
1
M0 ¼ r0 t 2 (16) which can be regarded as type I; the Y-shaped element I has
4 the a of 95� , which can be approximatively regarded as type
where t is the wall thickness of structures and r0 denotes I with a of 90� since the mixed type is difficult to calculate.
the flow stress of structural material which can be expressed The membrane energy of 3-plane element in type I is:
as follow [48]:
E3−plane
m
element I

r0 ¼ 0:92ru (17) � �!
4M0 H 2 tanðaÞ a
¼ � þ 2 tan
t tan ðaÞ þ 0:05=tanðaÞ =1:1 2
4.2. The prediction of membrane energy (19)
The cross-section of tubes is complex. In general, the cross- Tran et al. simplified the 6-panel corner element into two
section of the thin-walled energy absorbing structure is 4-panel elements; similarly, the 8-panel element was
12 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

Figure 11. The constituent elements of structures and simplified form:(a) Sribs; (b) SC tubes; (c) SO tubes; (d) SS tubes; (e) SCOS tubes.

simplified into four 3-panel elements I of a ¼ 45� as shown

as Figure 11(b) [44, 63]. In summary, the total of membrane
energy of Sribs can be described as:

4:4M0 H 2 5:31M0 H 2
ESribs
m ¼ Nc e � þ NT−s e I � þ NY−s e I
t t
12:4M0 H 2 7:5M0 H 2
� þ4�
t t Figure 12. The types of 3-plane element with different deformation patterns:
(20) (a) Type I (0� <a < 90� ); (b) Type II (120� <a < 180� ); (c) Mixed type
(90� <a < 120� ).
where the Nc e , NT−s e I and NY−s e I represent the numbers
of corner element, T-shaped element I and Y-shaped elem­ regarded as criss-cross element [64]:
ent I, respectively. 16M0 H 2
Ecriss−cross
m ¼ (21)
t
4.2.2. The membrane energy of SC Hence, the membrane energy of SC can be obtained as
For SC, there are two types of elements that been discussed, follow:
including Y-shaped element II and convex tree-shaped elem­ 12:4M0 H 2 16M0 H 2
ent which had been displayed in Figure 11(c). The former is ESC out
m ¼ Em þ NY−s e II � þ Nc t−s e � (22)
t t
regraded 3-planed element I with the a of 90� , the latter is
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 13

where NY−s e II and Nc t−s e denotes the numbers of Y- (20). The latter is equivalent to tree-shaped element with
shaped element II and convex tree-shaped element. The total c ¼ 85� . The total membrane energy of SCOS is:
membrane energy of outer wall Eout m is introduced here and2
Eout ¼ N � 4:4M0 H 2
þ N � 5:31M0 H 2
þ NY−s e I � 12:4Mt 0 H . 12:4M0 H 2 11:7M0 H 2
m c e t T−s e I t ESCOS
m ¼ Eout
m þ NY−s e II � þ NT−s e II �
t t
12:3M0 H 2 H2
4.2.3. The membrane energy of so þ Nt−s e I � þ Nt−s e II � 16M0
t t
For SO tubes, the 3-plane element I is approximatively
(27)
equivalent to type II, which can be obtained by the formula
as follow [59, 62]: where NY−s e II , NT−s e II , Nt−s e I and Nt−s e II denote the
� � � � � � number of Y-shaped element II (a ¼ 90� ), T-shaped element
3−plane element II 2M0 H 2 b b
Em ¼ 4tan þ 2sin þ 3sinðbÞ II (a ¼ 85� ), concave tree-shaped element (c ¼ 85� ) and con­
t 4 2
vex tree-shaped element of the inner nested structures.
(23)
where b ¼ 2(p−a), which is the central angle of 3-plane
element. 4.3. The prediction of MCF
The membrane energy of tree shaped element would be In order to predict the MCF of the thin-walled structures,
calculated as follow [65]: the total length of the cross-section profile Lc need to be
� � � �
2M0 H 2 c obtained first. For the corrugation square outer wall, the
Etress−shaped
m
element
¼ 4tan þ 2sin ð cÞ þ 3sinð 2c Þ side length is obtained by calculation formula of arc length,
t 2
(24) which has been displayed as follow:

where c is the included angle of branch and trunk shown in ð 20 sffiffiffiffiffiffiffiffiffi�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�
ffiffiffiffiffiffiffiffiffi�
ffiffiffi�
ffiffiffiffiffi
2
3 p
Figure 11 (d). The membrane energy of SO can be obtained S¼ 1þ psin x dx (28)
−20 8 8
as:
12:4M0 H 2 16M0 H 2 The total length of the outer wall can be got by multiply­
ESO out
m ¼ Em þ N3−p e I � þ Nc−c e ing the side length by four. The profile length of other
t t
2 internal nested structures can be obtained according to the
13:3M0 H 13:3M0 H 2
þ N3−p e II � þ Nt−s e � (25) corresponding calculation method, which will not be related
t t
here.
where N3−p e I , Nc−c e , N3−p e II and N−s e denote the num­ The H can be calculated under stationary condition
ber of 3-plane element I (a ¼ 90� ), criss-cross element, 3- which is:
plane element II (b ¼ 135� ) and tree-shaped element
(c ¼ 67.5� ) nested inside. @Pm
H¼ (29)
@H
4.2.4. The membrane energy of SS Submit the H, Eb and Em into Eq. (14), the theoretical
The constituent elements not mentioned of SS tube contain prediction values of MCF about thin-walled structures under
3-plane element II with b ¼ 90� and tree shaped element axial compression would be obtained. The Table 6 displays
with c ¼ 45� . Their membrane energy would be calculated the comparison of MCF between theoretical and FE model.
by using the Equations (22) and (23). The membrane energy It is found that the errors substantially are less than 10%.
of SS is: Because of the same type and quantity of constituent ele­
12:4M0 H 2 16M0 H 2 ments, SS1 and SS2 as well as SSID, SS2D, SS3D and SS4D
ESS out
m ¼ Em þ N3−p e I � þ Nc−c e report the identical values, which is applicable to the SCOS
t t
tubes. All in all, it is concluded that the theoretical predic­
12:1M0 H 2 12:1M0 H 2
þ N3−p e II � þ Nt−s e � (26) tion of MCF is reasonable and has good accuracy.
t t
where N3−p e I , Nc−c e , N3−p e II and Nt−s e denote the num­ Table 6. The comparison of MCF values about FE model and theoretical
ber of 3-plane element I (a ¼ 90� ), criss-cross element, 3- prediction.
plane element II (b ¼ 90� ) and tree-shaped element (c ¼ 45� ) No. Name FE TH Error No. Name FE TH Error (%)
nested inside. 1 Sribs 27.11 29.71 9.61% 12 SS2 33.50 35.71 6.56
2 SC1 34.02 35.49 4.32% 13 SS2D 46.12 49.52 7.37
3 SC1D 47.15 51.12 8.41% 14 SS3D 44.67 49.52 10.85
4 SO1 35.18 35.76 1.64% 15 SS4D 45.24 49.52 9.46
4.2.5. The membrane energy of SCOS 5 SO1D 50.00 51.12 2.24% 16 SCOS1 36.19 39.29 8.56
There are three not mentioned elements with the name of 6 SO2 33.68 36.00 6.88% 17 SCOS1D 50.32 53.91 7.13
7 SO2D 46.50 50.02 7.56% 18 SCOS2 36.55 39.29 7.49
Y-shaped element II, T-shaped element II and concave tree- 8 SO3D 48.99 51.69 5.51% 19 SCOS2D 50.07 53.91 7.67
shaped element. According to simplification principle, the 9 SO4D 47.17 50.96 8.03% 20 SCOS3D 49.49 53.91 8.93
first two can be equivalent to 3-plane element I with the 10 SS1 33.19 35.71 7.59% 21 SCOS4D 50.86 53.91 5.99
11 SS1D 45.64 49.52 8.50%
a ¼ 90� and a ¼ 85� , which can be obtained by Equation
14 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

Figure 13. The parameter of SO1D. (a) Gray line is the outer wall whose thickness is represented by tint; blue line is the middle wall whose thickness is represented
by tmid; green line is the inner wall whose thickness is represented by trib. (b) Dmid represents the diameter of circumscribed circle of the middle octagon wall.

Table 7. The values of each level of parameter investigated. compressed integrally at the level of 0.5 mm under tout and
Parameters Levels Values (mm) the folds become instable obviously when tout�1.25 mm. In
tint 7 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2 addition, the folds turned into large and non-progressive
tmid 7 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2 with the number of 2. For trib, the instable deformation
tout 7 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2
trib 7 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2
starts to engender when trib�1.5 mm.
Dmid 6 25, 27.5, 30, 31.1, 32.5, 35 When comparing the deformation patterns of SO1D hori­
zontally, it can be seen that tout and trib have the most ser­
5. Parameter analysis of SO1D ious effect to deformation, and tint has the least obvious
effect. In summary, the effect of wall thickness on SO1D is
To investigate the sensitivity of the ideal structure for the non-negligible, and careful consideration should be taken
parameters, a farther parametric analysis of SO1D is investi­ when selecting energy absorption structures.
gated. The thickness of constitutive structures which
includes tint, tmid, tout and trib and the size of middle wall
Dmid will be respectively study, which is shown in Figure 13. 5.1.2. Effect of wall thickness on crashworthiness criteria
According to Figure 13, parameters tint, tmid, tint and trib Figure 15(a) lists the values of IPCF with different parame­
represent the thickness of the outer structure, the middle ters t. It is obvious that IPCF and t reported linear relation­
structure, the inner structure and the ribs, respectively. ship. According to the gradient of curves, it can be
Table 7 lists the levels of t and Dmid. concluded that the effect of thickness on the IPCF of SO1D
As shown as Table 7, The parameters of thickness are is tout, trib, tmid and tint in order from high to low.
designed to 7 levels varying from 0.5 to 2.0 mm; the parame­ In addition, the effect of parameter t on the SEA also was
ters Dmid is designed to 5 levels varying from 25 mm to studied, which lists in the Figure 15(b). It is not difficult to
35 mm. These levels of wall thickness are mutually inde­ see from the image that the change of SEA shows an upward
pendent, which means that the other thickness is controlled trend with the increase of parameter t. In addition, the gra­
at 1 mm when the thickness investigated varies according to dient is highest when the values from 0.5 to 0.75 mm under
the levels. parameter tout and trib and then the gradient decreased with
the values increase. However, there is also an exception
when tout¼2.0 mm, in which the values of SEA reduce to
5.1. Effect of wall thickness on crash resistance of SO1D some extent. Reviewing the deformation, this result can be
5.1.1. Effect of wall thickness on deformation mode explained by the fact that SO1D presents irregular and
In order to study the effect of different thickness on the coarse deformation in that case. There is also a found that
deformation mode, 28 deformation modes of the structure the SO1D deformation under unideal modes when the par­
are generated by using FE numerical simulation (Because of ameter at large levels, but the values of SEA are high.
single variable, when the parameter t is 1 mm, their deform­ Therefore, for the selection of the optimal geometric size,
ation situations are the same). The deformation results of 28 both deformation mode and crashworthiness criteria must
structures with different thickness list in the Figure 14. be considered.
For tint and tmid, it can be seen that deformation patterns Figure 15(c) indicates the effect of thickness parameter
convert from ordered to disordered and unstable and the on the CFE. It is not hard to obverse that the values of CFE
amount of gradual deformation reduce when the t increases. all vary between from 0.75 to 0.9. When
Therefore, when the effect of inner and middle wall thick­ 0.5 mm � t � 1.0 mm, the highest CFE value occurs at tout;
ness is examined, the increase of t will enhance the instabil­ when 1.0 mm � t � 1.25 mm, it is tint. This result is opposite
ity of crushing form of the nested SO1D. For tint with the to the figure reported by SEA. When 1.5 mm � t � 1.75 mm,
level of 2.0 mm, the deformation is instable at the beginning CFE basically remains unchanged even decreases and when
of crushing and the deformation mode become global bend­ 1.75 mm � t � 2.0 mm, CFE decreases conspicuously. But the
ing as compression proceed. However, the SO1D is increase rate does not decrease under parameter tmid when
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 15

Figure 14. The deformation modes of structures with various wall thickness. From a longitudinal perspective, tint, tmid, tout and trib are varying respectively, while
the other three parameters remain at 1 mm. Horizontally, the wall thickness varies from 0.5 to 2.0 mm. The variable object and levels form a matrix of 4 times 7
finally.

Figure 15. The effect of wall thickness on three crashworthiness criteria: (a) IPCF; (b) SEA; (c) CFE.

t � 1.0 mm, and the highest CFE occurs at the level of

2.0 mm of tmid.

5.2. Effect of size of mid wall (Dmid) on crashworthiness

of SO1D
To explore the effect of geometric parameter on the energy
absorption, the size of middle octagon thin-wall (Dmid) is
designed as 5 levels and it is shown in Table 7. The size is
defined according to the diameter of the circumscribed cir­
cle of octagon which can be learned from Figure 13(b).
Figure 16 shows the change trends of IPCF, SEA and CFE
about Dmid ranging from 25 to 35 mm. Figure 16. The sensitivity analysis of Dmid for IPCF, SEA and CFE.
Consulting the figure, the increase of IPCF is linear with
the increase of Dmid, substantially. When Dmid ¼ 25 mm, the
value of IPCF is 59.02 kN, which is 7.23% lower than the other two criteria are examined, it was found that SEA and
Dmid with 35 mm. This found indicates that Dmid has a strik­ CFE represents same trend as the increasing of Dmid. When
ing effect on IPCF of nested SO1D and the smaller Dmid is, 25 mm � Dmid � 30 mm, the values of SEA and CFE report
the smaller the IPCF of the structure. In addition, when upward tendency; when 30 mm � Dmid � 35 mm, they start
16 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

to fall. The highest values of SEA and CFE appear when CRediT authorship statement
Dmid ¼ 30 mm, which are 8.13 and 22.61 kJ/kg, respectively,
Honghao Zhang: Conceptualization, Methodology, Writing
from which we can infer that 35 mm is the optimal geomet­
– original draft. Dongtao Yu: Data curation, Formal ana­
ric parameter of Dmid for crashworthiness criteria.
lysis, Investigation, Writing – original draft. Danqi Wang:
Therefore, the reasonable geometric dimension needs to be
Validation. Tao Li: Methodology. Yong Peng: Visualization.
considered when the structure nested twice SO1D is applied
Lin Hou: Conceptualization, Methodology. Zhixiang Li:
as energy absorption structure.
Data curation, Visualization.

6. Conclusion
Disclosure statement
In this study, a series of nested corrugation square energy-
No potential conflict of interest was reported by the author(s).
absorbing tubes (NCSTs) with circumferentially cosine pro­
file and different inner structures are proposed to investigate
the crashworthiness of nested structures. The inner struc­ Funding
tures of nested tubes are consisted of four types, including The work is supported by the National Natural Science Foundation of
circle, octagon, square and corrugation square, which is con­ China [grant number 52105523, 52075553]; the Natural Science
nected with outer wall by ribs. The numerical simulation is Foundation of Shandong [grant number ZR2021QE249]; the Hunan
conducted to study energy absorption characteristic by using Science Foundation for Distinguished Young Scholars of China [grant
FE software and the results are validated by experiment data number 2021JJ10059]; the Postgraduate Scientific Research Innovation
Project of Hunan Province [grant numbers QL20220042]; and the First
and theoretical prediction. In addition, TOPSIS method is Batch of 2021 MOE of PRC Industry-University Collaborative
used to determine the ideal tubes. Finally, the parameter Education Program [grant numbers 202107ZCJG05, Kingfar-CES
sensitivity analysis is conducted for the ideal structure. The ‘Human Factors and Ergonomics’ Program].
primary conclusions are as follow:
ORCID
1. Compared with non-nested corrugation square tubes,
the SEA values of nested tubes can be improved by up Lin Hou http://orcid.org/0000-0002-5577-0649
to 267.9%. Therefore, NCSTs can ameliorate effectively
energy absorption characteristic and has great potential
Data availability statement
to apply as absorption.
2. Due to different cross-section, the IPCF and SEA values No data was used for the research described in the article.
of thin-walled tubes under axial load changed from
17.95 to 65.92 kN and 6.08 to 22.37 kJ/kg, respectively.
References
Therefore, it can be seen that the shape and quantity of
inner structure are considerable for improving the 0[1] Z. Li, W. Ma, H. Zhu, G. Deng, L. Hou, P. Xu, and S. Yao,
crashworthiness of nested structures. Energy absorption prediction and optimization of corrugation-
reinforced multicell square tubes based on machine learning,
3. By using TOPSIS method, the tube SO1D is determined
Mech. Adv. Mater. Struct., vol. 29, no. 26, pp. 5511–5529, 2022.
as the ideal structure, this is because that SO1D is DOI: 10.1080/15376494.2021.1958032.
ranked first in 40% of result according to the weights 0[2] W. Guan, and G. Gao, Crashworthiness analysis of shrink cir­
under ten cases. cular tube energy absorbers with anti-climbers under multiple
4. The maximum errors occurred in SS3D between theor­ loading cases, Mech. Adv. Mater. Struct., vol. 30, no. 7, pp.
1453–1469, 2023. DOI: 10.1080/15376494.2022.2033892.
etical prediction and numerical simulation, which
0[3] Y. Peng, T. Li, L. Hou, K. Wang, G. Xie, and H. Zhang, Bio-
reaches 10.85%, and other error are all less than 10%. inspired cellular thin-walled structures under impact loading:
This result indicates that theoretical model has great modeling and multi-stage cooperative optimization algorithm,
accuracy. Mech. Adv. Mater. Struct., pp. 1–18, 2023. DOI: 10.1080/
5. The effects of thickness parameters tint, tmid, tout and trib 15376494.2023.2205848.
0[4] D. Wang, P. Xu, C. Yang, X. Xiao, and Q. Che, Crashing per­
on deformation mode and three crashworthiness criteria
formance and multi-objective optimization of honeycomb-filled
IPCF, SEA and CFE are extremely obvious. And when thin-walled energy absorber with axisymmetric thickness, Mech.
Dmid¼35mm, the SEA and CFE report optimal result. Adv. Mater. Struct., vol. 30, no. 11, pp. 2203–2220, 2023. DOI:
10.1080/15376494.2022.2053765.
The further research will be conducted from the follow­ 0[5] C. Gong, Y. Hu, and Z. Bai, Crashworthiness analysis and opti­
mization of lotus-inspired bionic multi-cell circular tubes,
ing two aspects: (1) determining the effect of the size of the Mech. Adv. Mater. Struct., vol. 30, pp. 4996–5014, 2022. DOI:
internal octagonal structure Dint on the SO1D endergonic 10.1080/15376494.2022.2111622.
performance and observing whether the interaction between 0[6] H. Liang, W. Hao, H. Sun, Y. Pu, Y. Zhao, and F. Ma, On
Dint and Dmid can significantly enhance the endergonic abil­ design of novel bionic bamboo tubes for multiple compression
ity, and (2) optimizing for SO1D combining the multi- load cases, Int. J. Mech. Sci., vol. 218, pp. 107067, 2022. DOI:
10.1016/j.ijmecsci.2022.107067.
objective optimization algorithm and MCDM determine the 0[7] Y. Peng, T. Li, C. Bao, J. Zhang, G. Xie, and H. Zhang,
appropriate geometric parameters and verify the effective­ Performance analysis and multi-objective optimization of bionic
ness of the framework. dendritic furcal energy-absorbing structures for trains, Int. J.
 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES 17

Mech. Sci., vol. 246, pp. 108145, 2023. DOI: 10.1016/j.ijmecsci. Non-Linear Mech., vol. 92, pp. 41–58, 2017. DOI: 10.1016/j.
2023.108145. ijnonlinmec.2017.03.001.
0[8] X. Chen, W. Wang, F. Jin, and H. Fan, Braided-textile rein­ [25] M. Yang, B. Han, P. Su, Q. Zhang, Q. Zhang, Z. Zhao, C. Ni,
forced thin-walled conical tubular structures: designing, manu­ and T.J. Lu, Crashworthiness of hierarchical truncated conical
facturing and testing, Thin-Walled Struct., vol. 174, pp. 109121, shells with corrugated cores, Int. J. Mech. Sci., vol. 193, pp.
2022. DOI: 10.1016/j.tws.2022.109121. 106171, 2021. DOI: 10.1016/j.ijmecsci.2020.106171.
0[9] Q. Li, L. Wu, L. Hu, X. Miao, X. Liu, and T. Zou, A sinusoidal [26] N.S. Ha, G. Lu, and X. Xiang, High energy absorption efficiency
beam lattice structure with negative Poisson’s ratio property, of thin-walled conical corrugation tubes mimicking coconut
Aerosp. Sci. Technol., vol. 133, pp. 108103, 2023. DOI: 10.1016/ tree configuration, Int. J. Mech. Sci., vol. 148, pp. 409–421,
j.ast.2022.108103. 2018. DOI: 10.1016/j.ijmecsci.2018.08.041.
[10] Y. Li, and Z. You, External inversion of thin-walled corrugated [27] W. Ma, S. Xie, Z. Li, Z. Feng, and K. Jing, Crushing behaviors
tubes, Int. J. Mech. Sci., vol. 144, pp. 54–66, 2018. DOI: 10. of horse-hoof-wall inspired corrugated tubes under multiple
1016/j.ijmecsci.2018.05.044. loading conditions, Mech. Adv. Mater. Struct., vol. 29, no. 22,
[11] Q. Gao, G. Yin, C.C.L. Wang, and W.-H. Liao, Crushing and pp. 3263–3280, 2022. DOI: 10.1080/15376494.2021.1892245.
theoretical analysis of multi-cell tube filled with auxetic struc­ [28] A. Sadighi, A. Eyvazian, M. Asgari, and A.M. Hamouda, A
ture under axial impact loading, Mech. Adv. Mater. Struct., vol. novel axially half corrugated thin-walled tube for energy
30, pp. 5034–5046, 2022. DOI: 10.1080/15376494.2022.2111730. absorption under Axial loading, Thin-Walled Struct., vol. 145,
[12] X. Deng, S. Qin, and J. Huang, Out-of-plane impact analysis pp. 106418, 2019. DOI: 10.1016/j.tws.2019.106418.
for a bioinspired sinusoidal honeycomb, Mech. Adv. Mater. [29] A. Sadighi, H. Salaripoor, and M. Asgari, Comprehensive study
Struct., vol. 29, no. 28, pp. 7259–7276, 2022. DOI: 10.1080/ on the crashworthiness of a new developed axially-half corru­
15376494.2021.1995547. gated aluminum tubes, Int. J. Crashworthiness., vol. 27, no. 3,
[13] Z. Ahmad, and D.P. Thambiratnam, Crushing response of pp. 633–650, 2022. DOI: 10.1080/13588265.2020.1836761.
foam-filled conical tubes under quasi-static axial loading, [30] Y. Lin, J. Min, Y. Li, and J. Lin, A thin-walled structure with
Mater. Des., vol. 30, no. 7, pp. 2393–2403, 2009. DOI: 10.1016/ tailored properties for axial crushing, Int. J. Mech. Sci.,
j.matdes.2008.10.017. vol. 157–158, pp. 119–135, 2019. DOI: 10.1016/j.ijmecsci.2019.
[14] M. Emadi, H. Beheshti, and M. Heidari-Rarani, Thickness effect 04.015.
study on the crushing characteristics of aluminum and compos­ [31] X. Deng, W. Liu, and L. Jin, On the crashworthiness analysis
ite tubes: numerical analysis and multi-objective optimization, and design of a lateral corrugated tube with a sinusoidal cross-
Mech. Adv. Mater. Struct., vol. 28, no. 24, pp. 2585–2594, 2021. section, Int. J. Mech. Sci., vol. 141, pp. 330–340, 2018. DOI: 10.
DOI: 10.1080/15376494.2020.1747667.
1016/j.ijmecsci.2018.03.001.
[15] Z. Li, W. Ma, L. Hou, P. Xu, and S. Yao, Crashworthiness ana­
[32] Z. Liu, W. Hao, J. Xie, J. Lu, R. Huang, and Z. Wang, Axial-
lysis of corrugations reinforced multi-cell square tubes, Thin-
impact buckling modes and energy absorption properties of
Walled Struct., vol. 150, pp. 106708, 2020. DOI: 10.1016/j.tws.
thin-walled corrugated tubes with sinusoidal patterns, Thin-
2020.106708.
Walled Struct., vol. 94, pp. 410–423, 2015. DOI: 10.1016/j.tws.
[16] X. Deng, S. Qin, and J. Huang, Multiobjective optimization of
2015.05.002.
axially varying thickness lateral corrugated tubes for energy
[33] Z. Li, W. Ma, P. Xu, and S. Yao, Crushing behavior of circum­
absorption, Mech. Adv. Mater. Struct., vol. 29, no. 25, pp.
ferentially corrugated square tube with different cross inner
4259–4272, 2022. DOI: 10.1080/15376494.2021.1924901.
ribs, Thin-Walled Struct., vol. 144, pp. 106370, 2019. DOI: 10.
[17] A. Sadighi, M. Mahbod, and M. Asgari, Bi-tubular corrugated
1016/j.tws.2019.106370.
composite conical–cylindrical tube for energy absorption in
[34] J. Zhou, C. Dong, Z. Wang, B. Chen, R. Qin, and X. Niu,
axial and oblique loading: analysis and optimization, J.
Approaching ideal energy absorption through the multicellular
Compos. Mater., vol. 54, no. 18, pp. 2399–2432, 2020. DOI: 10.
1177/0021998319897407. structure with gradient material distribution, Int. J. Mech. Sci.,
[18] M.B. Azimi, M. Asgari, and H. Salaripoor, A new homo-polyg­ vol. 225, pp. 107355, 2022. DOI: 10.1016/j.ijmecsci.2022.107355.
onal multi-cell structures under axial and oblique impacts; con­ [35] S. Wang, M. Zhang, W. Pei, F. Yu, and Y. Jiang, Energy-
sidering the effect of cell growth in crashworthiness, Int. J. absorbing mechanism and crashworthiness performance of
Crashworthiness., vol. 25, no. 6, pp. 628–647, 2020. DOI: 10. thin-walled tubes diagonally filled with rib-reinforced foam
1080/13588265.2019.1628461. blocks under axial crushing, Compos. Struct., vol. 299, pp.
[19] C. Shu, S. Hou, Y. Zhang, and Y. Luo, Crashworthiness analysis 116149, 2022. DOI: 10.1016/j.compstruct.2022.116149.
and optimization of different configurations multi-layered cor­ [36] W. Ma, Z. Li, and S. Xie, Crashworthiness analysis of thin-
rugated sandwich panels under crush loading, J Sandwich walled bio-inspired multi-cell corrugated tubes under quasi-
Struct Mater., vol. 23, no. 7, pp. 2901–2922, 2021. DOI: 10. static axial loading, Eng. Struct., vol. 204, pp. 110069, 2020.
1177/1099636220909755. DOI: 10.1016/j.engstruct.2019.110069.
[20] M. Asgari, A. Babaee, and M. Jamshidi, Multi-objective opti­ [37] A. Alavi Nia, and M. Parsapour, Comparative analysis of energy
mization of tapered tubes for crashworthiness by surrogate absorption capacity of simple and multi-cell thin-walled tubes
methodologies, Steel Compos. Struct., vol. 27, pp. 427–438, with triangular, square, hexagonal and octagonal sections, Thin-
2018. DOI: 10.12989/SCS.2018.27.4.427. Walled Struct., vol. 74, pp. 155–165, 2014. DOI: 10.1016/j.tws.
[21] A.A. Singace, and H. El-Sobky, Behaviour of axially crushed 2013.10.005.
corrugated tubes, Int. J. Mech. Sci., vol. 39, no. 3, pp. 249–268, [38] J. Zhang, D. Zheng, B. Lu, and T. Zhang, Energy absorption
1997. DOI: 10.1016/S0020-7403(96)00022-7. performance of hybrid cross section tubes under oblique loads,
[22] H. Zhang, D. Yu, T. Li, L. Wang, Z. Huang, and Y. Zhang, Thin-Walled Struct., vol. 159, pp. 107133, 2021. DOI: 10.1016/j.
Multi-Objective Optimization of Square Corrugation Multilayer tws.2020.107133.
Nested Structures, Appl. Sci., vol. 13, no. 17, pp. 9750, 2023. [39] J. Fu, Q. Liu, K. Liufu, Y. Deng, J. Fang, and Q. Li, Design of
DOI: 10.3390/app13179750. bionic-bamboo thin-walled structures for energy absorption,
[23] Z. Li, S. Yao, W. Ma, P. Xu, and Q. Che, Energy-absorption Thin-Walled Struct., vol. 135, pp. 400–413, 2019. DOI: 10.1016/
characteristics of a circumferentially corrugated square tube j.tws.2018.10.003.
with a cosine profile, Thin-Walled Struct., vol. 135, pp. 385– [40] L. Qin, S. Yang, H. Li, J. Dai, G. Wang, Q. Ling, and Z. Chen,
399, 2019. DOI: 10.1016/j.tws.2018.11.028. Crashworthiness design of bionic-shell thin-walled tube under
[24] S. Wu, G. Sun, X. Wu, G. Li, and Q. Li, Crashworthiness ana­ axial impact, J Mech Sci Technol., vol. 37, pp. 3427–3436, 2023.
lysis and optimization of Fourier varying section tubes, Int. J. DOI: 10.1007/s12206-023-0608-1.
18 MECHANICS OF ADVANCED MATERIALS AND STRUCTURES

[41] X. Deng, and W. Liu, Experimental and numerical investigation [54] Y.-Q. Shi, R.-X. Qin, and B.-Z. Chen, Crashworthiness design
of a novel sandwich sinusoidal lateral corrugated tubular struc­ and analysis for novel multi-corner square columns under axial
ture under axial compression, Int. J. Mech. Sci., vol. 151, pp. loading, Mech. Adv. Mater. Struct., vol. 29, no. 27, pp. 5843–
274–287, 2019. DOI: 10.1016/j.ijmecsci.2018.11.010. 5859, 2022. DOI: 10.1080/15376494.2021.1967529.
[42] H. Chen, Y. Zhang, J. Lin, F. Zhang, Y. Wang, and X. Yan, [55] H. Nikkhah, A. Baroutaji, and A.G. Olabi, Crashworthiness
Crushing responses and optimization of novel sandwich col­ design and optimisation of windowed tubes under axial impact
umns, Compos. Struct., vol. 263, pp. 113682, 2021. DOI: 10. loading, Thin-Walled Struct., vol. 142, pp. 132–148, 2019. DOI:
1016/j.compstruct.2021.113682. 10.1016/j.tws.2019.04.052.
[43] T. Tran, D. Le, and A. Baroutaji, Theoretical and numerical [56] _ Albak, Crashworthiness design for multi-cell circumferen­
E.I.
crush analysis of multi-stage nested aluminium alloy tubular tially corrugated thin-walled tubes with sub-sections under mul­
structures under axial impact loading, Eng. Struct., vol. 182, pp. tiple loading conditions, Thin-Walled Struct., vol. 164, pp.
39–50, 2019. DOI: 10.1016/j.engstruct.2018.12.072. 107886, 2021. DOI: 10.1016/j.tws.2021.107886.
[44] _ Albak, Crashworthiness design and optimization of nested
E.I. [57] S. Pirmohammad, and S. Esmaeili Marzdashti, Crashworthiness
structures with a circumferentially corrugated circular outer optimization of combined straight-tapered tubes using genetic
wall and inner ribs, Thin-Walled Struct., vol. 167, pp. 108219, algorithm and neural networks, Thin-Walled Struct., vol. 127,
2021. DOI: 10.1016/j.tws.2021.108219. pp. 318–332, 2018. DOI: 10.1016/j.tws.2018.01.022.
[45] S. Wu, G. Li, G. Sun, X. Wu, and Q. Li, Crashworthiness ana­ [58] B. Cheng, H. Fu, T. Li, H. Zhang, J. Huang, Y. Peng, H. Chen,
lysis and optimization of sinusoidal corrugation tube, Thin- and C. Fan, Evolutionary computation-based multitask learning
Walled Struct., vol. 105, pp. 121–134, 2016. DOI: 10.1016/j.tws. network for railway passenger comfort evaluation from EEG
2016.03.029.
signals, Appl. Soft Comput., vol. 136, pp. 110079, 2023. DOI:
[46] N. Jones, Some Phenomena in the structural crashworthiness
10.1016/j.asoc.2023.110079.
field, Int. J. Crashworthiness., vol. 4, no. 4, pp. 335–350, 1999.
[59] T. Chen, Y. Zhang, J. Lin, and Y. Lu, Theoretical analysis and
DOI: 10.1533/cras.1999.0110.
crashworthiness optimization of hybrid multi-cell structures,
[47] N. Jones, Several phenomena in structural impact and structural
Thin-Walled Struct., vol. 142, pp. 116–131, 2019. DOI: 10.1016/
crashworthiness, Eur. J. Mech. ASolids., vol. 22, no. 5, pp. 693–
j.tws.2019.05.002.
707, 2003. DOI: 10.1016/S0997-7538(03)00077-9.
[60] Y. Zhang, X. Xu, J. Wang, T. Chen, and C.H. Wang, Crushing
[48] X. Zhang, and H. Zhang, Axial crushing of circular multi-cell
analysis for novel bio-inspired hierarchical circular structures
columns, Int. J. Impact Eng., vol. 65, pp. 110–125, 2014. DOI:
10.1016/j.ijimpeng.2013.12.002. subjected to axial load, Int. J. Mech. Sci., vol. 140, pp. 407–431,
[49] T. Wierzbicki, and W. Abramowicz, On the Crushing 2018. DOI: 10.1016/j.ijmecsci.2018.03.015.
Mechanics of Thin-Walled Structures, J. Appl. Mech., vol. 50, [61] X. Zhang, and H. Zhang, Energy absorption of multi-cell stub
no. 4a, pp. 727–734, 1983. DOI: 10.1115/1.3167137. columns under axial compression, Thin-Walled Struct., vol. 68,
[50] C.-T. Chen, Extensions of the TOPSIS for group decision-mak­ pp. 156–163, 2013. DOI: 10.1016/j.tws.2013.03.014.
ing under fuzzy environment, Fuzzy Sets Syst., vol. 114, no. 1, [62] X. Zhang, and H. Zhang, Numerical and theoretical studies
pp. 1–9, 2000. DOI: 10.1016/S0165-0114(97)00377-1. on energy absorption of three-panel angle elements, Int. J.
[51] Z. Huang, H. Zhang, D. Wang, H. Yu, L. Wang, D. Yu, and Y. Impact Eng., vol. 46, pp. 23–40, 2012. DOI: 10.1016/j.ijimpeng.
Peng, Preference-based multi-attribute decision-making method 2012.02.002.
with spherical-Z fuzzy sets for green product design, Eng. Appl. [63] T. Tran, S. Hou, X. Han, W. Tan, and N. Nguyen, Theoretical
Artif. Intell., vol. 126, pp. 106767, 2023. DOI: 10.1016/j.engap­ prediction and crashworthiness optimization of multi-cell tri­
pai.2023.106767. angular tubes, Thin-Walled Struct., vol. 82, pp. 183–195, 2014.
[52] J. Li, Y. Zhang, Y. Kang, and F. Zhang, Characterization of DOI: 10.1016/j.tws.2014.03.019.
energy absorption for side hierarchical structures under axial [64] T. Tran, S. Hou, X. Han, N. Nguyen, and M. Chau, Theoretical
and oblique loading conditions, Thin-Walled Struct., vol. 165, prediction and crashworthiness optimization of multi-cell
pp. 107999, 2021. DOI: 10.1016/j.tws.2021.107999. square tubes under oblique impact loading, Int. J. Mech. Sci.,
[53] H. Zhu, S. Yao, Z. Li, J. Liu, P. Xu, and M. Liu, vol. 89, pp. 177–193, 2014. DOI: 10.1016/j.ijmecsci.2014.08.027.
Crashworthiness analysis of multilayered hexagonal tubes under [65] X. Zhang, and H. Zhang, The crush resistance of four-panel
axial and oblique loads, Mech. Adv. Mater. Struct., vol. 30, no. angle elements, Int. J. Impact Eng., vol. 78, pp. 81–97, 2015.
17, pp. 3608–3629, 2023. DOI: 10.1080/15376494.2022.2079031. DOI: 10.1016/j.ijimpeng.2014.12.004.