International Journal of Engineering Science 130 (2018) 12–32

Contents lists available at ScienceDirect

International Journal of Engineering Science

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

Approaching perfect energy absorption through structural

hierarchy
Hualin Fan a,∗, Yonghao Luo b, Fan Yang c, Weiwei Li a
a
Research Center of Lightweight Structures and Intelligent Manufacturing, State Key Laboratory of Mechanics and Control of Mechanical
Structures, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
b
Research Center of Lightweight Structures and Intelligent Manufacturing, State Key Laboratory for Disaster Prevention & Mitigation of
Explosion & Impact, Army Engineering University of PLA, Nanjing 210007, China
c
School of Aerospace Engineering and Applied Mechanics, Tongji University, Shanghai 200092, China

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

Article history: Energy absorbing efficiency of thin-walled tubular structures is restricted by the long-wave
Received 5 December 2017 folding elements, which makes the mean crushing force (MCF) of the thin-walled tubular
Revised 6 May 2018
structure usually much smaller than its yielding strength. Hierarchical lattice topology in-
Accepted 20 May 2018
creases the energy absorbing ability of tubular structure notably without increasing the
Available online 31 May 2018
weight. In an effort to reveal this advantage, hierarchical rectangular tubular structures are
Keywords: proposed in this paper. These proposed structures have multi-cellular structure and lattice
Hierarchical structure sandwich cellular walls. During crushing of these structures, three typical folding styles,
Perfect energy absorber i.e., macro-cell folding, micro-cell folding and hybrid folding were observed in both ex-
Folding style periments and finite element (FE) simulations. Micro-cell folding has relative short wave
Mean crushing force length depending on the dimension of the microstructure cells. Macro-cell folding has rela-
Theoretical prediction tive long wave length determined by the dimension of the overall tubular structure. Micro-
cell folding notably increases the mean crushing force (MCF) while macro-cell folding de-
creases the MCF. However, the maximum MCF is associated with the hybrid folding style
which is a transition from micro-cell folding to macro-cell folding. When the hierarchy
level of the tubular structure is gradually increased, the MCF approaches the full-plastic
strength of the matrix. Through high order hierarchical topology, extraordinary energy ab-
sorption can be achieved.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction

Thin-walled structures are most commonly used as crash-resistant energy absorbing devices. Abramowicz and Jones
(1984, 1986), Abramowicz and Wierzbicki (1989), Wierzbicki and Abramowicz (1983) and others proposed the crushing
model of circular and square tubes. Hong et al. (2013) and Sun and Fan (2017) observed a new folding element, named as
inward-contracted folding element, during the compression of triangle tubes, which is different from the two conventional
plastic collapse elements, extensional and in-extensional folding elements. Wang et al. (2015) studied the collapse mecha-
nism and energy absorbing ability of triangle tubes in lateral compression. These researches provide considerable insights
into the energy-absorbing mechanisms of tubular structures.

∗
Corresponding author.
E-mail address: fhl15@nuaa.edu.cn (H. Fan).

https://doi.org/10.1016/j.ijengsci.2018.05.005
0020-7225/© 2018 Elsevier Ltd. All rights reserved.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 13

A limitation of the thin-walled tubular structure is that the MCF is usually much smaller than the peak force (PF) in-
duced by long wave length of the folding. To increase the weight efficiency of tubular structures as energy absorbers, efforts
were dedicated to multi-cell tubular structures. Birman (2014) improves energy absorption of cylinder through functionally
graded structure. Chen and Wierzbicki (2001) studied single-cell, multi-cell and foam-filled thin-walled structures in energy
absorption. It was found that the interaction effects between the foam core and the column wall contribute to the total
crushing resistance by the amounts equal to 140% and 180% of the direct foam resistance for double cell and triple cell,
respectively. Hong et al. (2014) designed multi-cell tubes with triangular and Kagome lattices to realize the progressive col-
lapse mode and explore the folding mechanism of thin-walled multi-cell tubes. Zhang and Zhang (2014) studied the energy
dissipation mechanisms of circular multi-cell columns and proposed theoretical models to predict the mean crushing force.
Tran, Hou, Han, Tan, and Nguyen (2014) conducted crashworthiness optimization of multi-cell triangular tubes. Sun, Lai, and
Fan (2016a) decomposed the multi-cell trianglular tubes into 3-panel angle elements, 4-panel angel elements and 6-panel
angel elements, and derived the equations for the energy dissipation. Qiu et al. (2016) analyzed the mean crushing force for
four different hexagonal tubes with multiple cells based on the simplified super folding element (SSFE) theory.
Some researchers investigated energy absorptions of hybrid structures (An & Fan, 2016). Deshpande and Fleck (2003) in-
vestigated energy absorption of an egg-box material. Xu, Qiao, and Chen (2014) mitigated impact/blast energy via a novel
nanofluidic energy capture mechanism. Chen, Qiao, Shougen Zhao, Zhen, and Liu (2016) proposed a novel self-locked en-
ergy absorbing system to consume more impact energies. Zhang, Heyne, and To (2015) enhanced energy dissipation through
biomimetic staggered composites. Foam-filled tubes (Wang, Fan, Chen, & Liu, 2016; Yang, Meguid, & Hamouda, 2017; Zheng,
Wu, Sun, Li, & Li, 2014; Zheng et al., 2014) usually have better energy absorption than the linear summation of the compo-
nents. However, when the cells are full-filled, the weight efficiency is decreased.
Hierarchy can make thin-walled structures even more weight-efficient (Fan et al., 2008; Sun, Zheng, Fan, & Fang, 2017).
Lakes (1993) constructed hierarchical paper honeycomb whose plastic strength was 3.8 times higher. Fan et al. (2008) de-
scribed the mechanism of hierarchical structure in improving the stiffness and strength, and proved that hierarchy can
make thin-walled structures even more weight-efficient in energy absorption. Fan et al. (2012, 2013) and Zheng, Zhao, and
Fan (2012) made hierarchical honeycombs using ductile woven textile composites. Their specific energy absorption (SEA)
is even greater than those of metallic lattice trusses and honeycombs. Recently, progresses in manufacturing technology
enabled scientists fabricate nano-lattice structures (He, Wang, Zhu, Wu, & Park, 2017; Meza, Das, & Greer, 2014; 2015;
Schaedler et al., 2011; Zheng et al., 2014). Some of these lattices have excellent energy absorption profited from hierar-
chical lattice topology. But most of them are not ideal in energy absorption due to the brittleness of the matrix and the
three-dimensional lattice topology.
Two-dimensional sandwich-walled triangular lattice tubular structures were designed and tested by Sun et al. (2016a,
2016b) to explore the energy-absorbing mechanism. Based on the tests, three mechanisms contribute to the enhanced crush-
ing resistance of the hierarchical structure, i.e., hierarchical fold, shorter wave length and greater plastic bending moment
of sandwich wall. However, the improvement will be trade off by the transition of folding mode from progressive folding to
global bending. The MCF is still far below the full-plastic strength of the structure.
In order to further enhance the energy absorbing ability of tubular structures, higher order hierarchical lattice tubular
structures are proposed in this paper. Energy absorbing mechanisms relating to hierarchy are investigated experimentally,
numerically and theoretically.

2. Hierarchical lattice structures

2.1. Perfect energy absorbing structure

Here a structure can be considered as a perfect energy absorber if its mean crushing stress (MCS), σ m , defined by the
ratio of the MCF to the area of the structure, satisfies
Pm
σm = = ρ ∗ σ p or Pm =Pp = ρ ∗ σ p A, (1)
A
√
where, σ p is the full-plastic stress, σ p = (σy + σu )/2 or σ p = σy σu . σ y is the yield stress. σ u is the ultimate stress. Pp is
the full-plastic force of the structure. Pm is the MCF. A is the cross-section area of the tubular structure. s is the cross section
area of solid walls. ρ ∗ =s/A is the relative density. According to Eq. (1), the mass utilization efficiency is 100% for the perfect
energy absorber.
Aluminium foams are always deemed as excellent energy absorbing materials, with (Gibson & Ashby, 1997)
σm = 0.3ρ ∗1.5 σy , (2)
According to experiments of Kooistra, Deshpande, and Wadley (2004), MCS of aluminium tetrahedral lattice truss struc-
tures is given by Jiang, Sun, Zhang, and Fan (2017) as
σm =1.224ρ ∗1.5 σ p . (3)
Meza et al. (2015) gave MCS of polymer second-order hierarchical nano-lattice as
σm = 0.316ρ ∗1.36 σy . (4)
14 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 1. Definition of perfect energy absorbing structure.

Fig. 2. Typical lattice structures transferring from single-cell lattice, multi-cell lattice, hierarchical lattice to high-order hierarchical lattice structure.

Wierzbicki (1983) gave MCS of metal honeycomb as

σm = 3.03(ρ ∗ )5/3 σ p . (5)
Hong et al. (2013) suggested MCS of triangular tubes as
σm = 2.45(ρ ∗ )5/3 σ p . (6)
As shown in Figs. 1 and 3D hierarchical nano-lattices have high specific stiffness and strength and ultra-light density, but
their complex micro-structure compromises their energy absorbing abilities compared with the conventional 3D lattices. On
the other hand, 2D lattice structures always have better weight-efficiency in energy absorption than 3D lattice structures.
But they still cannot achieve the perfect absorption ability.
Based on the folding mechanism of 2D tubular and lattice structures, planar hierarchy is a potential applicant for achiev-
ing perfect energy absorption.

2.2. Hierarchical lattice tubular structures

For thin-walled tubular structures, lower post-failure strength limits their energy absorption. Optimized topology can
improve the post-failure strength. For this purpose, a multi-cell lattice tube (MLT) is proposed that is divided into micro-
cells with thinner and shorter walls as shown in Fig. 2. Each side of MLT has N segments and there are N2 micro-cells.
Another weight-efficient topology is hierarchical lattice tube (HLT), whose sidewall is made up of sandwich panels with
square lattice truss cores. The tubular structure has a hierarchical structure with self-similarity. Each side has N segments
and there are 4(N-1) micro-cells in each tube.
Based on the topology of multi-cell tube and hierarchy, super hierarchical lattice tube (SHLT), super multi-cell lattice tube
(SMLT) and hierarchical multi-cell lattice tube (HMLT) are designed, as shown in Fig. 2. SHLT is developed from MLT02 by
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 15

Table 1
Dimensions of thin-walled square lattice truss tubes.

Tube Side wall Cell Side length, L Cell dimension, B Wall thickness, t Tube length, h Cross-section
segment number, number (mm) (mm) (mm) (mm) area of wall, S
N (mm2 )

ST 1 1 120 120.0 3.0 100 1440

MLT02 2 4 120 60.0 2.07 100 1440
MLT03 3 8 120 40.0 1.52 100 1440
MLT04 4 16 120 30 1.2 100 1440
HLT04 4 12 120 30.0 1.343 100 1440
HLT05 5 16 120 24.0 1.256 100 1440
HLT06 6 20 120 20.0 1.20 100 1440
HLT07 7 24 120 17.14 1.17 100 1440
HLT08 8 28 120 15 1.145 100 1440
HLT09 9 32 120 13.33 1.13 100 1440
HLT10 10 36 120 12.0 1.11 100 1440
HLT12 12 44 120 10.0 1.09 100 1440
HLT20 20 76 120 6.0 1.05 100 1440
SHLT06 6 29 120 20 0.9 100 1440
SHLT08 8 41 120 15 0.834 100 1440
SHLT10 10 53 120 12 0.78 100 1440
SHLT12 12 65 120 10 0.76 100 1440
SHLT16 16 89 120 7.5 0.73 100 1440
SHLT20 20 113 120 6 0.72 100 1440
SMLT10 10 64 120 6 0.65 100 1440
SMLT16 16 112 120 7.5 0.585 100 1440
SMLT20 20 154 120 6 0.569 100 1440
HMLT13 13 105 120 9.23 0.52 100 1440
HMLT17 17 145 120 7.06 0.4857 100 1440
HMLT21 21 185 120 5.71 0.4667 100 1440

introducing truss-core sandwich wall into MLT for N = 2. SMLT is developed from MLT03 by introducing truss-core sandwich
wall into MLT for N = 3. HMLT is developed from MLT04 by introducing truss-core sandwich wall into MLT for N = 4. SHLT,
SMLT and HMLT have higher-order hierarchy.
As the hierarchy increases, the quantity of micro-cells is tremendously increased meanwhile the dimension of the micro-
cells is decreased, as shown in Fig. 2. If the folding is restrained within these micro-cells, the wavelength of each lobe will
be shortened and the quantity of the folding layers and micro-lobes will be enormously increased. Much more energies
could be absorbed by these enormous micro-lobes. If the folding takes place in the scale of macro-cell, the sandwich wall
has greater full-plastic bending resistance than the thin solid wall and the hierarchy still can improve the energy absorption
to a certain extent.

3. Axial compression behaviors

3.1. Axial compression experiments

Four tubular samples were designed, fabricated and compressed, as shown in Fig. 2. HLT04 has 12 micro-cells in the walls
for N = 4; HLT08 has 28 micro-cells for N = 8; and HLT10 has 36 micro-cells for N = 10, while SHLT10 has 49 micro-cells.
Corresponding square single-cell tube (ST) (N = 1) and high-order hierarchical lattice tube (SHLT10) (N = 10) with identical
mass and cross-sectional area were also designed, as listed in Table 1. Each tube is 120 mm wide and 100 mm long.
All specimens were wire cut out from solid steel Q235 rod. The tensile stress-strain curve of Q235 is shown in Fig. 3
according to GBT 228.1–2010. The Young’s modulus E is 210 GPa. The yield stress σ y is 343 MPa. The ultimate stress σ u is
475 MPa. The Poisson’s ratio υ is 0.3. Experiments were carried out on a 1500 kN universal testing machine at loading rate
of 2 mm/min. Deformation curves and crushing modes are displayed in Figs. 3 and 4, respectively.
For STs, the thickness of the wall is 3 mm. The peak forces (PFs) are 439 kN and 422 kN, respectively, as listed in Table 2.
After PF, strain hardening of the steel is not obviously displayed and the load drops dramatically. The MCFs are 156 kN and
154 kN, respectively. As shown in Fig. 4, the ST contains only one macro-cell fold during crushing. Energy absorption stability
factor (EASF) ξ defined by the ratio of the PF, Pmax , to the MCF, Pm , is given as

Pmax
ξ= . (7)
Pm

ξ is an important parameter to evaluate the energy absorption performance of thin-walled structure and it will be 1.0 for
perfect energy absorbers. For ST, EASFs are 2.81 and 2.74, respectively.
16 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 3. (a) Engineering stress-strain curve of steel Q235; compression curves of typical hierarchical lattice structures: (b) HLT and (c) SHLT; and (d) devel-
opment of MCF with N.

Macro-cell folding Micro-cell folding Hybrid folding Micro-cell folding

ST01 HLT04 HLT08 HLT10 SHLT10

Fig. 4. Crushing styles of typical hierarchical lattice structures.

3.3. Hierarchical tubular lattice structures

For HLT04, the yield loads are 341 kN and 331 kN, respectively, as listed in Table 2. Strain hardening increases the load
continuously to 452 kN and 441 kN, respectively. Then the load gradually decreases to a plateau value of 300 kN, with few
fluctuations. Three layers of micro-cell lobes formed in compression, as shown in Fig. 4. The MCFs are 289 kN and 304 kN,
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 17

Table 2
Measured data in axial compression experiments.

Tube Pm (kN) Py (kN) Pmax (kN) Py /Pm Pmax /Pm

ST 156 380 439 2.43 2.81

154 364 422 2.36 2.74
HLT04 289 341 452 1.17 1.56
304 331 441 1.08 1.45
HLT08 389 297 469 0.76 1.2
421 341 611 0.81 1.45
HLT10 402 292 539 0.72 1.34
383 272 511 0.71 1.33
SHLT10 414 350 581 0.84 1.4
404 318 520 0.78 1.28

respectively, which are close to the yield force. EASFs of HLT04 are 1.56 and 1.45, respectively, as listed in Table 2. If EASF is
P
taken as ξ = Pmy , it is very close to 1.0. Here Py denotes the yield force of the tube.
For HLT08, there are five layers of progressive micro-cell lobes during compression. The yield forces are 297 kN and
341 kN, respectively, as list in Table 2. Whereas the PFs are 469 kN and 611 kN, respectively, contributed by the strain hard-
ening effect. The MCFs of HLT08 are 389 kN and 421 kN, respectively, much greater than the yield force. EASFs of HLT08 are
1.2 and 1.45, respectively.
The displacement curves and the progressive collapse modes of HLT10 are shown in Figs. 3 and 4, respectively. The
folding style of HLT10 is different from those of HLT04 and HLT08. Firstly, the tube is crushed and the micro-cell lobes are
formed. Then a shear band appears in one side wall. Shear band reduces the stiffness of the sandwich wall and finally the
wall will be bent to form a macro-cell lobe. There are three side walls crushed at the macro-cell folding style and one
side crushed with layers of micro-cell folds. The yield forces are 292 kN and 272 kN, while the PFs are 539 kN and 511 kN,
respectively. The PFs of HLT10 are greater than those of HLT04 and HLT08. MCFs of HLT10 are 402 kN and 383 kN, close to
the that of HLT08.

3.4. Super hierarchical lattice tubular structure

As shown in Fig. 3, the plastic plateaus of SHLT10 and HLT10 keep on an even keel and MCFs of SHLT10 and HLT10 are
close to each other. The PFs of SHLT10 is 581 kN and 520 kN, respectively, as listed in Table 2. The ratio of the PF to the MCF
is about 1.4 and 1.28. Crushed with layers of micro-cell lobes, SHLT10 shows different folding style from HLT10, as indicated
in Fig. 4. Compared with HLT10, high order topology changes the folding style, turning it from the macro-cell folding mode
to micro-cell folding mode again.

3.5. Hierarchy effects

For ST, the width of each folding lobe is confined by the side length of the tubular structure. The lobe wavelength would
match the side length of the macro-cell, or even be longer. Compared with the macro-cell folding style of ST, HLT04 and
HLT08 have progressive micro-cell folding styles. The folding lobes of HLT are confined within each micro-cell. There would
be N lobes in each side skin as the side is divided into N segments. The width of the micro-cell lobe is just 1/N of the
macro-cell lobe of ST. Correspondingly, the wavelength is much shorter than the macro-lobe, leading to multi-layer lobes
along HLT. Hierarchy brings enormous increase in the number of micro-lobes. As N increases from 1 to 8, MCF of HLT
increases dramatically. Therefore, hierarchy can significantly change the folding style and improve the MCF.
As N increases to 10, the sandwich wall gets thin enough to induce macro-lobe. HLT10 is likely to transform from micro-
cell folding to macro-cell folding. The crushing style will finally turns to macro-cell folding and the MCF of HLT will be
reduced when the cell number N increases to a critical value.
Then high order hierarchy takes effect to keep the crushing in progressive micro-cell folding style and continually in-
crease the MCF.

4. Numerical analyses

4.1. Method

To further investigate the energy absorption mechanisms of HLTs, SHLTs and SMLTs with various geometrical parame-
ters, FE analysis using ABAQUS/Explicit were carried out. The cell wall was modeled by shell element S4R. The tubes were
crushed by two rigid planes. All the friction coefficients in the contact were set to 0.2. Contact properties are defined for
the interaction between the structure and the two rigid planes, and for the self-contact of the structure. It is found that
changing the friction coefficient from 0.15 to 0.25 has little influence to the result.
18 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

700
16
600
14

Computing time(day)
500 12
Force (kN)

400 10

Peak force 8
300
Computing time 6
200
4
100 2

0 0
2.50 2.25 2.00 1.75 1.50 1.25 1.00
element size (mm)

Fig. 5. Cell dimension effect to the simulated force of HMLT21.

To check the mesh independency, the mesh effect of HMLT21, the most complex structure in this research, is investigated.
When the element dimension varies from 2.5 mm, 2.0 mm, 1.5 mm to 1.0 mm, the variation of the simulated peak force is
not great and converges gradually. But when the cell dimension is 1.0 mm, the computing time is abruptly increased to 16
days, as shown in Fig. 5. So in this research dimension of the elements is 1.5 mm.
As shown in Fig. 6, tubes of the same weight with different side segments were investigated here, with the dimensions
listed in Table 1. All of these tubes have the same section area s, as 1440 mm2 . As the number of side segments increases,
the number of cells in the wall gets larger, while the thickness of the wall gets smaller. Quasi-static compression deformation
is simulated assuming that the total kinetic energy is negligible compared with the total internal energy during the period
of loading. According to the history-output data, the ratio of the kinetic energy to internal energy is less than 1.0% at first,
then gradually decreases to 0.01% during the yield period.
Comparisons between finite element modeling (FEM) and experiments are made, and the results are listed in Table 3 to
verify the FE simulations. The largest error of FEM relative to experiments for MCF is −16.7% and the smallest error is 2.3%.

4.2. Crushing forces

As shown in Fig. 6, the compression curve comprises of different stages including elastic deformation, yielding, strain-
hardening, softening, plastic deformation plateau and densification. It is obvious that strain-hardening effect can improve
the PF as it increases the number of the side wall segments.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 19

Fig. 6. Crushing behaviors of typical HLTs and transformation of folding modes.

Table 3
Comparisons between FEM and experiments.

Case Peak force (kN) Mean crushing force (kN)

FEM Experiment Error (%) FEM Experiment Error (%)

HLT04 515 452 13.9 275 289 −4.8

441 16.7 304 −9.5
HLT08 590 469 25.7 398 389 2.3
611 −3.4 421 −5.4
HLT10 560 539 3.8 335 402 −16.7
511 9.5 383 −12.5
SHLT10 544 581 −6.3 432 414 4.3
520 4.6 404 6.9
20 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 7. Crushing behaviors of typical SHLTs and transformation of folding modes.

Crushing force of the deformation plateau increases with N when N ≤ 8. After that increasing N will decrease the crush-
ing force of HLT. EASF decreases as N increases when N ≤ 6, and then increases as N further increases. According to the
simulation, HLT8 has the best energy absorbing ability except when N > 8.
As shown in Fig. 7, SHLT has crushing behavior similar to HLT. As the side wall segments number N increases from 6 to
12, the crushing force of the deformation plateau increases. After that, increasing N will decrease the crushing force. The
MCF of SHLT increases with N before N reaches the critical value of 10 and then start to decrease with further increase of
N. The critical segment number of SMLT is 12.
As shown in Fig. 8, among the SMLTs, SMLT16 has the greatest crushing force of the deformation plateau, which is 1.58
times greater than SHLT16 and 2.34 times greater than HLT16. The MCF of SMLT20 is 1.19 times greater than SHLT20 and
1.56 times greater than HLT20.
As shown in Fig. 9, among the HMLTs, HMLT17 has the greatest crushing force of the deformation plateau, while the
MCFs of HMLT13 and HMLT21 are much smaller. All these phenomena can be explained by the folding styles as will be
illustrated next.

4.3. Folding style

As shown in Fig. 6, when N changes from 1 to 6, layers of micro-cell folding becomes obvious during the compression.
When N = 3, there are two layers of folds, and the number increases to five when N = 6. Simultaneously, the MCF of HLT
changes from 193 kN to 398 kN. Increased folding layers shortens the wavelength and increases the quantity of the folding
elements, which improves the energy absorption. When N = 8 and N = 10, micro-cell folding and macro-cell folding are both
observed. Micro-cell folding increases the MCF while the macro-cell folding decreases the MCF. Usually the MCF of the HLT
takes maximum when the hybrid folding appears. When N > 10, the HLT is crushed at macro-cell folding style. The folding
wavelength is equal to the tube length. Different from the ST, the macro-cell fold indicates the folding of the sandwich wall.
The MCFs of these tubes are smaller than the tubes crushed at micro-cell folding style. Meanwhile, the bending moment of
the sandwich wall is much larger than that of the solid wall, leading to much larger MCF for these tubes than for ST.
The folding style transformation is well demonstrated in Fig. 6. Typical folding elements are observed, including the
sandwich corner element, sandwich wall element and solid corner element. All of these elements displayed are for HLT,
while these elements can also be observed in the compression of SHLT and SMLT.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 21

Fig. 8. Crushing behaviors of typical SMLTs and transformation of folding modes.

Fig. 9. Crushing behaviors of typical HMLTs and transformation of folding modes.

22 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 10. Variation of MCFs for structures evolving from HLT, SHLT, SMLT to HMLT.

As shown in Figs. 7–9, the transformation of folding style from micro-cell folding to macro-cell folding is also observed in
the compression of SHLT, SMLT and HMLT. When N = 6 and N = 10, the SHLTs are crushed at progressive micro-cell folding,
as shown in Fig. 7. When N = 12, the SHLT are crushed at hybrid folding. When N = 16 and 20, the crushed configuration of
tubes are similar to that of MLT02.
Similar trend is observed in Fig. 8. When N = 10, SMLT is crushed at progressive micro-cell folding. When N = 16, SMLT
is crushed at hybrid folding. When N = 20, SMLT is crushed at macro-cell folding. The MCF of SMLT16 is close to the PF of
HLT16, which is the strongest MCF among the investigated tubes.
As shown in Fig. 9, when N = 13, HMLT13 is crushed at micro-cell folding. When N = 17, HMLT17 is crushed at hybrid
folding. When N = 21, HMLT21 is crushed at macro-cell folding.

4.3. Hierarchy effect

As shown in Fig. 1, four kinds of hierarchical structures are designed. HLT is the high-order topology of ST01. SHLT is
the high-order topology of MLT02. SMLT is the high-order topology of MLT03. HMLT is the high-order topology of MLT04.
PFs of these hierarchical structures do not differ much. But improvement in MCF is observed when the hierarchy level of
the tubular structure increases. Each hierarchical structure has an optimal N corresponding to maximum energy absorption.
When N is larger than the optimal value, MCF will become smaller if the topology is kept unchanged. To further increase the
MCF, the hierarchy style of the structure must be improved. This can achieve continuous improvement of MCF as indicated
in Fig. 10, where the structure is transferred from HLT to SHLT, SMLT and finally HMLT. It can be seen that MCF of the HMLT
is very close to its PF.
Conclusions can draw from the above simulations: (a) Micro-cell folding is an effective way to absorb energy; (b) a
proper side wall segment number can maximize the energy absorption; (c) high-order hierarchical topology can change the
folding style and improve the energy absorption.

5. Theoretical analyses

5.1. Micro-cell folding model

A simplified super folding element (SSFE) theory was suggested by Abramowicz and Wierzbicki (1989). Instead of con-
structing a kinematically admissible model consisting of trapezoidal, toroidal and cylindrical surfaces with moving hinge
lines, the basic folding element only consists of extensional triangular elements and two stationary hinge lines. This method
has been adopted by Tran et al. (2014) and Sun, Lai, Fan, and Fang (2016b), Zhang and Zhang (2014) to analyze the crushing
behaviors of multi-cell tubes. The theory is further improved for HLTs.
FEM results indicate that there are three kinds of angle elements in HLT, including corner element, X-shape element and
T-shape element, as shown in Fig. 11. In a HLT with N side wall segments, there are 4 2-panel elements, (8N-16) 3-panel
elements and 4 4-panel elements. The general theoretical equation of the total energy absorbed during crushing, Ep is given
by,
     
E p = 4 Eb2 p + Em
2p
+ (8N − 16 ) Eb3 p + Em
3p
+ 4 Eb4 p + Em
4p
, (8)
where Eb is the bending energy and Ep is the energy absorbed by the angle element. 2p, 3p and 4p represent the corner,
T-shape and X-shape element, respectively.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 23

Fig. 11. Folding of (a) corner element, (b) T-shape element and (c) X-shape element.

5.1.1. Bending energy

L
The energy dissipated by two stationary hinge lines, whose lengths equal to 2N , is calculated by


4
L L
Eb = θi M 0 = 2 π M0 , (9)
2N 2N
i=1
√ σy +σu
4 σ pt and σ p = σy σu or σ p =
1 2
with M0 = 2 .

5.1.2. Membrane energy

The membrane energy of corner element is calculated based on the asymmetric or symmetric deformation mode in SSFE
theory. Asymmetric deformation comes from the extensional mode, which develops three additional part areas for each
web, as shown in Fig. 12(a). While symmetric deformation comes from in-extensional mode, which develops two additional
parts for each web, as shown in Fig. 12(b). Thus the energy dissipated in membrane during crushing of one wavelength of
each panel is calculated by integrating over the part areas. Then,

H1 2
asym
Em = σ0tds = M0 , (10)
s 2t
and

H1 2
sym
Em = σ0tds = M0 . (11)
s t

5.1.3. Corner element

In the work of Chen and Wierzbicki (2001), the energy dissipated in bending of each plane, Eb , was calculated by sum-
ming up the energy dissipated by two stationary hinge lines. Besides the two hinge lines, some other hinge lines are dis-
covered during the compression and the process of corner element folding. According to the FEM results, more than two
stationary hinge lines are observed in a single corner element. As shown in Fig. 12(c), the hinge lines developed in one side
of a wavelength of one corner element are simplified to AB, BC, BD, DE, AD and CD. Then,
π √ L π √
Ebcorner = 2Eb + M0 H2 + 2H2 M0 π = 2π M0 + M0 H2 + 2H2 M0 π , (12)
2 N 2
24 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 12. (a) Asymmetric membrane deformation mode, (b) symmetric membrane deformation mode and (c) simplified mode of bending hinge lines of
corner element.

The fold of corner element consists of two parts. The wave length of the first part is h1 , which is obviously shorter
than that of the second part, h2 . The folding of the two parts dissipates energy by different ways and the simplified mode
of corner element is displayed in Fig. 11(a). The energy dissipated by the first fold consists of bending contribution and
membrane contribution, while the energy dissipated by the second fold just consists of bending contribution. Then
L π √ h1
2
E 2 p = Eb2 p + Em
2p
= 2 π M0 + M 0 h 2 + 2h 2 M 0 π + M 0 . (13)
N 2 t
5.1.4. T-shape element
T-shape element is created by one right-corner element and one additional panel. The simplified folding model is shown
in Fig. 11(b). However, the energy dissipated by T-shape element is calculated by summing up each panel with asymmetric
deformation. There are three panels in each T-shape element. Thus,
 
  L H2
E 3p
=3 Eb3 p + 3p
Em = 3 π M 0 +M 0 , (14)
N 2t
with H = h1 + h2 .

5.1.5. X-shape element

The energy dissipated by X-shape element is similar to T-shape element. X-shape element has four panels. The simplified
folding model is shown in Fig. 11(c). The energy calculation for X-shape element follows similar way as for T-shape element,
as given in,
 
  L H2
E 4p
=4 Eb4 p + 4p
Em = 4 π M 0 +M 0 . (15)
N 2t

5.1.6. Mean crushing force

p
Introducing coefficient of effective crushing distance, k, the MCF, pm , is given by
 2

p Ep 4E 2 p + (8N − 16 )E 3 p + 4E 4 p M0 A Bh1 C h1
pm = = = − + + DH + E , (16)
Hk Hk k H H H
√
with A = 24(N − 1 )π L/N, B = E = (4 2 + 2 )π , C = 4/t, and D = (12N − 16 )/t. The wave length is determined by ∂ pm /∂ h1 =
p

0 and ∂ pm /∂ H = 0. Then,
p

  √ 
√
B 2 2+1 96(N − 1 )π Lt − 4 2 + 9 π 2 t 2 N
h1 = = tπ , H = . (17)
2C 4 16N (3N − 4 )
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 25

Fig. 13. MCF predicted by micro-cell folding and macro-cell bending mechanisms for (a) HLTs, (b) SHLTs, (c) SMLTs and (d) HMLTs and compared with
traditional method of Chen and Wierzbicki (2001).

6t
According to the simplified model, k is determined by k = 1 − . Thus
H
 
p M0 A B2
pm = − + DH + E . (18)
k H 4CH

In this case, L = 120mm, σy = 343MPa, σu = 475MPa. The results predicted by the theory are compared with FE simula-
tions. The model agrees well with the FEM, as shown in Fig. 13(a).

5.2. Macro-cell folding model

In macro-cell folding, the energy consists of two parts contributed by: corner element and sandwich element.

5.2.1. Corner element

The membrane energy of corner element is contributed by walls with two plies. Asymmetric deformation is still available
for macro-cell folding. Thus
H
2 t 2 H 2 M0
Ecorner = 4 σp = . (19)
2 2 t
26 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Fig. 14. Plastic bending model of sandwich wall.

5.2.2. Sandwich element

As shown in Fig. 6, the folding of the sandwich element includes two skins and ribs. The simplified model is shown in
Fig. 14. The energy dissipated by two skins can be calculated by
4M0 π BL
skin
Ebending = 4M0 π L + σ p π BtL = 4M0 π L + . (20)
t
r
The energy dissipated by each rib, Ebending , is given by

s
Ecore = π B2 t σ p . (21)

5.2.3. Mean crushing force

MCF of macro-cell folding mode, pGm , is given by
 
4 H2 4M0 π BL
pGm = 2 M0 + 4 π M0 L + + (N − 1 )π B2 σ pt . (22)
Hk t t

The wavelength H = 100 mm and k = 0.7, with B = NL . MCFs of HLT08, HLT10 and HLT12 are calculated by the theoretical
model. The results predicted by the theory are compared with FE simulations. The model agrees well with the FEM, as
shown in Fig. 13(a).

5.3. Hybrid folding model

MCF of the hybrid folding, phm , is given by

phm =η pm
p
+(1 − η ) pGm , (23)
where η represents the ratio of the progressive micro-cell folding in the hybrid folding framework. There are three side
walls folding in micro-cell folding mode and one side wall folding in macro-cell folding mode in HLT08, same as HLT10.
Thus three hybrid modes can be suggested: (a) Hybrid model A, in which one side follows progressive micro-cell folding
and the other three follow macro-cell folding; (b) Hybrid model B, in which two sides follow progressive micro-cell folding
and the other two follow macro-cell folding; (c) Hybrid model C, in which three sides follow progressive micro-cell folding
and the other one follows macro-cell folding. MCFs of these three hybrid modes, phA hB hC
m , pm and pm are given by

1 p 3 G
phA
m = pm + pm , (24)
4 4

1 p 1 G
phB
m = pm + pm , (25)
2 2
and,
3 p 1 G
phC
m = pm + pm . (26)
4 4
The three hybrid models result in MCFs that are between the micro-cell folding mode and the macro-cell folding mode.
Eq. (24) well predicts the transitional stage, as shown in Fig. 15. Combining with Eqs. (18), (22) and (24), the theory can
accurately predict the MCF and folding mode transition of HLT.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 27

Fig. 15. (a) MCF predicted by hybrid folding modes and (b) MCF variation with N.

5.4. Prediction of MLT

According to the micro-cell folding model, the energy dissipated by MLT02 is contributed by 4 corner elements, 4 T-shape
elements and 1 X-shape element. Thus the mean crushing force of MLT02 is given as,
 
M0 24π L  √  π h1 4 h1  √  2
8
pm
MLT 02 = − 4 2+2 + + H+ 4 2+2 π . (27)
k HN H tH t
with
√  √ 
B 2 2+1 48π Lt − 4 2 + 9 π 2 t 2
h1 = = tπ , H = . (28)
2C 4 32
Accordingly, pm
MLT 02
= 185.8kN. The MCF of MLT02 by FEM is 218 kN.

5.5. Prediction of SHLT

5.5.1. Micro-cell folding model

There are 4 corner elements, 8 X-shape elements, and 48 T-shape elements in SHLT06. According to the micro-cell folding
model, the MCF of SHLT06, pm , is given by
p 1  2p 
pm = 4E + 8E 4 p + 48E 3 p . (29)
Hk
Taking E2p , E3p and E4p into the equation, one gets
 
M0 184π L  √  h1 4 h1  √  2
88
p
pm = − 4 2+2 π + + H+ 4 2+2 π , (30)
k HN H tH t
with
√  √ 
B 2 2+1 736π Lt − 4 2 + 9 π 2 t 2 N
h1 = = tπ , H = , (31)
2C 4 352N
∂ pm
p
∂ pm
p
where h1 and H are determined by = 0 and = 0 jointly.
∂ h1 ∂H
p
For SHLT06, pm = 353.8kN. In conclusion, micro-cell folding model and macro-cell bending model can describe the mech-
anism of compression and can predict the MCF of the HTL and SHLT.
According to the count, each SHLT has 4 corner elements, 16 X-shape elements, and (12N-44) T-shape elements. Thus,
the MCF of SHLT is given by
 
M0 L  √  h1 4 h1 18N − 34  √ 
2
p
pm = (36N − 60)π − 4 2+2 π + + H+ 4 2+2 π . (32)
k HN H tH t
with
  √ 
√
B 2 2+1 48(3N − 5 )π Lt − 4 2 + 9 π 2 t 2 N
h1 = = tπ , H = , (33)
2C 4 8N (9N − 17 )
28 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

∂ pm
p
∂ pm
p
where h1 and H are determined by = 0 and = 0 jointly.
∂ h1 ∂H
The MCFs of SHLT predicted by FE simulations and theoretical models are compared. The theoretical model agrees well
with the FE simulations, as shown in Fig. 13(b).

5.5.2. Macro-cell folding model

As shown in Fig. 7, the folding style of SHLT16 is different from SHLT08. Compared with MLT02, they have similar fold-
ing mechanisms. Based on the FE numerical results, the macro-cell folding mechanism of SHLT is proposed. Based on the
plasticity theory and ignoring the bending effect of ribs, the fully plastic bending moment is calculated by
L 5L
M p = σ0 t , (34)
N 8N
while M0 represents the fully plastic bending moment on each unit, as given by,
Mp 5L
M0 = L
= σ0 t (35)
N
8N
Taking M0 into pm
MLT 02
, considering h1 = H,

M0 24π L 12H

pGm
SHLT = + , (36)
k NH t

2π Lt ∂ pGm 2
where, H = N , determined by ∂ H = 0 and k = 1 − N .
SHLT

The results predicted by the theory are compared with FE simulations. The model agrees well with the FE simulations,
as shown in Fig. 13(b).

5.6. Prediction of SMLT

5.6.1. Micro-cell folding model

The micro-cell folding of SMLT includes 4 corner elements, 8(2N − 11 ) T-shape elements and 36 X-shape elements. Thus,
the MCF of SMLT is given by
 2

p Ep M0 A Bh1 C h1
pm = = − + + DH + E , (37)
Hk k H H H
√
with A = 16(3N − 7 )π L/N, B = E = (4 2 + 2 )π , C = 4/t, and D = 4(6N − 15 )/t. The wave length is determined by ∂ pm /∂ h1 =
p

0 and ∂ pm /∂ H = 0. Then,
p

  √ 
√
B 2 2+1 64(3N −7 )π Lt − 4 2 + 9 N π 2 t 2
h1 = = tπ , H = . (38)
2C 4 32(6N − 15 )
The predicted values by the theory are compared with FE simulations. The model agrees well with the FE simulations,
as shown in Fig. 13(c).

5.6.2. Macro-cell folding bending model

The calculation of MCF for SMLT is similar to that for SHLT. Taking M0 into the expression of pm
MLT 03
, one gets,

M0 48π L 24H

pGm
SMLT = + , (39)
k NH t

2π Lt ∂ pGm
where, H = N , is determined by ∂ H = 0. Thus
SMLT

48M0 2π L
pGm
SMLT = , (40)
k Nt
The predictions of the theory are compared with FE simulations. The theoretical model agrees well with the FE simula-
tions, as shown in Fig. 13(c).

5.7. Prediction of HMLT

5.7.1. Micro-cell folding model

The micro-cell folding of HMLT has 4 corner elements, (20N − 148 ) T-shape elements and 64 X-shape elements. Thus,
the MCF of HMLT is given by
 2

p E p M0 A Bh1 C h1
pm = = − + + DH + E (41)
Hk k H H H
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 29

Fig. 16. Crushing mode and energy absorbing mechanism of typical hierarchical tubular structures.

√
with A = 60(N − 3 )π L/N, B = E = (4 2 + 2 )π , C = 4/t, D = (30N − 94 )/t and k = 1 − 6t/H. The wave length is determined
by ∂ pm /∂ h1 = 0 and ∂ pm /∂ H = 0. Then,
p p

  √ 
√
B 2 2+1 240(N − 3 )π Lt − 4 2 + 9 π 2 t 2 N
h1 = = tπ , H = . (42)
2C 4 4N (30N − 94 )

The predictions of the theory are compared with FE simulations. The theoretical model agrees with the FE simulations,
as shown in Fig. 13(d).

5.7.2. Macro-cell folding model

The calculation of MCF for HMLT is similar to that for SHLT. Taking M0 into pm
MLT 03
, one gets,

M0 80π L 40H

pGm
HMLT = + , (43)
k NH t

∂ pGm 2π Lt
∂ H = 0, H =
HMLT
where H is determined by N , thus

80M0 2π L
pGm
HMLT = (44)
k Nt

with k = 1 − 3/N. The predictions of the theory are compared with FE simulations. The theoretical model agrees with the
FE simulations, as shown in Fig. 13(d).
30 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Micro-cell folding model of HLT

Macro-cell folding model of HLT
Micro-cell fold model of SHLT
700 Macro-cell folding model of SHLT
Micro-cell folding model of SMLT
Macro-cell folding model of SMLT
Micro-cell folding model of HMLT
Macro-cell folding model of HMLT Perfect energy absorption
600

Mean crushing force (kN)

Full-Plastic Force Eq. (37) Eq. (41) Eq. (44)

HMLT
500 Micro-cell
folding of SMLT
Eq. (32)
400 Micro-cell Eq. (40)
folding of SHLT Macro-cell folding of SMLT
(a)
Eq. (36)
300 Macro-cell
Micro-cell folding of HLT folding of SHLT
Eq. (18) Eq. (22)
200 Macro-cell folding of HLT

100
4 8 12 16 20
N

1.2
Mean crushing force/Full-plastic force

Full-Plastic Force Perfect energy absorption

Hybrid folding of SMLT Hybrid folding of HMLT
1.0 Hybrid folding of SHLT

0.8 Micro-cell folding of SMLT Micro-cell folding of HMLT

Hybrid folding of HLT

0.6 (b)
Micro-cell folding of SHLT

0.4 Micro-cell folding of HLT

0.2
4 8 12 16 20 24
N

Fig. 17. Development of MCF of hierarchical lattice structures approaching perfect energy absorption: (a) Competition between micro-cell folding and
macro-cell folding and (b) competition between micro-cell folding and hybrid folding.

5.8. Comparisons

The approach proposed by Chen and Wierizbicki (2001) for progressive folding is also applied to predict the MCF. For
HLT,

1 M0 
pm = 24 1 − (N − 1 )2π L. (45)
N k
For SHLT,

5 M0 14
pm = 12 3 − 3N − π L. (46)
N k 3

For SMLT,

7 M0 7
pm = 16 3 − N− 4π L. (47)
N k 3

For HMLT,

3 M0 
pm = 60 1 − (N − 3 )5π L. (48)
N k
As compared in Fig. 13, the approach by Chen and Wierizbicki (2001) leads to much greater errors. The improved ap-
proach in this research is more appropriate in predicting the MCF of the hierarchical tubular structures.
 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32 31

6. Perfect energy absorbing structure

According to the above results, when the structure transits from ST to MLT, HLT, SHLT, SMLT, and finally to HMLT, the
crushing mechanism changes from macro-cell folding to micro-cell folding, resulting in shorter folding wavelength and
stronger MCF, as shown in Fig. 16. Hierarchy enhances the energy absorption ability. For each hierarchical structure, side
segment division greatly affects the crushing. As N increases, the folding evolves from micro-cell folding, hybrid folding to
macro-cell folding, accompanied by the first increase and then decrease of MCF. The maximum value of MCF inextricably
corresponds to hybrid folding.
Therefore, a feasible approach to achieve perfect energy absorbing ability is available. As shown in Fig. 17(a), when the
structural hierarchy level increases simultaneously from ST, MLT, HLT, SHLT, SMLT to HMLT or even higher order with more
segment divisions, MCF of tubular structure increases in a zigzag mode and finally approaches to the full-plastic force.
Considering the hybrid folding, MCF of tubular structure increases in a multi-step growing mode and finally approaches to
the full-plastic force, as shown in Fig. 17(b).

7. Conclusions

To achieve perfect energy absorption, hierarchical topology was applied to design novel lattice tubular structures. Through
experiments, FE simulations and theoretical analyses, advantages of hierarchical lattice tubular structures in energy absorp-
tion were manifested. This research indicates that through high-order hierarchy design, lattice tubular structures can have
MCFs approaching to its full-plastic strength, achieving perfect energy absorption.
Three folding styles are observed during the compression, including micro-cell folding, macro-cell folding and hybrid
folding. Micro-cell folding has shorter wave length dominated by the micro-cell dimension while macro-cell folding has
longer wave length depending on the macro-cell dimension of the tubular structure. As the number of cells in the wall in-
creases, the crushing mode transforms from micro-cell folding, hybrid folding to macro-cell folding. The MCF first increases,
approaches to its peak value at hybrid folding and then decreases.
When the hierarchy level of the tubular structure is gradually raised, the maximum MCF of the lattice tubular structure
can be gradually increased, approaching the full-plastic strength of the matrix. This provides us a feasible way to achieve
perfect energy absorption through high-order hierarchical lattice topology.

Acknowledgments

Supports from National Natural Science Foundation of China (11672130, 11372095) and State Key Laboratory of Mechanics
and Control of Mechanical Structures (MCMS-0217G03) are gratefully acknowledged.

References

Abramowicz, W., & Jones, N. (1984). Dynamic axial crushing of square tubes. International Journal of Impact Engineering, 2(2), 179–208.
Abramowicz, W., & Jones, N. (1986). Dynamic progressive buckling of circular and square tubes. International Journal of Impact Engineering, 4(4), 243–270.
Abramowicz, W., & Wierzbicki, T. (1989). Axial crushing of multi corner sheet metal columns. Journal of Applied Mechanics , 56(1), 113–120.
An, X. Y., & Fan, H. L. (2016). Hybrid design and energy absorption of luffa-sponge-like hierarchical cellular structures. Materials & Design, 106, 247–257.
Birman. , V. (2014). Mechanics and energy absorption of a functionally graded cylinder subjected to axial loading. International Journal of Engineering Science,
78, 18–26.
Chen, W. G., & Wierzbicki, T. (2001). Relative merits of single-cell, multi-cell and foam-filled thin-walled structures in energy absorption. Thin-Walled
Structures, 39, 287–306.
Chen, Y. L., Qiao, C., Shougen Zhao, S. G., Zhen, C. R., & Liu, B. (2016). A novel self-locked energy absorbing system. Journal of the Mechanics and Physics of
Solids, 87, 130–149.
Deshpande, V. S., & Fleck, N. A. (2003). Energy absorption of an egg-box material. Journal of the Mechanics and Physics of Solids, 51, 187–208.
Fan, H. L., Sun, F. F., Yang, L., Jin, F. N., & Zhao, D. J. (2013). Interlocked hierarchical lattice materials reinforced by woven textile sandwich composites.
Composites Science and Technology, 87, 142–148.
Fan, H. L., Zhao, L., Chen, H. L., Kuang, N., Yang, C. K., Huang, S. Q., & Jiang, Y. F. (2012). Ductile deformation mechanisms and designing instructions for
integrated woven textile sandwich composites. Composites Science and Technology, 72, 1338–1343.
Gibson, L. J., & Ashby, M. F. (1997). Cellular Solids: structure and properties. Cambridge: Cambridge University Press.
He, Z. Z., Wang, F. C., Zhu, Y. B., Wu, H. A., & Park, H. S. (2017). Mechanical properties of copper octet-truss nanolattices. Journal of the Mechanics and Physics
of Solids, 101, 133–149.
Hong, W., Jin, F. N., Zhou, J. N., Xia, Z. C., Xu, Y., Yang, L., Zheng, Q., & Fan, H. L. (2013). Quasi-static axial compression of triangular steel tubes. Thin-Walled
Struct, 62, 10–17.
Hong, W., Fan, H. L., Xia, Z. C., Jin, F. N., Zhou, Q., & Fang, D. N. (2014). Axial crushing behaviors of multi-cell tubes with triangular lattices. International
Journal of Impact Engineering, 63, 106–117.
Jiang, S., Sun, F. F., Zhang, X. R., & Fan, H. L. (2017). Interlocking orthogrid: An efficient way to construct lightweight lattice-core sandwich composite
structure. Composite Structures, 176, 55–71.
Kooistra, G. W., Deshpande, V. S., & Wadley, H. N. G. (2004). Compressive behavior of age hardenable tetrahedral lattice truss structures made from alu-
minium. Acta Materialia, 52, 4229–4237.
Lakes, R. (1993). Materials with structural Hierarchy. Nature, 361, 511–516.
Meza, L. R., Das, S., & Greer, J. R. (2014). Strong, lightweight, and recoverable three-dimensional ceramic nanolattices. Science, 345(6202), 1322–1326.
Meza, L. R., Zelhofer, A. J., Clarke, N., Mateos, A. J., Kochmann, D. M., & Greera, J. R. (2015). Resilient 3D hierarchical architected metamaterials. PNAS, 112(37),
11502–11507.
Qiu, N., Gao, Y. K., Fang, J. G., Feng, Z. X., Sun, G. Y., & Li, Q. (2016). Theoretical prediction and optimization of multi-cell hexagonal tubes under axial
crashing. Thin-Walled Structure, 102, 111–121.
32 H. Fan et al. / International Journal of Engineering Science 130 (2018) 12–32

Schaedler, T. A., Jacobsen, A. J., Torrents, A., Sorensen, A. E., Lian, J., Greer, J. R., Valdevit, L., & Carter, W. B. (2011). Ultralight metallic microlattices. Science,
334(6058), 962–965.
Sun, F. F., & Fan, H. L. (2017). Inward-contracted folding element for thin-walled triangular tubes. Journal of Constructional Steel Research, 130, 131–137.
Sun, F. F., Lai, C. L., & Fan, H. L. (2016a). In-plane compression behavior and energy absorption of hierarchical triangular lattice structures. Materials &
Design, 100, 280–290.
Sun, F. F., Lai, C. L., Fan, H. L., & Fang, D. L. (2016b). Crushing mechanism of hierarchical lattice structure. Mechanics of Materials, 97, 164–183.
Sun, F. F., Zheng, Q., Fan, H. L., & Fang, D. N. (2017). The mechanical properties of hierarchical truss-walled lattice materials. International Journal of Applied
Mechanics, 9(2), 1750027.
Tran, T., Hou, S. J., Han, X., Tan, W., & Nguyen, N. (2014). Theoretical prediction and crashworthiness optimization of multi-cell triangular tubes. Thin-Walled
Structure, 82, 183–195.
Wang, L., Fan, X. M., Chen, H., & Liu, W. Q. (2016). Axial crush behavior and energy absorption capability of foam-filled GFRP tubes under elevated and high
temperatures. Composite Structures., 149, 339–350.
Wang, P., Zheng, Q., Fan, H. L., Sun, F. F., Jin, F. N., & Qu, Z. X. (2015). Quasi-static crushing behaviors and plastic analysis of thin-walled triangular tubes.
Journal of Constructional Steel Research, 106, 35–43.
Wierzbicki, T. (1983). Crushing analysis of metal honeycombs. International Journal of Impact Engineering, 1(2), 157–174.
Wierzbicki, T., & Abramowicz, W. (1983). On the crushing mechanics of thin-walled structures. Journal of Applied Mechanics, 50, 727–734.
Xu, B. X., Qiao, Y., & Chen, X. (2014). Mitigating impact/blast energy via a novel nanofluidic energy capture mechanism. Journal of Applied Mechanics Solids,
62, 194–208.
Yang, F., Meguid, S. A., & Hamouda, A. M. S. (2018). Kinematically admissible folding mechanisms: For the progressive collapse of foam filled conical frusta.
International Journal of Mechanics and Materials in Design, 14, 105–126.
Zhang, P., Heyne, M. A., & To, A. C. (2015). Biomimetic staggered composites with highly enhanced energy dissipation: Modeling, 3D printing, and testing.
Journal of the Mechanics and Physics of Solids, 83, 285–300.
Zhang, X., & Zhang, H. (2014). Axial crushing of circular multi-cell columns. International Journal of Impact Engineering, 65, 110–125.
Zheng, G., Wu, S. Z., Sun, G. Y., Li, G. Y., & Li, Q. (2014). Crushing analysis of foam-filled single and bitubal polygonal thin-walled tubes. International Journal
of Mechanical Sciences., 87, 226–240.
Zheng, J. J., Zhao, L., & Fan, H. L. (2012). Energy absorption mechanisms of hierarchical woven lattice composites. Composites Part B: Engineering, 43,
1516–1522.
Zheng, X., Lee, H., Weisgraber, T. H., Shusteff, M., DeOtte, J., Duoss, E. B., et al. (2014). Ultralight, ultrastiff mechanical metamaterials. Science, 344(6190),
1373–1377.