International Journal of Solids and Structures xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

International Journal of Solids and Structures

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

Mechanical behavior of hexagonal honeycombs under low-velocity

impact – theory and simulations
L.L. Hu a,⇑, T.X. Yu b
a
Department of Applied Mechanics & Engineering, School of Engineering, Sun Yat-Sen University, Guangzhou 510275, PR China
b
Department of Mechanical Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong

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

Article history: Based on the cells’ collapse mechanisms of the hexagonal honeycombs revealed from the numerical sim-
Received 5 September 2012 ulations under the low-velocity impact, an analytical model is established to deduce the crushing
Received in revised form 16 April 2013 strength of the honeycomb and the stress at the supporting end both as functions of impact velocity, cell
Available online xxxx
size, cell-wall angle, and the mechanical properties of the base material. The results show that the hon-
eycomb’s crushing strength increases with the impact velocity, while the supporting stress decreases
Keywords: with the increase of the impact velocity. Combining with the dynamic predictions under the high-veloc-
Honeycomb
ity impact in our previous work (Hu and Yu, 2010), the crushing strength of the honeycombs can be ana-
Low-velocity impact
Crushing strength
lytically predicted over wide range of crushing velocities. The analytical expression of the critical velocity
Critical velocity is also obtained, which offers the boundary for the application of the functions of the honeycomb’s crush-
Supporting stress ing strength under the low-velocity and the high-velocity impacts. All of the analytical predictions are in
good agreement with the numerical simulation results.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction such as using a honeycomb block as an energy absorption layer in

Metallic based cellular materials, such as honeycombs, foams
and hollow spheres, have been developed and are growing in use
as new engineering materials. The advantages of these ultra-light
metal materials are not only the high relative stiffness and strength
but also the effective energy absorption during accidental impacts
while limiting the crushing force (Gibson and Ashby, 1997). This
fact recently triggered many interests in investigating the dynamic
properties of them.
Honeycombs with regular hexagonal cells are widely used in
industry. Extensive experimental (Hou et al., 2011) analytical
(Okumura et al., 2002; Zhu and Mills, 2000; Hu and Yu, 2010)
and numerical (Zou et al., 2009; Ruan et al., 2003) studies have
been conducted on their mechanical behaviors. By simply changing
the internal cell-wall angle of hexagonal cells, new cell configura-
tion can be obtained, which is easy to implement based on the
existing production process. The out-of-plane mechanical proper-
ties of hexagonal honeycombs with changed cell-wall angles were
studied by Yamashita and Gotoh (2005). The results show that
changing the cell structures by this simple method can enhance
the honeycomb’s out-of-plane crushing strength to 1.5 times com-
paring to the regular hexagonal honeycomb. In some applications,
⇑ Corresponding author.
E-mail addresses: llhu_80@163.com, hulingl@mail.sysu.edu.cn (L.L. Hu).
aircraft against bird or debris collision, the crushing could take
place along any direction of the honeycomb. Hence, the effect of
changing the cell-wall angle on the in-plane dynamic behaviors
of the honeycomb also needs to be known in addition to its out-
of-plane behaviors. The relevant work has been carried out by both
numerical simulations and experiments (Hu et al., 2013). More
theoretical analysis needs to be done on it.
Localized deformation is a characteristic feature of honeycombs
under in-plane compression (Chung and Waas, 2002; Papka and
Kyriakides, 1998). In quasi-static compression, collapse first takes
place at the weakest row or band of cells, which is dominated by
the distribution and extent of the initial imperfections, and gradu-
ally spreads to stronger areas of the structure (Hu et al., 2008; Pap-
ka and Kyriakides, 1994). The response to dynamic loadings is very
different from the quasi-static one, as the structural and inertial ef-
fects will significantly influence the crushing behaviors (Hou et al.,
2012). Localization bands would initiate at the loading end or the
supporting end of the honeycomb block once the impact velocity
exceeds a critical value, despite the initial imperfections distribute
among the honeycomb (Honig and Strong, 2002a). Honig and
Strong (2002b) derived the critical velocity, V cr , using the theory
of wave trapping:
Z sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ecr
1 dr
V cr ¼  de ð1Þ
0 q de

0020-7683/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
2 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

Nomenclature

b out-of-plane thickness of honeycomb V 0c critical velocity between LVDM and HVDM for the dou-
E Young’s modulus of base material ble-thickness honeycomb
Ep energy dissipated by the plastic hinges W work done by the extra forces acted on the representa-
h cell wall thickness of honeycomb tive block over a complete collapse period
Ho original height of the representative block b cell-wall angle defined in Fig. 1
Hf final height of the representative block ecr critical strain where dr=de ¼ 0 in the stress–strain curve
Lo original length of the representative block ed densification strain of cellular material
l cell wall length of the honeycomb t Poisson ratio of base material
l’ vertexes’ distance of the cell wall q density of cellular material
Mp fully plastic bending moment of cell walls, qs density of base material.
2
M p ¼ bh rys =4 qr x densification strain of cellular material
px01 initial momentum in the x1 direction rys yield stress of base material
pxf 1 final momentum in the x1 direction r0 quasi-static plateau stress
To initial kinetic energy r1 crushing stress suffered by the representative block
Tf final kinetic energy r 1 average crushing strength of single-thickness honey-
DT increase in the kinetic energy of the representative comb over a collapse period
block r 01 average crushing strength of double-thickness honey-
t0 time instant at the beginning of a collapse period of cells comb over a collapse period
tf time instant at the end of a collapse period of cells r2 supporting stress applied to the representative block
t relative time in a collapse period of cells r 2 average supporting stress of single-thickness honey-
V crushing velocity comb over a collapse period
Vc critical velocity between LVDM and HVDM for the sin- r 02 average supporting stress of single-thickness honey-
gle-thickness honeycomb comb over a collapse period

where ecr is the critical strain where the tangent modulus of the
stress–strain curve, dr=de, becomes zero, and q is the density of
the cellular material.
According to the shapes of localization bands, Ruan et al. (2003)
identified three deformation patterns in the uniaxially impacted
honeycombs by numerical simulations, which is ‘‘X’’ shape, ‘‘V’’
shape and ‘‘I’’ shape. The transition velocities at which the shapes
of localization bands changed were studied empirically based on
the massive numerical results.
Under high crushing velocity, the localization band caused
by cells’ collapse initiates at the loading end perpendicular to
the impact direction, and continues to propagate layer by layer
to the fixed end (i.e. ‘‘I’’ pattern defined by Ruan et al. (2003)),
which is similar to the propagation of a shock wave in a con-
tinuum bar. Therefore, by treating a cellular material as contin-
uum and employing the one-dimensional shock wave theory,
the dynamic crushing strength of cellular material rd was de-
rived under this ‘shock’ type deformation pattern (Reid and
Peng, 1997):
q 2
rd ¼ r0 þ V ð2Þ
ed
where r0 and ed are the quasi-static plateau stress and densification
strain of the cellular material, respectively; and V is the crushing
velocity. In Eq. (2), the plateau stress r0 and the densification strain
ed both need to be determined from the quasi-static stress–strain
curve of the cellular material.
By carefully studying the repeatable collapsing processes of the
cell-structure under high-velocity impact, Hu and Yu (2010) de-
rived an analytical formula on the dynamic crushing strength of
hexagonal honeycombs from a structural point-of-view instead
of continuum:
 2 
2 h 6 h q
rd ¼ rys þ pffiffiffi l sh V 2 ð3Þ
3 l 3 38 l
where rys and qs are the yield strength and the density of the
base material; h and l are the thickness and the length of the cell
walls of the honeycomb. It is evident from Eq. (3) that the dy-
namic crushing strength of hexagonal honeycombs depends on
the cell walls’ relative thickness h/l, the physical property of base
material and the crushing velocity. A further study (Hu and Yu,
2010) shows that this result based on a structural point-of-view
is consistent with that obtained from the theory of continuum
by Reid and Peng (1997) (i.e., Eq. (2)), which validates the consis-
tency of the two theories on predicting the dynamic strength of
cellular material.
Both Eq. (2) and Eq. (3) are established under the condition of
high-velocity impact, under which cell collapse occurs in a pro-
gressive manner from the impact end to the supporting end. How-
ever, the cells of the honeycomb could collapse in other manners
under low-velocity impact. Hence, the mechanical behavior of cel-
lular materials under lower-velocity impact has to be comprehen-
sively explored.
When honeycomb is used as a protective layer to absorb energy,
we also need to predict the stress and the energy transferring to
the protected object so as to evaluate the protection efficiency of
the honeycomb layer. Under high-velocity impact, this stress at
the supporting end is less affected by the impact velocity and is
much smaller compared with that at the impacting end, so it is
usually represented by the quasi-static crushing strength of the
honeycomb (Zou et al., 2009; Hu and Yu, 2010). The stress between
the honeycomb and the protected project should be a function of
the cell structure, the cell size, and even the crushing velocity un-
der some conditions, but those expressions have not been fully re-
ported in literatures.
The aim of this paper is to fill the gap in lacking analytical
expressions for both the crushing strength and the supporting
stress of the honeycombs under the low-velocity impact. The ana-
lytical models are established based on the cells’ collapse mecha-
nism of the honeycomb revealed from the numerical simulations.
The expression of the critical velocity between the low-velocity
and the high-velocity deformation modes of the honeycombs is
also deduced. All of the analytical predictions are compared with
the numerical simulation results.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 3

Upper right vertex Upper rigid plate

Honeycomb
y
Fig. 1. Configuration of a single cell.

2. Numerical simulations
Lower rigid plate

2.1. Deformation modes of honeycombs

Fig. 2. Numerical model in simulations.
To further understand the dynamic deformation behaviors of
the metal hexagonal honeycomb, a finite element model is built
using ANSYS-LSDYNA. Fig. 1 shows the structure of a single cell
of hexagonal honeycombs, where b is the angle of the honeycomb’s
bevel edge to the x direction, which is called the cell-wall angle. By
changing the cell-wall angle, five types of honeycombs with rela-
tive density qr ¼ q =qs ¼ 0:1 are obtained with b being equal to
15°, 30°, 45°, 60° and 75°, respectively. The edge length of the cells
l is taken as 4 mm, and then the cell-wall thickness can be obtained
according to the relative density as listed in Table 1. The cell wall
material is assumed to be elastic, perfectly plastic with
E = 68 GPa, rys ¼ 130 MPa, qs ¼ 2700 kg=m3 and t ¼ 0:3, where E
and t are the Young’s modulus and Poisson ratio of the base mate- (a) High-velocity deformation mode (b) Low-velocity deformation mode
rial, respectively.
Fig. 3. Deformation modes of honeycombs: (a) high-velocity deformation mode;
In the finite element simulations the hexagonal honeycomb
(b) low-velocity deformation mode.
block consists of enough cells (21 cells in the x-direction and 13
cells in the y-direction) to ensure the numerical results indepen-
dent on the cell number. Each cell-wall is meshed into 16 shell ele- Fig. 3(a). In the low-velocity deformation mode, cells collapse
ments with one layer elements along the out-of-plane direction along the ‘‘V’’ shape localization band layer by layer, forming a
(the z-direction). The out-of-plane displacement of the nodes with- stepped front of the densified region, as shown in Fig. 3(b). The de-
in the x–y plane (i.e. z = 0) is constrained so as to prevent the spec- tails on the spreading process of the deformation modes and the
imen from out-of-plane bulking. A single self-contact is defined on dependence of the deformation modes on the impact velocity
the honeycomb model, and a surface-to-surface contact is applied and the cell-wall angles were extensively described in another pa-
between the honeycomb specimen and the two rigid plates respec- per by the authors (Hu et al., 2013).
tively without considering the contact friction. In the simulations
the hexagonal honeycomb block is placed on a fixed rigid base
2.2. Cells’ collapse process under LVDM
(the lower plate) at one end and crushed along the x-direction by
a rigid plate (the upper plate) with a constant crushing velocity V
Under 10 m/s impact, the regular hexagonal honeycomb with
at the other end, as shown in Fig. 2.
b = 30° exhibits the low-velocity deformation mode. The strain
The honeycombs exhibit distinct deformation modes under the
and the node velocity of the cells on the third column within the
x-directional crushing, which are classified into two kinds: the
honeycomb are tracked and shown in Figs. 4 and 5, respectively.
high-velocity deformation mode (HVDM) and the low-velocity
In the figures, the horizontal axial denotes the compressed dis-
deformation mode (LVDM) (Hu et al., 2013). Under the high veloc-
placement of the honeycomb normalized by the cell height
ity impact, a localized crushing band is initiated at the loading edge
2l cos b so that the value of the ‘normalized displacement’ is di-
perpendicular to the impact direction and continues to propagate
rectly related to the total number of collapsed rows in the block.
layer by layer to the supporting end. The cell walls are packed clo-
‘Cell stain’ shown in Fig. 4 is defined from the relative shrink of
sely along the ripple front of the densified region, as shown in
the distance between the upper cell wall and the lower one of a
cell. The ‘node velocity’ in Fig. 5 means the velocity of the upper-
right vertex of the cell (see Fig. 1) and normalized with respect
Table 1 to the impact velocity V. The numbers beside the curves in both
Thickness of the cell walls for various honeycombs.
Figs. 4 and 5 indicate the number of the row where the cell lies
b (°) h (mm) h/l on, as marked in the honeycomb block in Fig. 4.
15 0.324 0.0810 Fig. 4 demonstrates the periodic characteristic of the cells’ col-
30 0.346 0.0865 lapse process. It is seen that the cell on the third row collapses first,
45 0.322 0.0805 whose strain increases rapidly, implying that the cell becomes
60 0.249 0.0623
unstable and the block’s deformation concentrates in this row,
75 0.136 0.0340
whilst the strains of the cells in other rows remain almost zero.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
4 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx
1
0.8
3 4 5 6
Cell strain
0.6
0.4
0.2
0 16
0 2
Fig. 4. Variation of cell strain with honeycomb’s displacement (b = 30°, V = 10 m/s).
1.2
1
0.8
Node velocity
5 7 9 10
0.6
6 8
0.4
0.2
0
-0.2
0 2 4 6 8 10 12 14 16 18 20 22 24
Normalized displacement
Fig. 5. Variation of node velocity with honeycomb’s displacement (b = 30°,
V = 10 m/s).
Once the cell strain exceeds 0.81, which means that the cell has al-
most been densified, most of the deformation instability shifts onto
the cell in the next row, while the strain of the cell on the third row
increases slowly. Thus the cells in the honeycomb block collapse
row by row from the impact end to the 10th row. Then the collapse
shifts to the cells near the supporting end of the block and the cells
collapse progressively in a repeated process until all the rows are
densified.
Fig. 5 indicates that the velocity of the cell’s node increases rap-
idly from zero during the cell’s collapse period and vibrates around
the crushing velocity after the end of the collapse period, indicating
that the cell will move down with the striker in the same velocity
after it is compressed to densification.
A repeated collapse process of the cells is exhibited in Fig. 6,
where t ¼ ttt 0
is the relative time in a collapse period of cells. t
f t 0
is the crushing time of the honeycomb block; t 0 and t f are the time
instants at the beginning and the end, respectively, of the cell’s col-
lapse period. Thus t ¼ 0 and t ¼ 1 denote the beginning and the
end of the collapse period, respectively. It is shown in Fig. 6 that
the cells lying in the same V-shape layer collapse synchronously
in a similar way. And considering the periodic characteristic of
the cells’ collapse process, most of the cells in the honeycomb
can be considered to experience the same collapse process, but
their collapse will not always take place simultaneously because
of the periodical collapse layer by layer. Therefore, a typical col-
lapsing cell will be employed to deduce the mechanical properties
of the honeycomb.
A typical cluster of cells during the collapse process, as marked
within the rectangles shown in Fig. 6, is focused on and exhibited
in Fig. 7. At the beginning of the collapse period, as shown in
Fig. 7(a), cell walls AB, BC and CK jointed the densified region
and moved downward at velocity V, which is equal to the striker’s
Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
1
18 17
2
3
10 4
7 8 9 19 16 5
6
7
8
9
10
12
14
4 6 8 10 12 14 16 18 20 18
Normalized displacement 20
velocity. Walls BD and KE have rotated almost to the horizontal
direction, whilst walls EF, FG, GH, HJ, IJ and EI almost remain their
original positions as sides of a hexagon. During the collapse pro-
cess, walls BD and KE rotate about points B and K respectively
(see Fig. 7(b)), until contact with AB and CK (see Fig. 7(c)), indicat-
ing the densification of the cell BDFEKC. During the whole collapse
process wall EF remains in the horizontal direction, while walls GH
and HJ almost hold still. The final state of cell EFGHJI shown in
Fig. 7(c) is similar to the initial state of cell BDFEKC shown in
Fig. 7(a). Thus, a period of the cells’ collapse process is completed,
and thereafter a new period starts.
Because of the repeated and periodical process of the cells’ col-
lapse, the final state of a collapse period is exactly identical to the
initial state of the next period. Therefore, for the initial state and
the final state of a collapse period, certain geometric and kinematic
conditions should be satisfied, as will be discussed in the following
sections.
3. Theoretical analysis
3.1. Representative block and assumptions
In view of the periodic characteristics of hexagonal honeycombs
both in the geometric configuration and in the cells’ collapse pro-
cess, a representative block across two hexagonal cells with six cell
walls is employed, as shown in Fig. 8. By periodically placing this
representative block along both x1 and x2 directions, the whole
honeycomb block can be constructed.
During a period of the repeated collapse process associated with
the propagation of the localization bands, the deformed configura-
tions of the representative block are shown in Fig. 9, which corre-
spond to the deformation configurations of cells in the numerical
simulation shown in Fig. 7. In Fig. 9(a), i.e., at the beginning instant
of the collapse period of the representative block, the cell walls DF,
EF and FG remain their respective original positions as the sides of
the hexagon without deformation or rotation, and the wall BD is in
the horizontal direction. In Fig. 9, r1 and r2 are the crushing stress
applied on the top of the representative block and the supporting
stress applied from its bottom, respectively. Since the densified re-
gion of the honeycomb moves at a constant velocity V, r1 is equal
to the contact stress between the striker and the honeycomb block,
which is exactly the crushing stress of the honeycomb that we are
interested in. r2 is equal to the supporting stress on the honey-
comb provided by the lower rigid plate because the un-collapsed
cells below the representative block are almost under static
equilibrium.
The displacements of the nodes as marked in both Fig. 7 and
Fig. 9 are tracked in the numerical simulations during the cell’s col-
lapse process as shown in Fig. 10, where the nodes’ displacements
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 5

(a) t = 0 (b) t = .021

(c) t = 0.43 (d) t = 0.65

(e) t = 0.88 (f) t = 1.09

Fig. 6. Repeated collapse process of cells in the numerical simulation: (a) t ¼ 0; (b) t ¼ 0:21; (c) t ¼ 0:43; (d) t ¼ 0:65; (e) t ¼ 0:88; (f) t ¼ 1:09.

A A

C D
C B D
A
B
E F C B D
K E F
K I G
I G E F G
J H J
J H K I H

(a) t = 0 (b) t = 0.43 (c) t = 1.09

Fig. 7. A typical period of the repeated collapse process of cells: (a) t ¼ 0; (b) t ¼ 0:43; (c) t ¼ 1:09.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
6 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

σ1 Besides, the length of the honeycomb’ cell walls is assumed to

remain constant l during the collapse process, and all of the plastic
hinges are assumed to take place at the cells’ vertexes. Thus, the
A work done by the crushing force is transferred into two parts:
the kinetic energy gained by the cell walls and the energy dissi-
β pated by the plastic hinges located at the cells’ vertexes.

B 3.2. Kinematic analysis and crushing strength

C
At the beginning instant of a collapse period of the representa-
tive block, t ¼ 0, as shown in Fig. 9(a), cell walls DF, EF and FG hold
still and remain their original positions as sides of the hexagon
x
Ho
without any deformation or rotation. Thus the kinetic energy and
D
the momentum of the three cell walls are all equal to zero at this
instant, i.e.,
F y T 0DF ¼ T 0EF ¼ T 0FG ¼ 0 ð6Þ
E
px0DF
1
¼ px0EF
1
¼ px0FG
1
¼0 ð7Þ
where T0 and px01
denote the initial kinetic energy and the initial
momentum in the x1 direction during a collapse period,
G respectively.
In Fig. 9(a), walls BC and BD lie in the same line paralleling to
σ2 the horizontal direction. Walls AB and BC have just joined the den-
sified region and become the front of the densified region. They re-
L
main their original configurations as sides of the original hexagon,
while have a translation from their original position. Since walls AB
Fig. 8. A representative block of honeycombs. and BC will move down with a constant velocity V during the cell’s
collapse period as shown in Fig. 9, their kinetic energy and momen-
tum will keep the same at the initial time and the final time of the
are normalized by the cell height 2l cos b, and the abscissa is the collapse period, respectively, i.e.,
cell’s collapse time normalized by the time of one collapsed period
1
of cells, 2l cos b=V. It is shown in Fig. 10 that the displacement T 0AB ¼ T 0BC ¼ T fAB ¼ T fBC ¼ q bhlV 2 ð8Þ
curves of the nodes A, B and C are overlapped together and linearly 2 s
related to the cell’s collapse time during the whole collapse period
px0AB
1
¼ px0BC
1
¼ pxfAB
1
¼ pxfBC
1
¼ qs hblV ð9Þ
since the cell walls AB and BC are part of the front of the densified
region in the honeycomb block. Thus, where Tf and pxf 1
denote the final kinetic energy and the final
~ momentum in the x1 direction during the collapse period,
VA ¼ ~
VB ¼ ~
VC ¼ V ð4Þ
respectively.
where ~ V A~
V B and ~
V C are the velocity of nodes A, B and C, respectively. Fig. 9(c) sketches the final configuration of the representative
Fig. 10 shows that the nodes of D, E and F almost have the same block at the end of a collapse period, which also represents the
displacement during the whole cell’s collapse period, so that the beginning of the next period. At this moment, cell walls BD, DF
motion of walls DF and EF can be assumed to be a translation; that and EF have formed the new stepped front and become a part of
is, the two walls are rigidly connected and translate together with- the densified region. Hence, all these walls have gained the same
in the plane without any rotation, while nodes D, E and F have the velocity V as the striker. Thus
same velocity during the cell’s collapse period, i.e., ~ VD ¼ ~VE ¼ ~
VF.
1
Thus, DF and EF always remain their original directions until they T fBD ¼ T fDF ¼ T fEF ¼ q bhlV 2 ð10Þ
2 s
become the new densification fronts, so the densification fronts are
parallel to the walls of DF and EF, i.e., BC//EF and AB//DF. Cell walls
pxfBD
1
¼ pxfDF
1
¼ pxfEF
1
¼ qs hblV ð11Þ
BC and EF are in the horizontal direction, while AB and DF have an
angle of b with the vertical direction during the whole collapse In Fig. 9(c), wall FG has rotated about node G to the horizontal
process as shown in Fig. 9. Therefore, at the initial instant of the direction at the end of the collapse period, which is the same as the
cell’s collapse process, t ¼ 0, the representative block possesses initial state of BD shown in Fig. 9(a), because of the periodicity in
the original width L0 ¼ lð1 þ sin bÞ and the original height both the geometry and the collapse process of the honeycomb.
H0 ¼ 3l cos b. Hence, the following relationships should be held:
It can be seen from Fig. 10 that the displacement of node G is
T 0BD ¼ T fFG ð12Þ
very small (less than 0.05) during the whole collapse period of
the cell, so node G is assumed to hold still and wall FG rotates
px0BD
1
¼ pxfFG
1
ð13Þ
about node G during the cell’s collapse process until it lies in the
horizontal direction, as shown in Fig. 9(c), which marks the end The final height of the representative block in Fig. 9(c) is now
of the collapse period and the beginning of the next period. Hence, (
during the collapse process, l cos b þ h ðb 6 p6 Þ
Hf ¼ l cos b þ maxðh; 2h sin bÞ ¼  
~ l cos b þ 2h sin b b > p6
VD ¼ ~
VE ¼ ~
V F ? FG ð5Þ
ð14Þ

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 7

σ1

A σ1
β VA
A

C B D β VA
D
β

Ho
VC VB B
C VD
F β
VC
VB F

H
E E

VE VF
G
G

σ2 σ2
Lo Lo

(a) t = 0 (b) 0 < t < 1

σ1

A(D)
β
Hf

VA (VD)
C(E) B(F) G

VC (VE) σ2
VB (VF)
Lo

(c) t = 1
Fig. 9. Configurations of the representative block during a collapse process of cells: (a) t ¼ 0; (b) 0 < t < 1; (c) t ¼ 1.

( 2
2bl r 1 ðcosb  2lh Þð1 þ sinbÞ ðb 6 p6Þ
W ¼ ðH0  Hf Þr
 1 bL0 ¼
2    
1 2bl r 1 cosb  l sinb ð1 þ sinbÞ b > p6
h

R tf ð15Þ
C
0.8 A r1 dt R1
where r  1 ¼ t t ¼ 0 r1 dt is the average crushing stress of the
t0
Normalized displacement

B B f 0
A honeycomb over a collapse period, which is called as the honey-
0.6 C comb’s crushing strength. This work is absorbed by the representa-
D tive block and is ultimately transformed into two parts: the energy
E F
E dissipated by the plastic hinges, Ep, and the increase of the kinetic
0.4
energy of the representative block, DT, i.e.,
F
D
0.2 G W ¼ Ep þ DT ð16Þ

G By considering a complete collapse period of the representative

0 block, the initial and the final kinetic energy of the representative
0 0.2 0.4 0.6 0.8 1 1.2 block are
Normalized time
T 0 ¼ T 0AB þ T 0BC þ T 0BD þ T 0DF þ T 0EF þ T 0FG ð17Þ
Fig. 10. Displacement of the cell’s vertexes during a cell’s collapse process.
and
T f ¼ T fAB þ T fBC þ T fBD þ T fDF þ T fEF þ T fFG ð18Þ
With regard to the representative block shown in Fig. 9, the respectively. By combining Eqs. (6), (8), (10), (12), (17), and (18), the
work done by the extra forces acted on the representative block increase in the kinetic energy of the representative block, DT, is
over the entire collapse period, W, is found to be

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
8 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

3 2 2
DT ¼ T f  T 0 ¼ qs bhlV ð19Þ 15°simulation
2
15°theory
During a complete collapse period from t ¼ 0 (Fig. 9(a)) to t ¼ 1 1.5

Stress(MPa)
(Fig. 9(c)), the total energy dissipated by the plastic hinges within
the representative block is 1
p  1
2
Ep ¼ 4Mp  b ¼ bh rys ðp  2bÞ ð20Þ 0.5
2 2
bh2 r
where M p ¼ 4 ys is the fully plastic bending moment of the cell 0
walls. 0 0.2 0.4 0.6 0.8 1
Finally, by substituting Eqs. (15), (19), and (20) into Eq. (16), the Strain
average crushing stress of the honeycomb over a collapse period, (a) β = 15 o
r 1 , is expressed as
8 1.5
>
2
rys ðhlÞ ðp2bÞþ3ðhlÞqs V 2   30°simulation
>
< 4ðcos b 6 p6
b h Þð1þsin bÞ 1.2
r 1 ¼ 2l
ð21Þ 30°theory
>
>
2
r ðhÞ ðp2bÞþ3ðhlÞqs V 2  

Stress(MPa)
: ys l h b> p
4ðcos b l sin bÞð1þsin bÞ 6 0.9

which provides an analytical expression of the crushing strength of 0.6

hexagonal honeycombs under lower-velocity impact in terms of the
cell-wall angle, the relative thickness of cells, the mechanical prop- 0.3
erties of the base material and the crushing velocity. As for a regular
0
hexagonal honeycomb (b ¼ p6), Eq. (21) is reduced to
0 0.2 0.4 0.6 0.8 1
 2 h
2p h q Strain
r 1 ¼ pffiffiffi  rys þ plffiffiffi s  V 2 ð22Þ (b) β = 30 o
9 3  hl l 3 3  hl

Similar to Eq. (3), the expression of the honeycomb’s crushing 45°simulation 45°theory
0.6
strength under low-velocity impact consists of two terms: the first 60°simulation 60°theory
term is related to the plastic collapse pattern of the cells, and the 0.5
75°simulation 75°theory
second term is resulted by the inertial effect.
Stress(MPa)

0.4
Fig. 11 compares the theoretical predictions of the crushing
0.3 β =45
strength obtained from Eq. (21) and the stress–strain curves ob-
tained from the numerical simulations under the crushing velocity 0.2
β =60
of 10 m/s, which verifies that Eq. (21) can well predict the ‘‘plateau 0.1
stage’’ of the honeycombs’ stress–strain curve under low-velocity β =75
impact. 0
The variation of the honeycombs’ crushing strength r  1 with the 0 0.2 0.4 0.6 0.8 1
Strain
crushing velocity and the cell-wall angle can be obtained from both
(c) β = 45 , β = 60 , β = 75o
o o
Eq. (21) and the numerical results by averaging the stress in the
‘‘plateau stage’’ of the stress–strain curve, as depicted in Fig. 12. Fig. 11. Comparison of the theoretical predictions by Eq. (21) with the numerical
Both the theoretical predictions and simulation results show that results (V = 10 m/s).
the crushing strength of the honeycombs increases with the crush-
ing velocity, but decreases with the increase of the cell-wall angle
while the relative density of the honeycombs remaining as con-
stant. Under 10 m/s impact, the crushing strength of the honey-
comb with the cell-wall angle 75° is diminished to only 10% of
that of a regular honeycomb. It reveals that the cell configuration
does have significant influence on the crushing strength of the
honeycombs under low-velocity impact, and this offers a guideline theoretical prediction
for the design of honeycombs’ cell-configuration. 1 numerical data
Moreover, it is shown in Fig. 12 that the theoretical prediction 45° β=15°
of Eq. (21) has a good agreement with the numerical results in 0.8 β=30°
the cases of cell-wall angle larger than 30°, while it underestimates β=45°
σ1 (MPa)

the honeycombs’ crushing strength when b P 30 . The deviation 0.6

β=60°
reaches 23% for the cases of b ¼ 15 and 20% for the cases of
b ¼ 30 . This is attributed to the assumption that all the plastic 0.4 β=75°
hinges take place at the cells’ vertexes, which neglects some plastic
hinges possibly forming within the cell walls. This issue will be 0.2
extensively discussed in Section 4.1. The error caused by this
assumption will further affect the validity of the prediction on 0
0 10 20 30 40 50 60
the honeycomb’s supporting stress as discussed in the following
Impact velocity (m/s)
section.
In our previous work (Hu and Yu, 2010), an analytical formula Fig. 12. Variation of honeycombs’ crushing strength with crushing velocity and
on the dynamic crushing strength of the regular hexagonal honey- cell-wall angle.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 9

8 0.8
LVDM HVDM 10m/s simulation
10m/s theory
Crushing strength (MPa)

6
critical velocity 0.6 30m/s simulation
Eq.(3) 30m/s theory

σ2(MPa)
4 0.4

2
0.2
Eq. (22)
0
0 50 100 150 0
V(m/s) 0 0.2 0.4
0.6 0.8 1
Strain
Fig. 13. Comparison of the honeycombs’ crushing strength between theoretical (a) Honeycombs crushed with the velocity of 10m/s and 30m/s respectively
predictions and numerical results (b ¼ 30 ).

comb is derived as Eq. (3), which is based on the cells’ collapse 0.8 20m/s simulation
mechanism under high-velocity deformation mode. For the regular 20m/s theory
hexagonal honeycombs, Fig. 13 compares the predictions of the 40m/s simulation
0.6 40m/s theory
crushing strength by Eqs. (3) and (22) with the numerical results
(marked by the red dots). It is denoted that when the crushing

σ2(MPa)
velocity is lower than a certain value (shown as the dotted vertical 0.4
line in Fig. 13), the result predicted by the LVDM theory, i.e., Eq.
(22), is much closer to the numerical dots than Eq. (3) deduced
based on the HVDM. Once the crushing velocity exceeds this criti- 0.2
cal value, the HVDM theory of Eq. (3) will be more valid. That is
what we expect. The critical velocity denotes the boundary for 0.0
the application of the LVDM and HVDM theories, which will be dis- 0 0.2 0.4 0.6 0.8 1
cussed in Section 3.3. Strain
Anyhow, by combining Eq. (3) and Eq. (22), the crushing (b) Honeycombs crushed with the velocity of 20m/s and 40m/s respectively
strength of hexagonal honeycombs is analytically predicted over
Fig. 14. Comparison of the supporting stress between the theoretical predictions by
a large variety of crushing velocities. It is clear in Fig. 13 that the
Eq. (25) with the numerical results (b ¼ 45 ).
crushing strength curve displays a quicker increase with the im-
pact velocity under the HVDM compared with that under the
LVDM, implying that the energy absorption capacity of the honey-  2 h
comb block is greatly enhanced under the high-velocity impact. 2p h q
r 2 ¼ pffiffiffi  rys  plffiffiffi s  V 2 ð26Þ
h
9 l 3 3 3  hl
l
3.3. Supporting stress
Fig. 14 is the comparison between Eq. (25) and the curves of
By reviewing the collapse period of the cells as shown in Fig. 9, supporting stress obtained from the numerical simulations for
and based on the conservation of linear momentum, the difference the honeycombs with b ¼ 45 , where the abscissa presents the
between the impulses of r1 and r2 in the time interval is equal to compression strain of the honeycomb block. It is seen that the
the change of the momentum in the x1 direction of the six cell numerical curves of the supporting stress show notable vibration,
walls involved in the representative block; that is, especially for the cases under the higher impact velocity. The big
Z tf   drops in the response on the supporting end for the three cases
bLo ðr1  r2 Þdt ¼ pxfAB
1
þ pxfBC
1
þ pxfBD
1
þ pxfDF
1
þ pxfEF
1
þ pxfFG
1
(20 m/s, 30 m/s and 40 m/s) are resulted from the reflection of
t0
 1  the elastic wave in the honeycomb. The honeycomb block is put
 px0AB þ px0BC
1
þ px0BD
1
þ px0DF
1
þ px0EF
1
þ px0FG
1
on the lower rigid plate without fixation, so that the lower plate
ð23Þ can only provide compressive force and cannot sustain tensile
force to the honeycomb. When the compressive elastic wave
Combined with Eqs. (7), (9), (11), and (13), Eq. (23) is simplified
reaches the supporting end, a part if it will be reflected, resulting
as
in an upward particle velocity near the boundary, so as to cause
ðH0  Hf Þ the ‘‘unloading’’ on the supporting end. Nevertheless, Fig. 14 shows
bLo ðr
1  r
 2Þ ¼ 3qs hblV ð24Þ
R tfV that the theoretical expression of Eq. (25) can provide a good pre-
r2 dt R1
where, r 2 ¼ t t ¼ 0 r2 dt is the average supporting stress of the
t0
diction on the average of the supporting stress. The average of the
f 0
honeycomb over a collapse period. supporting stress will help to estimate the impulse that the pro-
By substituting Eq. (21) into Eq. (24), the average supporting tected object is likely to bear when the honeycomb is used in the
stress acting on the honeycomb by the fixed rigid plate r  2 is ob- protective structures.
tained as, According to Eq. (25), the variation of r  2 with the crushing
8 h2 velocity as well as the cell-wall angle can be depicted in Fig. 15.
> rys ð Þ ðp2bÞ3ðhÞqs V 2  
>
< 4ðl cos b h Þð1þsinl bÞ b 6 p6 The numerical simulation data is also shown. Fig. 15 indicates that
2l
r 2 ¼  
ð25Þ Eq. (25) can well predict the average supporting stress of the hon-
> h 2 2
: rys ð l Þ ðph2bÞ3ð l Þqs V
h
> b > p6 eycomb when the cell-wall angle is larger than 30°. Both Eq. (25)
4ðcos b l sin bÞð1þsin bÞ
and the numerical simulations reveal that the average supporting
For a regular hexagonal honeycomb (b ¼ p6 ), Eq. (25) is reduced stress of the honeycomb decreases with the increase of the cell-
to wall angle as well as the increase of the crushing velocity, which

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
10 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

theoretical prediction For a regular hexagonal honeycomb (b ¼ p6 ), Eq. (28) is reduced

0.7
to
Average supporting stress (MPa)

numerical data
0.6 pffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffi
2p h rys
0.5 Vc ¼   ð29Þ
3 l qs
0.4 β=30°
Fig. 16 depicts a deformation mode map of the honeycombs
0.3 β=15°
β=45° based on the numerical simulations. The red line in the figure is ob-
0.2 tained according to Eq. (28). It is evident that Eq. (28) well predicts
β=60°
the critical velocity between LVDM and HVDM, except for the hon-
0.1 β=75° eycombs with small cell-wall angle, e.g. 15° and 30°, for which the
0 theoretical values are higher than the numerical results.
0 10 20 30 40 50 60
Actually during the crushing process the stress at the support-
V (m/s)
ing end of the honeycomb block displays a notable oscillation in-
Fig. 15. Variation of the honeycombs’ average supporting stress with the crushing stead of a steady plateau stress. The oscillation is more severe for
velocity and the cell-wall angle. the honeycombs with cell-wall angle of b ¼ 15 and b ¼ 30 , as
shown in Fig. 17. As the result of the oscillation of the supporting
stress, the transition of the honeycombs’ deformation mode occurs
is different from the varying tendency of the crushing strength r 1 at a lower critical velocity under the condition of r2 P 0.
as shown in Fig. 12. In the previous reports (Zou et al., 2009; Hu
and Yu, 2010), the crushing velocity is believed to have little effect
4. Discussion
on the supporting stress of the honeycomb. However, the present
study discovers that the honeycombs’ average supporting stress in-
4.1. Cells’ length during the collapse process
deed depends on the crushing velocity when the honeycomb is
crushed at a lower velocity (i.e., under LVDM), which decreases
In Sections 3.2 and 3.3, the theoretical predictions of r  1 and r
2
with the crushing velocity by a square law as described in Eq.
are lower than the numerical results in the cases of cell-wall angle
(25). It implies that when the honeycomb is used in protective
less than 30°. To explore the reason of the low prediction, the dis-
structures, within a certain range of the impact velocity, the pro-
tance between the vertexes of the three representative cell walls
tected object is likely to bear a higher impulse under the lower-
DF, EF and FG as marked in Fig. 7 is tracked during two collapse
speed crushing compared with that under a higher-velocity im-
periods of the cells, as shown in Fig. 18, in which the vertexes’ dis-
pact. More attention should be paid to this issue in honeycombs’
tance of the cell walls l0 is normalized by their original length l, and
applications.
the horizontal coordinate is the displacement of the impact plate in
the two collapse periods of cells normalized by the cell height
3.4. Critical velocity between LVDM and HVDM
2l cos b. After two periods of cell collapse, all of the three cell walls
become a part of the densified region. It is noted from Fig. 18 that
In the last section, the supporting stress is deduced as Eq. (25),
the vertexes’ distance almost remains unchanged during the cells’
which means the average stress of the honeycomb resisting against
collapse process, except for that of wall FG in the cases of b ¼ 15
the supporting rigid plate. The value of r2 should always not be
and b ¼ 30 , in which the shortening of the vertexes’ distance of FG
negative because the supporting plate cannot produce a tensile
is about 16% and 8.3% respectively after FG jointing into the densi-
force on the honeycomb. Thus, combining with r  2 P 0 Eq. (25)
fied region.
leads to
By zooming in the cell wall FG, as shown in Fig. 19, it is found
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p  2b h rys that the shortening of the vertexes’ distance of FG is resulted by
V6   ð27Þ the bending of the cell wall with two plastic hinges forming near
3 l qs
the vertexes. In the theoretical analysis above, the energy dissi-
which gives a limit condition for the application of Eqs. (21) and pated by these two plastic hinges is not considered, which finally
(25) obtained based on the LVDM. Thus the critical velocity be- results in the low prediction on both the crushing strength and
tween the LVDM and the HVDM is obtained as the supporting stress of the honeycombs in the cases of b ¼ 15
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi and b ¼ 30 .
p  2b h rys
Vc ¼   ð28Þ
3 l qs 1.5
15° 30° 45°
80
LVDM
1.2 60° 75° β =30o
Supporting stress (MPa).

HVDM
60 0.9
β =15o
V (m/s)

Eq. (28)
40 0.6
β =45o
20 0.3
β =60o
β =75o
0 0
0 15 30 45 60 75 0 0.2 0.4 0.6 0.8 1
° Strain

Fig. 16. Deformation mode map of honeycombs. Fig. 17. Variation of the honeycombs’ supporting stress with its strain (V = 1 m/s).

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 11

1.2 1.2

1 1

l '/l
l '/l
0.8 0.8
DF DF
0.6 EF 0.6 EF
FG FG
0.4 0.4
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Normalized displacement Normalized displacement
(a) β = 15o (b) β = 30 o
1.2
1.2

1 1

l '/l
0.8
l '/l

0.8
DF DF
0.6 EF 0.6 EF
FG FG
0.4 0.4
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Normalized displacement Normalized displacement
(c) β = 45o (d) β = 60o

1.2

0.8
l '/l

DF
0.6 EF
FG
0.4
0 0.5 1 1.5 2
Normalized displacement
(e) β = 75o

Fig. 18. Vertexes’ distance of the representative cell walls during two collapse period of cells: (a) b ¼ 15o ; (b) b ¼ 30 ; (c) b ¼ 45 ; (d) b ¼ 60 ; (e) b ¼ 75 .

sion of the shear and the membrane energy maybe is another rea-
son for the low prediction on both the crushing strength and the
supporting stress of the honeycombs in the cases of b > 45 . The
G effect of this exclusion will be notable for the honeycomb with lar-
ger cell-wall thickness ratio, which needs to be explored in the fur-
F ther study.

4.2. Crushing strength VS. supporting stress

Fig. 19. Zooming in the cell wall FG (b ¼ 30 ).
On the other hand, only the bending effect was considered in
calculating the strain energy of the cell walls in the above analysis,
with ignoring the shear and the membrane energy, which is rea-
sonable when the cell walls are sufficiently slender. It is shown
in Table 1 that the cell-wall thickness ratios of the honeycombs
are all less than 1/10. For the honeycombs with the cell-wall angle
less than 45°, the cell walls are very slender. For the honeycombs
with the cell-wall angle larger than 45°, the cell-wall thickness ra-
tio is about 0.08, in which the bending energy is still the dominant
factor compared with the shear and the membrane ones, though
the later also has contribution on the total strain energy. The exclu-
A comparing of Eq. (21) and Eq. (25) shows that the difference
of the two equations is only at the sign of the inertial item
3qs ðhlÞV 2 in the numerator. In the above analysis, the lower rigid
plate (see Fig. 2) is regarded to be fixed and the upper rigid plate
is movable, i.e., the reference frame is fixed on the lower plate.
Now let’s consider again the same problem described in Fig. 2 by
setting the reference frame on the upper plate, which means the
upper plate is regarded as still while the lower plate and the hon-
eycomb move together towards to the upper plate with a velocity
of V, as shown in Fig. 20. Thus the initial velocity of the honey-
comb’s cell walls is equal to V. Once the cell walls join in the den-
sified region, their velocity becomes to 0.
The stresses suffered by the honeycomb from the upper plate
and the lower plate are still called as r1 and r2 , respectively. It is
clear that the stress and the deformation mode of the honeycomb
will not be affected by only changing the reference frame. Review-

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
12 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

2h
Upper rigid plate
h h

x
h
V h

2h
Honeycomb block y

V Fig. 21. Sketch of the double-thickness honeycomb.

Lower rigid plate

walls possessing the same thickness as discussed in the above sec-
Fig. 20. New description on the model shown in Fig. 2 with the reference frame on
tions is called as ‘‘single-thickness honeycomb’’. Following the
the upper plate. method used in Section 3 (the details is shown in Appendix A),
the crushing strength and the supporting stress of the double-
thickness honeycomb are expressed as
ing the collapse process of the cells as shown in Fig. 7 and Fig. 9, h2 
only r2 does work over the collapse period in the new reference rys ðp  2bÞ þ 4qs hl V 2
r 01 ¼ l  ð35Þ
frame. Thus, the work done by the extra forces acting on the repre- 4 cos b  hl ð1 þ sin bÞ
sentative block over a collapse period is rewritten as
( 2     and
2bl r 2 cos b  2lh ð1 þ sin bÞ b 6 p6
W ¼ ðH0  Hf Þr
 2 bL0 ¼     h2 
2
2bl r 2 cos b  hl sin b ð1 þ sin bÞ b > p6 rys ðp  2bÞ  4qs hl V 2
r 02 ¼ l  ð36Þ
ð30Þ 4 cos b  hl ð1 þ sin bÞ

Within the representative block as shown in Fig. 9, the velocity respectively. Moreover, the critical velocity between the low-veloc-
of the cell walls DF, EF and FG are all equal to V at the beginning of ity deformation mode and the high-velocity deformation mode is
the collapse period, thus their kinetic energy is rewritten as obtained as
1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

T 0DF ¼ T 0EF ¼ T 0FG ¼ q bhlV 2 ð31Þ rys ðp  2bÞ hl
2 s V 0c ¼ ð37Þ
4qs
Cell walls AB and BC belong to the densified region over the
complete collapse period, while walls BD, DF and EF join in the which is smaller than that of the single-thickness honeycomb as gi-
densified region at the end of the collapse period. Thus, ven by Eq. (28).
T 0AB ¼ T 0BC ¼ T fAB ¼ T fBC ¼ 0 ð32Þ By comparing Eq. (35) with Eq. (21), it is obvious that the crush-
ing strength of the double-thickness honeycomb is higher than
T fBD ¼ T fDF ¼ T fEF ¼ 0 ð33Þ that of the single-thickness one, and the difference between them
is related with the crushing velocity. The ratio of r 01 to r
 1 achieves
By substituting Eqs. (12), (31), (32), and (33) into Eqs. (17) and 0
the upper limit with V ¼ V c and the lower limit with V ¼ 0 respec-
(18), the increase in the kinetic energy of the representative block, r 0
tively, thus the variation range of r 11 can be obtained, as denoted
DT, is rewritten as with the shadow consisting of the dashed lines in Fig. 22 for the
3 2 honeycombs with hl ¼ 20 1
if these two kinds of honeycomb have
DT ¼ T f  T 0 ¼  qs bhlV ð34Þ the same material parameters and geometry parameters as de-
2
scribed in Section 2.1 except for the thickness of the cell walls par-
Since the energy dissipated by the plastic hinges within the rep-
allel to the y direction.
resentative block will have no change in the new reference frame,
Eq. (20) remains valid. Thus by substituting Eqs. (20), (30), and (34)
into Eq. (16), the expression of r 2 again can be obtained as Eq. (25). 1.25
Hence, the term 3qs ðhlÞV 2 in Eq. (25) can be regarded as the inertial
1.174 1.178 1.168
effect, which is the essential reason resulting in the reduction of 1.160 1.152
the honeycomb’s supporting stress with the increase of crushing 1.15 V = Vc′
velocity.
σ 1′
σ1
4.3. Unequal thickness 1.05
V =0
Most of the metal honeycombs and Nomex honeycombs in 1.027 1.031 1.022 1.015 1.008
broad applications are extended from the equidistantly-glued me- 0.95
tal foil or Nomex foil, so the thickness of the cell walls parallel to 0 15 30 45 60 75 90
the y direction is double of that unparallel to the y direction, as β (°)
sketched in Fig. 21. Here, we call this kind of honeycomb as ‘‘dou-
r 0
ble-thickness honeycomb’’, and the honeycomb with all of the cell Fig. 22. Variation range of r 11 for the honeycombs with hl ¼ 20
1
.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx 13

Fig. 22 reveals that these two kinds of honeycomb have similar 3

crushing strength under very low crushing velocity, while the dou- 40m/s
ble-thickness honeycomb shows obviously higher crushing 2.5
30m/s

Absorbed energy (J)

strength than the single-thickness one when the crushing velocity 30m/s
2 20m/s
increases. However, the enhancement is no more than 20% under 40m/s
10m/s
the LVDM. In our previous work (Hu and Yu, 2010), the dynamic
1.5
crushing strength of the regular hexagonal honeycomb was ana-
20m/s
lyzed when the honeycomb deformed with the HVDM, under 1
which the crushing strength of the double-thickness honeycomb 10m/s
is about 1.3 times of that of the single-thickness honeycomb since 0.5
the dynamic crushing strength of the honeycomb is dominated by
the inertia effect. 0
0 0.2 0.4 0.6 0.8 1
Normalized displacement
4.4. Non-constant crushing velocity
Fig. 24. The absorbed energy by the honeycombs in the non-constant velocity
In the above study, the crushing velocity is assumed to be a con- cases.
stant during the whole crushing process of the honeycombs, which
is close to the reality when the impact mass is sufficient large com-
Table 2
pared with the honeycomb itself. However, in these cases, the ini-
Change of the honeycombs’ crushing strength in the non-constant velocity cases.
tial kinetic energy would not remain constant for different impact
velocities, so it is not clear to discuss the dependence of the energy ðVÞ0 m/s ðr
 1 Þ0 MPa ðVÞ0:92 m/s ðr
 1 Þ0:92 MPa Change of r
1%

absorption of a honeycomb block on the impact velocity. 40 0.5194 18.00 0.3368 35.16
With the initial kinetic energy of the drop hammer (i.e., the im- 30 0.4192 14.92 0.3223 23.13
pact plate) remaining as constant of 2.5 J, the honeycombs with the 20 0.3477 11.16 0.3082 11.34
10 0.3047 6.00 0.2956 3.00
cell wall angle of 45° were crushed numerically with the initial
velocity of 10 m/s, 20 m/s. 30 m/s and 40 m/s, respectively. The
variety of the velocity of the impact plate and the absorbed energy
by the honeycombs during the crushing process of the honey- 40
combs are plotted in Figs. 23 and 24, respectively, in which the ab-
scissa means the crushing displacement of the impact plate
Crushing load (N)

30
normalized by the initial height of the honeycomb. The absorbed
energy by the honeycomb block is obtained by integrating the
20
load–displacement curve of the honeycomb.
It is shown in Fig. 23 that the impact velocity put on the honey-
10
comb decreases notablely during the crushing process of the hon-
eycomb when the initial kinetic energy of the impact plate is not
large enough. The crushing strength of the honeycombs at the ini- 0
0 30 60 90 120
tial time of the impact and at the time before densification with the
Displacement (mm)
normalized displacement 0.92, can be estimated by Eq. (21), as
shown in Table 2, where the subscripts of 0 and 0.92 denote the (a) The case with the initial crushing velocity of 10m/s
normalized displacement of the honeycomb. It is shown that the
crushing strength has little change (only 3%) for the 10 m/s case, 40
implying that the crushing load of the honeycomb oscillates almost
around a constant value during the crushing process, as shown in 30
Crushing load (N)

Fig. 25(a). Therefore, the absorbed energy by the honeycomb is al-

most linearly proportional to the displacement for the 10 m/s case, 20
as shown in Fig. 24. With the increase of the initial crushing veloc-
ity, however, the decrease of the crushing load becomes more obvi-
10
ous during the impact process, as shown in both Table 2 and

0
50 0 30 60 90 120
40m/s Displacement (mm)
40 30m/s
Crushing velocity(m/s).

(b) The case with the initial crushing velocity of 40m/s

40m/s 20m/s
30 10m/s Fig. 25. Variety of the crushing load with the displacement in the non-constant
30m/s velocity cases: (a) The case with the initial crushing velocity of 10 m/s; (b) The case
with the initial crushing velocity of 40 m/s.
20 20m/s

10 10m/s
Fig. 25(b). For the case of 40 m/s, for instance, the honeycomb’s
0 crushing strength with the normalized displacement 0.92 is re-
0 0.2 0.4 0.6 0.8 1 duced to 65% of that at the initial time. Thus, the energy curve
Normalized displacement has a decreasing slope with the increasing crushing displacement,
which is observed in the cases of 30 m/s and 40 m/s, as shown in
Fig. 23. Variety of the crushing velocity in the non-constant velocity cases. Fig. 24.

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017
14 L.L. Hu, T.X. Yu / International Journal of Solids and Structures xxx (2013) xxx–xxx

It is shown in Fig. 24 that the honeycomb suffered from a higher px0BC

1
¼ pxfBC
1
¼ pxfEF
1
¼ 2qs hblV ðA:2Þ
initial crushing velocity can absorb more energy before densifica-
tion, although the input energy (the initial kinetic energy of the where h is the thickness of the cell walls unparallel to the y direc-
drop hammer) is the same in all the cases. It means that the energy tion, so the thickness of the cell walls parallel to the y direction is
absorption capacity of the honeycombs is indeed affected by the 2h. Thus Eq. (19) and Eq. (24) will become
impact velocity, which increases with the crushing velocity. DT ¼ T f  T 0 ¼ 2qs bhlV
2
ðA:3Þ

5. Summary and conclusion and

In the present work, the analytical models are established based
on the cells’ collapse mechanisms revealed from the numerical
simulations to study the mechanical properties of hexagonal hon-
eycombs under low-velocity crushing. An analytical expression of
the honeycombs’ crushing strength is obtained based on the en-
ergy method, which is in good agreement with the numerical sim-
ulation results. Thus combining with the dynamic predictions
under the high-velocity impact in our previous work (Hu and Yu,
2010), the crushing strength of the honeycombs can be analytically
predicted over wide range of crushing velocities. The results show
that both the crushing strength and the energy absorption capacity
of the honeycombs increase with the crushing velocity.
By employing different methods, the average of the supporting
stress of honeycombs can be deduced to the same expression as Eq.
(25). Both the analytical expression and the numerical results dis-
cover that the supporting stress of honeycombs does depend on
the impact velocity, which is different from the belief in literatures
(Zou et al., 2009; Hu and Yu, 2010). This issue should be paid more
attention in the honeycombs’ applications. However, this result is
valid only when the honeycomb deforms in the LVDM, since the
strain energy possibly also depend on the impact velocity under
the HVDM and therefore the methods employed in the present pa-
per will be not active. The dependences of both the crushing
strength and the supporting stress of the honeycomb on the impact
velocity are attributed to the inertia effect of the cell walls, since
the walls have to change their status with gaining a velocity of V
from still during the impact.
The critical velocity between the LVDM and the HVDM is de-
duced, which not only offers a boundary for the application of
the functions of the honeycomb’s crushing strength shown as
Eqs. (3) and (22), but also gives a limit condition for the application
of the expression of the supporting stress obtained in the present
paper.
Acknowledgements
The authors would like to thank the support from the National
Natural Science Foundation of China under Grant Nos. 11172335
and 10802100. The support from the Fundamental Research Funds
for the Central Universities No. 3161263 is also gratefully
acknowledged.
Appendix A. Mechanical properties of double-thickness
honeycombs
For the double-thickness honeycomb, the representative block
shown in Fig. 9 and Eqs. (6)–(13) are still tenable, except for the ki-
netic energy and the momentum of walls BC and EF, which are re-
written as
2
T 0BC ¼ T fBC ¼ T fEF ¼ qs bhlV
ðH0  Hf Þ
bLo ðr
1  r
 2Þ ¼ 4qs hblV ðA:4Þ
V
respectively.
The final height of the representative block in Fig. 9(c) is re-
placed by
Hf ¼ l cos b þ 2h ðA:5Þ
Thus, the work done by the extra forces acted on the represen-
tative block over a complete collapse period, W, is re-written as
 
h
 1 bL0 ¼ 2bl2 r
W ¼ ðH0  Hf Þr  1 cos b  ð1 þ sin bÞ ðA:6Þ
l
Then the crushing strength, the supporting stress, and the crit-
ical velocity of the double-thickness honeycomb are expressed as
Eqs. (35)–(37), respectively.
References
Chung, J., Waas, A.M., 2002. Compressive response of honeycombs under in-plane
uniaxial static and dynamic loading, Part1: Experiments. AIAA J. 40 (5), 966–
973.
Gibson, L.J., Ashby, M.F., 1997. Cellular Solids: Structure and Properties, second ed.
Cambridge University Press.
Honig, A., Strong, W.J., 2002a. In-plane dynamic crushing of honeycomb. Part I
Crush band initiation and wave trapping. Int. J. Mech. Sci. 44, 1665–1696.
Honig, A., Strong, W.J., 2002b. In-plane dynamic crushing of honeycomb. Part II:
Application to impact. Int .J. Mech. Sci. 44, 1697–1714.
Hou, B., Ono, A., Abdennadher, S., et al., 2011. Impact behavior of honeycombs under
combined shear-compression. Part I: Experiments. Int. J. Solids Struct. 48 (5),
687–697.
Hou, B., Zhao, H., Pattofatto, S., Liu, J.G., Li, Y.L., 2012. Inertia effects on the
progressive crushing of aluminium honeycombs under impact loading. Int. J.
Solids Struct. 49 (19–20), 2754–2762.
Hu, L.L., Yu, T.X., 2010. Dynamic crushing strength of hexagonal honeycombs. Int. J.
Impact Eng. 37, 467–474.
Hu, L.L., Yu, T.X., Gao, Z.Y., Huang, X.Q., 2008. The inhomogeneous deformation of
polycarbonate circular honeycombs under in-plane compression. Int. J. Mech.
Sci. 50, 1224–1236.
Hu, L.L., You, F.F., Yu, T.X., 2013. Effect of cell-wall angle on the in-plane crushing
behaviour of hexagonal honeycombs. Mater. Des. 46, 511–523.
Okumura, D., Ohno, N., Noguchi, H., 2002. Post-buckling analysis of elastic
honeycombs subject to in-plane biaxial compression. Int. J. Solids Struct. 39,
3487–3503.
Papka, S.D., Kyriakides, S., 1994. In-plane compressive response and crushing of
honeycombs. J. Mech. Phys. Solids 42, 1499–1532.
Papka, S.D., Kyriakides, S., 1998. In-plane crushing of a polycarbonate honeycomb.
Int. J. Solids Struct. 35, 239–267.
Reid, S.R., Peng, C., 1997. Dynamic uniaxial crushing of wood. Int. J. Impact Eng. 19,
531–570.
Ruan, D., Lu, G., Wang, B., Yu, T.X., 2003. In-plane dynamic crushing of honeycombs
– finite element study. Int. J. Impact Eng. 28, 161–182.
Yamashita, M., Gotoh, M., 2005. Impact behavior of honeycomb structures with
various cell specifications – numerical simulation and experiment. Int. J. Impact
Eng. 32, 618–630.
Zhu, H.X., Mills, N.J., 2000. The in-plane non-linear compression of regular
honeycombs. Int. J. Solids Struct. 37, 1931–1949.
Zou, Z., Reid, S.R., Tan, P.J., Li, S., Harrigan, J.J., 2009. Dynamic crushing of
honeycombs and features of shock fronts. Int. J. Impact Eng. 36, 165–176.
ðA:1Þ

Please cite this article in press as: Hu, L.L., Yu, T.X. Mechanical behavior of hexagonal honeycombs under low-velocity impact – theory and simulations. Int.
J. Solids Struct. (2013), http://dx.doi.org/10.1016/j.ijsolstr.2013.05.017