Accepted Manuscript

Deformation mode evolutional mechanism of honeycomb structure when un-

dergoing a shallow inclined load

Zhonggang Wang, Zhaijun Lu, Song Yao, Yiben Zhang, David Hui, Luciano
Feo

PII: S0263-8223(16)30098-8
DOI: http://dx.doi.org/10.1016/j.compstruct.2016.02.057
Reference: COST 7270

To appear in: Composite Structures

Please cite this article as: Wang, Z., Lu, Z., Yao, S., Zhang, Y., Hui, D., Feo, L., Deformation mode evolutional
mechanism of honeycomb structure when undergoing a shallow inclined load, Composite Structures (2016), doi:
http://dx.doi.org/10.1016/j.compstruct.2016.02.057

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Deformation mode evolutional mechanism of honeycomb
structure when undergoing a shallow inclined load
Zhonggang WANGa,b*, Zhaijun LUa,b, Song YAOa,b, Yiben Zhanga,b,
David HUIc, Luciano FEOd
a. Key laboratory of Traffic Safety on Track, Ministry of Education; Changsha, China.
b. School of Traffic & Transportation engineering ,Central South University, Changsha,
China.
c. Department of Mechanical Engineering, University of New Orleans, USA
d. Department of Civil Engineering, University of Salerno, 84084, Fisciano (SA), Italy
* wangzhonggang@yahoo.com

!Inclined loading conditions cannot be avoided and strongly influence the

Abstract!
mechanical response of honeycomb structures. In present study, deformation modes as
well as dynamic responses of hexagonal honeycomb structures undergoing non-ideal
oblique impact loadings have been investigated considering as significant parameters
of the load angle (ranging from 0 to 10°) and the impact velocity (ranging from 3 to 70
m/s). Some new deformation phenomena and response modes have been observed.
Evident influence of asynchronous loading as well as impact speed on the deformation
mode evolution has been determined. Corresponding mode classification maps were
constructed. The present study provides a significant advancement to the
comprehensive understanding of the dynamic response of honeycombs, which could be
used to develop more valuable guidelines for design purposes in cellular energy
absorbing devices.
Keywords: Honeycomb structure; inclined load; deformation mode map; dynamic
response

ͳ–”‘†—…–‹‘
Aluminum honeycomb structures have been widely used in kinds of industrial
fields, such as high-speed train, aerospace, automobile, due to their excellent
mechanical performance and high-energy absorption capacity [1,2]. In the past decades,
many investigations have been carried out extensively on its stiffness, performance and
mechanical behaviors by means of theoretical [3-4] and experimental [5-6]
methodologies. Recently, numerical simulations have become popular, by means of
which many findings have been achieved and presented in the literature (see, for
instance, Wang et al. [7], Aktay et al. [8], Sun et al. [9], Papka & Kyriakides [10], Fan
et al. [11], Liu et al. [12], Chen & Yan [13]). From these constructive works, the
honeycomb structure has been turned out to show completely different behaviors at
different loading directions. When loaded at out-of-plane impact directions, the
honeycomb structure cells collapse progressively, whereas, when loaded at in-plane
direction, three typical patterns of localization bands can be identified at different
impact velocities, which have been summarized in Cricrì’s experimental research [14],
Ruan’s numerical investigation [15], and Hu’s work [16]. These three typical patterns
of localization bands are called “X” pattern, “V” pattern and “I” pattern, respectively.
At low impact velocity, localized deformation appears in the form of ‘‘X’’ bands (‘‘X’’
mode), whereas, at high impact speed, vertical localized bands (‘‘I’’ mode) are
generated perpendicularly to the loading direction. A transitional ‘‘V’’ mode occurs at
moderate impact velocities. The localized band moves towards the loaded edge for
increasing values of the impact velocity.
However, such modes occur under ideal loading conditions, which never happen
in the real-life situations, especially in complex engineering applications. In fact,
honeycomb structures do not undergo ideal axial impacts, but oblique loading
situations often happen. The investigation of the mechanical behavior of honeycomb
composite structures under non-ideal loading conditions has been a popular research
topic in recent years. Some significant research works were performed by Reyes et al.
[17-19], Ahmad et al. [20], Zarei & Kroger [21], Greve et al. [22]. The oblique loading,
an unavoidable or even common situation, represents a major threat to safety. It may
cause mutation of dynamic response mode, loss of energy-absorption capacity, or even
worse, it leads to an unpredictable deformation process, especially for long tandem
honeycomb series as presented in Eskandarian & Marzougui’s research work [23].
These phenomena, which can be collectively referred to as “oblique effects” have
attracted a high attention in engineering applications. Keshavanarayana & Thotakuri
[24] investigated off-axis compression behavior of honeycomb core with different
off-axis angles. According to their investigations, the crush initiation load and crush
stress decrease as the off-axis angle increasing. A more extensive study of this problem
was carried out by Hong et al. [25]. In their works, detailed quasi-static and dynamic
out-of-plane uniaxial crush experiments under dominant inclined loads were introduced,
and empirical formulae of normalized normal crush strengths as functions of the impact
velocity as well as the orientation angle were derived. All these excellent works provide
a good insight into the behavior of honeycomb under oblique loading, but have a
limited predictive capability for analyzing the deformation mode evolution. As a matter
of fact, an effective analysis of cellular honeycomb materials subjected to compressive
loadings, because of the slenderness of their longitudinal walls should include the
influence of strong nonlinearities in both geometric and constitutive properties
occurring at different scales. In order to account for the effects of microstructural
properties of these materials on their non-linear mechanical behavior, non-linear
homogenized constitutive models are widely adopted (see [26-28]). More powerful
approaches able to capture the actual geometrical and/or material nonlinearities of such
heterogeneous structures are the so-called multiscale approaches, grouped in
semi-concurrent types (see [29-31]) and concurrent types (see [32-34]), whose
accuracy is the same as the fully microscopic models, which often are not pursed in
engineering practice due to the required large computational effort.
As is well known, a stable and orderly buckling process is preferable for
increasing the energy absorption, especially in aluminum honeycomb. Thus,
determining honeycomb’s performance and mechanical behavior under oblique loading
has become an urgent issue.
This paper focuses on the hexagonal aluminum honeycomb’s deformation mode
under oblique impact loading at different velocities ranging from 3 to 70 m/s and
different inclination angles ranging from 0 to 10°. The scopes of this paper are limited
to explicitly obtain the state mode evolution for honeycomb structures at oblique
loading conditions and to construct complete mode classification maps. The rest of the
paper is organized as follows. Section 2 establishes the detailed full-scale finite element
model. In addition, it reports the model validation conducted by means of comparisons
with experiments. Section 3 presents the results obtained by the parametric numerical
simulations, with a special attention to the three uniaxial responses. The related
mechanical properties are also derived here. Section 4 is devoted to the construction of
the mode classification maps. Section 5 discusses the factors influencing the mode
evolution with particular reference to angle and loading velocity. Section 6 draws some
major conclusions.
ʹ ‹‹–‡‡Ž‡‡–‘†‡Žƒ†‡š’‡”‹‡–ƒŽ˜ƒŽ‹†ƒ–‹‘
2.1 Finite element model
All the numerical simulations have been conducted using explicit finite-element
analysis program package LS-DYNA3D®. The corresponding detailed numerical
models have been established, in which the cell walls of hexagonal honeycomb have
been meshed with Belytschko-Tsay shell elements. Fig.1 shows the honeycomb’s
geometric structure and the repeating cell. In each model of this study, the thickness t of
cell wall has been set as 0.06 mm. The edge length h and the width l of the cell have
been set as 4 mm. The core consists respectively of 15¯15 cells for in-plane conditions
(along X1 and X2 directions) and of 15¯13 for out-of-plane conditions (along X3
direction). The overall thickness of the specimen along X3 direction is of 5 mm for
in-plane loading conditions, and of 60 mm for out-of-plane loading conditions to
capture more collapsed folds in the latter case. In order to analyze the crushing behavior
along X2 direction, for example (see Fig. 1(c)), the honeycomb specimen is placed on a
fixed rigid plate and crashed by another moving rigid plate inclined by an angle ș and
subjected to a constant vertical velocity v. Analogous loading conditions have been
applied to study the crushing behavior along X1 and X3 directions. An hourglass control
algorithm has been employed in the following numerical computations. The binding
material between cell walls has been ignored in this paper since the adhesive has full
capability to attach every cell wall. More generally, for some kinds of cellular-like
materials, the debonding between cell walls within the layer of the binding material
could be considered. As a matter of fact, debonding mechanisms in both static and
dynamic loading conditions (see [35-37]) may lead to a premature failure of the
composite structure thus compromising its energy absorption capacity. Meanwhile, this
kind of debonding has not been observed in aluminum honeycomb yet, according to the
existing experiments (see [4,7,15]).
 Fig. 1. Honeycomb structure and schematic of the oblique loading conditions
(load directed along the X2 axis): (a) honeycomb structure; (b) cell; (c) schematic of the
oblique loading conditions.

The mechanical properties of cell wall material are: density ρ = 2700 kg/m3 ,
Young’s modulus E = 68.97 GPa , Poisson’s ratio µ = 0.35 , yield stress σ 0 = 292 MPa ;
moreover, the stress-strain response is idealized by a bilinear law with a post-yield
modulus set as E/100, and its behavior is treated as rate-independent, as was done by
Aktay et al. [8], Fan et al. [11], Ruan et al. [15], Hu et al.[16], Yamashita & Gotoh [38],
Liu & Zhang [39]. An automatic single surface-to-surface contact algorithm has been
applied to avoid penetration between cell walls, whereas the dynamic friction factor has
been set as 0.20. A convergence study with different element sizes has been carried out
before the reported numerical simulations, after which the optimal element size has
been deduced and set as 1 mm in each FE model.

2.2 Experimental validation

In order to assess the validity of the proposed numerical model, a uniaxial
compression test is performed along the X3 direction by employing the universal testing
machine INSTRON 1342. The geometrical and mechanical parameters of the specimen
as well as the loading speed and the boundary conditions used in the numerical
simulations are set equal to that used in the experiments. Fig. 2 shows the results
obtained by the experiments and the corresponding simulations, from which it can be
clearly seen that the experimental deformed configuration is in agreement with that
resulting from the numerical simulation, as shown respectively in Fig. 2(a-b) and Fig.
2(c-d), for two values of compression ratio İ, i.e. ε = 0 .3 and ε = 0.6 . In this study the
mean stress is defined as σ = F / A , in which F is dynamic force acting on the
honeycomb surface area A . It should be pointed out that for oblique loading conditions,
A changes gradually during the compression. Thus, σ represents the nominal stress
calculated with reference to the initial area A0 . The experimental curve representing

the dynamic response in terms of mean stress as a function of the compression ratio
depicted in Fig. 2(e) shows an excellent agreement with the numerical one. Fig. 2(f)
presents the internal energy absorption and the hourglass energy obtained by simulation,
as well as the energy absorption obtained by experiment as a function of the
compression ratio, from which a perfect coincidence can be clearly observed.
Additionally, the amount of the hourglass energy remains very small compared to the
internal energy, thus confirming the validation of the obtained numerical results.

Loading pressure [MPa]

Fig. 2. Comparison between experiment and numerical simulation in terms of

deformed configuration and mechanical response (along X3): (a) experimental
deformed configuration at for ε = 0 .3 ; (b) experimental deformed configuration for
ε = 0 .6 ; (c) simulated deformed configuration for ε = 0 .3 ; (d) simulated deformed
configuration for ε = 0.6 ; (e) mechanical response; (f) energy absorption behavior.

In addition, the simulated in-plane dynamic response under a aligned compressive

load (i.e. at a zero inclination angle) for X1 and X2 directions (shown in Sections 3.1 and
3.2) is consistent with that presented in Ref. [15], providing a further validation of the
numerical model developed in this work.
Based on the above-mentioned arguments, the numerical results obtained for
oblique loading conditions by means of finite element simulations are expected to be
likely as they involve similar models and boundary conditions with respect to the case
of aligned load.

͵›ƒ‹…”‡•’‘•‡ˆ‘”‘„Ž‹“—‡Ž‘ƒ†‹‰…‘†‹–‹‘•
Unlike the previous case, for oblique loading conditions, the simulated deformed
shape evolves in a complex manner during the deformation history, notably affecting
the overall structural response of the considered honeycomb.

3.1 Load acting along X1 direction

Fig. 3 shows the comparison between the cases θ = 0 $ and θ = 10 $ in terms of
deformation history; in the numerical models, both loaded at a speed of 3 m/s, periodic
wave phenomena can be observed, running perpendicularly to the loading direction.
With reference to the case θ = 0 $ , the typical “X” band occurs at t = 10 ms, and
subsequently, the right wing of the “X” tends to vanish during the loading process as
the densification occurs; finally the “X” band completely vanishes at the full
densification stage.

Fig. 3. Deformation histories for differently inclined loads acting along the X1
direction: (a) θ = 0 $ ; (b) θ = 10 $ .

Compared with the aligned load case, a different deformation pattern appears for
case θ = 10 $ . The localized “X” band disappears, being replaced by the “Half-X” mode.
As shown in Fig. 3(b), a “Half-X” pattern can be clearly observed at 10 ms, which is
doubled at 15 ms; after that, several localized “Half-X” can be observed as the impact
loading progresses. Densification delay was also observed in the evolutionary
processes. Moreover, this densification mainly occurs in the neighborhood of the firstly
impacted edge portion. This irregular pattern induces negative influence on energy
absorption properties.
Fig. 4 shows the mean stress and the energy absorption against the time at a speed
of 3 m/s. The stress curves depicted in Fig. 4(a) shows that the load inclination does not
affect the load-carrying capacity at plateau stage. The peak force σ " , essentially

related to the initial buckling stage at θ = 0 $ , disappears for the case θ = 10 $ . Moreover,
the curves in Fig. 4(b) show that the energy absorption for θ = 10 $ is not always less
than for θ = 0 $ . As shown in Fig. 3, in the range between 20 and 30 ms, the
densification area for the case θ = 10 $ is wider than for θ = 0 $ , and therefore, in the
former case, the energy absorption in this time interval is larger.
Loading pressure [MPa]

Fig. 4. Response curves for differently inclined loads acting along the X1 direction:
(a) mean stress versus time; (b) energy absorption versus time.

3.2 Load acting along X2 direction

Similarly to Fig. 3, Fig.5 shows the comparison between the cases θ = 0 $ and
θ = 10 $ in terms of deformation history, under a load acting at X2 direction at a speed of
3 m/s. As shown in Fig. 5, at t = 6 ms for the case θ = 0 $ , a “V” localized band initially
appears at the waist of the honeycomb core, followed by other bands which are stacked
up until the densification stage, occurring at t = 24 ms. This mode can be defined as
“Waist-V” mode, in which the collapsed waves flow to the waist.
 Fig. 5. Deformation histories for differently inclined loads acting along the X2
direction: (a) θ = 0 $ ; (b) θ = 10 $ .

On the other hand, for case θ = 10$ , the first visible oblique “I” band occurs at the
top of the edge close to the loading region, until a “V” localized band appears in the
waist at 12 ms. After that, the hexagonal cells are crushed and new “I” localization
bands appear near the loaded edge as the applied displacement increases. It can be
obviously observed at 18 ms. This mode can be defined as “Edge I with waist V” mode.

3.3 Load acting along X3 direction

As mentioned above, the dynamic behavior of honeycomb under out-of-plane
loading conditions is different with respect to the in-plane case. It shows a collapse
mode characterized by a progressive cell buckling. Fig. 6 presents the dynamic
response under oblique loading with two different values of the inclination angle, i.e.
θ = 0 $ and θ = 10 $ , and impact velocity set as 10 m/s. Unlike the cases of loading acting
along X1 and X2 directions, the honeycomb behavior under loads acting along the X3
direction is stable. In particular, for θ = 10 $ , fold waves appear regularly and propagate
along the oblique plane. When close to densification (at 5.8 ms), the right cell foils
densify completely whereas left foils continue to collapse. The densification process
gradually shifts to left. There exists a separator line between these two stages# as
highlighted in Fig. 6(b) (the fifth frame). Finally, it is work noting that for θ = 10$ the
densification process is slower than for θ = 0$ .
 Fig. 6. Deformation histories for differently inclined loads acting along the X3
direction: (a) θ = 0 $ ; (b) θ = 10 $ .

3.4 Strength properties

The previously described evolutionary processes occurring in honeycombs loaded
in the X3 direction strongly influence their mechanical properties. Table 1 reports both
the peak stress σ p and the mean stress at plateau stage σ m when the honeycomb is

impacted at a speed of 70 m/s with different inclination angles. It can be found in Table
1 that σ p has a remarkable drop as the inclination angle increases from 0° to 2°, no

matter at what direction. The larger the angle θ , the smaller the peak stress σ p ,

whereas, σ m remains almost constant at varying inclination angles. When θ is greater

than 4 $ , σ p turns to be very close to σ m . This phenomenon produce a more stable

behavior at the initial state of the structural response. The response curves of Fig. 7,
which are zoomed between 0 and 0.3 ms, illustrate this behavior in a more effective
manner, showing also that as the loading angle increases the fluctuation of the loading
pressure decreases. The remaining parts of these six curves are very similar to the
curves depicted in Fig.4 (a) in the range between 10 ms and 30 ms. Moreover, the
considered specimens exhibit completely stable and ordered deformation processes.
Furthermore, as the plateau stress remains almost constant, it follows that, even if the
peak stress tends to strongly decrease for small inclination angles, the overall energy
absorption capacity is not reduced. Ultimately, the honeycomb shows an excellent
ability to carry slightly oblique loads.
Table 1. Strength properties under loads acting with different θ at a speed of 70 m/s.
(Unit: MPa )
Direction θ = 0$ θ = 2$ θ = 4$ θ = 6$ θ = 8$ θ = 10 $
X1 3.456 0.666 0.482 0.535 0.334 0.437
σp X2 4.417 1.087 0.575 0.744 0.344 0.653
X3 9.824 6.916 4.465 3.995 3.816 3.468
X1 0.249 0.253 0.266 0.257 0.248 0.242
σm X2 0.364 0.362 0.332 0.374 0.367 0.302
X3 2.772 3.214 3.158 3.135 3.123 3.115
Loading pressure [MPa]

Fig. 7. Response curves for differently inclined loads acting along the X3 direction
( v = 70 m/s ).

Ͷ‘†‡…Žƒ••‹ˆ‹…ƒ–‹‘ƒ’•
This section discusses the effects of the impact velocity and the inclination angle
on the overall mechanical behavior of honeycomb structures; the impact velocity varies
in the range 3-70 m/s, whereas the inclination angle varies between θ = 0 $ and θ = 10 $ ,
respectively. Two mode classification maps for the in-plane failure are created
collecting all the localization band shapes observed during the whole impact processes.

4.1 Mode classification map for honeycombs loaded along the X1 direction
Fig. 8 reports the key frames showing the typical mode transition in honeycombs
loaded along the X1 direction. Comparing it with Fig. 5#it can be clearly found that the
honeycomb maintains the “X” and “Half-X” modes at low impact velocities (see Fig.
8(a-d)). However, it reverts to the stable “I” mode at the loaded edge perpendicular to
the loading direction associated with a blurry transitional “V” band for increasing
values of the impact velocity, as shown in Fig. 8(e-h). For v ≥ 70 m/s , the transitional
“V” band is not observable, and therefore only the “I” localization band appears. A
progressive buckling occurs until final collapse at which the folding waves are
propagated to the opposite site. A concomitant “Half-X” band appears for δ = 69 mm
and θ = 10 $ .

Fig. 8. Key frames showing the typical mode transition (load directed along the X1
direction).

The above-discussed deformation modes can be summarized by using a

classification map, as shown in Fig. 9. In addition to the traditional “X” mode and “I”
mode, other three zones can be identified, i.e. a “Half-X” mode, a “V” mode, and a
“Half-X following I” mode. These three zones just serve as transitional modes from the
“X” to the “I” mode. For v ≤ 10 m/s and θ < 6$ , the honeycomb shows a “X” mode; the
same mode occurs for a smaller inclination range ( θ ≤ 4$ ) for 10 m/s ≤ v ≤ 30 m/s . The
“Half-X” dominates the zones in which v ≤ 10 m/s and θ ≥ 6$ or 10 m/s ≤ v ≤ 30 m/s
and θ ≥ 4$ . For increasing values of the impact velocity, the “V” mode appears, and
occupies the areas in which 30 m/s ≤ v ≤ 50 m/s and θ ≤ 4 $ or 50 m/s ≤ v ≤ 70 m/s and
θ ≤ 6$ . When the honeycomb is loaded with larger inclination angles at the same speed
range, the transition to the “Half-X following I” mode occurs. Finally, the pure “I”
mode can be observed only for values of velocity greater than 70 m/s.

Fig. 9. Mode classification map for honeycomb loaded along the X1 direction.

4.2 Mode classification map for honeycombs loaded along the X2 direction
In a similar way, the key deformation frames as well as the mode classification
map for honeycombs loaded along the X2 direction are depicted in Figs. 10 and 11,
respectively. From the analysis of Fig. 10, reporting the key frames showing the typical
modes, it can be observed that both pattern and location are different with respect to
those obtained by Cricrì [14] and Ruan [15], referring to aligned loading conditions.
The usual transitional “V” mode no longer appears close to the loading area, but
originates near the waist. Thus, the “V” mode is replaced by a “Waist-V” localization
band, as shown in Fig. 10(a). For oblique loading conditions, three typical modes can be
observed in Fig. 10. A “Waist-V” mode appears under oblique impact loadings at a
speed of 3 m/s for the case θ = 2$ , whereas, for θ = 10 $ , it evolves to the “Edge I with
waist V” mode, as shown in Fig. 10(a-d). For v ≥ 30 m/s , the “Edge I with waist V”
mode dominates until v reaches the value of 50 m/s, at which the honeycomb shows a
pure “I” mode.
 Fig. 10. Key frames showing the typical mode transition (load directed along the
X2 direction).

Fig. 11. Mode classification map for honeycomb loaded along the X2 direction.

From the mode classification map in Fig. 11, it can be found that the “Waist-V”
mode appears for v ≤ 10 m/s and θ < 8$ , or 10 m/s ≤ v ≤ 30 m/s and θ < 6$ . The typical
“I” mode dominates in the regions defined by the following conditions:
30 m/s ≤ v ≤ 50 m/s and θ ≥ 8$ , or 50 m/s ≤ v ≤ 70 m/s and θ ≥ 2$ , or v ≥ 70 m/s . The
stair-shaped intermediate transitional zone is occupied by the “Edge I with waist V”
pattern.
It should be pointed out that unlike the cases of X1 and X2 directions, the
deformation mode for honeycombs loaded along the X3 direction is insensitive to both
the impact velocity and the inclination angle (if small enough), and therefore no further
details are provided in this paper about this aspect.
ͷ‹•…—••‹‘
The first topic of discussion is the reason why the honeycomb structure shows
different behaviors under aligned and oblique loading. The main aspect concerns the
asynchronous loading. In detail, the honeycomb structure shows different performances
when loaded along the X1 and X2 directions, which is a direct consequence of the
different geometric configurations assumed in these directions. For loads acting along
the X1 direction, the elastic buckling does not occur. Consequently, the cell walls of
honeycomb bear a pure compression. At this situation, a sequential collapse of the cell
walls occurs though adjacent hexagonal cells, starting from those located at the free
ends of the loaded edge, according to the sequence depicted in Fig. 12(a) (i.e. from core
$ to core % passing through core &), thus progressively forming the right branch of
the “X” localization band.
Conversely, under oblique loading conditions, the bottom layer cells collapse
earlier than the top ones at the same compression level. Thus, only half of the right
branch of the “X” band collapses, as shown in Fig. 12(b). This phenomenon will lead to
insufficient collapse for the right top cells and shift down the central rhombus of “X”
mode. In this case, the honeycomb structure experiences an unexpected rotational effect.
At the same time, the deformation mode comes to change from symmetrical to
asymmetrical, and the loading carrying capacity of the structure suffers a considerable
loss. This behavior persists for different impact velocities, until the value of 70 m/s is
reached.
On the other hand, for the case of honeycombs loaded along the X2 direction, the
cell walls which are parallel to the loading direction, undergo axial crushing. At the
initial crushing stage, the maximum deflection occurs at middle area, leading to the
collapse of the cells located “in the waist”. As a consequence, the resulting localization
mode is the so-called “Waist V” mode.
 Fig. 12. Schematic of the “X” mode nucleation: (a) aligned load; (b) oblique load.

The second topic of discussion is the reason why the mode changes with
increasing values of the loading speed. This is due to the inertial effect, as already
studied by Ruan [15] for aligned loads. This effect also occurs for oblique loading
conditions. In fact, by analyzing the obtained results, it can be found that the
asynchronous loading and the impact velocities are the main factors determining the
deformation mode evolution under the action of oblique loads, as can be verified in the
mode classification maps reported in Section 4.

͸‘…Ž—•‹‘•
Based on the above-discussed numerical simulation results, the following major
conclusions can be drawn:
(1) Honeycomb structures show a different dynamic response depending on
whether they are loaded obliquely or not. In particular, both the impact velocity and the
inclination angle are proved to exert a strong influence on their dynamic behavior.
(2) Compared to aligned loading conditions, oblique loads lead to new patterns for
the localization bands. In detail, for honeycombs loaded along the X1 direction, the
oblique loading lead to the mode transition from “X”, “ V”, and “I” modes to “Half-X”,
“Half-X following I” ,“ V” and “I” modes; on the other hand, for honeycombs loaded
along the X2 direction, the “ V” and “I” modes remain, and a new transition mode
occurs for different impact velocities. Interestingly, the “I” mode dominates for large
impact velocities (greater than 70 m/s) in honeycombs loaded along both the X1 and X2
directions.
(3) For honeycombs loaded along the X3 direction, oblique collapse modes, which
are maintained perpendicular to the loading direction, appear and progress until the full
densification stage. A moving separator line can be depicted between the collapse stage
and the densification stage. A notable result obtained in this study is that honeycomb
structures show an excellent ability to carry slightly oblique loads.
In summary, although not considered in common practice, oblique loading
conditions cannot be avoided and strongly influence the mechanical response of
honeycomb structures. Given that, the present study provides a significant
advancement to the comprehensive understanding of the dynamic response of
honeycombs, that could be used to develop more valuable guidelines for design
purposes as well as to realize more effective cellular energy absorbing devices.

…‘™Ž‡†‰‡‡–
This work was financially supported by The National Natural Science Foundation
of China (51505502), (51275532) and the Key Union Project of National Natural
Science Foundation of China on High-Speed Railway (U1334208). The authos would
like to express their thanks.

‡ˆ‡”‡…‡•
[1] Gibson LJ and Ashby MF. Cellular Solids-Structures and Properties, 2nd ed.,
Cambridge University Press, Cambridge,1997.
[2] Lu GX and Yu TX. Energy Absorption of Structures and Materials, Woodhead
Publishing, Cambridge, 2001.
[3] Zhang X, Zhang L, Zhang P. Equivalent constitutive equations of honeycomb
material using micro-polar theory to model thermo-mechanical interaction. Compos
Part B-Eng2012; 43: 3081-3087.
[4] Mahmoudabadi MZ and Sadighi M. A theoretical and experimental study on metal
hexagonal honeycomb crushing under quasi-static and low velocity impact loading.
Mat Sci Eng A Struct 2011; 528 (15): 4958-4966.
[5] Roy R, Nguyen KH, Park YB, et al. Testing and modeling of Nomex™ honeycomb
sandwich Panels with bolt insert. Compos Part B-Eng2014; 56:762-769.
[6] Hazizan Md Akil, Cantwell WJ. The low velocity impact response of an aluminum
honeycomb sandwich structure. Compos Part B-Eng2003; 34:679-687.
[7] Wang Z, Tian H, Lu Z, Zhou W. High-speed axial impact of aluminum
honeycomb-Experiments and simulations. Compos Part B-Eng 2014; 56:1-8.
[8] Aktay L, Johnson AF and Kroplin BH. Numerical modeling of honeycomb core
crush behavior. Eng Fract Mech2008;75 (9):2616-2630.
[9] Sun DQ, Zhang WH and Wei YB. Mean out-of-plane dynamic plateau stresses of
hexagonal honeycomb cores under impact loadings. Compos Struct 2010;92
(11):2609-2621.
[10] Papka S D and Kyriakides S. Experiments and full-scale numerical simulations of
in-plane crushing of a honeycomb. Acta mater 1998;46(8): 2765-2776.
[11] Fan X, Ignaas V, and Dirk V. Finite element analysis of out-of-plane compressive
properties of thermoplastic honeycomb. Sandw Struct Mater 2006;8(5):437-458.
[12] Liu L, Meng P, Wang H, Guan Z. The flatwise compressive properties of Nomex
honeycomb core with debonding imperfections in the double cell wall. Compos Part
B-Eng 2015; 76:122-132.
[13] Chen Z, Yan N. Investigation of elastic moduli of Kraft paper honeycomb core
sandwich panels. Compos Part B-Eng 2012;43:2107-2114.
[14] Cricrì G, Perrella M, Calì C. Honeycomb failure processes under in-plane loading.
Compos Part B-Eng 2013; 45(1): 1079-1090.
[15] Ruan D, Lu G, Wang B and Yu TX. In-plane dynamic crushing of honeycombs—a
finite element study. Int J Impact Eng 2003;28(2):61-182.
[16] Hu LL and Yu TX. Dynamic crushing strength of hexagonal honeycombs. Int J
Impact Eng2010; 37(5):467-474.
[17] Reyes A, Langseth M and Hopperstad OS. Crashworthiness of aluminium
extrusions subjected to oblique loading: experiments and numerical analyses. Int J of
Mech Sci2002; 44(9):1965-1984.
[18] Reyes A, Langseth M and Hopperstad OS. Square aluminium tubes subjected to
oblique loading. Int J Impact Eng 2003;28 (10):1077-1106.
[19] Reyes A, Hopperstad OS and Langseth M. Aluminium foam-filled extrusions
subjected to oblique loading: experimental and numerical study. Int J of Solids Struct
2004;41 (5):1645-1675.
[20] Ahmad Z, Thambiratnam DP and Tan ACC. Dynamic energy absorption
characteristics of foam-filled conical tubes under oblique impact loading. Int J Impact
Eng2010; 37 (5):475-488.
[21] Zarei H and Kroger M. Optimum honeycomb filled crash absorber design. Mater
Design 2008; 29(1):193-204.
[22] Greve L, Pickett AK, Payen F. Experimental testing and phenomenological
modelling of the fragmentation process of braided carbon/epoxy composite tubes under
axial and oblique impact. Compos Part B-Eng 2008; 39: 1221-1232.
[23] Eskandarian A, Marzougui D, Bedewi NE. Finite element model and validation of
a surrogate crash test vehicle for impacts with roadside objects. Int J
Crashworthiness1997; 2 (3):239-258.
[24] Keshavanarayana S and Thotakuri MV. Off-axis compression behavior of
honeycomb core in WT-plane. Int J Crashworthiness 2009; 14 (2):173-181.
[25] Hong ST, Pan J, Tyan T et al. Dynamic crush behaviors of aluminium honeycomb
specimens under compression dominant inclined loads. Int J Plasticity 2008;
24(1):89-117.
[26] Greco F. A study of stability and bifurcation in micro-cracked periodic elastic
composites including self-contact. Int J Solids Struct 2013; 50: 1646-1663.
[27] Greco F, Luciano R. A theoretical and numerical stability analysis for composite
microstructures by using homogenization theory. Compos Part B 2011; 42(3): 382-401.
[28] Bruno D, Greco F, Luciano R, Nevone Blasi P. Nonlinear homogenized properties
of defected composite materials. Comput Struct 2014; 134: 102-111.
[29] Greco F. Homogenized mechanical behavior of composite micro-structures
including micro-cracking and contact evolution. Eng Fract Mech 2009, 76(2): 182-208.
[30] Bruno D, Greco F, Lonetti P, Nevone Blasi P. Influence of micro-cracking and
contact on the effective properties of composite materials. Simul Model Pract Theory
2008, 16(8): 861-884.
[31] Greco F, Leonetti L, Nevone Blasi P. Non-linear macroscopic response of
fiber-reinforced composite materials due to initiation and propagation of interface
cracks. Eng Fract Mech 2012; 80: 92-113.
[32] Greco F, Leonetti L, Lonetti P, Nevone Blasi P. Crack propagation analysis in
composite materials by using moving mesh and multiscale techniques. Comput Struct
2015; 153: 201-216.
[33] Greco F, Leonetti L, Nevone Blasi P. Adaptive multiscale modeling of
fiber-reinforced composite materials subjected to transverse microcracking. Compos
Struct 2014; 113: 249-263.
[34] Feo L, Greco F, Leonetti L, Luciano R. Mixed-mode fracture in lightweight
aggregate concrete by using a moving mesh approach within a multiscale framework.
Compos Struct 2015; 123: 88-97.
[35] Bruno D, Greco F, Lonetti P. A fracture-ALE formulation to predict dynamic
debonding in FRP strengthened concrete beams. Compos Part B 2013; 46: 46-60.
[36] Greco F, Leonetti L, Lonetti P. A novel approach based on ALE and Fracture
Mechanics for multilayered composite beams. Compos Part B-Eng 2015; 78: 447-458.
[37] Bruno D, Greco F, Lo Feudo S, Nevone Blasi P. Multi-layer modeling of edge
debonding in strengthened beams using interface stresses and fracture energies. Eng
Struct 2016; 109: 26-42.
[38] Yamashita M and Gotoh M. Impact behavior of honeycomb structures with various
cell specifications-numerical simulation and experiment. Int J Impact Eng 2005;
32(1):618-630.
[39] Liu Y and Zhang XC. The influence of cell micro-topology on the in-plane
dynamic crushing of honeycombs. Int J Impact Eng 2009; 36(1):98-109.
A list of table and figure captions
Table 1 Strength properties under loads acting with different θ at a speed of 70 m/s.
Honeycomb structure and schematic of the oblique loading conditions (load
Fig.1.
directed along the X2 axis):
Comparison between experiment and numerical simulation in terms of
Fig.2.
deformed configuration and mechanical response (along X3):
Deformation histories for differently inclined loads acting along the X1
Fig.3.
direction:
Fig.4. Response curves for differently inclined loads acting along the X1 direction:
Deformation histories for differently inclined loads acting along the X2
Fig.5.
direction:
Deformation histories for differently inclined loads acting along the X3
Fig.6.
direction:
Response curves for differently inclined loads acting along the X3 direction
Fig.7.
( v = 70 m/s ).
Key frames showing the typical mode transition (load directed along the X1
Fig.8.
direction).
Fig.9. Mode classification map for honeycomb loaded along the X1 direction.
Key frames showing the typical mode transition (load directed along the X2
Fig.10.
direction).
Fig.11. Mode classification map for honeycomb loaded along the X2 direction.
Fig.12. Schematic of the “X” mode nucleation: