Original Article

Journal of Sandwich Structures & Materials

0(0) 1–21
Investigation of sandwich ! The Author(s) 2018
Reprints and permissions:
composite failure under sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1099636218791162
three-point bending: journals.sagepub.com/home/jsm

Simulation and
experimental validation

Sudharshan Anandan, Gurjot Dhaliwal,

Shouvik Ganguly and
K Chandrashekhara

Abstract
A sandwich structure consists of a two thin and strong facesheets, bonded to a thick
lightweight core material. The mechanical response of a sandwich structure depends on
the properties of its constituents. A numerical model and experimental validation of the
three-point bending test of sandwich composites are presented in this study. The core
material is aluminum honeycomb. The facesheets are made of IM7/Cycom5320-1,
which is a carbon fiber/epoxy prepreg system. A comprehensive model of the failure
under flexural loading was developed. Facesheet failure was modeled using Hashin’s
failure criteria. A detailed meso-scale model of the honeycomb core was included in the
model. The experiments indicated that failure initiation was due to local buckling in
the honeycomb core. Failure propagation was in the form of core failure, facesheet
compressive failure, and interlaminar failure. The developed meso-scale model was able
to accurately simulate failure initiation and propagation in the composite sandwich
structure. The effect of elevated temperature on the three-point bending behavior
was studied numerically as well as experimentally. An increase in test temperature
to 100
C resulted in a drop of 9.2% in flexural strength, which was also predicted by
the numerical model.

Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology,
Rolla, USA
Corresponding author:
K Chandrashekhara, Missouri University of Science and Technology, 188 Toomey Hall, Rolla, Missouri 65409,
USA.
Email: chandra@mst.edu
2 Journal of Sandwich Structures & Materials 0(0)

Keywords
Sandwich composite, meso-scale modeling, finite element method

Introduction
Background
Honeycomb sandwich structures consist of a thick honeycomb core material which
is “sandwiched” between two relatively thin composite facesheets. The composite
facesheet is bonded to the core using a film adhesive. Sandwich composites have
better bending stiffness compared to composite laminates, and a high strength to
weight ratio, due to which they are used in aerospace structures [1]. The honey-
comb core is a cellular material which has reduced density, resistance to out-of-
plane loads, high flexural stiffness, and energy absorption. The composite face-
sheets resist in-plane loading while the core provides bending stiffness. Structural
response of sandwich composites is dependent on the properties of the constitu-
ents. After mechanical characterization of the constituent materials, behavior of
sandwich structures can be simulated using numerical models. A complete exper-
imental approach to characterize the mechanical behavior of sandwich structures
can be time consuming and expensive. A numerical model can be helpful to eval-
uate the mechanical properties of sandwich structures and help select the desired
combination of sandwich structure constituent materials for further testing. Such
“virtual testing” methods can be very useful in studying the mechanical behavior of
composites and sandwich structures [2]. Aerospace sandwich structures can be
exposed to extreme conditions in service. The sandwich constituents, honeycomb
core and facesheet, are affected differently when exposed to elevated temperatures.
This behavior can also be studied effectively using numerical models.
The mechanical behavior of honeycomb cores has been characterized by Gibson
and Ashby [3]. Numerical simulation methods have been used to study the behav-
ior of various cellular core structures and optimize core geometries using dynamic
finite element analysis (FEA) [4]. Li et al. studied the mechanical behavior of
cellular cores with irregular cell shapes using numerical modeling [5]. Giglio
et al. conducted finite element simulations to study the behavior of sandwich
structures under flatwise compression and three-point bending [6,7]. Nomex hon-
eycomb cores with aluminum facesheets were investigated. The effect of friction
between the puncher and the facesheet was analyzed using numerical models.
A detailed meso-scale model of a Nomex honeycomb core was developed [8,9].
For core compression, it was observed that a single layer of shell elements was
capable of representing the mechanical behavior of the Nomex core.
Zhou et al. studied the damage in sandwich panels subjected to bending and the
effect of indenter shape, facesheet thickness, and core density [10]. Nomex and
Anandan et al. 3

aluminum cored sandwich structures were evaluated. It was found that

damage mechanisms depend on the shape of the indenter and facesheet thickness.
When a hemispherical indenter is used, failure is in the form of core crushing and
facesheet delamination. Three-point bending of sandwich specimens has also been
modeled using homogenized core parameters [11]. The linear portion of the load–
displacement curve was modeled and good agreement was found between simula-
tion and experiments. Analytical solutions for homogenized core properties have
been reported in [12,13]. Strain energy-based homogenization method has been
used to simulate honeycomb mechanical behavior [14]. Meso-scale modeling
approach has been used to simulate the behavior of sandwich composites with
honeycomb cores. Sun et al. developed numerical models to study the crashwor-
thiness of aluminum honeycombs with aluminum facesheets [15]. Three-point
bending and edgewise compressive loading was simulated and the models were
validated experimentally. The effect of geometric and material parameters was
quantified. Xang et al. used a finite element model along with a response surface
analysis to evaluate energy absorption in thin-walled honeycombs [16]. Numerical
models have also been used to evaluate the effect of facesheet thickness, core
parameters on the quasi-static indentation response, and used to optimize sand-
wich parameters [17]. Zhou et al. studied the mechanical behavior of structures
manufactured using two Nomex honeycombs. Finite element models were used to
predict the location of buckling and collapse stress in the structure [18]. The effect
of entrapped air in sandwich composites with Nomex honeycomb cores was inves-
tigated [19]. The authors used unit cell as well as a multi-cell modeling approach
and observed that the multi-cell approach provided curves smoother than the unit-
cell model.

Failure in three-point bending

Failure mode maps for three-point bending tests have been developed, using an ana-
lytical approach, based on core and facesheet properties [20]. Under three-point bend-
ing, sandwich panel failure can take place due to various modes. Facesheet failure can
occur due to in-plane stresses, facesheet wrinkling, and intra-cell dimpling.
Facesheet failure occurs when the in-plane stresses, in either of the facesheets,
reach the critical value. Under three-point bending, the maximum bending
moment is at the mid-span of the beam. For symmetrical sandwich specimens,
the stress is the same on tension and compression sides. However, failure in com-
posite facesheets is generally on the compressive side. The facesheet stress at the
point of failure of sandwich composites can be calculated using equation (1) [21]

Pmax S
r¼ (1)
2tf ðts þ tc Þb

where Pmax is the load to failure, tf is the facesheet thickness, ts is the total sand-
wich thickness, tc is the core thickness, and b is the breadth of the sandwich
4 Journal of Sandwich Structures & Materials 0(0)

structure. The term S stands for the support span during the three-point
bending test.
Under axial loads, facesheet wrinkling takes place due to skin buckling with a
wavelength greater than the honeycomb cell width [22–24]. The critical stress for
facesheet wrinkling can be calculated for honeycomb sandwich constructions with
composite facesheets based on equation (2) [1,25]

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 tf Ec Efx Efy
rw ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (2)
3tc 1 
xy
yx

where Ec stands for core modulus, Efx is the facesheet modulus in the axial direc-
tion, and Efy is the facesheet modulus in the transverse direction.
xy and
yx are
facesheet Poisson’s ratios.
Honeycomb sandwich structures with very thin facesheets can fail due to
intra-cell dimpling of the facesheets [20]. This mode of failure occurs due to
in-plane stresses in the facesheets, in the region which is unsupported by the
walls of the honeycomb. Classical models consider the facesheets to be supported
on elastic foundations. Thomsen and Banks developed an improved model for
intra-cell buckling in honeycomb sandwich structures with very thin face-
sheets [26].
Core failure can also occur under three-point bending loads. Common modes of
failure are core shear and failure due to local indentation. Shear forces can lead to
core shear failure. The core shear failure mode is dependent on geometric param-
eters of the sandwich, loading span, and core density. Since the honeycomb core
are orthotropic, the orientation of the honeycomb cells also affect core failure.
According to ASTM C393, core shear failure occurs if

2rfmax t
S  (3)
kFs

where S is the support span, rfmax is the expected facesheet ultimate strength, k is
the core shear strength factor to ensure facesheet failure (recommended value is
0.75), and Fs is the core shear strength.
Honeycomb sandwich structures under three-point bending can exhibit failure
due to local indentation at the point of loading. Failure is due to core crushing
under the puncher. If the area of contact between the puncher and sandwich is
known, the critical failure stress of this mode can be calculated. It is also
assumed that the load transfer is uniform over this contact area. However, for
cylindrical puncher indenting a flat facesheet surface, the contact area is difficult
Anandan et al. 5

to estimate. This mode of failure is also affected by material irregularities, local-

ized shear stresses, puncher shape, and other imperfections in the material.
Previous studies on Nomex cores with aluminum facesheets have shown core
crushing failure [10]. Local indentation failures can be simulated by using a
full three-dimensional finite element model of the honeycomb core.

Current work
Modeling the honeycomb core using homogeneous orthotropic solid geometries
can result in low computational times. The mechanical stiffness can be modeled
using a set of elastic moduli, which are dependent on cell geometries and honey-
comb density [27]. The response of honeycombs to in-plane and out-of-plane loads
have been described previously [3,13]. Numerical homogenization can also be
performed using FEA, instead of analytical solutions, to obtain core effective
properties [7,14]. Catapano and Montemurro used a numerical homogenization
technique to evaluate effective elastic properties of a honeycomb [28]. A Genetic
Algorithm (GA)-based optimization scheme was used to determine optimal face-
sheet and core parameters for a honeycomb structure. While homogenized models
can provide accurate representation of the linear load–displacement behavior, they
are not capable of predicting local core crushing and meso-scale damage. The
current work focuses on developing a meso-scale numerical model of the three-
point bending test on sandwich composites. The core mechanical behavior was
simulated using a meso-scale model. Composite facesheets were used in the sand-
wich structure and the failure propagation in the facesheet was included in the
model. The effect of elevated temperatures was incorporated and all simulations
were validated using experiments. Results show good agreement between experi-
mental results and model predictions at room temperature and elevated
temperatures.

Numerical modeling
Facesheet failure
Failure in the facesheets can be modeled using Hashin’s criteria. The following
modes of failure can be simulated using these criteria: (a) fiber rupture due to
tensile loads; (b) fiber buckling and kinking due to compressive loads; (c)
matrix cracking under transverse tension and shearing; and (d) matrix crushing
under transverse compression and shearing [29]. The initial behavior of the
undamaged facesheet is linear elastic [30]. Once the failure stress is reached,
damage propagation in the facesheet is modeled using a linear softening criterion.
The elastic constants are reduced using a damage parameter denoted by
6 Journal of Sandwich Structures & Materials 0(0)

d. The constitutive equations for facesheet mechanical behavior are shown in

equations (4) to (6)
2 3
ð1  df ÞE1 ð1  df Þð1  dm Þ
21 E1 0
e6
6
7
7
r¼ 6 ð1  df Þð1  dm Þ
12 E2 ð1  dm ÞE2 0 7
D4 5
0 0 ð1  ds ÞG12 D
(4)

D ¼ 1  ð1  df Þð1  dm Þ
21
12 (5)
    
ds ¼ 1  1  dtf 1  dcf 1  dtm 1  dcm
21
12 (6)

Initially, the facesheet is undamaged and the parameter d is equal to zero. This
results in linear elastic behavior. Once failure load is reached, damage initiation in
the facesheets is controlled by Hashin’s criteria. Failure initiates when one of the
criteria in equations (7) is satisfied. The quadratic stress failure criteria are used.

Fiber tension ðr11 > 0Þ : Fiber compression ðr11 <0Þ :

 2  2
r11 r11
Ftf ¼ T Fcf ¼
X Xc
(7)
Matrix tension ðr22 > 0Þ : Matrix compression ðr22 <0Þ :
" 2 #
 2  2    2  2
r22 s12 r22 2 Yc r22 s12
F tm ¼ T þ Fm ¼
c
þ  1 þ
Y SL 2S T 2ST Y c SL

where Ftf denotes the fiber tensile failure index, Fcf is the fiber compressive failure
index, Ftm is the matrix tensile failure index, Fcm is the matrix compressive failure
index, r11 and r22 are the longitudinal and transverse stresses, respectively, and s12
is the shear stress. XT ; YT ; Xc ; Yc , SL , and ST are longitudinal tensile strength,
transverse tensile strength, longitudinal compressive strength, transverse compres-
sive strength, longitudinal shear strength, and transverse shear strength, respec-
tively (Table 1). Temperature dependence of material properties were incorporated
based on data provided in the manufacturer datasheet.
Once failure initiates at any integration point in the facesheet, damage evolution
takes place according to a linear softening law. For all modes of failure, stiffness
degradation takes place according to equation (8)

dfeq ðdeq  d0eq Þ

d¼ (8)
deq ðdfeq  d0eq Þ
Anandan et al. 7

Table 1. Temperature dependent properties of the facesheet material [31].

Elastic properties

Temperature Longitudinal Transverse Poisson’s Shear moduli

(
C) modulus (E1) modulus (E1) ratio (
12) (G12, G13)

24 156 GPa 9.3 GPa 0.30 5.5 GPa

121 158 GPa 8.3 GPa 0.32 4.9 GPa
Strength properties

Temperature (
C) XT YT Xc Yc

24 2503 MPa 75.9 MPa 2078 MPa 165 MPa

121 2477 MPa 62.4 MPa 2016 MPa 135 MPa

Figure 1. Schematic of linear softening behavior [29].

where d is the damage variable, deq is the current equivalent displacement, dfeq is the
equivalent displacement at failure, and d0eq is the equivalent displacement at failure
initiation. The equivalent displacement and stresses are represented schematically
in Figure 1. The constitutive law for each element can be expressed as a relation
between the equivalent displacement and equivalent stress.
8 Journal of Sandwich Structures & Materials 0(0)

Table 2. Fracture energies of the facesheet material [32].

Longitudinal Longitudinal Transverse Transverse

tension compression tension compression

81.5 kJ/m2 106.5 kJ/m2 0.28 kJ/m2 0.79 kJ/m2

Figure 2. Honeycomb cell modeling: (a) commercial honeycomb cell and (b) shell section
definitions.

For a completely degraded facesheet, d ¼ 1. In this failure model, the fracture

energy Gc must be specified for each mode of failure. The fracture energy is the
area described by OAB (Figure 1). Gc for each failure mode of failure is extracted
from the properties of a similar material system reported by Zhang et al. [32]
(Table 2).

Honeycomb core model

The honeycomb cells have a very large height to thickness ratio and can be mod-
eled using shell elements. A single honeycomb cell of a commercial aluminum
honeycomb is shown in Figure 2(a). In a typical honeycomb core, the cell walls
oriented along the ribbon directions have twice the wall thickness. This difference
in wall thicknesses was included in the shell section definitions (Figure 2(b)).
Failure can take place due to plastic buckling in the cell walls or shear failure in
the core. An elastic–plastic model was used to describe the behavior of the alumi-
num core. The modulus of the honeycomb constitutive material, aluminum 5052,
was 70 GPa and yield stress was 255 MPa. Poisson’s ratio was 0.3.
Anandan et al. 9

Figure 3. Honeycomb core showing regions of fine and coarse mesh.

Each cell wall was modeled using a single layer of shell elements. The failure in
the honeycomb core is affected by the mesh size. In order to capture local failure
phenomena accurately, a fine mesh region was included in the central area of the
core. The meso-scale model of the honeycomb core is shown in Figure 3.
The central region of the core had a mesh size of 0.5 mm, which results in four
elements along each cell wall of the honeycomb. The outer regions had a mesh size
of 1 mm, corresponding to two elements along the length of the honeycomb cell
wall. Correspondingly, the facesheet was also divided into coarse and fine mesh
regions. The central region of the facesheet, above the fine mesh region of the
honeycomb core, had a mesh size of 1 mm, while the outer regions had a mesh
size of 2 mm.

Modeling of three-point bending test

The numerical model of the three-point bending test was built corresponding to the
experimental setup. The specimen configuration was according to the requirements
in ASTM D7250 and ASTM C393 (Figure 4). The facesheets were modeled using a
single layer of shell elements (S4R) and had a [0
/90
]s composite layup. The
supports and puncher, which were cylinders of 25 mm diameter, were modeled
using discrete rigid elements. The motion of the supports was controlled using a
reference point and constrained along all degrees of freedom. The puncher was free
to move along the z axis, which is perpendicular to the face of the sandwich com-
posite. Motion along all other degrees of freedom was constrained. A general
contact interaction was incorporated to model the contact between all surfaces
of the numerical model. The Coulomb friction model was used at the interface
between the puncher/supports and the facesheet. The coefficient of friction
between the sandwich faces and surfaces of supports and puncher was set as 0.3.
The facesheet and core components were bonded using kinematic tie constraints.
The analysis was performed using Abaqus/Explicit software package. The
stable time increment was 1E-9 which leads to extremely high computational
10 Journal of Sandwich Structures & Materials 0(0)

Figure 4. Schematic of the three-point bending test.

costs. In order to reduce computational requirements, mass scaling technique was

used. In addition, a reduced width of the sample was simulated. The specimen
width in the simulation was equal to the width of four honeycomb cells. This
smaller sample width was used (12.7 mm), compared to a standard sample
(50.8 mm) to reduce computational time. Results were scaled up to correspond
to the strength of the test samples used in experiments.

Materials and manufacturing

Composite sandwich panels were manufactured using an out-of-autoclave prepreg
process. This process is a low-cost alternative to autoclave curing and can result in
significant reductions in operating costs [33]. The facesheets were made of
IM7/Cycom 5320-1 prepregs. The selected facesheet material is a toughened aero-
space epoxy system which can be cured out-of-autoclave. The layer orientation was
[0
/90
]s. The core designation was CRIII-1/8-5052-0.002, manufactured by
Hexcel. The term 1/8 in the core designation stands for the core cell size which
was 1/8 in. (3.175 mm). Aluminum 5052 is the constituent material of the honey-
comb and 0.002 in. (0.0508 mm) is the thickness of the foil gauge. It corresponds to
the thickness of a single cell wall in the honeycomb. The core density was 8.1 pcf
(127 kg/m3). Core and facesheet were bonded using Cytec FM300-2 film adhesive.
The layup including composite facesheets, film adhesive, and the honeycomb core
was co-cured under atmospheric pressure at recommended cure cycle of 250
F
(121
C) for 2 h. Samples were then post cured at 350
F (177
C) for 2 h. Test
specimens measuring 10 in.  2 in. (254 mm  50.8 mm) were cut from the manu-
factured sandwich panel for three-point bending experiments. Two samples were
used and tested at room temperature and three at the elevated temperature.
Anandan et al. 11

The edges were polished with sand paper in order to remove imperfections which
may affect the test results.

Experiments
Three-point bending tests were conducted on sandwich specimens according to
ASTM D7250 and ASTM C393 [21,34]. Load was applied using an Instron
5585 testing machine with a 100 kN load cell. The experimental setup is shown
in Figure 4. The cylinders used as supports and puncher had a diameter of 25 mm.
The support span length was 8 in. (203.2 mm). Tested sandwich panels had the
ribbon-direction of the honeycomb, perpendicular to the axis of the puncher.
The uppermost layer of the facesheets had fibers oriented perpendicular to the
axis of the puncher. For high temperature testing, the entire setup was moved
into an environmental chamber, available at Missouri S&T. The elevated temper-
ature chosen for this study is 100
C. The chamber was heated to the required
temperature. Samples and fixture were conditioned inside the chamber for 2 h
prior to testing. Three-point bending tests were conducted as per ASTM standard
C393. A loading rate of 3 mm per minute was maintained during the test. The
application of load was continued well after failure in order to capture the stabi-
lized failure propagation zone. All tested specimens exhibited a similar general
trend in the load–displacement curve. Samples tested at room temperature exhib-
ited failure loads of 6293 N and 6599 N, respectively. Three samples were tested at
100
C, and the corresponding failure loads were 5689, 5977, and 5878 N, respec-
tively. A representative load–displacement curve for the test at room temperature
is shown in Figure 5. Three zones can be observed in the experimental curve.
The first zone is an initial linear elastic zone where the sandwich is undamaged.
The load increases until a peak is observed, and then failure initiates. It is followed
by a progressive failure zone. Finally, a stabilized damage propagation region is
observed. At this point, the facesheet has already failed and energy is absorbed
mainly through the failure of the honeycomb core and damage propagation in
the facesheet.

Results and discussion

The numerical model was used to predict sandwich behavior under three-point
bending. The load–displacement curve from the numerical model is also shown in
Figure 5. The simulation was performed until the puncher displacement reaches
14 mm. There is good agreement between experiments and simulations in the elas-
tic region of the curve. The peak load obtained in the simulation is a little higher in
comparison to the experimental results. In the numerical simulations, the damage
propagation is more catastrophic in nature, while experiments showed a gradual
failure propagation zone. The reasons behind this behavior will be explored in this
study. The stabilized load-carrying capacity of the panel after failure is predicted
accurately by the numerical model.
12 Journal of Sandwich Structures & Materials 0(0)

Figure 5. Representative curve of load–displacement behavior under three-point bending.

Failure under three-point bending

When subjected to flexural loads, sandwich panel failure can take place due to
various modes such as facesheet failure, facesheet wrinkling, intra-cell facesheet
dimpling, core shear, and local failure due to indentation [20]. Based on the char-
acteristics of the sandwich structure, critical failure stresses under various failure
modes can be calculated. The in-plane stresses in the facesheets are not high
enough to cause facesheet wrinkling according to equation (2). For the span
length and sandwich parameters considered in this study, expected mode of failure
can be due to facesheet failure under compression or core crushing. The maximum
facesheet stress at the failure load of 6598 N can be calculated to be 260 MPa. This
corresponds to the stress in the outermost layer of the facesheet, in contact with the
puncher. Figure 6 shows the evolution of the compressive failure criterion in the
upper facesheet, which is in contact with the puncher, as predicted by the numer-
ical model. Facesheet failure occurs when the failure criterion reaches the value
of 1. Sandwich failure is observed at a displacement prior to the failure of the
composite facesheet. This indicates that the failure initiation is due to other mech-
anisms such as core crushing in the honeycomb sandwich which is then followed by
progressive facesheet failure.
Figure 7 depicts sandwich failure under three-point bending. Local core com-
pressive failure is observed in the region under the puncher. Significant cell buck-
ling is observed in this region followed by folding of cell walls. Previous studies
Anandan et al. 13

Figure 6. Damage evolution in the facesheet during three-point bending. HSNFCCRT stands for
Hashin’s Fiber Compression Failure Criterion.

Figure 7. Failure in composite sandwich structures under three-point bending. HSNFCCRT

stands for Hashin’s Fiber Compression Failure Criterion. Damaged regions are colored red.

have observed a densification phenomenon once the cell walls come in contact
with each other [7]. In the current study, the puncher displacement was stopped
before such a point was reached. The composite facesheets show a variety of failure
modes. Compressive failure is clearly observed in the region directly under the
puncher. In addition to compressive failure, interlaminar failure is also observed
in the facesheet. This interlaminar failure propagates down the length of the sand-
wich composite, away from the region under the puncher. Figure 7 shows the
location of failure predicted by the FEA model. Facesheet failure due to compres-
sion is observed in the location beneath the puncher. Location of facesheet com-
pressive failure observed in the experiments corresponds well with the predictions
14 Journal of Sandwich Structures & Materials 0(0)

Figure 8. Progression of failure in the honeycomb core. (a) initiation of failure (b-f) failure
progression.

by the numerical model. Total facesheet failure is seen in the regions directly under
the puncher. From the experiments it was observed that the facesheet failure
occurred due to the combination of additional failure modes. A combination of
compressive failure and propagation due to interlaminar failure was observed.
The interlaminar failure was not observed in the simulations because the facesheets
were modeled using a single layer of shell elements.
The meso-scale model for the honeycomb allows simulation of localized cellular
failure. Initiation of failure in the sandwich composite was due to core compression
in the region under the puncher. The core materials in composite sandwich struc-
tures are subjected to shear and normal loads, when loaded in a three-point bend-
ing configuration. Gdoutos and Daniel reported that in long span beams under
three-point bending, a complex biaxial state of stress can exist in the core, which
leads to failure [23]. Figure 8 shows the detailed images of progression of failure in
the honeycomb core. Failure initiation is observed in Figure 8(a) and (b) by local
buckling of the honeycomb cell walls. On continued loading, crushing of the core is
observed as in Figure 8(b) to (d). The localized failure leads to stress concentra-
tions which initiates failure in the composite facesheet. Failure progresses due to
plastic buckling in the core. Significant plastic failure is observed, as shown in
Figure 9. The regions of plastic buckling correspond to plastic buckling failure
of aluminum honeycombs under out-of-plane loads as mentioned by Gibson and
Ashby [3]. A comparison of experimental and numerical failure due to core crush-
ing is shown in Figure 9. The mode of failure observed in experiments corresponds
well with the core failure progression predicted by the numerical model.
Anandan et al. 15

Figure 9. Core failure in three-point bending. Comparison between experimental and simula-
tion results is shown.
PE: plastic strain.

Figure 10. Temperature dependent variation in the properties of honeycomb and facesheet.

Effect of elevated temperatures

The numerical model was used to predict three-point bending behavior of sand-
wich structures at elevated temperatures. The adhesive is not included in the
numerical model because failure is dominated by facesheet compressive failure
16 Journal of Sandwich Structures & Materials 0(0)

and core crushing, rather than core-facesheet debonding. The test temperature
chosen in this study was 100
C. The temperature falls within the service temper-
ature range of the film adhesive used in this study. Figure 10 shows the variation of
honeycomb core and facesheet properties with temperature. The core properties
were obtained from the Hexcel product datasheet [35]. Around 12% reduction in
strength is observed when the ambient temperature is increased to 100
C. Previous
studies have reported that the properties of the composite laminates are dependent
on the service temperature [36]. The thermoset matrix dominated properties tend
to be sensitive to temperature changes. The longitudinal properties are not as
sensitive to changes in temperature changes because they are fiber dominated. It
is expected that 90
compression, 90
tension, and interlaminar shear strength will
reduce when temperature is increased. In this study, facesheet properties at room
temperature and 121
C were obtained from the Cycom 5320-1 datasheet.
Intermediate values were obtained using linear interpolation. Figure 10 shows
the variation of facesheet compressive strength in the 90
direction with tempera-
ture. A reduction of 14% is observed as temperature is increased to 121
C from
room temperature. Temperature dependent properties were incorporated into the
numerical model. The “predefined field” option in Abaqus was used to define the

Figure 11. Comparison of load–displacement curves at room temperature and high tempera-
ture (100
C). Experimental results are shown in the figure.
Anandan et al. 17

test temperature. Simulations were then utilized to predict the drop in strength of
the honeycomb sandwich structure at elevated temperatures.
Representative curves of experiments at room temperature and elevated tem-
perature are shown in Figure 11. The results suggest that the strength of the
sandwich structure drops by 9.2% when the temperature is increased from room
temperature to 100
C. However, the slope of the elastic region does not show a
significant variation when the temperature is increased. This can be due to the
relative temperature insensitivity of the facesheet moduli as shown in Table 1. In
addition, the variation in modulus of aluminum is also minimal in this temperature
range. As a result, a change in temperature has a significant effect on the load-
bearing capacity but a minimal effect on sandwich stiffness. The progressive
damage propagation in the facesheet is more significant at room temperature com-
pared to elevated temperature. One possible reason for this observation can be a
reduction in interlaminar shear strength of the facesheet, which reduces the energy
absorbed by this failure mode. The loads at stabilized damage propagation are
very similar in both cases.
Results of simulation and experiments at elevated temperatures are shown in
Figure 12. There is good agreement between experimental and numerical results

Figure 12. Experimental and numerical load–displacement curves at elevated temperature

of 100
C.
18 Journal of Sandwich Structures & Materials 0(0)

with reference to load-carrying capacities of the sandwich. Failure under high

temperature tests shows a similar general trend as compared to the room temper-
ature simulations. There is also good agreement between the numerical results and
simulation in the progressive failure zone.

Effect of puncher location

As mentioned by Giglio et al., the exact position of the puncher with respect to
honeycomb cells may affect the results of the numerical model [6]. The force dis-
placement behavior can vary with the position of the central axis of the puncher
with respect to the center of a cell. The relative position of the puncher was varied
in the numerical model. Three different positions were evaluated as shown in
Figure 13(a). P1 corresponds to the position used in the baseline numerical
model, where the axis of the puncher is in line with the center of a honeycomb

Figure 13. Effect of relative position of the puncher on results of the three-point bending tests.
(a) Point of contact of the puncher relative to the honeycomb cell. (b) Effect of position on load–
displacement curves.

Figure 14. Variation of localized damage with a change in puncher location.

Anandan et al. 19

cell. In P3, the axis is in line with the end of a honeycomb cell. For this study,
the span length was kept constant. Figure 13(b) shows the variation in load–
displacement behavior. The overall trend shows little change. There is a small
variation in the damage propagation region, which is affected by the local failure.
The localized failure is difficult to examine experimentally but can be easily
investigated using simulations. Figure 14 shows the variation of localized failure
in the core as the puncher is moved to the right. It can be observed that the point of
maximum deformation shifts as the puncher is moved. However, this change in
localized failure did not affect load to failure or the linear portion of the load–
displacement curve. Therefore, changes in the relative position of the puncher
relative to the honeycomb cells do not affect the results of the test in a significant
manner.

Conclusions
This paper focused on modeling of a composite sandwich structure under three-
point bending. The sandwich structure had facesheets made of carbon/epoxy com-
posites and the core material was aluminum honeycomb. Hashin’s failure model
was used to simulate damage propagation in the facesheets. Experimental valida-
tion was also performed using three-point bending tests on sandwich structures
with carbon/epoxy facesheets and aluminum honeycomb core. Simulation results
showed good accuracy in predicting the load-carrying capacity as well as overall
damage behavior in the sandwich composite. Local core crushing was simulated
accurately. Failure of cell walls due to local buckling was observed in simulations
as well as experiments. Samples tested at room temperature exhibited failure at an
average load of 6446 N. When the temperature was increased to 100
C, the
strength of the sandwich structure decreased by 9.2% to 5848 N, while the stiffness
did not change significantly. This effect was also captured effectively by the numer-
ical models. Simulation models predicted a failure load of 7257 N at room tem-
perature and 5968 N at 100
C. The location of the puncher axis relative to cell
geometry is a parameter difficult to determine using experiments and was evalu-
ated using the numerical model. It was observed that the sandwich in the relative
position of the puncher affects the local failure modes while the load–displacement
behavior is not affected significantly.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, author-
ship, and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for the research, author-
ship, and/or publication of this article: This research is sponsored by the Industrial
20 Journal of Sandwich Structures & Materials 0(0)

Consortium of the Center for Aerospace Manufacturing Technologies (CAMT) at Missouri

University of Science and Technology.

ORCID iD
K Chandrashekhara http://orcid.org/0000-0001-5973-2132

References
1. Vinson J. The behavior of sandwich structures of isotropic and composite materials.
Lancaster: Technomic, 1999.
2. Okereke M, Akpoyamare A and Bingley M. Virtual testing of advanced composites,
cellular materials and biomaterials. Composites Part B 2014; 60: 637–662.
3. Gibson L and Ashby M. Cellular solids, structure and properties. Cambridge: Cambridge
University Press, 1999.
4. Heimbs S. Virtual testing of sandwich core structures using dynamic finite element
simulations. Comput Mater Sci 2009; 45: 205–216.
5. Li K, Gao X and Wang J. Dynamic crushing behavior of honeycomb structures with
irregular cell shapes and non-uniform cell wall thickness. Int J Solids Struct 2007;
44: 5003–5026.
6. Giglio M, Gilioli A and Manes A. Numerical investigation of a three point bending test
on sandwich panels with aluminum skins and Nomex honeycomb core. Comput Mater
Sci 2012; 56: 69–78.
7. Giglio M, Manes A and Gilioli A. Investigations on sandwich core properties through
an experimental–numerical approach. Composites Part B 2012; 43: 361–374.
8. Seeman R and Krause D. Numerical modelling of Nomex honeycomb sandwich cores
at meso-scale level. Compos Struct 2017; 159: 702–718.
9. Liu L, Wang H and Guan Z. Experimental and numerical study on the mechanical
response of Nomex honeycomb core under transverse loading. Compos Struct 2015;
121: 304–315.
10. Zhou G, Hill M, Loughlan N, et al. Damage characteristics of composite honeycomb
sandwich panels in bending under quasi-static loading. J Sandwich Struct Mater 2006;
8: 55–90.
11. Stocchi A, Colabella L, Cisilino A, et al. Manufacturing and testing of a sandwich panel
honeycomb core reinforced with natural-fiber fabrics. Mater Des 2014; 55: 394–403.
12. Masters I and Evans K. Models for elastic deformation of honeycombs. Compos Struct
1997; 35: 403–422.
13. Malek S and Gibson L. Effective elastic properties of periodic hexagonal honeycombs.
Mech Mater 2015; 91: 226–240.
14. Ijaz H, Saleem W, Zain-Ul-Abdein M, et al. Finite element analysis of bend test of
sandwich structures using strain energy based homogenization method. Adv Mater Sci
Eng 2017. DOI: https://doi.org/10.1155/2017/8670207
15. Sun G, Huo H, Chen D, et al. Experimental and numerical study on honeycomb sand-
wich panels under bending and in-panel compression. Mater Des 2017; 133: 154–168.
16. Xang C, Xu P, Yao S, et al. Optimization of honeycomb strength assignment for a
composite energy-absorbing structure. Thin Walled Struct 2018; 127: 741–755.
17. Sun G, Chen D, Huo X, et al. Experimental and numerical studies on indentation and
perforation characteristics of honeycomb sandwich panels. Compos Struct 2018;
184: 110–124.
Anandan et al. 21

18. Zhou H, Xu P, Xie S, et al. Mechanical performance and energy absorption properties
of structures combining two Nomex honeycombs. Compos Struct 2018; 185: 524–536.
19. Feng H, Liu L and Zhao Q. Experimental and numerical investigation of the effect of
entrapped air on the mechanical response of Nomex honeycomb under flatwise com-
pression. Compos Struct 2017; 182: 617–627.
20. Petras A and Sutcliffe M. Failure mode maps for honeycomb sandwich panels. Compos
Struct 1999; 44: 237–252.
21. ASTM Standard C393. Standard test method for core shear properties of sandwich
constructions by beam flexure. ASTM International, www.astm.org (2016, accessed
17 July 2018).
22. Harris B and Crisman W. Face wrinkling mode of buckling of sandwich panels. J Eng
Mech Div-ASCE 1965; 91: 93–111.
23. Gdoutos E and Daniel I. Failure modes of composite sandwich beams. Theor Appl
Mech (Belgr) 2008; 35: 105–118.
24. Yussuff S. Theory of wrinkling in sandwich construction. J R Aeronaut Soc 1955;
59: 30–36.
25. Zalewski B, Dial W and Bednarcyk B. Methods for assessing honeycomb sandwich
panel wrinkling failures. NASA Report, NASA/TM-2012-217697, October 2012.
26. Thomsen O and Banks W. An improved model for the prediction of intra-cell buckling
in CFRP sandwich panels under in-plane compressive loading. Compos Struct 2004;
65: 259–268.
27. Soliman H and Kapania R. Equivalent constitutive behavior of sandwich cellular cores.
J Sandwich Struct Mater 2015; 19: 424–455.
28. Catapano A and Montemurro M. A multi-scale approach for the optimum design of
sandwich plates with honeycomb core. Part I: homogenisation of core properties.
Compos Struct 2014; 118: 664–676.
29. Lapczyk I and Hurtado J. Progressive damage modeling in fiber-reinforced materials.
Composites Part A 2007; 38: 2333–2341.
30. ABAQUS online users manual. Version 6.12, Simulia, 2015.
31. Cytec Solvay. Cycom 5320-1 epoxy resin system. Datasheet, 2015. Available at: www.
cytec.com
32. Zhang T, Yan Y and Li J. Experiments and numerical simulations of low-velocity
impact of sandwich composite panels. Polym Compos 2015; 38: 646–656.
33. Centea T, Grunenfelder L and Nutt S. A review of out-of-autoclave prepregs – material
properties, process phenomena, and manufacturing considerations. Composites Part A
2015; 70: 132–154.
34. ASTM Standard D7250. Standard practice for determining sandwich beam flexural and
shear stiffness. ASTM International, 2016. Available at: www.astm.org
35. Hexcel Corporation. Hexweb honeycomb attributes and properties. Datasheet, 2017.
Available at: www.hexcel.com
36. Guo Z, Feng J, Wang H, et al. A new temperature-dependant modulus model of glass/
epoxy composite at elevated temperature. J Compos Mater 2012; 47: 3303–3310.