Professional Documents
Culture Documents
The manuscript was received on 2 September 2010 and was accepted after revision for publication on 22 March 2011.
DOI: 10.1177/1464420711406777
Abstract: In this contribution, the impact response and failure modes of square sandwich test
panels subjected to low-velocity impact were investigated. The two main groups of test speci-
mens investigated were: (a) bonded specimens – sandwich panels with skins glued to the hon-
eycomb core; (b) unbonded specimens – sandwich panels with skins not glued to the honeycomb
core. The focus of the study is on the effect of the adhesive and the various modes of failures on
the impact response of sandwich panels. A relatively detailed experimental investigation was
conducted in conjunction with a detailed finite element analysis. The experiments were con-
ducted using a standard dart test machine and the finite element analysis was carried out using a
commercially available finite element software. The results of maximum contact force, contact
duration, and failure modes are presented, compared, and discussed in this technical article.
densification and decreased back plate deflection, From force transducer on impactor
while increasing the plate thickness was also found Steel impactor
to decrease back plate deflection, although the panels φ 13.1 mm
then had a substantially higher overall mass.
Zhou [2, 15] have investigated almost all parame- sensor Dynamic Data
Acquisition Card
ters in the failure mode and energy absorption char-
acteristic using an extensive experimental study. In Specimen
their study, the change of the indenter nose shape is
shown to result in a change in the damage mecha-
φ 76.4 mm
nisms and have the most significant effect on energy
Clamp plate
absorption. Increasing the core density has a very 100 mm
Computer
little effect on the ultimate loads and energy-absorb-
ing capacity. The effect of the support conditions on Fig. 1 Schematic of the test set-up
the damage and energy-absorbing characteristics has
also shown to be small. Furthermore, different core
materials with a similar density show little difference
in either the damage or the energy-absorbing charac- was used for the drop-weight tests. In this test, the
teristics due to the limited contribution of transverse impact mass is kept at a constant of 2.65 kg through-
shear resistance. out the tests. Upon release, the free-falling impactor
The bonding between skins and the core is vital would fall along two smooth guided columns, and
for sandwich structures to make most of the advan- through the centre hole of the clamp plate of diame-
tage of their high structural stiffness. The load ter 76.4 mm to strike the specimen.
transfer mechanism between skins and the core The specimen was positioned between the top and
was studied using the finite element method by bottom clamp plates, with the mid-point of the plate
Burton and Noor [9] and a closed-form formula directly located underneath the impactor. The sup-
for calculating the stress in adhesive layers in sand- port fixture for the specimen facilitated circular
wich beams was deduced by Huang [10]. clamped condition. After the first impact, the
References on the bonding process and the rebound brake was activated to support the cross-
amount of adhesive in sandwich structures are head, and thus the impactor was only allowed to
also available [11, 12]. Although the delamination strike the specimen once. Transient response of the
crack along the skin/core has been reported by samples included the velocity and deflection of the
experiment [11, 13–16], the investigation into the impactor, as well as load, as a function of time. The
effect of adhesive on the mechanical properties of total displacement of impactor and bottom skin
sandwich panels is still absent. deflection was recorded as a function of time with a
The purpose of this article is to address the lack of data acquisition system. Graphical plots were then
research and publications on the role of adhesive in generated using data collected from the data acquisi-
impact loading environment. The study will explore tion system.
the role of adhesive in sandwich panel by comparing For the sandwich panels in these experiments,
the structural impact responses of bonded sandwich skins were made of acrylic material, while honey-
panels with those of unbonded sandwich panels. comb core was made of Nomex paper. The mechan-
Acrylics skins are used in place of composite skins ical properties of acrylic were obtained using
in sandwich panels, because the transparency of standard tests. While the compression property of
acrylics allows one to see the damage and failure Nomex paper has been investigated using standard
mechanisms in the core and also those occurring compression tests, these data were published in Foo
between the skin and the core. The study of the et al. [17]. Two groups of sandwich specimens were
impact response of the brittle monolithic acrylic conducted by drop-weight tests: (a) the skins were
panel is included as a comparison in this bonded with the core with Araldite 2012; (b) the
contribution. skins and the core were unbonded. Each specimen
is made of two acrylic plates of size 100 100 mm2
and a honeycomb core made of Nomex paper with
2 EXPERIMENTAL PROCEDURES a height 15 mm. The cell size of each honeycomb core
is 6 mm, while acrylic plates of thicknesses 1.5, 2, and
Figure 1 shows an illustration of the experimental set- 3 mm are used, respectively. The specimens were
up of the impact testing with digital data acquisition. impacted with velocities 0.7, 0.8, 0.9, 1.7, and 2.3 m/s.
The Instron Dynatup 8250 impact testing machine In total, there were more than 30 specimens tested,
as listed in Table 1. Each configuration was tested panels. In the first stage, the contact force increases
at least twice to ensure that the test results were linearly with impact displacement until the force
repeated and thus consistent with some degree of reaches the initial threshold force. After that, the con-
reliability. tact force usually increases slightly and non-linearly
For bonded and unbonded sandwich panels as to reach the maximum force. In the last stage, the
well as the monolithic acrylic panel, curves of contact contact force decreases dramatically with impact
force with impactor displacement are shown in Figs 2 displacement.
and 3. It can be seen in these figures that there are In the case of the unbonded sandwich panels, the
three stages for the response of the contact force with initial response of the force versus displacement
impact displacement for the bonded sandwich curve is linear until it reaches a level that radial crack-
ing of the acrylic is initiated. After that, the load drops
and then the force increases and fluctuates with
increasing displacement until either the circumferen-
Table 1 Test configuration for Dynatup drop-weight
test
tial failure is achieved or the impactor rebounds.
These responses were similar to those of the mono-
Velocity (m/s) Thickness of acrylic (mm) Specimen type
lithic acrylic panels. The details of the low-velocity
0.7 1.5 2 3 Bonded impact response of acrylic panels have recently
1.5 2 3 Unbonded
0.8 1.5 2 3 Bonded
been reported [15].
1.5 2 3 Unbonded
0.9 1.5 2 3 Bonded
1.5 2 3 Unbonded
3 FINITE ELEMENT ANALYSIS
1.7 1.5 2 3 Bonded
1.5 2 3 Unbonded To investigate the impact response of failure process
2.3 1.5 2 3 Bonded
1.5 2 3 Unbonded in both the bonded and unbonded sandwich panels
(a) 0.6
(a) 0.7
Initial threshold force Bonded
0.6 0.5
Maximum force
Force, kN
0.4 0.3
0.3
0.2 Acrylics
0.2 Unbonded
0.1
0.1 Acrylics
0
0 0 1 2 3 4 5
0 2 4 6 8 10 12
Displacement of Impactor, mm
Displacement of Impactor, mm
(b) 1 (b) 1.2
0.9 Bonded
Bonded
1
0.8
0.7 0.8
Force, kN
0.6
Unbonded
Force, kN
0.5 0.6
0.4 Unbonded
0.4
0.3 Acrylics
0.2 0.2
0.1
0
Acrylics 0
0 2 4 6 8 10 0 1 2 3 4
Displacement of Impactor, mm Displacement of Impactor, mm
Fig. 2 Contact force versus displacement of impactor Fig. 3 Comparison of contact force versus displace-
for impact velocity 0.7 m/s ment plots
hc
(c) 3D view
hf
using finite element analysis, the drop-weight tests The mechanical property of Nomex core was
were simulated using ABAQUS/Explicit. A finite specified to be elastic–plastic with elastic modulus
element model of a sandwich panel subjected 3.40 GPa, Poisson’s ratio 0.4, and yield stress
to a low-velocity mass impact is shown in Fig. 4. 30 MPa. The dimensions of the core are kept constant:
Figure 4(a) shows the plan and side view of the core cell size of 6 mm and core thickness hc ¼ 15 mm.
finite element model; it is seen that clamped regions Acrylic material was specified as a linear elastic frac-
outside the radius of 38.2 mm are assigned with fully ture mechanics with an elastic modulus 3.13 GPa,
fixed constraints. Figure 4(b) gives the isometric view Poisson’s ratio 0.32, and the fracture stresses 55, 51,
of the solid model with and without the top skin. and 46 MPa for values of thickness hf 1.5, 2, and 3 mm,
The bonded and unbonded sandwich panels were respectively, obtained from standard ASTM D3039
modelled with real dimensions and geometries with standard tests [18].
the details shown in Fig. 4. The impactor was mod- Initial velocity was specified to the impactor and
elled as a hemispherical volume of radius 6.55 mm as the contact force, displacements, stress, and strain
in the experimental set-up. The input density of this of the system were output. For interaction property,
volume is equivalent to impact mass 2.65 kg. To general contact was defined between sandwich panel
save computational time, the impactor was placed and the impactor so that contact would take place
just on top of the sandwich panel and an initial veloc- automatically as soon as the skin failed.
ity was specified. The prescribed velocities are listed In the finite element models, the difference
in Table 1. between unbonded and bonded sandwich panels is
For compatibility, the core was meshed with four- that skins and the core were tied together for bonded
node thick shell reduced integration elements (S4R) sandwich panel while contact with friction was
and the skin was meshed with eight-node linear solid defined instead for unbonded sandwich panel.
reduced integration elements (C3D8R). The mesh Typical finite element results of the bonded and
near to contact area was refined and there were four unbonded sandwich panels are shown in Fig. 5.
elements in the thickness direction for skin after The comparisons showed the presence of adhesive
conducting mesh sensitivity analysis. on the load versus displacement response, the
deformation process, and the failure modes of the initial threshold force at least 1.5 times higher when
sandwich panel. subjected to impact load.
1.5
unbonded sandwich panel is used
Fbcrt 1
¼ crt ð1Þ
Funb
0.5
where Fbcrt and crt
Funb
are the respective threshold loads
0
of bonded and unbonded sandwich panels subjected 0.7 0.8 0.9 1.7 2.3
to impact load. The ratios of initial threshold loads Impact Velocity, m/s
with different velocities were plotted in Fig. 6.
Fig. 6 The ratios of initial threshold force of bonded
It can be seen in Fig. 6 that the ratio ranged from sandwich panel to that of unbonded sandwich
1.5 to 2.5. This means, for a sandwich panel with brit- panel
tle skins, bonding the skins to the core produces an
(b)
(a) 0.6
Bonded
0.5
0.4 A
Unbonded panel (Point A)
Force, kN
0.3
0.2 B Unbonded
0.1
0
0 2 4 6 8 10 12
Displacement of impactor, mm
Fig. 5 Finite element results for sandwich panels (skin thickness 1.5 mm and velocity 0.7 m/s)
Table 2 Maximum force values for bonded and unbonded sandwich panels
Velocity (m/s)
Thickness (mm) Skin conditions 0.7 0.8 0.9 1.7 2.3
1.5 Bonded 0.46 0.60 0.55 0.73 0.73
Unbonded 0.21 0.40 0.45 0.59 0.53
2 Bonded 0.63 0.66 0.73 0.84 0.95
Unbonded 0.32 0.43 0.45 0.89 0.93
3 Bonded 0.99 1.17 1.15 1.22 1.19
Unbonded 0.65 0.51 0.74 1.19 1.48
1.8
unbonded sandwich panel. The structural stiffness
1.6 initial threshold force of bonded sandwich panel increases with increasing
maximum force thickness of the skin, which agrees well with former
1.4 research work. It is interesting to note that the stiff-
ness of unbonded sandwich panel decreases with
1.2 increasing thickness of the skin. The fact that the
stiffness of unbonded sandwich panel decreases
1
0.5 1 1.5 2 2.5 with increasing thickness of skins is due to the sliding
Impact velocity, m/s of the core and hence, the inability of the sandwich
panel to transfer the stress in the core thickness
Fig. 7 The ratio of maximum force of bonded sand- direction. This is observed in the finite element
wich panel to that of unbonded sandwich panel result of unbonded sandwich panel which shows
for different impact velocities
the core under the area of contact slants. After
deformation, the core deforms significantly and
thereby the structural stiffness decreases significantly
thereafter.
the maximum force for both cases can be regarded as The downward movement of the impactor is com-
the same. posed of bending and shear deformation as well as
The ratio of maximum force for bonded sandwich the indentation of the top skin. Bending and shear
panel, Fbmax , to that of the unbonded sandwich panel, stiffnesses of bonded sandwich panel are given by
max
Funb , is calculated using the following [15, 17]
The ratios and for each impact velocities were 4Gc ðhc þ hf Þ2
Ks ¼ ð4Þ
averaged and compared, as shown in Fig. 7. The hc ð1 þ 2 ln Rp =Rc Þ
figure shows that for low velocity, the maximum
force of bonded sandwich panel is nearly double where E and
are the elastic modulus and Poisson’s
that of the unbonded sandwich panel. However, for ratio of skin and Gc the shear modulus of the core; hc
higher impact velocity, there is little difference and hf the height of the core and thickness of the
between the two. In contrast, the ratios for initial skins; Rc Rp and the contact radius and boundary dis-
threshold force hovers around the factor of two tance of the panels.
The relationship of indentation to contact force is
throughout the range of impact velocity studied.
quite complex and has been investigated extensively
[6, 15] and in the linear elastic region, the slope of
4.3 Structural stiffness of sandwich panels load–indentation adopted here was from Yigit and
Christoforou [19]
As discussed in the earlier section, the initial stage of
load–deflection curve is rather linear and the slope of K ¼ 20 Ri ð5Þ
where 0 is the yield stress for ductile material and The transverse stiffness for the plate resting on
fracture stress for brittle material. elastic foundation is also given [20]
The structural stiffness of the bonded sandwich pffiffiffiffiffiffiffi
panels, K , is therefore determined as Ktop ¼ 8 kD ð8Þ
3
f
Eh
1 1 1 1 where D ¼ 12ð1
2 Þ is the bending rigidity of skin. The
¼ þ þ ð6Þ bottom skin which is subjected to uniform pressure
K K Kb Ks
from the elastic foundation approximately and the
The structural stiffness of unbonded sandwich transverse stiffness can be written as
panel can be deduced from a circular plate resting
64D
on elastic foundation, as shown in Fig. 8. To explore Kbottom ¼ ð9Þ
the relationship between the contact force with the Rp2
displacement of the impactor, the top skin is mod- Considering that the load acting on the unbonded
elled as a plate resting on elastic foundation with a sandwich panel can be transferred to boundary from
modulus of foundation in thickness direction given 0
top skin to bottom skin, the structural stiffness of K
by Timoshenko and Krieger [20] can be written as
Ec 0
K ¼ Ktop þ Kbottom ð10Þ
k¼ ð7Þ
hc
Substituting equations (7) to (9) into equation (10),
where and Ec are the relative density and Young’s it yields
modulus of the core. The top skin is subjected to a sffiffiffiffiffiffiffiffiffiffiffiffi
central concentrated load, P, and resting on the foun- 0 Ec D 64D
dation, while the bottom skin is subjected to the pres- K ¼ 8 þ ð11Þ
hc Rp2
sure from the elastic foundation.
The results obtained using equations (6) and (11)
are compared with data obtained from the experi-
ments, and these are shown in Fig. 9.
P In bonded sandwich panel, the core and the skins
are perfectly merged together and the stiffness would
Skin
decrease only if the core buckles or the skins fail.
However, in unbonded sandwich panel, the core
and the skins are independent; thus, the shear trans-
Core k ferring from top skin to the core and to bottom skin is
lacking. In most sandwich structures, the strength of
core is much less than the skins and this is also true
Skin
for their stiffness. As a result, the core does not con-
tribute significantly in unbonded sandwich panels
Fig. 8 Sketch of an unbonded sandwich panel with a with thicker skin. As shown in Fig. 9, the stiffness of
central concentrated load
bonded sandwich panel is more than 3.5 times larger
1.8 1.8
1.6 1.6
Structural Stiffness, MN/m
Structural Stiffness, MN/m
1.4 1.4
1.2 Bonded 1.2 Bonded
1 Unbonded 1 Unbonded
Eq.(6) Eq.(6)
0.8 Eq. (11) 0.8 Eq. (11)
0.6 0.6
0.4 0.4
0.2 0.2
0 0
1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2
Thickness of skin, mm Thickness of skin, mm
Fig. 9 The structural stiffness of bonded and unbonded sandwich panels in comparison with
analytical solution (symbols represent experimental data and lines theory)
than that of the unbonded sandwich panel with the causes the core to buckle. At the final failure stage,
same skin thickness. circumferential cracks appear in top skin and the
contact force decreases sharply.
In the finite element results for the bonded sand-
4.4 Failure mechanisms
wich panel, the conjunct nodes will not separate
Looking at the failure mechanism involved, it can be during impact. However, for the unbonded sandwich
seen that both bonded and unbonded sandwich panel, separation appears. Take the specimen of the
panels failed in a similar fashion. At first, the panels unbonded sandwich panel with skin thickness 3 mm;
deform linearly at the onset of contact. With the fur- for example, a representative node, N, which is a node
ther downward movement of the impactor, radial at the common point between the core and skin, was
cracks start to appear on the bottom surface of the selected to examine the sliding of the skin over the
top skin due to tensile stress exceeding the fracture core. A node in the core facing the impactor is
stress. Contact force increases nearly linearly with the selected and the compressive stress of this node was
displacement of impactor during this stage. Then, the compared for both the bonded and unbonded sand-
top skin deforms significantly into the core and wich panels, as shown in Fig. 10. The strain of the core
18 Exp-Bonded
x x 1.4 Exp-Unbonded
16 FEA-Bonded
FEA-Unbonded B
onset of U
Compressive Stress, MPa
14 1.2
fracture
12
1
10
Force, kN
0.8
8
6 0.6
4
Bonded 0.4
2
Unbonded
0 0.2
0 0.2 0.4 0.6 0.8 1 1.2
Displacement of impactor, mm 0
0 1 2 3 4 5 6
Fig. 10 Compression stress of the core for bonded and Displacement of Impactor, mm
unbonded sandwich panels in elastic stage
Fig. 12 Load–deflection curves of test panels with
for skin thickness 3 mm and impact velocity
thickness 3 mm and velocity 1.7 m/s
2.3 m/s
1.8
1.2 Exp-Bonded
Exp-Bonded Exp-Unbonded B
Exp-Unbonded 1.6 FEA-Bonded
FEA-Bonded FEA-Unbonded
1 U
1.4
FEA-Unbonded
0.8 1.2
Force, kN
Force, kN
1
0.6
0.8
0.4 0.6
0.4
0.2
0.2
0 0
0 2 4 6 8 10 0 2 4 6 8
Displacement of Impactor, mm Displacement of Impactor, mm
Fig. 11 Load–deflection curves of panels with thick- Fig. 13 Load–deflection curves of test panels with
ness 2 mm and velocity 2.3 m/s thickness 3 mm and velocity 2.3 m/s
Fig. 14 Fracture for bonded panel with skin thickness 2 mm and impact velocity 0.9 m/s
in bonded sandwich panel increases nearly linearly 1. The acrylic skin in bonded sandwich panels failed
but the strain in unbonded sandwich panel increases locally, which is very similar to that of composite
non-linearly. sandwich panel.
The obvious differences between the failure mech- 2. The skin in unbounded sandwich panel failed with
anism in bonded and unbonded sandwich panels are a larger area of radial and circumferential cracks.
as follows. The core does not affect the skin failure.
Fig. 16 Fracture for unbonded panel with skin thickness 2 mm and impact velocity 1.7 m/s
3. The structural stiffness of bonded sandwich panel response of honeycomb sandwich panels subjected
is more than three times that of unbonded sand- to blast loading. Mater. Des., 2010, 31, 1887–1899.
wich panel. Thicker skin enhances the structural 9 Burton, W. S. and Noor, A. K. Structural analysis
of the adhesive bond in a honeycomb core sand-
stiffness of bonded sandwich panel distinctly but
wich panel. Finite Elem. Anal. Des., 1997, 26,
has little influence on the structural stiffness of 213–227.
unbonded sandwich panel. 10 Huang, S. J. An analytical method for calculating the
4. The failure modes are well predicted though the stress and strain in adhesive layers in sandwich
paths to failure are not satisfactorily predicted. beams. Compos. Struct., 2003, 60, 105–114.
11 Jen, Y. M., Ko, C. W., and Lin, H. B. Effect of the
New but simple expressions to calculate the initial amount of adhesive on the bending fatigue strength
stiffness of the bonded and unbonded panels were of adhesively bonded aluminum honeycomb sand-
also developed and presented in this contribution. wich beams. Int. J. Fatigue, 2009, 31, 455–462.
The comparisons of results show the usefulness of 12 Rion, J., Leterrier, Y., and Jan-Anders, X. Prediction
these expressions. It is anticipated that the effect of of the adhesive fillet size for skin to honeycomb core
impact velocity and the use of the actual force–dis- bonding in ultra-light sandwich structures.
Composites: Part A, 2008, 38, 1547–1555.
placement response of the contact point should be 13 Can, W., Hao-ran, C., and Zhen-kun, L.
included in these expressions. Experimental investigation of interfacial fracture
behavior in foam core sandwich beams with visco-
elastic adhesive interface. Compos. Struct., 2010, 92,
ß Authors 2011
1085–1091.
14 Goswamia, S. and Becker, W. The effect of face-
REFERENCES sheet/core delamination in sandwich structures
under transverse loading. Compos. Struct., 2001,
1 Zhang, T. G. and Stronge, W. J. Theory for ballistic 54(4), 515–521.
limit of thin ductile tubes hit by blunt missiles. Int. J. 15 Zhou, D. W. and Stronge, W. J. Dynamic indentation
Mech. Sci., 1996, 18, 735–752. of lightweight sandwich panels. In Proceedings of
2 Zhou, D. W. and Stronge, W. J. Mechanical properties the 11th European Conference on Composite mate-
of fibrous core sandwich panels. Int. J. Mech. Sci., rials, 31 May–3 June 2004, Rhodes, Greece.
2005, 47, 775–798. 16 Mines, R., McKown, S., and Birch, R. S. Impact of
3 Abrate, S. Impact on composite structures, 1998 aircraft rubber tyre fragments on aluminium alloy
(Cambridge University Press, UK). plates: I Experimental. Int. J. Impact Engng., 2007,
4 Michelle, S., Fatt, H., and Sirivolu, D. A wave propa- 34, 627–646.
gation model for the high velocity impact response of 17 Foo, C. C., Chai, G. B., and Seah, L. K. Mechanical
a composite sandwich panel. Int. J. Impact Engng., properties of Nomex material and Nomex honey-
2010, 37, 117–130. comb structure. Compos. Struct., 2007, 80(4),
5 Shaker, M., Ko, F., and Song, J. Comparison of the low 588–594.
and high velocity impact response of Kevlar fiber- 18 Zhu, S. and Chai, G. B. Ductile and Brittle Material
reinforced epoxy composites. J. Compos. Techn. Res., Failures in Low-Velocity Impact. Proceedings of the
1999, 21(4), 1–6. Institution of Mechanical Engineers, Part L: Journal
6 Styles, M., Compston, P., and Kalyanasundaram, S. of Materials Design and Application, 2010, 224(4),
The effect of core thickness on the flexural behaviour 162–172.
of aluminium foam sandwich structures. Compos. 19 Yigit, A. S. and Christoforou, A. P. On the impact
Struct., 2007, 80, 532–538. between a rigid sphere and a thin composite lami-
7 Dai, G. M. and Zhang, W. H. Size effects of basic cell in nate supported by a rigid substrate. Compos. Struct.,
static analysis of sandwich beams. Int. J. Solids Struct., 1995, 30, 169–177.
2008, 45, 2512–2533. 20 Timoshenko, S. and Krieger, S. W. Theory of
8 Chi, Y., Langdon, G. S., and Nurick, G. N. The influ- plates and shells, 2nd ed., 1959 (McGraw-Hill,
ence of core height and face plate thickness on the New York).