Professional Documents
Culture Documents
G. Burriesci
University of Palermo, Italy
Paper No. 2774. Manuscript received 19 July 2002, revised version received 11 November 2002, accepted 2 January 2003.
176 T H E A ERONAUTICAL JOURNAL M ARCH 2003
(a) (b)
in-plane and out-of-plane properties. The theory assumes that the in-
plane properties are described assuming the cell walls behaving like
Euler-Bernoulli or Timoshenko beams under combined bending and
axial deformation. The geometric parameters of the unit cell play an Figure 2. Geometric parameters of a honeycomb unit cell.
important role into the definition of the honeycomb’s mechanical
properties. Further theoretical developments(11) underline the fact
that hinge mechanisms should also be taken into account to predict
the in-plane properties. A combined numerical and experimental
entrant unit cell configuration, with a subsequent comparison with
analysis of the monoaxial loading of auxetic honeycombs has
results given by the analysis of Zhang and Ashby(17). The transverse
demonstrated the validity of the theoretical framework(12).
shear moduli have been derived from simulation of shear tests ASTM
A possible interest on auxetic honeycombs lies upon the fact that
C273-00, with theoretical and experimental validations, together with
the flexural rigidity and buckling load of a sandwich composite with
unit cell implicit FEM models to verify the transverse shear modulus
a negative Poisson’s ratio core could be significantly increased
behaviour on the TW plane versus the gauge thickness.
compared to a similar structure with a regular hexagonal
The electromagnetic behaviour of a honeycomb composite is
honeycomb(5). This fact could affect positively the vibroacoustic
governed by the effective permittivity of the material (for non-
design of sandwich shells with lowered cut-of frequency to increase
magnetic composites)(15). Permittivity is a measure of the material's
the transmissibility loss factors in selected frequency ranges(6).
electrical polarizability. The effective relative permittivity in the
Moreover, an auxetic honeycomb panel assumes a sinclastic curva-
three spatial directions (the honeycomb is uniaxially anisotropic) has
ture under deformation, thus leading to an increased resistance to
been calculated using a finite difference time domain technique (FD-
buckling loads and easy manufacturing of curved sandwich struc-
TD)(16,23). FD-TD is used to predict the reflection coefficient of the
tures(2). Sandwich panels are still largely in use for aircraft radome
honeycomb for the three unit cell orientations. These data are then
constructions and satellite antennae, given their reduced modal
inverted to obtain the honeycomb’s effective permittivity(16).
densities at middle frequency ranges(28). Sandwich composites may
Homogenized permittivity data from the honeycomb have been vali-
also be employed in structural aerospace components that require
dated by comparison with measured data obtained using a rectan-
low radar scattering characteristics (so-called stealth)(22). Low radar
gular wave-guide technique(16).
cross section (RCS) performance is achieved by including within the
The data from the prediction methods have been then used to
composite a mechanism for absorbing radar waves(15,22). The use of
derive parametric curves of mechanical properties versus dielectric
honeycomb in radar applications requires the composite to be truly
permittivities with different unit cell geometry layouts of the honey-
multifunctional i.e. the composite is required to exhibit enhanced
comb. These graphs can be used at the design stage to select honey-
performance in two quite distinct domains — the mechanical and
comb configurations for combined mechanical and EM target perfor-
electromagnetic.
mance. Moreover, structural and electromagnetic engineers, thus
In this paper the authors consider a compared analysis between
allowing a true multifunctional design of the sandwich structure, can
mechanical and electromagnetic properties of auxetic honeycomb
use the graphs at the same time.
structures with different unit cell geometric parameters. The
mechanical characteristics considered in this work are the Young’s
modulus and compressive strength in the transversal direction, and
the two shear moduli in the X1X3 and X2X3 planes (Fig. 2) or, alterna- 2.0 DESCRIPTION OF MECHANICAL
tively, in the LW (ribbon) and TW (transverse-to-ribbon) PROPERTIES
planes(13,18). These mechanical characteristics are essential in order
to define the fundamental mechanical performance of a sandwich The in-plane and out-of-plane mechanical properties of auxetic and
structural element(5). general honeycombs can be adequately described using the cellular
The Young’s modulus and the compressive strength in the trans- material theory (CMT) by Gibson and Ashby(5). The theory is based
versal direction have been determined through a numerical simula- on the knowledge of the mechanical properties of the constituent
tion of the ASTM C393-00 standard for flatwise compression of phase of the core material, as well as geometric parameters of the
non-metallic honeycomb cores(13). The simulation has been unit cell layout, as shown in Fig. 2. The in-plane properties are
performed using an explicit finite element model set-up with the LS- modelled considering the cell walls behaving like Euler or Timo-
DYNA 3D code(14). The model has been then prepared for a re- shenko beams in combined flexular-axial deformation. The Euler-
SCARPA E T A L M ECHANICAL AND ELECTROMAGNETIC BEHAVIOUR OF AUXETIC HONEYCOMB STRUCTURES 177
Figure 3. In-plane Poisson’s ratio v12 versus internal cell angle and Figure 4. Comparison between theoretical, FEM and experimental
wall aspect ratio. values for the transverse Young’s modulus E3.
Bernoulli approximation is considered valid for low relative density higher values of β the following formula has to be implemented(27):
values (β < 0⋅2), while for increased wall thickness shear correction
factors proper of Timoshenko beam theory are required(5). The out-
áCosí − â
of-plane transverse Young's modulus is calculated scaling the phase è lim = Sin −1 . . . (5)
core material modulus by the area of the load-bearing section. The 2Cosí
two transverse shear moduli are computed using the theorems of
minimum potential energy and minimum complementary energy. Figure 3 shows the behaviour of the Poisson’s ratio ν12 versus the
The theorems give upper and lower bound values of the Voigt and internal cell angle, for various wall aspect ratio values, and a relative
Reissner type(5,21), which coincide with the shear modulus G23. The density β of 0⋅12. While the negative internal cell angle dictates the
other transverse shear modulus G23 presents two different values for auxetic behaviour of the honeycomb, the absolute values of the
the upper and lower bounds, and a numerical technique is required to Poisson’s ratio are significantly affected also by the cell wall aspect
compute an unique value considering the actual shear stress distribu- ratio α. As an example, for α = 0⋅5 and θ = –5°, the theoretical value
tion over the gauge thickness. for ν12 would be –9⋅43. For the corresponding positive internal cell
angle (θ = 5°), the in-plane Poisson’s ratio value would be 6⋅63. The
effect of shear deformation for the cell walls leads to an increase of
2.1 In-plane Poisson’s ratio values the Poisson’s ratio magnitude. For the same cases, and considering a
From the CMT results, the in-plane linear elastic mechanical proper- honeycomb core made of NOMEX aramid paper, with a Poisson’s
ties obey to the classic orthotropic relation: ratio νc = 0⋅4(17), the ν12 values would be respectively –9⋅70 and 6⋅82,
with an average increase of the 2⋅8%. The shear deformation of the
E1 í 21 = E2 í 12 . . . (1)
cell walls is particularly evident for small internal cell angles, while
for larger θ values the effect is less relevant(10). All these results indi-
Considering both in-plane flexural and axial deformation behaviour, cate that, when loading in the 1 direction, the transverse displace-
the two in-plane Poisson’s ratio can be described as following(5): ments are larger for auxetic honeycombs. This feature is in agree-
ment with the theoretical increase in deformation that a solid
Cos 2è 1 + (1 ⋅ 4 + 1 ⋅ 5ν c )â 2 composed by a positive strain energy homogeneous material with
ν12 =
( )
Sinè (á + Sinè ) 1 + 2 ⋅ 4 + 1 ⋅ 5ν c + Cot 2 θ â 2
. . . (2) negative Poisson's ratio values would exhibit when subjected to
loading(2). Applications for this large deformation behaviour of re-
entrant structures have been sought in MEMS (micro electro-
mechanical systems) technologies and microsensor components(25),
Sinè (α + Sinè ) 1 + (1 ⋅ 4 + 1 ⋅ 5 í c )â 2
ν 21 = where displacement forces are an important scale factor.
Cos è2
( )
1 + 2 ⋅ 4 + 1 ⋅ 5í c + 2 á Cos 2 θ â 2
. . . (3)
Table 1
Comparison between theoretical, numerical and experimental values of collapse stress due to buckling
Table 2
Comparison between theoretical and numerical values of collapse stress due to brittle fracture
the end-conditions (simply supported or fully clamped) of the unit 2.3 Transverse shear moduli G13 and G23
cell walls. One has to note that for unitary values of α the value of
The overall bending stiffness of sandwich beams and plates is
the compressive strength given by Equation (9) is the same predicted
affected by the shear core resistance(5), while the face sheets
by Zhang and Ashby method(17). The numerical simulation gives a
contribute to increment the inertia moment of the structure cross
compressive stress value closer to the experimental one, although the section. A generic honeycomb material is a special orthotropic solid
predictedcollapse level still results lower than in experimental tests. with different shear moduli in the X1X3 and X2X3 planes(5). Accord-
This is probably due to the selected failure criteria, which is very ingto the CM Theory(8), the transverse shear modulus G13 is given
conservative and assumes that the resistance of the elements drops to by:
zero when the collapse stress is reached. A better understanding of
the failure mechanism could contribute to obtain more realistic Cosè
G13 = Gc β . . . (11)
predictions. Another factor could also be the high value for the á + Sinè
contact friction coefficient which has been introduced to enforce the
quasi-clamped conditions of the cell walls over the basement. In Fig. Also in this case, auxetic configurations lead to an increase of the
5 the stress contour maps for the above model are reported at shear modulus for some combinations of unit cell shapes. Figure 7
selected stages of the test simulation, with the progressive order shows that for small internal cell angles and reduced wall aspect
going from top-down, left-right. In particular, the maximum prin- ratio values, an auxetic honeycomb would have a steep increase in
cipal stress at the middle section is represented. the shear modulus value compared to regular hexagonal configura-
From the first figure of the sequence it is possible to observe that tion (α = 1⋅0, θ = 30º). For higher a values, the value of the shear
failure first occurs at the external segment of double thickness, modulus would be substantially constant. The values plotted in Fig.
where the stress level reduces after failure. During the test sequence, 7 come from Equation (11) and FEM simulations of the ASTM
stress concentrate in two in-plane sections close to the free edges, C273-00 shear test for non-metallic honeycomb materials(13). Also in
where the buckles due to the Eulerian load develop more signifi- this case, the FEM models represent honeycomb samples with the
cantly. With the increase of load and deformation, the stress inten- minimum overall dimensions allowed by the test standard. As in the
sity accumulates in the cell segments with double thickness. The flatwise-compressive test case, implicit quadrilateral four nodes
described stress distribution is regular and extended to the entire shells have been used to model the cell walls, with meshes having a
specimen, with exception of the external cells, especially along the total number of elements varying between 2,400 and 4,800 elements.
re-entrant sides. These arrays of cells, being partially unconstrained, The general agreement between the results is very good. The simula-
are freer to deform and tend to concentrate the stress at the median tions have been performed for honeycomb samples having a gauge
and external regions of the specimen. It is worth noticing that in the thickness ratio of 8. However, one has to bear in mind that the gauge
case of general honeycomb, previous experiences(27) showed a less thickness does affect (theoretically) the value of the transverse shear
uniform distribution, characterised by stresses concentration along modulus G13. On the contrary, for the shear modulus G23 the height
the median line, in the direction of the double thickness segments of the honeycomb affects the shear stress distribution over the cell
(see Fig. 6 and Table 1). walls(20,21). The shear modulus value G23 is contained between an
For thick brittle honeycombs the collapse stress is given by brittle upper limit (Reissner bound) and a lower value (Voigt bound)(21). A
failure(17), which is directly proportional to density of the honey- semi empirical formula to relate the two bounds into a unique value
comb and the brittle fracture stress of the core material σbf: is the following(21):
ñ
ó *3 = ⋅ ó bf
ñc
. . . (10) G23 = G23
low
+
K
bl
(
up
G23 − G23
low
) w
. . . (12)
Equation (10) leads to some interesting features for auxetic honey- 0 ⋅ 787 è ≥ 0 0
combs, because their density increases for wall aspect ratios up to 2 for where K =
1 ⋅ 342 è < 0
0
the same relative density β(11). Equation (10) has been validated also
with the same explicit FEM model developed for the buckling compres-
sive load, but this time with a relative density β of 0⋅2. Table 2 shows The values of K have been calculated through a nonlinear least
the comparison between the results from Equation (10) and the numer- square fitting algorithm from 2,400 data sets coming from FEM
ical simulation. As one can observe, the convergence is very good. homogenisation process of unit cells with different layouts. The
180 T H E A ERONAUTICAL JOURNAL M ARCH 2003
Figure 6 (c-b). Flatwise compression deformation of an auxetic honeycomb at different time steps (left-right, top-down).
SCARPA E T A L M ECHANICAL AND ELECTROMAGNETIC BEHAVIOUR OF AUXETIC HONEYCOMB STRUCTURES 181
are three independent design variables (α, θ and β), compared to one conventional honeycomb of increasing β or εc is simultaneously to
(β) in a conventional honeycomb; thus the radar signature designer increase =ε 11, =ε 22 and =ε 33; =ε 11, =ε 22 and =ε 33 cannot be varied indepen-
has much greater freedom to choose an optimised permittivity, and dently. In a re-entrant honeycomb, θ and α are not fixed. From the
therefore an optimised radar signature. The effective relative permit- electromagnetic design viewpoint, control of θ and α facilitates a
tivity of a honeycomb is Ref. 23: much greater control over the honeycomb’s effective permittivity
182 T H E A ERONAUTICAL JOURNAL M ARCH 2003
Figure 10. Out-of plane honeycomb mechanical properties versus Figure 12. Honeycomb dielectric permittivities versus nondimensional
in-plane Poisson’s ratio. transverse Young’s modulus.
4.0 COMMENTS
The results from literature and tests carried out by the present
workers on the rectangular waveguide outline the validity of the
numerical simulations about the compressive and shear mechanical
properties of the honeycombs, as well as the finite difference-time
domain technique developed to predict the dielectric properties of
the cellular solids. Honeycomb materials are used in absorber and
radome applications(22) and, more general, in airframe constructions.
The possibility of increasing the degree of anisotropy both from the
mechanical and dielectric point of view using auxetic honeycombs
Figure 11. Honeycomb dielectric permittivities versus in-plane allows the aerospace structural and electromagnetic analyst to intro-
Poisson’s ratio. duce more degrees of freedom into the integrated design of aero-
space components.
Figure 10 shows the behaviour of the out-of-plane mechanical prop-
erties of general honeycombs (Young’s modulus in the X3 direction
than is possible in a conventional honeycomb. and transverse shear moduli G13 and G23) with wall aspect ratio α =
A FD-TD -based method utilising periodic boundary conditions 2⋅4 and relative density β = 0.12 versus the honeycomb Poisson's coef-
has been developed to predict the effective permittivity of a dielectric ficient ν21. The normalised gauge thickness is 3. For increasing nega-
honeycomb(15,23). The reflection coefficient is predicted for the case tive Poisson’s ratio values the transverse shear modulus G23 increases,
with a minimum value around ν21 = 0, corresponsing to a rectangular
of an infinite planar layer of dielectric honeycomb subject to
unit cell honeycomb(8). The transverse shear modulus G23 tends to
planewave illumination from three mutually orthogonal directions.
increase once again for positive Poisson’s ratio values. It is worth
The reflection data are then inverted to obtain data on the honey- noticing that the magnitude of the transverse shear modulus values G23
comb’s effective permittivity. The data inversion process is based on remain higher in the negative Poisson's ratio field. This behaviour is
a conventional planewave transmission line model of the honeycomb. less marked for high gauge thickness, because the shear modulus
The inversion process homogenises the unit cell, replacing the inho- approaches the lower bound, which decreases for negative internal cell
mogeneous anisotropic material with an equivalent homogeneous angles(21). The transverse shear modulus G13 shows a slight increment
material. Using the homogenous model, the unit cell parameters can for negative Poisson’s ratio values, but these variations are negligible
be chosen to yield an appropriate permittivity characteristic for a for wall aspect ratios higher than 1⋅5(21). The Young's modulus shows
given radar absorption or radar window problem. Moreover, by large increase in magnitude for negative Poisson's ratio values, with
combining the electromagnetic and mechanical homogenised models, increments up to the 230% compared to cell layouts with correspon-
global optimisation for all engineering functions can take place. dent positive internal cell angles(21). However, the increase of trans-
The results from the FD-TD technique have been validated against verse Young’s modulus is accompanied by a correspondent increase in
experimental data gathered on NOMEX and cast epoxy resin samples honeycomb density(11). In any case, all linear elastic transverse
tested in a rectangular waveguide facility(16,23). The effective core mechanical properties scale linearly with the relative density β.
SCARPA E T A L M ECHANICAL AND ELECTROMAGNETIC BEHAVIOUR OF AUXETIC HONEYCOMB STRUCTURES 183