Professional Documents
Culture Documents
Experimental and Numerical Evaluation of The Mechanical Behaviour of GFRP Sandwich Panels
Experimental and Numerical Evaluation of The Mechanical Behaviour of GFRP Sandwich Panels
www.elsevier.com/locate/compstruct
Abstract
This work deals with the analysis of the mechanical behaviour of a class of sandwich structures widely employed in marine construc-
tions, constituted by fiber-glass laminate skins over PVC foam or polyester mat cores. In detail, a systematic experimental study and
numerical simulations have shown that the theoretical prediction of the strength and the actual failure mechanism of these sandwich
structures can be affected by significant errors, specially in the presence of prevalent shear loading. Moreover, because of the low shear
stiffness and the elastic constants mismatch of the skins and core material, failure modes and strength are strongly influenced by eventual
stresses orthogonal to the middle plane of the sandwich. In particular, for the sandwich structures with a PVC foam core, such a stress
interaction leads to early skin–core delamination failure, whereas for those with a polyester core it leads to core shear-cohesive failure. By
means of accurate non-linear simulations, accurate failure criteria, that can be used at the design stage in the presence of complex load-
ing, have also been developed.
Ó 2006 Elsevier Ltd. All rights reserved.
*
Corresponding author.
As shown in the literature [8–10], the failure of a sand-
E-mail addresses: russo@dima.unipa.it (A. Russo), zuccarello@ wich structure can occur through several damage mecha-
dima.unipa.it (B. Zuccarello). nisms such as:
0263-8223/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2006.10.007
576 A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586
(1) skin compressive/tensile failure [13]; N is the shear stiffness provided by the following formulas
(2) core shear failure [14,15]; [22]:
(3) delamination, i.e. skin–core debonding [16];
ðh3 h3c Þ
(4) local skin wrinkling [17–19]; and D ¼ Es ð3Þ
(5) core indentation failure, related to high specific pres- 12ð1 v2s Þ
sure [20,21]. Gc ðh þ hc Þ2
N¼ ð4Þ
4hc
In general, in the case of failures like type (1) and (2) it is
It has to be noted that because the compressive strength of
possible to perform a theoretical strength prediction analy-
the skins is in general higher than the tensile strength, un-
sis [1] from the knowledge of the mechanical characteristics
der pure bending the failure of the skin subjected to tension
of the skin and the core. In more detail, considering a gen-
prevents the damaging of the opposite skin subjected to
eric loading condition constituted by a bending moment M
compression.
and a shear load T (see Fig. 1a) the composite’s rupture fol-
From Eqs. (1) and (2) it follows that, in the presence of a
lows the skin failure if the maximum normal stress r0x;max
complex loading (bending + shear), the rupture of the
acting in the external skin face exceeds the skin tensile
sandwich composite is caused by the skin tension failure
strength rs,u. It follows the core rupture if the maximum
if the load ratio r = M/T is higher than rmax = Mmax/Tmax,
shear stress sxz,max acting in the middle plane of the core
i.e. if:
exceeds the core shear strength sc,u. Using the notation
reported in Fig. 1, as well as the widely used Classical Lam- M Gc 2D rs;u
inate Theory and the First Order Shear Deformation The- r¼ > rmax ¼ ð5Þ
T Es Nh sc;u
ory [1] to compute r0x;max and sxz,max, the skin failure and
the core failure occur when M and T exceed the corre- Otherwise, if r is less than rmax, the rupture is caused by the
sponding values Mmax and Tmax given by: core failure. As an example, if the sandwich structure is
2D subjected to a three point bending (TPB) in which r is equal
M max ¼ rs;u ðskin tension failureÞ ð1Þ to half the span length a1 [22], then it fails due to the skin
Es h
failure if a1 satisfies the condition:
N
T max ¼ sc;u ðcore shear failureÞ ð2Þ
Gc a1 > 2rmax ðTPBÞ ð6Þ
where Es is the Young modulus of the skins, Gc is the shear Otherwise the composite’s rupture is caused by the core
modulus of the core, whereas D is the bending stiffness and shear failure. Similarly, for a four point bending (FPB)
with distance between the two loads equal to half the span
length a2 [22], the rupture is caused by the skin failure if the
ts
span length a2 satisfy the condition:
zs
a2 > 4rmax ðFPBÞ ð7Þ
h hc
T T Eqs. (6) and (7) show that for increasing span length the
core failure mechanism moves from the core shear failure to
M M the tensile skin failure. Obviously, if a1 and a2 equal the
z skin corresponding critical values ac1 ¼ 2rmax and ac2 ¼ 4rmax ,
a then the skin failure and the core failure occur at the same
x time.
BACK Unfortunately, because of various parameters affect the
failure properties (the skin–core adhesion, loading interac-
tions, contact stresses, etc.) that can lead also to failures
type (3)–(5), such theoretical predictions of the failure
FRONT
modes (Eqs. (5)–(7)) and corresponding strengths (Eqs.
b (1) and (2)) are in general not accurate. Also, the theoreti-
cal prediction of these failure modes (3)–(5) is a difficult
task and, consequently, a reliable strength evaluation of a
sandwich structure often requires an accurate experimental
analysis and/or numerical simulations [23,24].
In order to improve the structural design of naval com-
ponents, two types of sandwich composites widely
c employed in this field have been studied. Various experi-
mental tests have been carried out on skins and core
Fig. 1. (a) Scheme of a sandwich panel and general notations, (b) pictures (tensile, compressive and shear tests) as well as on the sand-
of the PVC foam core and (c) the polyester mat core. wich structures (tensile, compressive, shear and bending
A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586 577
tests). Successive non-linear FEM simulations have also 4.1. Skins characterization
been carried out to verify whether reliable estimations of
failure modes and mechanical strength can be obtained The experimental characterization of the skins has been
using accurate numerical results. Finally, proper failure cri- performed using tensile, compressive and shear tests. The
teria that can be used in the presence of the delamination or tensile test has been carried out following the ASTM
the core shear failure under complex loading, have been D3039/D3039M-93 standard [25], by using specimens
developed from numerical results. properly instrumented with an HBM DD1 extensometer
to measure the longitudinal strain, and a MM CEA-06-
3. Materials description 125UW-350 strain gauge to measure the transverse strain.
The tests have shown that the skins exhibit a linear elastic
For both examined types of sandwich panels, furnished behaviour up to about 4000 micro strain and an ultimate
by the AiconÒ Italian shipbuilder, the skins are made from strain et,u 13000 micro strain. Moreover, all specimens
3 mm thick fiber-glass laminates manufactured by hand have shown an XGM type failure [25], i.e. a sudden failure
lay-up using an E-glass mat (randomly oriented fibres, with localized at the central zone of the specimen, with the frac-
density of 450 g/m2) and an orthophthalic polyester resin ture surface perpendicular to the specimen axis (see
(CRYSTIC 446 LVPAÒ produced by Scott Bader), with Fig. 2a).
fibre volume fraction of Vf = 0.17. The core of one sand- The compressive test has been carried out following the
wich composite type is constituted by a ‘‘closed cell’’ ASTM D3410-87 standard [26]. Two opposite longitudinal
PVC foam DivinycellÒ H 100 (having thickness MM CEA-06-125 UW-350 strain gauges were mounted on
hc = 10 mm and density of 100 kg m3); it is a GRID- the specimens. Under compressive loading the material
SCORED type core, with longitudinal and transverse cuts exhibits a linear elastic behaviour up to about 4000 micro
on a face and a fibre-glass reinforcing grid placed on the strain, with an ultimate strain ec,u 16000 micro strain.
other face (see Fig. 1b). The core of the other sandwich Unlike tensile tests, the failure is due to the shear rupture
composite type is constituted by a 4 mm thick thermoset of the matrix and debonding phenomena (see Fig. 2b).
polyester mat having a grid of holes (see Fig. 1c), commer- The shear test has been carried out following the ASTM
cially denominated COREMATÒ; it is the so-called ‘‘open D 4255-83 standard [27], by performing the so-called
cells’’ material that absorbs resin during the lamination ‘‘three-rail shear test’’ on specimens with two rectangular
process. Therefore the sandwich panel with the PVC core strain gauge rosettes type MM CEA-06-250UR-350. The
has a total thickness h = 16 mm, whereas the one with experimental shear stress–strain curves have shown that
the polyester core has a total thickness h = 10 mm. under shear loading the skin exhibits a linear elastic behav-
iour up to a shear strain of about 5000–6000 micro strain,
whereas the failure corresponds to an ultimate shear strain
4. Experimental analysis of es,u 25000 micro strain. Moreover, the skin rupture is
mainly due to the shear failure along one or more vertical
Experimental tests have been carried out by using a specimen sections (see Fig. 2c).
mechanical testing machine (HOUNSFIELD H 20 kN) In accordance to the ASTM standards, all the elastic
and a hydraulic testing machine (MTS 350 kN), both inter- constants of the skins have been determined by considering
faced to a data acquisition system type HBM UPM 100. the bilinear stress–strain approximation, i.e. the tensile
Such a device allows the user to acquire the applied loads Young’s modulus (E0s;1 ; E0s;2 ), the compressive Young’s
along with the corresponding strains of the specimens, modulus (E0s;1 ; E0s;2 ), the Poisson ratio vs and the shear mod-
measured by electrical strain gauges (SG) or LVDT ulus Gs. These values are reported in the upper part of
transducers. Table 1, along with the tensile strength rs,u, the compres-
In the following sections detailed descriptions of the sive strength r0s;u , the shear strength ss,u and the corre-
tests carried out on skins, cores and sandwich panels are sponding tensile transition stress rs,t, the compressive
reported. transition stress r0s;t and the shear transition stress ss,t.
Fig. 2. Skins characterisation: pictures of the specimens after the rupture under: (a) tensile, (b) compressive and (c) shear loading.
578 A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586
Table 1
Mechanical properties determined from the experimental tests
Tensile test (MPa) Compressive test (MPa) Shear test (MPa) Flexural test (MPa)
Skins rs,u = 95 r0s;u ¼ 120 ss,u = 49 –
rs,t = 45 r0s;t ¼ 58 ss,t = 24 –
Es,1 = 8580 E0s;1 ¼ 9000 Gs,1 = 2950 –
Es,2 = 5650 E0s;2 ¼ 5700 Gs,2 = 1600 –
ms = 0.38 – –
PVC foam core rc,u = 3.7 r0c;u ¼ 1:45 – –
rc,t = 2.5 E0c;1 ¼ 35 – –
Ec,1 = 120 E0c;2 ¼ 0:4 – –
Ec,2 = 75 – –
Polyester core rc,u = 3.5a r0c;u > 30a – –
rc,t = 2.55a r0c;t ¼ 9:6a – –
Ec,1 = 610a E0c;1 ¼ 390a – –
Ec,2 = 140a E0c;2 ¼ 210a – –
Sandwich structure with the PVC foam core rd = 1.65 – sd ¼ 1:65 sc,u = 2
E = 120 – G = 78.6 Gc = 49.8
– – Gc = 48.5 –
Sandwich structure with the polyester core ru = 3.5 r0u > 30 sc;u ¼ 3:65 sc,u = 4
rt = 2.55 r0t ¼ 9:6 G = 670 Gc = 394
E1 = 610 E 0 = 390 Gc = 390 –
E2 = 140 E = 210 – –
a
Values determined from flat-wise tensile and compressive tests performed on the sandwich structure, as prescribed by the ASTM Standards [27,28].
The elastic constants (the tensile Young’s modulus Ec failure (at ±45°) propagating toward the lower skin–core
and the compressive Young’s modulus E0c ), as well as the interface with a successive delamination (see Fig. 6a and
strengths rc,u (tensile), r0c;u (compressive) and the corre- c). On the contrary, in accordance to the theoretical failure
sponding transition stresses (the tensile transition stress mode prediction, the longer specimens have failed after the
rc,t, the compressive transition stress r0c;t ) are reported in lower skin tensile failure (see Fig. 6b and d).
Table 1. From these values it is seen that the compressive For both types of sandwich structures the mean value of
stiffness and compressive strength of the PVC foam core the actual bending moment at failure M max is reported in
are much lower than those corresponding to tensile Table 2 along with theoretical estimations (see Eq. (1)).
loading.
Following the ASTM standards, mechanical character- 4.3.2. Four points bending test
istics of the polyester core have indirectly been determined The FPB test has been carried out in accordance with
from tests carried out on the corresponding sandwich the ASTM C393-62 standard [22]. In detail, to observe
structures (see the following section). These characteristics, the core failure mechanisms, the following specimens have
in fact, can not be determined directly by using specimens been tested with span length of a2 = 330 mm for the sand-
made by the polyester mat, that is a fibrous material with wich structure with the PVC core and with a2 = 166 mm
no initial bending and transversal stiffness. for the sandwich structure with the polyester core. In fact,
both of these span length values satisfy Eq. (7)
4.3. Sandwich structures characterization (4rmax = 508 mm and 280 mm respectively for the sand-
wich panel with the PVC core and the polyester core, see
The characterization of both types of sandwich panels Table 2) and therefore the failure of these two sandwich
has been carried out by means of three points bending structures should follow the core shear rupture.
(TPB) tests, four points bending (FPB) tests, shear tests, In Fig. 7a it is seen that the sandwich structure exhibits a
flatwise tensile tests and flatwise compressive tests. significant deviation from linearity for midspan deflections
more than about 25–30% of the corresponding failure
4.3.1. Three points bending test
TPB tests have been carried out following the ASTM
C393-62 standard [22]. To observe the possible failure
mechanisms of the skins, specimens with span length of 1600
a1 = 380 and 330 mm for the sandwich panel with the a1=330 mm
PVC core and a1 = 300 and 166 mm for the sandwich panel 1400 C
a1=380 mm
with the polyester core have been tested. In fact, all speci- D
1200 a1=166 mm
mens satisfy the condition expressed by Eq. (6)
(2rmax = 254 mm and 140 mm, respectively for sandwich A
1000
structures with the PVC core and the polyester core, see
Load P1 [N]
Table 2
Theoretical and experimental values of Mmax, Tmax and rmax defined by Eqs. (1), (2) and (5)
Theoretical values Experimental values % Error
Sandwich structure with the PVC foam core Mmax (N) 3491 M max (N) 3576 2.3
Tmax (Nm m1) 33.8 T max (Nm m1) 25.3 33.6
rmax (mm) 127 – – –
Sandwich structure with the polyester core Mmax (N) 1689 M max (N) 1989 15
Tmax (Nm m1) 49 T max (Nm m1) 24.34 101
rmax (mm) 70 – – –
580 A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586
Fig. 6. Pictures of the TPB specimens after the rupture: (a) the sandwich panel with the PVC core having a1 = 330 mm and (b) 380 mm, (c) the sandwich
panel with the polyester core having a1 = 166 mm and (d) 300 mm.
2000 inclined at about 45° (PVC core, Fig. 7b) or placed at the
1800 a B core middle plane (polyester core, Fig. 7c) is observed.
For both types of sandwich structures the mean value of
1600
the actual failure shear load T max is reported in Table 2
1400 along with that estimated theoretically by Eq. (2). Also, fol-
lowing the ASTM standard [22], the core shear strength as
1200
Load P2 [N]
A
well as the core shear modulus Gc have been computed (see
1000 A. Specimen with the polyester core Table 1).
800 B. Specimen with the PVC core
4.3.3. Shear test
600 The shear test has been performed following the ASTM
400 C273-61 standard [31], by using an LVDT transducer type
HBM W 10 TK (see Fig. 8a) to measure the relative dis-
200
placement of the two opposite faces of the sandwich struc-
0 ture. Fig. 8b shows the obtained s–c curves. It can be
0 5 10 15 20
observed that the sandwich panel with the polyester core
Midspan deflection w2 [mm]
exhibits a linearity and a failure shear stress higher than
those of the sandwich panel with the PVC core. Using
b the relationships provided by the ASTM standard [31]
the shear modulus G of the composite has been computed
and the core shear modulus Gc, already estimated by the
FPB test, has been confirmed (see Table 1).
Core shear failure Moreover, the experimental analysis has shown that,
unlike the usual theoretical prediction of the core shear
failure, the actual rupture of the sandwich panel with the
c
PVC core is caused by the delamination of the skin–core
interface where the core reinforcing grid is located (see
Fig. 8c). In accordance with the theoretical predictions,
the failure of the polyester core sandwich panel is caused
Core shear failure by the core rupture (Fig. 8d).
It is important to note that both the delamination of the
Fig. 7. FPB tests: (a) load–deflection curves, pictures of the sandwich PVC core sandwich composite and the core rupture of the
structures with (b) the PVC core and (c) the polyester core after the
rupture.
other sandwich composite type happen at a shear stress
level less than the value previously estimated by the corre-
sponding FPB test (see Table 1). In particular, the delami-
value, like in the previous TPB tests. Also, it is shown in nation corresponds to an apparent ultimate delamination
Fig. 7b (sandwich panel with the PVC core) and Fig. 7c stress sd ¼ 1:65 MPa significantly less than the core shear
(sandwich panel with the polyester core) that in accordance strength sc,u = 2 MPa (see Table 1), whereas the core rup-
to the theoretical prediction, the failure of both sandwich ture corresponds to an ultimate core shear stress
structures follows the core rupture localized in the region sc;u ¼ 3:65 MPa less than the core shear strength value
located between the section where the load is applied and sc,u = 4 MPa (see Table 1). As shown by the successive
the support section. A typical cohesive failure surface numerical simulations reported in the next section, the dif-
A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586 581
LVDT 2.5
[MPa]
2
B
τ
l 1.5
0.5
A. Specimen with the polyester core
B. Specimen with the PVC core
hinge pin 0
0 0.05 0.1 0.15
x
γ [mm]
z
Fig. 8. Shear tests: (a) loading device, (b) stress–strain curves and pictures of the specimens with (c) the PVC core and (d) the polyester core after the
rupture.
by spurious transverse tensile stresses existing in the shear Sandwich panel with the PVC core
test; such spurious stresses are related to the unavoidable 450 2
misalignment between the actual loading direction and 3
P2/2 8
7 a
6
5
4 τ xz
Stress [MPa]
3
2 τxz τxz
1
1200 0 σz
experimental
-1
-2
σz
1000 FEM
-3
-4
0 10 30 50 70 90 110 130
800
x [mm]
Load P2 [N]
1.1
600
1 numerical-experimental data
1
2
b
linear criterion (eq.8)
0.9
400 3
0.8 4
0.7
200
τ xz / τu
0.6
0.5
0
0 1 2 3 4 5 6 0.4
1. FPB test;
Midspan deflection w2 [mm] 0.3
5
2. shear test;
Fig. 11. Numerical simulations: (a) the model used to simulate the FPB 0.2
3. 4. 5. “shear/ tensile”
test of the sandwich structure with the polyester core and (b) the 0.1 tests;
comparison between numerical and experimental load–midspan deflection 6
0
curves. 6. tensile test.
-0.1
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
achieved with deviations less than 4%. Similar FEM mod- σz /σu
els have been used to simulate the remaining experimental Fig. 12. Sandwich structure with the polyester core: (a) stress distributions
tests, and similar deviations have been observed. in the middle plane obtained numerically and (b) resistance domain under
complex shear–tensile loading.
6.1. Numerical results and failure criteria under shear–tensile
loading
imen. Additional numerical simulations of the shear test,
The results of the simulation of the TPB and FBP tests, performed by increasing the Young’s modulus of the two
not reported in this paper for simplicity sake, have corrob- plates constituting the test device, have corroborated that,
orated the experimental results in terms of failure modes in accordance with the theoretical analysis of the adhesive
and corresponding strengths. joints [32], the non-uniformity of such stress distributions
Relevant results are those provided by the simulations decrease if the shear stiffness of the system increases. How-
of the standard shear tests, carried out by using accurate ever, such significant tensile transverse (peel) stresses rz
FEM models that simulate the specimens and the loading next to the two specimen edges are responsible for the pre-
device in the actual experimental conditions. As it is shown mature core failure observed experimentally at an ultimate
in the following paragraphs, such results have permitted to mean shear stress of su = 3.6 MPa, lower than the core
develop, for both types of sandwich structures examined, shear strength of sc,u = 4 MPa, estimated experimentally
accurate failure criteria that allow the user reliable strength with the FPB test. In other words, the interaction between
predictions under complex loading. shear stresses sxz and peel stresses rz leads to core cohesive-
shear failure (mixed mode). To develop a failure criterion
6.1.1. Sandwich structure with the polyester core that can be used to take into account such a stress interac-
Considering the sandwich structure with the polyester tion, the numerical simulation of the experimental ‘‘tensile–
core Fig. 12a shows the distribution of the stresses sxz shear’’ test has also been performed. The relative results, in
and rz at the middle plane of the core, where the failure terms of non-dimensional mean shear stress sxz =sc;u and
occurs (see also Fig. 8d) . From this figure it is possible non-dimensional mean transverse stress r z =rc;u (see
to observe that the shear stress distribution is not uniform Fig. 12a) are reported in Fig. 12b. In this figure the failure
and, most importantly, non-negligible tensile transverse points are defined by the mean stress components provided
stresses rz are observed close to the two edges of the spec- by FEM when the applied load is equal to the failure load
584 A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586
8 8
a b
6 6
4 4
σ~ z
τ xz [MPa]
σz [MPa]
2 2
0
τ~xz
0
-2 -2
a a
-4 -4
-6 -6
0 10 30 50 70 90 110 130 150 170 190 0 10 30 50 70 90 110 130 150 170 190
x [mm] x [mm]
Fig. 13. Numerical results relative to the shear test of the sandwich structure with the PVC foam core: stress distribution of: (a) rz and (b) sxz at the skin–
core interface.
measured experimentally. It is seen that these results are in high stress concentration phenomena) applied by consider-
practice well aligned with the straight line (deviations less ing an integration interval a (see Fig. 13) equal to two times
than 5–6 %) between the two limit points (1, 0) and (0, 1), the core thickness hc, i.e. a = 2hc (a = 20 mm for the exam-
representing respectively the cohesive core failure under ined sandwich structure).
pure flatwise tensile loading and the core shear failure From Fig. 14, that shows the numerical results in terms
under pure shear loading. Consequently, according to data of non-dimensional stresses r ~z =rd and ~sxz =sd (the symbol
presented in Fig. 12b, the failure of the polyester core is indicates the average stress computed in the a integration
accurately predicted by the following simple failure interval), it is shown that the failure points follow the
criterion: Ye’s criterion, i.e. they are well fitted by the well known
jsxz j rz Ye formula:
þ ¼1 ð8Þ 2 2
sc;u rc;u ~sxz r
~z
þ ¼1 ð9Þ
sd rd
6.1.2. Sandwich structure with the PVC core in which rd = 1.65 MPa is the transverse tensile strength of
For the sandwich structure with the PVC core, Fig. 13 the sandwich structure provided by the flatwise tensile test
shows the distributions of rz (Fig. 13a) and sxz (Fig. 13b) (see Table 1) whereas sd = 2.2 MPa is the delamination
at the skin–core interface where, as it is shown by the strength of the sandwich structure under a pure shear load-
experimental shear tests, the delamination failure occurs. ing, provided by the same fitting procedure. It is important
From this figure it is possible to observe that, although
the transverse load is small, the low shear stiffness and
the high elastic constants mismatch of the skin and the core 1.1
(see Table 1) lead to a rz stress distribution that diverges 1 6
next to the free edges; in particular, rz is positive at the left 0.9
edge and negative at the right edge. This particular stress 0.8
distribution justifies the premature delamination phenom- 1. tensile test.
5
3
2
0.7 4
ena experimentally observed at the left edge of the shear 2. shear test;
0.6
test specimen (see Fig. 8c). In fact, the tensile stress rz pro-
τ~xz / τd
0.5 3. 4. 5. “shear/tensile”
duces a peeling force at the skin–core interface that leads to tests;
0.4
the delamination. In order to develop a failure criterion
6. FPB test;
that takes into account the interaction between peel (rz) 0.3
joints) along with the so-called average stress criterion Fig. 14. Delamination resistance domain of the sandwich structure with
[34] (commonly used for composite materials subjected to the PVC foam core under complex loading.
A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586 585
to note that the latter characteristic stress cannot be esti- Cerami for their contribution to the specimens manufac-
mated experimentally with the FPB test, because it is high- turing and the successive instrumentation.
er than the core shear strength of sc,u = 2 MPa and,
consequently, in absence of transverse stresses rz the core
shear failure prevents the skin–core delamination. References
7. Conclusions [1] Vinson JR. The behaviour of sandwich structures of isotropic and
composite materials. Westport: Technomic; 1999.
[2] Mines RAW, Worrall CM, Gibson AG. Low velocity perforation
In the present work the actual mechanical behaviour
behaviour of polymer composite sandwich panels. Int J Impact Engng
and the peculiar failure mechanisms of two types of sand- 1998;21(10):855–79.
wich structures widely used in marine constructions (con- [3] Torre L, Kenny JM. Impact testing and simulation of composite
stituted by fibre-glass laminate skins over a PVC foam or sandwich structures for civil transportation. Compos Struct
a polyester mat core) have been studied by means of exper- 2000;50:257–67.
[4] Kootsookos A, Burchill PJ. The effect of the degree of cure on the
imental investigations and successive numerical analyses
corrosion resistance of vinyl ester/glass fibre composites. Composites
under different loading conditions. The experimental anal- A 2004;35:501–8.
ysis has shown that both types of sandwich structures exhi- [5] Allard JF. Propagation of sound in porous media: modelling sound
bit a significant non-linearity, with different behaviour absorbing materials. Elsevier Applied Science; 1993.
under tensile, compressive and shear loading. Also, in the [6] Sahraoui S, Mariez E, Etchessahar M. Mechanical testing of
polymeric foams at low frequency. Polym Test 2001;20:93–6.
presence of prevalent bending and shear loading the theo-
[7] Kalaprasad G, Pradeep P, George Mathew, Pavithran C, Sabu
retical prediction of the possible failure modes (skin tension Thomas. Thermal conductivity and thermal diffusivity analyses of
failure or core shear failure) are in general not accurate. low-density polyethylene composites reinforced with sisal, glass and
The corresponding theoretical strength predictions are in intimately mixed sisal/glass fibres. Compos Sci Technol
general not accurate as well. In particular, if the failure 2000;60:2967–77.
[8] Anderson, Melvin S. Optimum proportions of truss core and web-
of the sandwich structure is caused by the skin tension rup-
core sandwich plates loaded in compression. NASA TN D-98; 1959.
ture, then the theoretical predictions underestimate the [9] Steeves CA, Fleck NA. Collapse mechanisms of sandwich beams with
actual strength with errors up to 15%; on the contrary, composite faces and a foam core, loaded in three-point bending. Part
if the failure of the sandwich structure is caused by the core I: analytical models and minimum weight design. Mech Sci
shear rupture, then the predictions overestimate the actual 2003;46:561–83.
[10] Steeves CA, Fleck NA. Material selection in sandwich beam
strength with errors up to 100%.
construction. Scripta Mater 2004;50:1335–9.
Under flatwise tensile loading the sandwich panel with [11] Danielsson M, Grenestedt JL. Gradient foam core materials for
the PVC foam exhibits a very low transverse tensile sandwich structures: preparation and characterisation. Composites A
strength associated with a skin–core delamination caused 1998;29:981–8.
by the presence of the reinforcing fibre-glass grid. How- [12] Lin Hong-Ru. The structures and property relationships of commer-
cial foamed plastics. Polym Test 1997;16:429–43.
ever, the transverse stresses significantly influence also the
[13] Vaikhanski L, Nutt SR. Fiber-reinforced composite foam from
core shear failure of the sandwich panel with the polyester expandable PVC microspheres. Composites A 2003;34:1245–53.
core. [14] Kim J, Swanson R. Design of sandwich for concentrated loading.
The non-linear FEM simulations have allowed us to Compos Struct 2001;52:365–73.
study the influence of the transverse stresses on the actual [15] Belingardi G, Cavatorta MP, Duella R. Material characterization of
a composite-foam sandwich for the front structure of a high speed
failure modes of the sandwich composites, as well as to
train. Compos Struct 2003;61:13–25.
develop accurate failure criteria that can be used at the [16] Corigliano AZ, Rizzi E, Papa E. Experimental characterization and
design stage. In practice, for the sandwich structure with numerical simulations of a syntactic-foam/glass-fibre composite
the polyester core, a reliable estimation of the cohesive- sandwich. Compos Sci Technol 2000;60:2169–80.
shear core failure can be obtained by using a simple linear [17] Mouritz AP, Thomson RS. Compression, flexure and shear properties
of a sandwich composite containing defects. Compos Struct
criterion involving the mean stresses, whereas for the sand-
1999;44:263–78.
wich structure with the PVC core, the skin–core delamina- [18] Gdoutos EE, Daniel IM, Wang KA. Compression facing wrinkling of
tion can be accurately predicted by applying the Ye’s composite sandwich structures. Mech Mater 2002;35:511–22.
delamination failure criterion to the average interface stres- [19] Hadi BK, Matthews FL. Development of Benson–Mayers theory on
ses computed from the FEM results by using the average the wrinkling of anisotropic sandwich panels. Compos Struct
2000;49:425–34.
stress method with a characteristic distance a equal to
[20] Sokolinsky VS, Shen H, Vaikhanski L, Nett SR. Experimental and
two times the core thickness hc. analytical study of nonlinear bending response of sandwich beams.
Compos Struct 2003;60:219–29.
Acknowledgements [21] Petras A, Sutcliffe MPF. Indentation failure analysis of sandwich
beams. Compos Struct 2000;50:311–8.
[22] ASTM C393-62, standard test method for flexural properties of flat
The authors thank the Aicon Italian shipbuilder for pro-
sandwich constructions; 1962.
viding the examined materials, as well as Dr. Bartlomiej [23] Mines RAW, Alias A. Numerical simulations of the progressive
Benedikt (LANL) that has read the paper. The authors collapse of polymer composite sandwich beams under static loading.
wish to thank also Giuseppe Giacalone and Giacomo Composites A 2002;33:11–26.
586 A. Russo, B. Zuccarello / Composite Structures 81 (2007) 575–586
[24] Borsellino C, Calabrese L, Valenza A. Experimental and numerical [30] Mines RAW, Jones N. Approximate elastic–plastic analysis of the
evaluation of sandwich composite structures. Compos Sci Technol static and impact behaviour of polymer composite sandwich beams.
2004;64:1709–15. Composites 1995;26:803–14.
[25] ASTM D3039/D3039-93, Standard test method for tensile properties [31] ASTM C273-61, Standard test method for shear properties in flatwise
of polymer matrix composite materials; 1993. plane of flat sandwich constructions or sandwich cores; 1961.
[26] ASTM D3410-87, Standard test method for compressive properties of [32] Shin KC, Lee JJ. Bond parameters to improve tensile load bearing
unidirectional or crossply fiber-resin composites; 1987. capacities of co-cured single and double lap joints with steel and
[27] ASTM D4255-83, Standard guide for testing shear properties of carbon fiber-epoxy composite adherends. J Compos Mater
composites laminates; 1983. 2003;37(5):401–20.
[28] ASTM C297-61, Standard test method for tensile strength of flat [33] Ye L. Role of matrix resin in delamination onset and growth in
sandwich constructions in flatwise plane; 1961. composite laminates. Compos Sci Technol 1988;33:257–77.
[29] ASTM C365-57, Standard test method for flatwise compressive [34] Whitney JM, Nuismer RJ. Stress fracture criteria for laminated
strength of sandwich cores; 1957. composites. J Compos Mater 1974;8:253–65.