Rakenteiden Mekaniikka (Journal of Structural Mechanics)

Vol. 56, No. 1, 2023, pp. 24–40

https://rakenteidenmekaniikka.journal.fi/index
https://doi.org/10.23998/rm.120756
© 2023 The Authors
Open access under the license CC BY 4.0

Durability of sandwich structures with a maximized natural

raw material basis: comparison of expanded polystyrene,
cellulose foam and polylactic acid subjected to UV-rain
aging
P. Hakala1 , O. Orell, E. Sarlin, E. Pääkkönen, L. Jutila and M. Kanerva

Summary In this study, alternative core materials to commercial cork were searched for. Ad-
ditionally, aging of new core materials was studied. Bio-based materials in sandwich structures
can be useful for sports equipment, transportation, and furniture with much less impact on the
environment in comparison with their synthetic counterparts. In this paper, sandwich panels
made of fiber-reinforced polymer (FRP) skins and various sustainable core materials with a core
thickness of 6 mm were studied. The core materials were: cork, expanded polystyrene (EPS),
cellulose foam, and 3D-printed polylactic acid (PLA) honeycomb lattice. FRP composites made
of flax fibre reinforcement and bio-based epoxy resin (30% bio-content) were used to manufac-
ture the skins to compete the glass fibre reinforced rival composite. The experimental analysis
of the panel performance focused on the out-of-plane behavior and aging due to conditioning
in an ultraviolet (UV) irradiation-rain cabinet. The results showed that under mechanical and
environmental loading, the sandwich composite with cellulose foam had comparable or even
better mechanical performance under shearing load, including UV-rain effects, than structures
with EPS cores. Failure localization was studied using digital image correlation (DIC). The
3D-printed PLA honeycomb sandwich structures had a high absolute flatwise tensile strength
and shear strength but also greatest degradation by the UV-rain aging.

Keywords: bio composite, composite sandwich structure, thermoset resin, tensile strength, shear,
failure

Received: 4 August 2022. Accepted: 30 January 2023. Published online: 2 May 2023.

Introduction
The use of sandwich structures with skin sheets made of fibre-reinforced polymer (FRP)
and lightweight core material leads to the most weight-optimized components under bend-
ing loads. The sandwich concepts have been shown to be very effective in increasing overall
strength and bending stiffness of structures [1]. The FRP skins can resist out-of-plane
impacts and in-plane stresses under flexural deformations, while the core is required to re-
sist shear stresses and prevent instability of thin skin sheets. The current partly biobased
FRP sandwich structures in the traditional sandwich structure applications, such as water
1
Corresponding author: [email protected]

24
sport equipment or transportation, are gaining a lot of attention in the recent literature
[1]. Nowadays, engineers aim, not only to improve strength-to-weight ratio in products,
but also various environmental aspects.
To be more truly environmentally conscious, conventional materials and structures
need to be re-evaluated to determine how they can become more sustainable through
proper design procedures and reach smaller environmental impact during the manufac-
turing, installation, service and disposal phase of the product lifetime [2, 3, 4]. FRP
composites made using synthetic fibres, such as glass or carbon fibres, are often applied
to the skins of sandwich panels. The concept of using plant-based natural fibres, such as
flax or hemp fibres, has also been increasingly explored for composites skins [5, 6]. Natural
fibres have a lower strength than their synthetic counterparts but their density-normalized
specific strength values are high. In addition, for some applications, the excellent damping
properties of natural fibre FRPs are an advantage [7].
Sandwich panels might experience a multitude of types of failure modes, e.g., core
shear, bonding failure, skin crushing or rupture and numerous instability modes [8]. Core
materials that are commonly studied include synthetic foams and different honeycombs.
Core also affects the damage tolerance often required for design [9]. Although flax has been
studied previously for the use in sandwich structures, the combination of flax FRP with
the latest bio-based core materials has yet to be studied in detail and analysed in terms
of mechanical performance not to mention durability against environmental stress. In
future, combinations of sandwich panel raw materials that are almost entirely composed
of a natural stock (such as plant fibres or wood) can lead to a lesser impact on the
environment compared to their synthetic rivals.
The aim of the study is to experiment, evaluate and compare the durability of bio-
based and synthetic core materials in sandwich panels with FRP skin sheets. The test
specimens are subjected to ultraviolet (UV) irradiation and moisture by using an UV-
rain cabinet. The aging in the cores is measured in terms of changes in out-of-plane
tension and shear performance. These panels are targeted to represent an alternative
material system to conventional sport equipment and transportation vehicles where glass
fibre and carbon fibre reinforced plastics are typical solutions for sandwich panels with
foam cores based on the petrochemical industry. When compared to metallic cores, the
manufacture and recycling of metals requires great amounts of energy due to the high
enthalpy of fusion. On the other hand, poly(lactic acid) (PLA), cork, and cellulose can
be industrially composted. Expanded polystyrene is well recyclable.

Materials and methods

Core splices
The material systems studied in this work are so-called sandwich panels. Sandwich panels
are layered structures, where monolithic composite panels (i.e., skin sheets) confine a core
splice in the middle so that a three-layer sandwich is formed. In this work, the main
interest is the mechanical performance of material used for the core part.
The reference material system (REF-G), representative of rival application, is a sand-
wich panel with glass reinforced polyester skin sheets and cork core (see Table 1 for cores).
This sandwich is a cost-efficient solution but rather heavy due to the relatively high density
of the cork core splice and glass fibre reinforcement. Alternative core and skin materials
were studied to compete the reference in terms of sustainability and light-weightiness.
Here, sustainability refers to bio-based raw materials as well as the ability to recycle. Of

25
all the core materials (Table 1), cork represents entirely natural material with the least
processing. The cellulose (CS) foam is natural material yet outcome of a laboratory-scale
processing method [10]. EPS is a polystyrene foam but can be almost completely recycled.
PLA represents a type of biopolymer that can be produced of biomass and composted in
specialized processes.
The densities of the selected cores can be varied. Cork, being tree bark, can be
harvested and certain batches can be found with different densities (here, the grade NL20
of 200 kg/m3 was selected). Foam-formed cellulose structure was prepared in a lab scale.
Bleached softwood kraft pulp fibres (Metsä Fibre Oy, mean fibre length 2.2 mm) were
mixed with water, polyvinyl alcohol (PVA Poval 6-88, Kuraray, dosage 0.1 g/g fibre)
and 3 w-% of chitosan. Pulp slurry was mixed 3000 rpm to the 32 % air content. Wet
foam was poured into a 43 × 43 cm sheet mould with a mesh on the bottom. Cellulose
foam was dried over night at 70 ◦ C in an oven. The dry sheet was rewetted up to 50 %
moisture content and pressed to the target thickness of 20 mm and dried between plates
in the oven. The final dry sheets’ properties were measured after conditioning (23◦ C,
50 % RH), and were: grammage 1320 g/m2 , thickness 20 mm and density 66 kg/m3 . The
PLA core was 3D-printed honeycomb core. The PLA core was 3D-printed using PLA
strand (type 1.24 g/cm3 , EasyFil PLA, Formfutura). Honeycomb is certain (common)
three-dimensional geometry typical of high-performance sandwich cores. Honeycomb is
characterized by its cell dimensions (Fig. 1). Different dimensions were studied—finally
capping without an angle was used.

Table 1: The core material candidates used in this study. All the core splices had a splice
thickness of 6 mm.

Core Grade Manufacturer, notes Density

material
Cork Corecork NL20 Amorim Cork Composites 200 kg/m3
EPS Symbra Bewi 30 kg/m3
Cellulose Foamed Laboratory-developed [10] 66 kg/m3
PLA EasyFil PLA 3D-printed honeycomb 547 kg/m3

Composite skin sheets

For the REF-G series, the sandwich panels were manufactured by using filament winding
in order to the mould and directly bond composite skins to the core surface. Commercial
glass fibre yarn (E6CR17-2400-386, Jushi) and an unsaturated polyester (UP) resin system
(M300 TBR, Ashland Composite Polymers) were used.
Representing a more sustainable option compared to REF-G sandwiches, flax rein-
forced composite skins were prepared for the candidate sandwich panels. For the flax
reinforced composite skins, commercial flax reinforcement (2/2 twill, 200 g/m2 , Compos-
ites Evolution (Biotex)) and partly bio-based (30% bio-content) epoxy resin (Super Sap
One, Entropy Resins) with hardener (Super Sap ONF, Entropy Resins) were used with
a mixing ratio of 100:43 by weight. The skin laminates were prepared under vacuum
bag (vacuum pressure -0,6 bar) and post-cured for 24 h at 50 ◦ C. The flax-epoxy skin
laminates were prepared with three layers of flax (lay-up [90/45/90]).
The PLA matrix composite skins were prepared by hot pressing. 4 layers of flax was
used and pre-pressed PLA layers in between. The stack was pressed by using a hot press

26
 Figure 1: Characteristics of the PLA honeycomb core.

(MKH press, Tarkmet), using 2 bar of pressure, and at a temperature of 180 ◦ C. The hold
period was 10 minutes. Reinforcement lay-up was adjusted according to wetting by PLA
and epoxy, and resulting thicknesses of skin laminates.

Core bonding
The core splices were bonded to the skin sheets by using three alternative methods. The
raw materials and the different bonding methods formed the basis of test series. Test
series of this study are shown in Table 2. The first method (Co) was direct co-lamination
of the skin composite on top of core; the bonding process is faster when bonding and cure
of the skin sheet occur at the same time. The rival (REF-G) and EF-EPS-Co series were
co-bonded.
The second method (Ad) was adhesive bonding with the resin used as matrix in the
composite skin sheets. The resin (Super Sap One, Entropy Resins) was applied to surfaces
of the threaded skins and cores. The core and skins were bonded under vacuum (–0.6
bar).
The third method (Pr) was studied to apply to the PLA honeycomb cores and PLA-
matrix based skins. This series (PF-PLA-Pr) was bonded by using a hot press (parame-
ters: 185 ◦ C, 2.0 bar). During hot pressing, only the outer surfaces of PLA-core reached
melting temperature for bonding. It should noted that the PLA core did not heat up to
185 ◦ C (melting) anywhere else than on the bond surfaces of the core splice.

27
 Table 2: Test series in this study.

Series name Matrix type Fibre type Core material Bonding method
(skin) (skin)
REF-G UP Glass Cork Filament-wound on core
(co-bonded)
EF-EPS-Co Epoxy Flax EPS Co-bonded within
skin lamination
EF-EPS-Ad Epoxy Flax EPS Cured skin sheets
adhesively bonded
EF-CS-Ad Epoxy Flax Cellulose foam Cured skin sheets
adhesively bonded
EF-PLA-Ad Epoxy Flax PLA (3D-printed) Cured skin sheets
adhesively bonded
PF-PLA-Pr PLA Flax PLA (3D-printed) Hot press bonded after
hot pressing of skins

UV-rain cabinet used for an aged condition of specimens

Environmental conditioning following the standard ASTM G155-13 was run except that
the water spraying step was applied without radiation (ultraviolet (UV) radiation was
applied in a separate step). The total test time was 500 hours meaning 250 two-hour
cycles (a two hour cycle in Table 3). Specimen rotation in the conditioning chamber was
performed to evenly apply conditioning to age the specimens from all sides. The cabinet
was of model Q-sun Xe-3 (Q-lab). The specimens were aged without free-edge sealing to
gain the most severe effect on the specimens used for testing (i.e., the aging effect here is
test specimen geometry-dependent).

Table 3: UV radiation and rain spray used in this study for aging the specimens.

Step Mode Radiation Black panel Air RH Time

intensity temperature temperature (%) (min)
(W/m2 ) (◦ C) (◦ C)
1 Radiation 0.35 63 48 30 102
2 Water spraying 0 38 21 90 18

Flatwise tensile testing

The Flatwise Tensile Strength (FTS) specimens were fabricated by following the standard
ASTM C 297. From the sandwich panels, pieces were water jet-cut to a 20 mm × 20 mm
size. Aluminium load blocks were glued (DP460, 3M) to the pieces for test jig connection.
Specimens schematic is presented in Fig. 2 The bond surfaces were sanded and cleaned
for the gluing. The glue was post-cured in the oven (50 ◦ C, 24 hours). FTS specimens
were tested by using a universal testing machine (5967, Instron) with computer control
and data acquisition and with a 30 kN load cell. FTS testing was performed at a constant
cross-head displacement rate of 1.0 mm/min. Displacement, load and time data were
recorded at 25 Hz. Five specimens of every series reference (no aging) and UV-rain aged

28
Figure 2: The schematic of the specimens for flat-wise tensile test resembling ASTM
C297. All dimensions are in millimeters.

series per material system were tested. The (averaged) failure stress was calculated using
force and cross-sectional area measured for each specimen. Failure strain was estimated
based on the displacement data and was defined as the point of maximum stress (i.e.,
failure stress). Toughness was integrated as the area under the stress-strain curve; the
change was calculated by comparing non-aged and aged specimens’ results.

Notched single-lap shear testing

The Notched Single-lap Shear (NSS) specimens were manufactured by following the stan-
dard ASTM D 5868. Pieces of a size of 20 × 160 mm were water jet cut from the bonded
panels. Crack slots for both sides (skins) were cut using a circular bench saw, adjusted to
the thickness of the skin (Fig. 3). All the samples were stabilized in 50 ◦ C, 24 hours. The
testing was performed using a test machine (5967, Instron), at a constant cross-head speed
of 1 mm/s. From sandwich panels, pieces were water jet cut to a 20 mm × 160 mm size.
Five specimens of every series reference (no aging) and UV-rain aged series per material
system were prepared for testing. Ultimate (average) shear stress was calculated using
force and cross-sectional area (between slots, nominal slot distance 25 mm) measured for
each specimen. Shear strain (γ) was calculated based on:

γ = arctan (∆d/t) , (1)

where t is the core thickness and ∆d is the axial displacement subjected to the test spec-
imen by the test device. Here, it is presumed that the deformation in the stiff composite
skin sheets is insignificant compared to core’s deformation.

DIC system application for strain field imaging

2D-Digital Image Correlation (DIC) technique [13] was used to study the deformation
of the core in flatwise tensile tests. The equipment consisted of a 5 Mpix CCD camera
(Basler piA2400-17gm) and a bilateral telecentric lens. Before testing, a black-and-white
random speckle pattern was applied on the specimens by spraying.

29
Figure 3: The schematic of the specimens for the NSS test resembling ASTM D 5868. All
dimensions are in millimeters.

Recording rate of 5 frames per second was used. The captured images were processed
with Davis 10.2.0 software (LaVision, Germany) into full-field strain distribution maps in
the loading direction (’axial’). The strain fields at the onset of the specimen failure were
used to investigate the localization of the damage in the specimen.

SEM imaging
Scanning electron microscopy (SEM) images were taken with JEOL (Variable pressure
tungsten-filament SEM, model IT500). The samples were gold-sputtered except for EF-
CS-Ad that was studied with the low vacuum operation mode without coating.

Results and Analysis

Out-of-plane tensile behavior
The cork-core of the reference series (REF-G) did not suffer significant aging in terms
of FTS but mainly in terms of failure strain, i.e., cork was comparatively durable core
material in the UV-rain environment. The series with EPS and adhesive bonding (EF-
EPS-Ad) had similar behavior compared to the REF-G series but FTS was lower. The
aging resulted in a similar decrease of FTS (–10 %) but greatly increased failure strain
(+85 %). This is a good result for the EPS core, when considering the open edges of
the specimens in this study. When the bonding technique was different (co-bonded), the
results were very similar (decrease of FTS –7 % and increased failure strain +56 %). This
suggests similar failure mode (cohesive failure) for both bonding techniques in these EPS
series. One reference specimen of the series EF-EPS-Ad indicated brittle behavior due
to core-skin sheet debonding indicated brittle behavior due to core-skin sheet debonding
that is presumably flaw during manufacture of the specimens (Fig. 4 c). The increase in

30
the failure strain indicates increased plasticization of core material, in the case of cohesive
failure, related to the aging by UV and rain.
When the core splice was changed into the cellulose foam, an interesting effect was
observed. The naturally moisture-sensitive cellulosic core material restored the initial
stiffness very well as indicated by the (decreased) series-average failure strain and a minor
change in FTS (–4.4 %). This is partly due the fact that the foam is very compliant and
the stress-strain curves tend to have plateau or repeated stress peaks. By looking at the
curves overall, the non-aged and aged specimens have essentially similar behavior.

Table 4: Tensile out-of-plane test (FTS) results for sandwich specimens.

Series Tensile failure Tensile strain at Toughness change

stress failure stress due to aging
(MPa) (m/m) (%)
REF-G 1.55 ± 0.25 0.07 ± 0.01
REF-G (aged) 1.64 ± 0.31 0.08 ± 0.01 -11.5
EF-EPS-Ad 0.41 ± 0.04 0.05 ± 0.01
EF-EPS-Ad (aged) 0.37 ± 0.12 0.09 ± 0.02 -9.5
EF-EPS-Co 0.48 ± 0.27 0.05 ± 0.02
EF-EPS-Co (aged) 0.45 ± 0.11 0.08 ± 0.02 -2.2
EF-CS-Ad 0.036 ± 0.02 0.40 ± 0.19
EF-CS-Ad (aged) 0.034 ± 0.02 0.20 ± 0.08 -48.2
EF-PLA-Ad 0.91 ± 0.53 0.03 ± 0.03
EF-PLA-Ad (aged) 0.56 ± 0.23 0.08 ± 0.01 -5.9
PF-PLA-Pr 0.75 ± 0.34 0.05 ± 0.02
PF-PLA-Pr (aged) 0.12 ± 0.08 0.04 ± 0.02 -85.3

The PLA honeycomb cores turned out to be very stiff and the failure mode of the
sandwich specimens was mainly debonding at the interface of the skin sheet and core
splice. The UV-rain aging resulted in a multiple debonding processes that included several
onset points over a rather low stress level—FTS was in average 39 % lower due to aging.
The change of the epoxy-based composite skins into PLA-matrix skins (and core bonding
by hot press method) did not affect the non-aged specimens’ behavior greatly (–17 % lower
FTS). However, the interface between the PLA-flax reinforced composite skins and the
PLA honeycomb splice, bonded by hot pressing, suffered significantly due to the UV-rain
(decrease of FTS –84 %).
During the UV-rain aging, humidity significantly degraded the hot-pressed joint (Fig. 4
f) and caused failure at much lower forces. The failure surface is essentially the same,
the polymer has simply weakened and got more brittle [17]. For thermosets as well as
thermoplastics (e.g., PLA) hydrolysis is the expected main type of degradation. It is also
presumed that the hot press method may have caused thermal residual stresses at the
bond interface and the moisture during conditioning resulted in crack propagation due to
residual stresses within the weakened polymer.
The toughness values indicated degradation for all the aged specimen series (–2 to
–85 %). The most significant decrease in toughness was suffered by the EF-CS-Ad and
PF-PLA-Pr series. Aged EF-EPS-Co, EF-EPS-Ad and EF-PLA-Ad series had less than
10 % effect on toughness due to degradation. These results essentially agree with the
degradation of bio-materials, such as those based on cork and PLA [16] [17].

31
 a) b)

c) d)

e) f)
Figure 4: Tensile out-of-plane stress-strain behavior for different sandwich specimens: a)
REF-G; b) EF-EPS-Ad; c) EF-EPS-Co; d) EF-CS-Ad; e) EF-PLA-Ad; f) PF-PLA-Pr.
The code ’Ref-series’ in the graphs refers to the non-aged specimens.

32
 It should be noted that, for many of the FTS specimens, the core splice failed so
that tiny strings or shreds were left unbroken and they could carry slight load after the
specimen was practically failed. This resulted in essentially horizontal curves after failure
in the data.

Shear behavior
The out-of-plane shear performance in general indicated similar tendency as the out-of-
plane tensile behavior (in terms of FTS) for the sandwich structures. For the reference
(REF-G) series, the shear behavior in terms of the shear failure stress degraded –15 % due
to aging. For the specimens with EPS cores, the shear failure stress level was in general
very low and the degradation due to UV-rain was ≈2 %. The bonding technique for the
specimens with EPS cores had basically no effect when comparing intact and UV-rain
conditioned specimens (change in shear failure stress –11 % and –23 %, for adhesive-
bonded and co-bonded series, respectively).
In contrast, the series with the cellulosic foam splice suffered greatly due to the UV-rain
conditioning (the decrease in failure shear stress –53 %). However, it is very interesting
to note that the shear performance of CS core was better compared to the absolue values
for the sandwich with EPS (both bonding techniques included). The shear failure stress
values are higher for non-aged and aged CS test series than those of EPS series (100 %
higher and 15 % higher for the non-aged and aged series, respectively). Behavior in
cellulose foam is not isotropic because it has layered structure and chitosan addition
might have increased durability against rain-UV. [14] [15]
For the PLA honeycomb cores, the shear behavior of the sandwich specimens had a
similar tendency as it was for the FTS tests. The shear failure stress decreased –39 %
and –62 %, for the epoxy-resin co-bonded and hot press-bonded specimens, respectively.
Interestingly, the hot press-bonded sandwich specimens have relatively low experimental
scatter (coefficient of variation 8 % and 6 %, for non-aged and aged series, respectively)
in the failure stress values.

Table 5: Shear out-of-plane test (NSS) results for sandwich specimens.

Series Shear failure Shear strain at

stress failure stress
(MPa) ( ◦)
REF-G 1.50 ± 0.05 13.7 ± 1.20
REF-G (aged) 1.28 ± 0.04 11.6 ± 0.96
EF-EPS-Ad 0.28 ± 0.01 11.6 ± 0.52
EF-EPS-Ad (aged) 0.25 ± 0.06 11.3 ± 4.25
EF-EPS-Co 0.29 ± 0.01 15.7 ± 1.80
EF-EPS-Co (aged) 0.22 ± 0.01 15.6 ± 4.48
EF-CS-Ad 0.57 ± 0.11 8.37 ± 1.97
EF-CS-Ad (aged) 0.27 ± 0.13 6.73 ± 0.71
EF-PLA-Ad 2.26 ± 0.47 14.6 ± 0.83
EF-PLA-Ad (aged) 1.37 ± 1.10 6.73 ± 2.99
PF-PLA-Pr 6.8 ± 0.51 9.3 ± 1.55
PF-PLA-Pr (aged) 2.3 ± 0.15 8.1 ± 1.30

33
 a) b)

c) d)

e) f)
Figure 5: Shear out-of-plane test results for different sandwich specimens: a) REF-G;
b) EF-EPS-Ad; c) EF-EPS-Co; d) EF-CS-Ad; e) EF-PLA-Ad; f) PF-PLA-Pr. The code
’Ref-series’ in the graphs refers to the non-aged specimens.

34
Localization of damage in flatwise tensile tests
The qualitative DIC analysis for the axial strain fields at the onset of the failure in flatwise
tensile test is presented in Figure 3. The figure depicts characteristic differences in the
strain localization between the different core materials. Of the tested materials, the cork
shows the most even strain distribution. The material deforms relatively uniformly and
no single location can be named for the failure initiation spot.

Hi
gha
xia
lst
rai
n

Lowa
xia
lst
rai
n

a)

b) c)

d) e)
Figure 6: Qualitative axial strain (out-of-plane) distributions based on DIC imaging at
the point of visual onset of damage: a) REF-G; b) EF-EPS-Ad; c) EF-EPS-Co; d) EF-
CS-Ad; e) EF-PLA-Ad. The ultimate (high, low) strain levels as color codes are given
above.

The specimens with EPS core show evident locations inside the core where the crack(s)
begin to initiate. In addition to the high strain concentration spots in the core itself, it is
also seen that the deformation localizes also at the skin-core interface. The deformation
in the CS specimen localizes in a horizontal layer of the core, indicating heterogeneous
mechanical properties of the material in the out-of-plane direction.

35
 For the 3D-printed PLA honeycomb, the deformation is localized largely at the junc-
tion of the cell walls and the cell cap. The location of strain localization [11] can indicate
the mode of failure in composite materials. The exact failure modes are discussed and
analysed in the next section.

Micro-structure of core splices after testing

In Fig. 7 a, there is the cork granules’ failed surface, which is full of air-filled voids between
and inside the granules. Non-harmed granule surface is shown on the background. The
image shows the morphology of natural cork’s cellular structure and ruptured granule
surfaces. Cork represents a cellular structure formed by thin-walled prismatic cells. A
detailed analysis on the cellular structure reveals that the cellular voids represent most
of the cork volume, which justifies its low density and most of its unique properties [12].
This structure is anticipated to bring in the even strain distribution observed in Fig. 6 for
series REF-G. The cohesive rupture has been shown to be typical for cork-core sandwiches
[17].
The EPS foam’s fractured surface is seen in Fig. 7 b. The image is a holistic view of
the fractured EPS structure in a sandwich system. In the middle of the image, epoxy
resin (droplet) that has impregnated inside the foam structure can be seen. In addition to
infused epoxy, there are air-filled voids of the ruptured EPS beads. This type of fracture
surface represent the vicinity of the bond line, naturally.
The macro-structure of the failed CS surface can be seen in the SEM micrograph in
Fig. 7 c. The cellulose fibres are irregular and the average distance and number of fibre-
crossings (presumed contact points) between the fibres vary throughout the foam micro-
structure. Overall, the cellulosic fibres are long—in the order of hundreds of microns [10].
Fibre sizes (diameter) are in the order of 20–40 µm. It should be noted that the integrity
of foam-formed structure is largely based on inter-fibre bonds and the effect of additives.
In Fig. 7 d, a fractured PLA-honeycomb wall structure is seen for a specimen with
failure in the core (cohesive). Failed strings of PLA—originating from the 3D-printing
process—is seen on fracture surface (Fig. 7 d, 8). Mostly the surface is suggested to
represent the interface between subsequent 3D-printing layers.

36
 a) b)

c) d)
Figure 7: SEM analysis of the fracture surfaces after out-of-plane tensile testing of sand-
wich specimens: a) REF-G (scale bar 500 µm); b) EF-EPS-Ad (scale bar 500 µm); c)
EF-CS-Ad (scale bar 500 µm); d) EF-PLA-Ad (scale bar 100 µm).

Figure 8: EF-PLA-Ad test series’ typical core failure mode. The honeycomb cell structure
failed at its root so that the 3D-printed flat base of the core splice is left attached to the
metallic loading block. Dimensions in centimeters.

37
Conclusions
This study investigated the aging performance of sandwich structures with bio-based
raw materials and ’green’ design concepts. The reference structure was a glass reinforced
polyester skin laminates and cork core. The candidate sandwich series were manufactured
using partly bio-based epoxy and flax reinforcement for the skins as well as polylactic
acid with flax reinforcement. The candidate core materials were expanded polystyrene,
cellulose foam and 3D-printed honeycomb. The main goal was to study the effects of
aging due to UV and rain conditioning.
The aging effect in the reference structures with cork core splices and glass fibre re-
inforced composite skins indicated a moderate aging effect overall (<16 %). The results
for cellulose foam-integrated sandwich showed that these structures had comparable or
even better mechanical performance under shearing load, including the degradation by
UV-rain, than the structures with the EPS core. This finding was valid when the EPS
core splice had approximately half of the density of cellulose foam. The 3D-printed PLA
honeycomb sandwich structures had a high flatwise tensile strength (0.75–0.91 MPa) and
shear failure stress (2.3–6.8 MPa) but also a big effect by the aging (even 84 % of de-
crease). The significant aging effect mainly affected the bond of the honeycomb PLA to
the flax fibre-reinforced composite skins.
Further testing of bending and out-of-plane impact performance, as well as numerical
simulation works, are necessary to continue the study of the promising bio-based core
materials for composite sandwich structures.

Acknowledgments
This study was funded by the KAUTE foundation, and the ecosystem project SmartRail
(Business Finland). This work made use of Tampere Microscopy Center facilities at
Tampere University. The support from Mari Honkanen is also acknowledged.

References
[1] P.R. Oliveira, M. May, T.H. Panzera, S. Hiermaier. Bio-based/green sand-
wich structures: A review. Thin-Walled Structures, 177:109426, 2022.
https://doi.org/10.1016/j.tws.2022.109426

[2] R.J. Tapper, M.L. Longana, A. Norton, K.D. Potter, I. Hamerton. An evalua-
tion of life cycle assessment and its application to the closed-loop recycling of car-
bon fibre reinforced polymers. Composites Part B: Engineering, 184:107665, 2020.
https://doi.org/10.1016/j.compositesb.2019.107665

[3] P. Glavic, R. Lukman. Review of sustainability terms and their

definitions. Journal of Cleaner Production, 15(18):1875-1885, 2007.
https://doi.org/10.1016/j.jclepro.2006.12.006

[4] T. Bayerl, M. Geith, A.A. Somashekar, D. Bhattacharyya. Influence of fibre architec-

ture on the biodegradability of FLAX/PLA composites. International Biodeteriora-
tion & Biodegradation, 96:18-25, 2014. https://doi.org/10.1016/j.ibiod.2014.08.005

[5] S. Alsubari, M.Y.M. Zuhri, S.M. Sapuan, M.R. Ishak, R.A. Ilyas, M.R.M.
Asyraf. Potential of natural fiber reinforced polymer composites in sandwich

38
 structures: A review on its mechanical properties. Polymers, 13(3):423, 2021.
https://doi.org/10.3390/polym13030423

[6] F. Bensadoun, I. Verpoest, J. Baets, J. Müssig, N. Graupner, P. Davies, M. Gomina,

A. Kervoelen, C. Baley. Impregnated fibre bundle test for natural fibres used in
composites. Journal of Reinforced Plastics and Composites, 36(13):942-957, 2017.
https://doi.org/10.1177/0731684417695461

[7] Xiaoning Tang, Xiong Yan. A review on the damping properties of fiber rein-
forced polymer composites. Journal of Industrial Textiles, 49(6):693-721, 2018.
https://doi.org/10.1177/1528083718795914

[8] I.M. Daniel, E.E. Gdoutos, K.A. Wang, J.L. Abot. Failure modes of com-
posite sandwich beams. Journal of Damage Mechanics, 11(4):309-334, 2002.
https://doi.org/10.1106/105678902027247

[9] M. Niu. Composite Airframe Structures: practical design information and data. 3rd
Edition. Hong Kong Conmilit Press Ltd, 2010. ISBN-13: 978-9627128069.

[10] T. Hjelt, J.A. Ketoja and H. Kiiskinen, A.I. Koponen, E. Pääkkönen. Foam forming
of fiber products: a review. Journal of Dispersion Science and Technology, 3(5):12-15,
2021. https://doi.org/10.1080/01932691.2020.1869035

[11] N. Pournoori, G. Corrêa Soares, O. Orell, S. Palola, M. Hokka, M. Kanerva. Adi-

abatic heating and damage onset in a pultruded glass fiber reinforced composite
under compressive loading at different strain rate. International Journal of Impact
Engineering, 147:103728, 2021. https://doi.org/10.1016/j.ijimpeng.2020.103728

[12] H. Pereira. The Rationale behind Cork Properties: A Review of Structure and Chem-
istry. BioResources, 10(7):2-5, 2015. https://doi.org/10.1177/1528083718795914

[13] Jones, E.M.C, Iadicola, M. A. A Good Practices Guide for Digital Image Correlation.
International Digital Image Correlation Society, 94, 2018.

[14] D. Alonso, M. Gimeno, R. Olayo, H. Vâzquez-Torres, J. D. Sepûlveda-Sânchez, K.

Shirai. Cross-linking chitosan into UV-irradiated cellulose fibers for the prepara-
tion of antimicrobial-finished textiles. Carbohydrate Polymers, 77(3):536-543, 2009.
https://doi.org/10.1016/j.carbpol.2009.01.027

[15] A. Ottenhall, T. Seppänen, M. Ek. Water-stable cellulose fiber foam with antimicro-
bial properties for bio based low-density materials. Cellulose, 25, 2599-2613, 2018.
https://doi.org/10.1007/s10570-018-1738-y

[16] S. Prabhakaran, V. Krishnaraj, H. Golla, M. Senthilkumar. Biodegradation be-

haviour of green composite sandwich made of flax and agglomerated cork. Polymers
and Polymer Composites, 30, 2022. https://doi.org/10.1177/09673911221103602

[17] MC. Mistretta ,FP. La Mantia,V. Titone, L. Botta, M. Pedeferri, M. Mor-

reale. Effect of ultraviolet and moisture action on biodegradable polymers and
their blend. Journal of Applied Biomaterials and Functional Materials, 18, 2020.
https://doi.org/10.1177/2280800020926653

39
Pauli Hakala, Olli Orell, Essi Sarlin, Mikko Kanerva
Tampere University
Materials Science and Environmental Engineering unit
Korkeakoulunkatu 10, Tampere, Finland
[email protected], [email protected], [email protected], [email protected],
[email protected]
Elina Pääkkönen
VTT Technical Research Centre of Finland Ltd
Solutions for Natural Resources and Environment
Koivurannantie 1, FI-40400 Jyväskylä, Finland
[email protected]
Lauri Jutila
CSI Composites Oy
Virtaintie 15, Mänttä-Vippula, Finland
[email protected]

40