Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Construction and Building Materials 275 (2021) 122062

Contents lists available at ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Mechanical properties and reinforcement mechanisms evaluation


of closed-cell polymer foams reinforced by recycled glass beads
Shunze Cao a, Yuwu Zhang b,⇑, Yang Lu c
a
Department of Engineering Mechanics, Tsinghua University, Beijing 100084, People’s Republic of China
b
College of Liberal Arts and Science, National University of Defense Technology, Changsha, Hunan 410073, People’s Republic of China
c
College of Architecture and Environment, Sichuan University, Chengdu 610065, People’s Republic of China

h i g h l i g h t s

 The GBs can increase the mechanical properties and energy-absorption capacity of polymer foams.
 The novel simulation strategy enables to model both cells and GBs explicitly.
 The numerical predictions show good agreements with experimental and analytical results.
 The numerical results indicated the hardening stress increased with the decrease of GBs size.

a r t i c l e i n f o a b s t r a c t

Article history: Particle reinforced (PR) polymer foams are promising materials for lightweight structures. In this
Received 14 July 2020 research, compression tests were conducted on glass beads (GBs) reinforced linear low-density polyethy-
Received in revised form 12 December 2020 lene (LLDPE) foams. The experimental results indicated that embedding GBs could increase the yield
Accepted 15 December 2020
strength, elastic modulus and energy absorption capacity of bulk material. Analytical predictions of
Available online 7 January 2021
GBs reinforced foams were also used to compare with experimental results. To investigate the microme-
chanics of neat foams and glass PR foams under compression, the finite element (FE) simulations were
Keywords:
conducted. The analysis of neat foams indicated that the strain localization occurred at right and left sides
GBs reinforced polymer foams
Three-dimensional finite element analysis
of hollow cells during the plateau stage. For GBs reinforced foams, the foam matrix around the particle
Analytical predictions reinforcements was nearly densified while the foam matrix without GBs remained undensified due to
Parametric study the stabilisation of glass particles. The parametric studies reasonably explained why hardening stress
of foam increases with the decrease of particle size.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction foams under deformation are also different. During compression,


the deformation of foams in the elastic region is controlled by
At present, various polymer foams have been widely used as the either cell wall bending for open cell foams or cell face stretching
core materials of sandwich panels such as slabs and walls in con- for closed cell foams. Based on different mechanisms of yielding,
struction of buildings and infrastructures. In addition to the low- the foams can be categorized in three groups: elastomeric foams
cost and lightweight properties, the functions such as thermal [5], elastic–plastic foams [6], and brittle foams [7].
and acoustic insulation are also superior characteristics [1,2], In the construction field, the polymer foams may not be strong
despite the high flammability that could limit the behavior of poly- enough to sustain strong wind loads or intense seismic loading. To
mer foams used in civil engineering [3]. Based on different charac- strengthen the neat foams, reinforcements such as particles have
teristics of cells, the foams can be classified into open-cell foams been used [8–10]. Guden et al. [11] and Elbir et al. [12] embedded
and closed-cell foams [4]. For the closed-cell foam, the faces in cells SiC particles (6% 20% volume fraction (VF)) into foam matrix with
are solid, so that each of the cells is sealed off from the neighbours. a variety of volume fractions. In comparison to the non-reinforced
However, for the open-cell foam, each cell is interconnected among foams, embedding SiC reinforcements increased the compressive
its neighbours. The micro-responses of closed-cell and open-cell strength. For the PR aluminum (Al) foams, Al2 O3 [13,14]{\text
{TiB}}_2TiB2 , and Y2 O3 [15] were selected as the reinforcements.
⇑ Corresponding author. With the rise of volume fraction (VF) of Al2 O3 particles, the com-
E-mail address: zhangyuwu12@nudt.edu.cn (Y. Zhang). pressive strength and elastic modulus of reinforced foams were

https://doi.org/10.1016/j.conbuildmat.2020.122062
0950-0618/Ó 2020 Elsevier Ltd. All rights reserved.
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

increased. Additionally, the efficiency of energy absorption and 2. Materials and manufacturing
plateau stress also increased as the VF of particles increased to
2%, but this trend was reversed when the VF of reinforcements The currently studied PR polymer foams are composed of poly-
was over 20%. The similar trends were also found when embedding mer foam matrix and porous glass beads (i.e. particle reinforce-
the TiB2 and Y2 O3 particles [14,15]. For the Al alloy foams, which ments). The foam matrix, which is made from LLDPE powder
have shown better mechanical and thermal properties than the supplied by Dalian Bona Biological Technology Co., Ltd., contains
neat Al foams [16], the ceramic particles were also embedded into two parts, i.e. LLDPE polymer (solid region) and spherical hollow
the foam matrix for reinforcing purposes. However, the reinforcing cells (voids within foam). The LLDPE powder is defined by a density
effect was not ideal. The foams became more brittle, and stress of qs ¼ 0:935 g=cm3 , an elastic modulus of Es ¼ 450 MPa, and a
fluctuations in the plateau region was more obvious compared to yield strength of rs ¼ 27:5 MPa. The recycled GBs with parent
the neat Al alloy foams [17–19]. The phenomena could be caused material density of qp ¼ 0:35g=cm3 are supplied by the Quietstone
by the inhomogeneous microstructures after embedding particles.
UK Ltd. Based on the random sampling, the size of GBs have the
Unlike the unstable mechanical properties of particles rein-
size dp ranging from 1.41 mm to 2.25 mm after sieving by a mesh
forced metal foams, polymer foams reinforced by nano-particles
with 2.5 mm circular pores. The particle size follows a normal dis-
or micro-particles showed apparent reinforcing effects. Jayavard-
tribution dp ~ N (dpm =1.75 mm, l2pm =0.0408 mm2) with dpm and l2pm
han and Doddamani [20] developed the glass micro-PR high den-
sity polyethylene foams. The enhancements of yield strength, and representing the mean size and standard deviation, respectively, as
elastic modulus were obvious compared to the non-reinforced shown in Fig. 1 (a). Fig. 1 (b) plots the compressive force versus dis-
HDPE foams. For the nanoscale particles, the reinforcements placement relationship of a single glass particle with the size of
played an important role in the modification of the foam dp = 1.75 mm under uniaxial quasi-static compression. The force
microstructures. Saha et al. [21] embedded TiO2 nanoparticles into fluctuations of GBs under compression could be caused by the
polyurethane foams. The compressive strength and elastic modu- nonuniform local and brittle damage within the internal porous
lus increased due to the stabilisation of cells wall and the modifi- structure before the catastrophic collapse. According to the curve
cation of cells size. fitting on the compressive response of the single particle via
The glass beads used in this research were made from recycled numerical calculations (see the Section 4.2 for FE analyses), the
glass. The beads were sintered and expanded at a temperature of compressive strength of the parent material of glass particle is cal-
750 °C with the benefits of lightweight, sound absorption, fire ibrated to be rp 9.87 MPa. Based on the assumption of elastic
proofing, and thermal insulation [22]. Aygün and McCann [23] Herzian contact between the glass particle and the loading
investigated the structural performances of recycled glass bead machine [32,33], the elastic modulus of this particle can be esti-
panels. The results indicated that the recycled glass-based struc- mated as Ep = 387.37 MPa using the following equation
tures showed the potential applications in both building structures 0:5
4dp s 1:5
and noise barriers in highway due to the high tension, bending Ep ¼ ð Þ ð1Þ
resistances, and the excellent sound absorption capabilities. 3PE ð1  mp Þ 2
To understand the micromechanics of foams, several numerical where P E and s denote the compressive force and displacement of
studies have employed the Laguerre and Voronoi tessellation tech- glass particles, respectively, as shown in Fig. 1 (b). mp is the Poisson’s
niques to build the 3D FE models of foam materials, which repli-
ratio of glass particles, which is calibrated to obtain mp = 0.
cated the internal microstructures of foams well [24–27]. Ghazi
The manufacturing methodology of the GBs reinforced polymer
and his co-authors established 3D FE models of foams based on
foams is described as follow. The polymer powder, glass particles
the X-ray CT technics, and the integrated Representative Volume
as well as the azodicarbonamide blow agent (supplied by the
Elements (RVEs) computation strategies, which showed accurate
Shuntai Technology CO. Ltd.) were mixed with the required pro-
predictions of mechanical responses and deformation mechanisms
portions and stirred in a blender for 5 min. The volume fractions
of closed-cell foams [28–30]. Cho et al. [31] created a 3D finite ele-
of polymer polymer (F) and glass particle reinforcements (f) were
ment model to understand the tensile responses of GBs reinforced
defined to quantify the relative amount of their constituents to
foams. The glass particles were inserted into the homogenous
the bulk material. Hence, F ¼ V p =V T and f ¼ V f =V T , where V p , V f
matrix randomly. The FE analysis suggested that the yielding of
and V T represent the volume of polymer polymer, the volume of
foam matrix occurred at the beginning of the particle breakage.
glass particle reinforcements and the volume of bulk specimen,
However, the previous investigations about the interactions
respectively. In order to demould, the release agent was spread
between hollow cells and particle reinforcements during deforma-
on surfaces of a mild steel mould where the mixture was then
tion were limited. Owing to the stochastic nature of the internal
evenly loaded from the blender with the speed of 500r/min in
structure of foam matrix, it might be over idealized to regard the
the room temperature. Then, the steel mould with the mixture
foams as the homogeneous matrix. In this study, the 3D FE models
was heated in the sealed mould at 210 °C for an hour, followed
of GBs reinforced foams were built to predict the compressive
by a cooling process at the room temperature for 24 h.
responses under uniaxial quasi-static compressive tests. The hol-
Due to the fact that there is insufficient polymer powder to pro-
low cells and glass particles were created explicitly in FE models
duce a foam specimen when F<0.214 and the density is too high
to understand the complex interactions during compression.
for specific applications when F>0.428, four types of neat foam
The outline of this paper is described as follow. The characteri-
specimens with F = 0.214, 0.267, 0.321 and 0.428 were manufac-
sations of foam and glass particles and the manufacturing method
tured for experimental measurements. In addition, six types of
are introduced in the Section 2; the experimental methodology is
GBs reinforced foams with the combinations of F = 0.214, 0.321
reported in the Section 3; the FE models of GBs reinforced foams
and f = 0.1, 0.2, 0.3 were also manufactured. It should be noted that
that are built based on the modelling strategy is described in the
a homogenous porous foam was unable to be achieved for f>0.3.
Section 4; the results obtained by experiments, analytical models,
In order to analyze the internal porous architectures of foams,
and FE analyses are analysed in the Section 5.
the two-dimensional images of sectional neat foam specimens by

2
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 1. (a) Normal distribution of particle sizes, and (b) the experimentally measured and numerically predicted compressive force as a function of displacement for a single
glass bead particle.

a bandsaw (Rikon 10–305) were obtained using an optical micro- of foam specimen. There were three repeats for each type of poly-
scope (OLYMPUS BX53M). The microscopic images of the neat mer foam.
foams at four different polymer volume fractions are shown in
Fig. 2, which indicated the internal characteristic of closed-cell. 4. Finite element modelling and simulation
The typical schematic of the internal architecture is drawn in
Fig. 2 (c) where e is the clear distance between two adjacent cells, Finite element calculation was conducted to predict the uniax-
and Dc the cell size. The images of sectioned neat polymer foams ial compressive behavior and to investigate the reinforcing mech-
with different F obtained from the optical microscope reveal the anism of glass particles within polymer foams. The finite element
characteristics of close cells, as shown in Fig. 2 (c) and (f). In order modeling can be explained by the following three parts.
to understand the distribution of cell size, the cell sizes were mea-
sured, and the distribution diagram of each type of neat foam is 4.1. Generation of geometry of the foam matrix
plotted in Fig. 2 as well. With the increase of F, the cell size
decreases, e.g. the cell sizes are in the range of 0.464 ~ 3.824 mm For the foam geometry, both porous glass particles and cells
for the foam with F = 0.214 while the size are 0.659 ~ 2.356 mm within the foam specimen were explicitly modelled as spheres
for the foam with F = 0.267. The data in the diagrams suggests that and inserted into the foam matrix randomly. Based on the
the distribution of cell size of foams can be regarded as the normal microstructural analysis in Section 2, it was assumed that both
distribution, i.e. Dc ~ N (Dcm , l2cm ), where Dcm is the mean size of the particle size and cell size followed the normal distribution of
cells and l2cm represents the variance. The distribution parameters dp ~ N (dpm , l2pm ) and Dc ~ N (Dcm , l2cm ), respectively. The following
of hollow cells at different polymer volume fractions have been algorithms were used to arrange the locations of spherical inser-
summarized in Table 1. tions, namely the glass particles and the hollow cells.
The microscopic image of sectioned PR foam with F = 0.321 and The locations of spherical insertions in the cubic domain satisfy
f = 0.2 is shown in Fig. 3 (a). It indicates that the presence of glass two conditions
particle reinforcements within foam matrix does not influence the (1) The coordinate (X, Y, Z) of spherical centers follow the rela-
cell size significantly. Fig. 3 (b) shows an X-ray computed tomogra- tion below
phy (CT) image of a foam with F = 0.321 and f = 0.2. The black
region represents the foam matrix, and the grey spheres represent ðX; Y; Z Þ 2 fðX; Y; Z Þjr 6 X 6 l  r; r 6 Y 6 l  r; r 6 Z 6 l  r g ð2Þ
glass particles. From the CT image shown in Fig. 3 (b), the glass par- where r is radius of a spherical insertions.
ticles are uniformly distributed within foam matrix, which guaran- (2) All the spheres are independent and do not overlap with
tees the repeatability of the compression tests on PR foams. each other
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2
X i  X j þ Y i  Y j þ Z i  Z j  r i þ rj þ e; 8i; j 2 n; i–j ð3Þ
th
3. Experimental methodology where i and j denote parameters associated with the i and the jth

sphere, respectively, and e is the mean distance between any two
In order to characterize the mechanical properties of the GBs adjacent cells, which has been listed in Table 1 for foams with dif-
reinforced polymer foams, quasi-static uniaxial compression was ferent volume fractions. n the number of all spheres.
performed based on the method described by ASTM 1621-04a The geometry of GBs reinforced polymer foam was generated
[34]. Both the neat foam and GBs reinforced foam specimens were via a Python script in environment of the finite element package
cubic with the edge length of l = 50 mm. The compressive ABAQUS Ò [35], and the geometry generation can be divided into
responses of the specimens were measured by a universal screw- three steps which are schematically illustrated in Fig. 4. The foam
driven testing machine (MTS815) with a constant loading rate of matrix was firstly created (Fig. 4 (a)), then the spherical insertions
5 mm/min. The compressive force PF and displacement Dl of foams that have the identical size with the glass particles and cells were
were detected by the 50kN load cell and the linear variable differ- inserted into the cubic foam matrix randomly (Fig. 4 (b)). Next, the
ential transformer (LVDT), respectively. Hence, the engineering inserted spheres were removed from the foam domain via a Boo-
compressive stress and strain can be calculated as r ¼ P F =A and lean operation (Fig. 4 (c)), and the hollow spheres representing
e ¼ Dl=l, respectively, where A is the original cross-sectional area particles were created and inserted into the same location where
3
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 2. The corresponding distributions of particle size for each type of neat foam are shown in (a), (b), (d), and (e), respectively. The microscopic images of neat polymer
foams with F = 0.214, 0.267, 0.321, 0.428 are shown in (c), and (f), respectively.

the solid spheres occupied (Fig. 4 (d)). The Cartesian coordinate It is known that there are plenty of cells with different size
system employed in the numerical simulation is shown in Fig. 4 scales among polymer foams, thus the FE modelling considering
as well, i.e. the compressive orientation of foams is parallel to Z all cells could be computationally expensive. Fig. 5 shows the
axis, both X axis and Y axis are perpendicular to the loading microscopic comparison of neat foam specimen with F = 0.321
orientation. before compression and at the nominal compressive strain of

4
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Table 1 by the eight-node brick elements (Element C3D8R) with reduced


Characteristics of cells. integration. The numerical studies confirmed that the maximum
Volume fraction Mean size of Variance of hollow Mean edge length of the element was 1/26 the size of the corresponding
of polymer (F) hollow cells (mm) cells size (mm2) separation particle, and the number of elements varied from 830 to 1290 were
(lm) required to achieve the converged results.
0.214 1.752 0.642 19.31 Prior to the numerical compression tests on the form speci-
0.267 1.282 0.145 36.96 mens, uniaxial compression test simulations were performed on
0.321 0.992 0.054 124.81
0.428 0.572 0.014 213.36
single GBs to obtain their load–displacement curves (one of which
was presented in Fig. 1(b)) for model calibration purposes. The
simulation process associated with a single GB was the same as
that for the polymer foams. In both cases, the 3D FE models of sin-
0.15, the dashed circles represent the outline of cell geometries. It
gle GBs and foams were clamped between two rigid plates (dis-
indicates that only the large cells give rise to significant deforma-
cretized with the four-node rigid elements, Element R3D4). One
tion, whereas the small cells have negligible deformation. Hence,
plate was fully fixed and another plate was restricted to move
it can be assumed that the large size cells dominate the compres-
along Z axis direction only. During the compression, the moveable
sive deformation of a foam, and only the large cells were modelled
plate imposed loading on the specimen at a constant velocity. To
explicitly to simplify the FE simulation and avoid heavy computa-
ensure that the numerical simulation was quasi-static, the applied
tion. The FE model of foam matrix with low density enables to be
velocity was controlled so that the kinetic energy was under 5% of
obtained via inserting hollow cells into the foam matrix with
the total energy of the model system. For all calculations, a penalty
higher density, and the modelling strategy for creating the FE
contact approach with friction coefficient 0.2 was used to simulate
model of neat foam specimen has been schematically illustrated
the interactions among all surfaces, including foam matrix-particle
in Fig. 6. Take the neat foam with F = 0.214 (Fig. 6 (a) and (b))
surfaces, foam matrix-rigid plate surfaces, and beads-rigid plate
for example, the cell sizes of this type of foam are measured to
interfaces. It should be noted that the numerical results were
be in the range of 0.464 ~ 3.824 mm that follow the normal distri-
insensitive to the friction coefficient according to the FE calcula-
bution of Dc ~ N (Dcm =1.752 mm, l2cm =0.641 mm2). On the other
tions. Ascribe to the considerably different stiffness between foam
hand, the measured cell sizes of the neat foam with F = 0.267 are
matrix and particle reinforcement, the interactive cohesive effect
in the range of 0.659 ~ 2.356 mm that follow the normal distribu-
between these two materials can be neglected.
tion of Dc ~ N (Dcm =1.286 mm, l2cm =0.145 mm2). Hence, the neat
foam with F = 0.214 can be modelled by inserting cells with size 4.3. Constitutive models and material parameters
in the range of 2.120 mm to 3.660 mm followed the normal distri-
bution of Dc ~ N (Dcm =3.0 mm, l2cm =0.063 mm2), into the neat foam 4.3.1. The polymer foam
with F = 0.267. In the elastic range of behaviors, the isotropic foam matrix can
be modelled as being linearly elastic. In the plastic region, the
4.2. Finite element model behaviors are different. The yield function U for isotropic harden-
ing model is adopted, given by:
The FE calculations were conducted via the explicit version of U ¼ rq  ry ¼ 0 ð4Þ
commercial FE package ABAQUS. Fig. 7 shows the FE model for
simulating the cubic GBs reinforced polymer foam specimen. The where ry is the strength in uniaxial tension/compression, and rq
foam matrix was modelled with four-node tetrahedral elements the equivalent stress, which can be defined as a function of the
(ABAQUS element C3D4). The foam matrix model consisted of over mean stress rm and the von Mises stress re [36],
250,000 elements, and the maximum edge length of the element r2e þ h2 r2m
was 1/20 edge length of foam domain for achieving converged r2q ¼ ð5Þ
1 þ h2 =9
results. For the particle reinforcements, each particle was modelled

Fig. 3. The (a) microscopic image and (b) CT image of PR foam with F = 0.321 and f = 0.2.

5
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 4. Generation procedure of the PR foam geometry.

Fig. 5. The FE model creation for neat foams with F = 0.214.

where h is the shape factor of yield surface, which can be calculated calculations, the Poisson’s ratio of polymer foam was assumed as
from the plastic Poisson’s ratio v p _p

0. In this study, we use the equivalent plastic strain rate e to
 
9 1  2v p describe the plastic deformation
h2 ¼   ð6Þ
2 1 þ vp

The isotropic model defines the yield stress ratio ki ¼ rf =pf as _p


 r : e_ pl
e ¼ ð8Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
rcf
ki ¼ 3 1  v p ð7Þ

where rf denotes the initial yield stress in uniaxial compression, where r denotes Cauchy stress tensor, rcf uniaxial compressive
and pf the initial yield stress in hydrostatic compression. In the FE stress, and e_ pl plastic strain rate.
6
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 6. (a) The normal distributions of cells size for the neat foams with F = 0.267, F = 0.214, and the inserted cells. (b) The cells were inserted into the foam matrix with
F = 0.267 to create the foam matrix with F = 0.214.

Fig. 7. Three-dimensional FE model of a PR foam specimen.

4.3.2. The glass particles


The initiation of cracks was modelled by the Rankine Criterion
[37], i.e. crack initiation occurred while the principal stress reached
the tensile strength of the constitutive material. The fracture
energy was used to describe the post-failure behavior. The stress
at a material point decreases linearly with the increasing of crack
opening displacement ut , as shown in Fig. 8. The W cr was defined
as the required energy to open a unit area of a crack

Z utm
W cr ¼ rt dut ð9Þ
0

where utm is the crack opening while the tensile stress rt decreases
to 0. In FE calculations, the meshes were deleted via the technique Fig. 8. The degradation of tensile stress against the crack-opening displacement.

7
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

of element deletion when the deformation of element reached to Fig. 9 (b) indicates that the analytical predictions obtained from
utm to avoid excessive mesh distortion. the Gibson-Ashby model present a good agreement with the exper-
imental measurements. As for the elastic modulus Ef of the foams,

Gibson and Ashby [38] suggested that Ef can also be linked with q,
5. Results and discussion expressed as follow

5.1. Compressive responses of polymer foams Ef 2 


¼ C 1 u2 q þ C 2 ð1  uÞ q ð11Þ
Es
Figs. 9 and 10 show the mechanical responses of the neat poly- where C 1 and C 2 are constants which can be obtained using curve
mer foams and GBs reinforced polymer foams, respectively, under fitting against experimental results, and u is defined as the fraction
uniaxial compressive tests. The measured mechanical properties of of solid contained in separation. Here, assuming that u is a constant
the foams are summarized in Table 2. In addition, the typical com- which is insensitive to the relative density of foams. Eq. (11) can be
pressive stress–strain curve of the polymer foams is plotted in therefore rewritten as follow via curve fitting against the measured
Fig. 11. The relationship of compressive stress versus strain can elastic modulus of foams
be classified into three stages which are elastic stage, plateau stage
and densification stage, corresponding to the curves of OA, AB and Ef 2 
¼ 0:9565q þ 0:186 q ¼ 0:9565F 2 þ 0:186F ð12Þ
BC, respectively, as highlighted in Fig. 11. The mechanical proper- Es
ties and the capacity of energy absorption of different types of The fitting curve shown in Fig. 9 (c) suggests that the elastic
polymer foam have been discussed as follow. modulus of the neat foams is quadratic with polymer volume
fraction.

5.1.1. Compressive responses of neat foams


5.1.2. Compressive responses of GBs reinforced foams
As shown in Fig. 9 (a), the elastic modulus and yield strength of
Fig. 10 illustrates the reinforcing effects of GBs on the compres-
neat foams increase with the rise F. Gibson and Ashby [38]
 sive responses of LLPDE foams. Compared to the neat foams with
reported the relation of yield strength rf and relative density q the same F, the foams reinforced by particles show stronger hard-
(equals F in this paper) of man-made closed-cell foam ening behavior at the plateau stage and can sustain higher stress
levels at densification stage, see Fig. 10 (a) and (b). The elastic
rf 2
 0:05q ð10Þ modulus and yield strength of foams are enhanced by particles
Es and are roughly positive related to the f, as shown in Fig. 10 (c)

Fig. 9. (a) The experimental measurements of uniaxial compressive responses of neat polymer foams with different F. The comparisons of experimental measurements and
analytical models of (b) yield strength and (c) elastic modulus with different F.

8
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 10. The experimental measurements of uniaxial compressive responses of polymer foams of (a) F = 0.214 and (b) F = 0.321 at selected f. The comparisons of experimental
measurements and analytical models of (c) yield strength and (d) elastic modulus as functions of f.

Table 2
The elastic modulus and yield strength of PR foams measured in experiment.

Foams F f Elastic modulus (MPa) Yield strength (MPa)


Neat foams 0.214 0 37.0 1.05
0.267 0 44.5 1.64
0.321 0 82.8 2.15
0.428 0 111.6 3.88
PR foams (F = 0.214) 0.214 0.1 41.2 1.09
0.214 0.2 51.5 1.37
0.214 0.3 83.8 1.51
PR foams (F = 0.321) 0.321 0.1 92.3 1.87
0.321 0.2 112.5 3.02
0.321 0.3 153.7 3.50

and (d). However, the enhancement in yield strength excludes the


GBs reinforced foam with F = 0.321 and f = 0.1 as shown by the
stress–strain curve in Fig. 10 (b) and the analytical prediction in
Fig. 10 (c). This is due to the fact that the foam with higher F pos-
sesses a higher yield strength, thus the particle reinforcements
with low f is not able to support the foam matrix before the crush-
ing of hollow cells, leading to the early failure of particles at low
compressive force. Besides, the enhancements in yield strength
for the foams with F = 0.321 and f = 0.2, 0.3 are higher than the ana-
lytical predictions apparently as shown in Fig. 10 (c), which relates
the well reinforcing performance provided by relatively high vol-
ume fractions of GBs. The analytical formulas for the predictions
of yield strength and elastic modulus are presented in the
following.
The analytical model for the yield strength of GBs reinforced
foams is based on the Reuss lower bound (inverse rule of mixtures)
[39] can be expressed as follow Fig. 11. The typical compressive stress–strain curve of a polymer foam under
uniaxial quasi-static compression.

9
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

rp rf 0:05rp Es q2
rc ¼ ð13Þ rc ¼ ð14Þ
ð1  f Þrp þ f rf 0:05fEs q2 þ ð1  f Þrp

where rc , rp , and rf represent the yield strengths of GBs reinforced It indicates that the analytical prediction of yield strength has a
foams, particle reinforcements, and foam matrix, respectively. The good agreement with experimental measurements at low f. Isai
yield strength of GBs reinforced foam can be further evaluated as and Cohen [40] proposed an analytical model (see Eq. (15)) for pre-
follow by substituting Eq. (10) into Eq. (13) dicting the elastic modulus of GBs reinforced foams

^ 1 (0 6 e 6 0:5) and (b)


Fig. 12. The absorbed energy per unit volume of PR foams with F = 0.214 and (b) F = 0.321 normalized by that of the corresponding neat foams, (a) W
W^ 2 (0:5 6 e 6 0:8), as functions of f.

Fig. 13. Comparisons of engineering stress–strain curves of foams with polymer volume fraction of 0.214 and f = 0, f = 0.1, and f = 0.2 between experiments and simulations.

10
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

" #
f forcements to improve the compression resistance. The FE simula-
Ec ¼ 1 þ 1=3
Ef ð15Þ tion in Section 5.2.2 suggests that the foam starts to be densified
m=ðm  1Þ  f
when upper and lower surfaces of the large hollow cells start to
with contact with each other, corresponding to the compressive strain
of 0.5 (Point B in Fig. 11); the foams can be regarded to be fully
m ¼ Ep =Ef ð16Þ
densified and particles are crushed when upper and lower surfaces
where Ec is the elastic modulus of GBs reinforced foam. are in full contact, corresponding to the compressive strain of 0.8
In addition, another analytical prediction, which considers the (Point C in Fig. 11). Hence, the absorbed energy per unit volume
R 0:5
micromechanics, suggested by Halpin and Tsai [41] is given as W 1 , W 1 ¼ 0 rde, is calculated to characterize the influence of
follow particle stabilization; the absorbed energy per unit volume of
R 0:8
1 þ dgf W 2 , W 2 ¼ 0:5 rde, is used to characterize the influence of particles
Ec ¼ Ef ð17Þ
1  gf crushing.
To understand the effect of particle reinforcements on the
with
energy absorption capacity of polymer foams, the absorbed energy
Ep =Ef  1 per unit volume of GBs reinforced foams W 1r ð0 6 e 6 0:5Þ and
g¼ ð18Þ
W 2r ð0:5 6 e 6 0:8Þ are normalized by the absorbed energy per
Ep =Ef þ d
unit volume of neat foams W 1p ð0 6 e 6 0:5Þ and
where, d is a non-dimensional parameter dependent on the geome- ^ 1 ¼ W 1r =W 1p
W 2p ð0:5 6 e 6 0:8Þ, respectively, expressed as W
try of reinforcement. Specifically, d ¼ 2 for the spherical particle
and W ^ 2 ¼ W 2r =W 2p . Fig. 12 (a) and (b) plot the W
^ 1 and W
^ 2 as
[42]. As plotted in Fig. 10 (d), both the two analytical models have
good agreements with the experimental measurements, especially a function of particle volume fraction at polymer volume fraction
at low particle volume fractions, i.e. f 6 0:2. of 0.214 and 0.321, respectively. The normalized energy absorption
of GBs reinforced foams is normally higher than that of the neat
5.1.3. Energy absorption capacity foams for the same volume fractions, which suggests the particle
Normally, the neat foams enable to absorb most energy via the reinforcement enables to improve the energy absorption capacity
deformation of hollow cells before foam densification. For the GBs of foams. However, the foams with F = 0.321 and f = 0.1 are the
reinforced foams, the cells can be stabilized by the particle rein- exceptions due to the earlier crushing of particle reinforcements

Fig. 14. FE predictions of deformation process of the neat polymer foam with F = 0.214 under uniaxial compression.

11
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

than crushing the hollow cells, as discussed in Section 5.1.2. In compressive stress when e P 0:4 for the foams with f = 0.2. Signif-
addition, the particle reinforcements, especially at lower particle icant mesh distortion of foam matrix occurred when e>0.5, which
volume fraction (f = 0.1), have more significant enhancement on resulted in the end of calculations. The fluctuation of curves
the energy absorption capacity for foams with F = 0.214 than the obtained from FE predictions for the foam with f = 0.1 and
foams with F = 0.321. f = 0.2, which were highlighted by the dashed circles, could be
caused by the crushing of glass particles.
5.2. Finite element simulation of polymer foams

5.2.2. Deformation mechanism of neat foam


5.2.1. Compressive responses of polymer foams
Fig. 14 shows the deformation process of a cross-section of the
To investigate micromechanics of the foams under compres-
neat polymer foam with F = 0.214, which is at the centre of the 3D
sion, the FE models of F = 0.214 with f = 0, f = 0.1, and f = 0.2 were
FE model parallel to the z axis. The coordinates and the loading
built. The selection of the appropriate size of FE model is signifi-
direction are also shown in Fig. 14, which are used throughout
cant, as the numerical predictions could be either over- or under-
the FE simulations. The deformation was quantified by the equiva-
estimated when using unsuitable model sizes. Roberts and Gar-
p R t _ p
boczi [43] defined a size parameter Ls with Ls ¼ l=Dc to evaluate lent plastic strain e ¼ 0 e dt. Four states were selected to analyse
the size effects, where l is the foam edge length and Dc is the cell the responses of foam matrix and hollow cells under uniaxial com-
size. The stable results were obtained when l=Dc P 5. In this study, pression, i.e. initial state (e ¼ 0) as shown in Fig. 14 (a), yielding
the edge length of foam matrix in FE model was chosen as 10 mm. (e ¼ 0:05) as shown in Fig. 14 (b), plateau stage (e ¼ 0:15) as
As the mean size of cell size of polymer foam with F = 0.267 was shown in Fig. 14 (c), and the onset of densification (e ¼ 0:5) as
Dc ¼ 1:286, thus l=Dc ¼ 7:776, which could achieve the stable shown in Fig. 14 (d). In comparison with the initial state before
results. compression (e ¼ 0 shown in Fig. 14 (a)), the yielding occurred
Fig. 13 (a), (b) and (c) show the comparisons of engineering at the adjacent hollow cells due to strain localization highlighted
stress–strain curves of polymer foams of F = 0.214 with f = 0, by the dashed circles when e ¼ 0:05 (Fig. 14 (b)). Then, at the
f = 0.1, and f = 0.2 between experiments and simulations, respec- plateau stage e ¼ 0:15 (Fig. 14 (c)), the phenomenon of strain
tively. The agreements were good in both the elastic stage and pla- localization connected all hollow cells horizontally, while the foam
teau stage, although the FE prediction slightly overestimated the matrix at the upper and lower sides of hollow cells still remained

Fig. 15. FE predictions of deformation process of the polymer foam with F = 0.214 and f = 0.2 under uniaxial compression.

12
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

undeformed. In addition, the shape of hollow cells changed from


spheres to the oblate spheroidal spheres. When e ¼ 0:5, the upper
and lower sides of an oblate sphere were nearly in contact with
each other, as highlighted by the dashed circles in Fig. 14 (d),
which indicated the initiation of densification. The foam matrix
was densified at the regions of high strain localization, e.g. the light
green regions at two sides of deformed hollow cells.

5.2.3. Deformation mechanism of GBs reinforced foam


As f = 0.2 provided apparent reinforcing effect compared with
f = 0.1 via the experiments, as shown in Fig. 10 (a), the deformation
mechanism of GBs reinforced foams was analysed via the FE model
with F = 0.214 and f = 0.2, as shown in Fig. 15. The states selected
from the mid of 3D FE model parallel to the z axis were the same as
those for the neat polymer foam. The glass particles were repre-
sented by the grey spheres. Fig. 15 (a) shows the initial state of
GBs reinforced foam before deformation. When e ¼ 0:05, the
yielding occurred due to strain localization highlighted by the
dashed circles, as shown in Fig. 15 (b). In the plateau stage Fig. 17. Effect of particle sizes with d0 = 1.4 mm, 1.75 mm, 2.00 mm on the
compressive response of foams with F = 0.214 and f = 0.1.
(e ¼ 0:15), strain localization locates at the regions surrounding
the particle reinforcements and right and left sides of hollow cells.
The region without particles and hollow cells show no deformation mean sizes of glass particles employed in the FE simulation are
as highlighted in Fig. 15 (c), which is reinforced by the glass parti- dpm = 1.40 mm, 1.75 mm and 2.00 mm, respectively. Fig. 17 shows
cles. At the initiation of densification (e ¼ 0:15), the foam matrix the predicted response of the foams reinforced by the above three
was densified, as shown the contact of upper and lower sides of types of particles and also the experimentally measured response
oblate spheres highlighted by the dashed circles. In addition, the of foam reinforced by the particles with mean size of 1.75 mm.
collapse of glass particles is also obtained from numerical predic- The comparison indicates that the elastic modulus of foam is
tions, as shown in Fig. 15 (d). insensitive to the particle size. However, both the yield strength
To understand the behavior of glass particles within the foam and hardening stress of foam increase with the decrease of particle
matrix during the compression and initial damage, the deforma- sizes, which may suggest the particles with smaller size play a
tion process of 3D FE model of glass particles is shown in Fig. 16. more significant role in reinforcing.
The von Mises stress re is used to show the stress states of parti- Fig. 18 shows the 3D FE models of particles with
cles. Fig. 16 (a) shows the initial state and the layout of the glass
dpm = 1.40 mm and 2.00 mm, which are embedded in the foam
particles. The initiation of damage is shown in Fig. 16 (b) when
with F = 0.214 and f = 0.2 when damage occurs. For the foam
e ¼ 0:25 highlighted by the yellow box. At the onset of densifica- with dpm = 1.40 mm, the damage occurs at e ¼ 0:19, as shown
tion, some particles were completed crushed, as shown in Fig. 16
in Fig. 17 (a). In contrast, the damage occurs at e ¼ 0:29 for the
(c).
foam with dpm = 2.00 mm. This indicates that the glass particles
with smaller size could bear higher compressive force than the
5.2.4. Size effect of particle reinforcements
larger particles at the same deformation, which is the reason
In order to understand the size effects of glass particles on the
why the hardening stress of foam increase with the decrease of
compressive behavior of GBs reinforced foams, the numerical sim-
particle sizes, as plotted in Fig. 17.
ulation was performed on the foam with F = 0.214 and f = 0.1. The

Fig. 16. The 3D FE predictions of deformation process of glass particles with the foam matrix of F = 0.214 and f = 0.2 under uniaxial compression.

13
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

Fig. 18. The compressive responses of particles with (a) dpm = 1.40 mm and (b) dpm = 2.00 mm.

6. Conclusions of matrix material. In the future studies, we will compare the cur-
rent strategy with RVE method about the computational expenses
The uniaxial compressive behaviors of neat polymer foams and and accuracy.
glass GBs reinforced polymer foams were studied through analyt-
ical analyses, FE analyses, and experimental measurements. The
internal porous microstructures of polymer foams were examined Declaration of Competing Interest
via the optical microscopes and X-ray CT scanning, which provided
a rational reference for selecting appropriate modelling parameters The authors declare that they have no known competing finan-
of the FE analysis. cial interests or personal relationships that could have appeared
The experimental measurements have demonstrated that to influence the work reported in this paper.
embedding glass particles can increase the macroscopic elastic
modulus, yield strength and capacity of energy absorption com-
References
pared to the neat foams. For the interpretation purpose, the analyt-
ical predictions of elastic modulus and yield strength of neat foams [1] S. Kaewunruen, C. Ngamkhanong, M. Papaelias, C. Roberts, Wet/dry influence
and GBs reinforced foams were calculated, which showed good on behaviors of closed-cell polymeric cross-linked foams under static, dynamic
and impact loads, Constr. Build. Mater. 187 (2018) 1092–1102, https://doi.org/
agreements with the experimental results.
10.1016/j.conbuildmat.2018.08.052.
In order to capture the deformation mechanism of polymer [2] S.J. Kumar, S.J.A. Kumar, Low-velocity impact damage and energy absorption
foams during uniaxial compression, FE simulations modelled the characteristics of stiffened syntactic foam core sandwich composites, Constr.
hollow cells and glass particles based on the observed internal Build. Mater. 246 (2020) 118412–118439, https://doi.org/10.1016/
j.conbuildmat.2020.118412.
microstructures explicitly. The FE predictions showed good agree- [3] V. Realinho, L. Haurie, M. Antunes, J.I. Velasco, Thermal stability and fire
ments with the experimental results. In addition, the deformation behavior of flame retardant high density rigid foams based on
mechanism was well captured in the simulations: during the pla- hydromagnesite-filled polypropylene composites, Compos. Part B-Eng. 58
(2014) 553–558, https://doi.org/10.1016/j.compositesb.2013.11.015.
teau stage, strain localization occurred at the regions surrounding [4] W.O. Soboyejo, T.S. Srivatsan, Advanced structural materials: properties,
the particle reinforcements and both right and left sides of hollow design optimization, and applications, first ed., CRC Press, 2006.
cells. The region without particles and hollow cells showed almost [5] L. Gong, S. Kyriakides, W.-Y. Jang, Compressive response of open-cell foams.
Part I: Morphology and elastic properties, Int. J. Solids Struct. 42 (5-6) (2005)
no deformation. The undeformed region was reinforced by the 1355–1379, https://doi.org/10.1016/j.ijsolstr.2004.07.023.
glass particles. It was shown that the foam matrix was densified [6] N.A.D. Muhalim, M.Z. Hassan, Y. Daud, Elastic-Plastic Behaviour of Ultrasonic
as a result of the contact of upper and lower sides of oblate spheres Assisted Compression of Polyvinyl Chloride (PVC) Foam, IOP Conf. Ser.: Mater.,
344 (2018) 12009. https://doi.org/10.1088/1757-899x/344/1/012009.
in the densification stage. The parametric study on the size effect of [7] S.C. Warburton, A.M. Donald, A.C. Smith, The deformation of brittle starch
glass particles indicated that the hardening stress of foams foams, J. Mater. Sci. 25 (9) (1990) 4001–4007, https://doi.org/10.1007/
increased with the decrease of particle sizes. BF00582472.
[8] F. Beer, J.E.R. Johnston, J. DeWolf, D. Mazurek, Mechanics of Materials, sixth ed.,
As a pioneering study on the explicitly modelling of cells and
McGraw-Hill Education, 2015.
particles simultaneously within the FE models of PR foams, there [9] I. Duarte, J. Ferreira, Composite and Nanocomposite Metal Foams. Materials 9
are still several issues that need to be addressed in the future stud- (79) (2016), 1-34. https://doi.org/10.3390/ma9020079.
ies. Although the cells were modelled using spheres, ignoring their [10] S. Cao, W. Zhu, T. Liu, Bio-inspired self-healing polymer foams with bilayered
capsule systems, Compos. Sci. Technol. 195 (2020) 108189, https://doi.org/
non-smooth geometry, the FE model was able to predict the 10.1016/j.compscitech.2020.108189.
mechanical behavior of interactions between internal microstruc- [11] M. Guden, S. Yüksel, SiC-particulate aluminum composite foams produced
tures of foams and GBs explicitly. The explicitly modelling of poly- from powder compacts: foaming and compression behavior, J. Mater. Sci. 41
(13) (2006) 4075–4084, https://doi.org/10.1007/s10853-006-7645-x.
hedral cells and particles simultaneously awaits further [12] S. Elbir, S. Yilmaz, A.K. Toksoy, M. Guden, I.W. Hall, SiC-particulate aluminum
investigation. Besides, in this study, we considered the real sce- composite foams produced by powder compacts: Foaming and compression
nario that the reinforcements did not intersect with the boundary behavior, J. Mater. Sci. 38 (2003) 4745–4755, https://doi.org/10.1023/
a:1027427102837.

14
S. Cao, Y. Zhang and Y. Lu Construction and Building Materials 275 (2021) 122062

[13] M. Alizadeh, M. Mirzaei-Aliabadi, Compressive properties and energy [27] Youming Chen, Raj Das, Mark Battley, Effects of cell size and cell wall thickness
absorption behavior of Al–Al2O3 composite foam synthesized by space- variations on the stiffness of closed-cell foams, Int. J. Solids Struct. 52 (2015)
holder technique, Mater. Des. 35 (2012) 419–424, https://doi.org/10.1016/ 150–164, https://doi.org/10.1016/j.ijsolstr.2014.09.022.
j.matdes.2011.09.059. [28] A. Ghazi, P. Berke, K. Ehab Moustafa Kamel, B. Sonon, C. Tiago, T.J. Massart,
[14] A.R. Kennedy, S. Asavavisitchai, Effects of TiB2 particle addition on the Multiscale computational modelling of closed cell metallic foams with
expansion, structure and mechanical properties of PM Al foams, Scripta detailed microstructural morphological control, Int. J. Eng. Sci. 143 (2019)
Materialia 50 (1) (2004) 115–119, https://doi.org/10.1016/j. 92–114, https://doi.org/10.1016/j.ijengsci.2019.06.012.
scriptamat.2003.09.026. [29] A. Ghazi, P. Berke, C. Tiago, T.J. Massart, Computed tomography based
[15] N.Q. Zhao, B. Jiang, X.W. Du, J.J. Li, C.S. Shi, W.X. Zhao, Effect of Y2O3 on the modelling of the behaviour of closed cell metallic foams using a shell
mechanical properties of open cell aluminum foams, Mater. Lett. 60 (13-14) approximation, Mater. Des. 194 (2020) 108866, https://doi.org/10.1016/
(2006) 1665–1668, https://doi.org/10.1016/j.matlet.2005.11.088. j.matdes.2020.108866.
[16] T. Shi, X. Chen, Y. Cheng, Y. Liu, H. Zhang, Y. Li, Y.,, Microstructure and [30] A. Ghazi, C. Tiago, B. Sonon, P. Berke, T.J. Massart, Efficient computational
Compressive Properties of Aluminum Foams Made by 6063 Aluminum Alloy modelling of closed cell metallic foams using a morphologically controlled
and Pure Aluminum, Mater. Trans. 59 (2018) 625–633, https://doi.org/ shell geometry, Int. J. Mech. Sci. 168 (2020) 105298, https://doi.org/10.1016/j.
10.2320/matertrans.m2017300. ijmecsci.2019.105298.
[17] T. Geramipour, H. Oveisi, Effects of foaming parameters on microstructure and [31] Y.J. Cho, W. Lee, Y.H. Park, Finite element modeling of tensile deformation
compressive properties of aluminum foams produced by powder metallurgy behaviors of iron syntactic foam with hollow glass microspheres, Materials 10
method, Trans. Nonferrous Met. Soc. China 27 (7) (2017) 1569–1579, https:// (10) (2017) 1201–1216, https://doi.org/10.3390/ma10101201.
doi.org/10.1016/S1003-6326(17)60178-X. [32] A. Samimi, A. Hassanpour, M. Ghadiri, Single and bulk compressions of soft
[18] S. Esmaeelzadeh, A. Simchi, D. Lehmhus, Effect of ceramic particle addition on granules: Experimental study and DEM evaluation, Chem. Eng. Sci. 60 (2005)
the foaming behavior, cell structure and mechanical properties of P/M AlSi7 3993–4004, https://doi.org/10.1016/j.ces.2005.02.036.
foam, Mater. Sci. Eng., A 424 (1-2) (2006) 290–299, https://doi.org/10.1016/j. [33] C. Thornton, Z. Ning, A analytical model for the stick/bounce behaviour of
msea.2006.03.013. adhesive, elastic-plastic spheres, Powder Technol. 99 (2) (1998) 154–162,
[19] Sirong Yu, Jiaan Liu, Yanru Luo, Yaohui Liu, Compressive behavior and https://doi.org/10.1016/s0032-5910(98)00099-0.
damping property of ZA22/SiCp composite foams, Mater. Sci. Eng., A 457 (1- [34] ASTM Standard D1621 04a, Standard Test Method for Compressive Properties
2) (2007) 325–328, https://doi.org/10.1016/j.msea.2006.12.089. of Rigid Cellular Plastics, ASTM International, West Conshohocken, PA, 2004,
[20] M.L. Jayavardhan, M. Doddamani, Quasi-static compressive response of DOI: 10.1520/D1621-04A, www.astm.org.
compression molded glass microballon/HDPE syntactic foam, Compos. Part [35] ABAQUS, ABAQUS analysis user’s manual, Version 6.14 ABAQUS, Inc., 2014.
B-Eng. 149 (2018) 165–177, https://doi.org/10.1016/j.compositesb.2018. [36] V.S. Deshpande, N.A. Fleck, Isotropic constitutive models for metallic foams, J.
04.039. Mech. Phys. Solids 48 (2000) 1253–1283, https://doi.org/10.1016/S0022-5096
[21] M.C. Saha, Md.E. Kabir, S. Jeelani, Enhancement in thermal and mechanical (99) 00082-4.
properties of polyurethane foam infused with nanoparticles, Mater. Sci. Eng., A [37] L. Dahmani, C.S.H. Mohand, Modélisation and influence of shear retention
479 (1-2) (2008) 213–222, https://doi.org/10.1016/j.msea.2007.06.060. parameter on the response of cast iron beam, Procedia Eng. 10 (2011) 19–24,
[22] T.O. Ogundairo, D.D. Adegoke, I.I. Akinwumi, O.M. Olofinnade, Sustainable use https://doi.org/10.1016/j.proeng.2011.04.005.
of recycled waste glass as an alternative material for building construction – A [38] L.J. Gibson, M.F. Ashby, Cellular Solids: Structure and Properties, third ed.,
review, IOP Conf. Ser.: Mater., Sci. Eng., 640 (2019), 012073. http://dx.doi.org/ Cambridge University Press, 1999.
10.1088/1757-899X/640/1/012073. [39] Y. Chen, Micro-mechanical Modelling of Polymeric Closed-cell Foams: Effect of
[23] H. Aygün, F. McCann, F., Structural and acoustical performance of recycled Microstructural Variability on the Mechanical Properties, Ph.D. The University
glass bead panels, Constr. Build. Mater., 258 (2020) 119581. https://doi.org/ of Auckland, 2016.
10.1016/j.conbuildmat.2020.119581. [40] O. Ishai, L.J. Cohen, Elastic properties of filled and porous epoxy composites,
[24] Claudia Redenbach, Microstructure models for cellular materials, Comput. Int. J. Mech. Sci. 9 (8) (1967) 539–546, https://doi.org/10.1016/0020-7403(67)
Mater. Sci. 44 (4) (2009) 1397–1407, https://doi.org/10.1016/ 90053-7.
j.commatsci.2008.09.018. [41] J.C. Halpin, S.W. Tsai, Effects of Environmental Factors on Composite Materials,
[25] C. Redenbach, I. Shklyar, H. Andrä, Laguerre tessellations for elastic Air Force Technical Report AFML-TR (1969) 67–423.
stiffness simulations of closed foams with strongly varying cell sizes, Int. [42] Wei Yang, Wei Shi, Zhong-Ming Li, Bang-Hu Xie, Jian-Min Feng, Ming-Bo Yang,
J. Eng. Sci. 50 (1) (2012) 70–78, https://doi.org/10.1016/j. Mechanical Properties of Glass Bead-Filled Linear Low-Density Polyethylene, J.
ijengsci.2011.09.002. Elastomers Plast. 36 (3) (2004) 251–265, https://doi.org/10.1177/
[26] H.X. Zhu, P. Zhang, D. Balint, S.M. Thorpe, J.A. Elliott, A.H. Windle, J. Lin, The 0095244304042989.
effects of regularity on the geometrical properties of Voronoi tessellations, [43] A.P. Roberts, E.J. Garboczi, Elastic properties of model random three-
Phys. A: Statistical Mech. Appl. 406 (2014) 42–58, https://doi.org/10.1016/ dimensional open-cell solids, J. Mech. Phys. Solids 50 (1) (2002) 33–55,
j.physa:2014.03.012. https://doi.org/10.1016/S0022-5096(01)00056-4.

15

You might also like