Polymer Testing 122 (2023) 108014

Contents lists available at ScienceDirect

Polymer Testing
journal homepage: www.elsevier.com/locate/polytest

Deformation analysis in impact testing of functionally graded foams by the

image processing of high-speed camera recordings
Márton Tomin a, Dániel Török a, Tamás Pászthy b, Ákos Kmetty a, c, *
a
Department of Polymer Engineering, Faculty of Mechanical Engineering, Budapest University of Technology and Economics, Műegyetem rkp. 3, H-1111, Budapest,
Hungary
b
Department of Information Technology, University of Miskolc, Egyetem út 17, H-3515, Miskolc - Egyetemváros, Hungary
c
ELKH–BME Research Group for Composite Science and Technology, Műegyetem rkp. 3, H-1111, Budapest, Hungary

A R T I C L E I N F O A B S T R A C T

Keywords: We developed an image processing algorithm and applied it on high-speed camera recordings to characterize the
Impact test deformation response of three-layered density-graded foam structures subjected to drop weight testing. Different
High-speed camera densities (30, 40, 50 and 70 kg/m3) of weakly cross-linked polyethylene foam sheets were laminated together to
Image processing
achieve varying density distributions along the thickness, and the effect of layer order on the shock absorption
Density gradient
Polymer foam
capability was evaluated. Foam structures with a higher density top layer and a negative density gradient showed
enhanced energy absorption in the initial stage of deformation, which resulted in lower maximum reaction
forces. The positive effect of layer order modification was more dominant at higher impact energies. We provided
a detailed explanation of the tendencies by investigating the differences in deformation propagation and the
changes in the diameter of the deformation zone. The presented method can be utilized to design sports and
packaging foam products.

1. Introduction high reaction forces (densification zone) [13]. As a result, focus has
shifted to developing functionally graded foams that have varying den­
Weight reduction to decrease material costs and the environmental sity distribution along the thickness [14]. With the use of a non-uniform
footprint of transportation is becoming an increasingly important goal for cell structure, the stress level of the plateau zone can be increased, and the
engineers. Therefore, polymeric foams are of paramount importance, as start of the densification zone can be shifted to higher strain levels,
they contribute to weight reduction and have excellent thermal and allowing more energy to be absorbed with lower reaction forces.
mechanical properties at the same time [1]. Foams are important not only Several studies in recent years aimed to produce foams with non-
in the packaging industry [2] but in several other applications as well. uniform density distribution. The approaches presented so far include
They are used in the construction industry for the thermal insulation of syntactic foaming [15–17], batch foaming [18,19], compression mold­
buildings [3] and as soundproofing walls for noise reduction [4]. More­ ing [20], and injection molding of structural foams [21,22], which all
over, their advanced energy-absorbing capacity is also exploited in result in a continuously varying density.
automotive [5,6] and sports applications [7,8], where they can protect In the case of syntactic foaming, researchers use micro-balloons
the passengers/athletes from injuries by reducing the shocks during distributed in a polymer matrix [15], which mostly results in precisely
collisions/impacts. Due to their cellular structure, polymer foams show a controlled particle distribution, thus a quasi-homogenous cell structure.
special material response to loads, especially under compression (see However, the achievable mass reduction is relatively low. Gupta [16]
Fig. 1.), when they can absorb a huge amount of energy in the so-called used glass micro-balloons and epoxy resin to produce 500–700 kg/m3
plateau region through cell wall bending and buckling [9,10]. density foams while using the same processing technology, Higuchi et al.
However, in some cases, a homogenous density foam structure is not [17] achieved 720 and 930 kg/m3 densities. Both studies came to the
resistant enough to absorb the energy of the impact, and the resulting conclusion that by changing the density distribution along the thickness,
excessive cell compaction leads to undesirable material response with the energy absorption capacity of foams measured in compression tests

* Corresponding author. Department of Polymer Engineering, Faculty of Mechanical Engineering, Budapest University of Technology and Economics, Műegyetem
rkp. 3, H-1111, Budapest, Hungary.
E-mail address: [email protected] (Á. Kmetty).

https://doi.org/10.1016/j.polymertesting.2023.108014
Received 9 December 2022; Received in revised form 14 March 2023; Accepted 1 April 2023
Available online 3 April 2023
0142-9418/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
M. Tomin et al. Polymer Testing 122 (2023) 108014

simulation of foam behavior at high strain rates requires high compu­

tational capacity and complicated parameter fitting [25], only one study
is available that investigated in detail the effect of the density gradient
on the shock-absorbing capacity at different impact energies. Cui et al.
[26] performed finite element simulations for the drop weight tests of
functionally graded closed-cell foams. In the impact energy range of
15–30 J, structures with decreasing density from top to bottom showed
better shock absorption capacity compared to uniform density foam
models. This difference disappeared at higher impact energies as the
foam models were completely compressed. Their conclusion assumed
that the foam should absorb all the impact energy in the plateau range
before densification starts. If this region is reached, the foam will
transmit the majority of the energy as a propagating stress wave to the
user/product, which should be protected. However, it is important to
note that the simulation results were not validated by tests, and the exact
deformation response of a varying-density foam structure to impact
loads was not studied.
In addition to the production of continuously graded structures in
one step during foaming, the generation of discretely graded structures
is also a possibility in a two-step manufacturing process. In the case of
packaging and sports, sheet extrusion [27] is the most widespread in­
Fig. 1. Typical stress–strain response of a polymer foam to compressive load
dustrial technology to produce normal foam sheets in the first step. Then
with the three well-distinguished regions: (a) linear elasticity, (b) plateau, (c)
a multi-layered structure is generated in the second step by bonding or
densification (reproduced with general permission of SciELO [11] and
permission of Elsevier [12]). welding laminates of different densities together [28]. Shimazaki et al.
[29] investigated three-layer structures created by laminating
ethylene-vinyl-acetate foam sheets to reduce the loads transmitted by
can be increased [16,17].
shoe soles to the user during running and walking. By varying the
Similar density reduction was achieved by Cusson et al. [20], who
concentration of the azodicarbonamide blowing agent (8, 12 and 16
applied different temperatures on both sides of a compression molding
phr), three different density layers (230, 170 and 110 kg/m3) were
system to produce linear low-density polyethylene foams using azodi­
foamed and laminated together and a 3-layer sandwich structure was
carbonamide as a chemical blowing agent.
produced with a total thickness of 15 mm. The shock absorption effi­
Much higher porosity can be obtained with supercritical gas as a
ciency of the multilayer sandwich structures was evaluated by applying
physical blowing agent [18,19]. Mohyeddin and Fereidoon [18]
cyclic loads of 1000 N per second and placing pressure sensors in the top
implemented a solid-state batch process, which resulted in foams with a
and bottom boundary layers. Their results showed that the use of
graded porosity foam core and an integral solid skin layer. Another
functionally graded foams could advantageously modify the shock ab­
innovative solution was proposed by Yu et al. [19], who batch foamed
sorption of the system. In their case, samples with increasing density
polymers on a substrate of anodized aluminum oxide film modified with
from top to bottom performed better, with the bottom layer taking the
fluorinated silane. First, polymer particles were compressed on the film
load first during running [29]. However, the deformation mechanisms in
in a hydraulic press to produce ~1 mm thick plates, and then they were
the foam layers were not investigated, so an accurate explanation of the
batch foamed in a high-pressure vessel with a scCO2 blowing agent. The
results from a structural point of view is missing.
cell size of the graded foam near the substrate decreased to a minimum
In summary, several modeling studies and experimental data have
of 2 μm, gradually increasing in an upward direction.
shown that density-graded polymeric foams can show enhanced energy
Compared to structural foams with uniform cells, we can also pro­
absorption compared to single-density uniform structures. However, the
duce products with far higher strength by injection molding structural
reason for the positive effect of varying the density distribution is not
foams, which have a sandwich structure. Both the foam core and rigid
fully understood, and the theories of previous research are not supported
shell layer of these structures are formed during the manufacturing
by experimental data. Furthermore, the typical testing methods of foams
process. Since the wall of the mold is cold, the polymer melt, which gets
(compression test [30], drop weight impact test [31,32], Split Hopkin­
into contact with the mold surface, solidifies quickly, so the gas cannot
son Bar tests [33,34]) are not sufficient alone to give a detailed expla­
expand the polymer there. Most studies using this approach set different
nation of tendencies.
temperatures on the two cavity surfaces and investigate the effect of this
This study aims to investigate the impact loading of multi-layered
temperature difference on the thickness ratio of the skin and core layers.
functionally graded foam structures by developing a new measure­
Although the transition between the morphology of the shell and core
ment method based on the image processing of high-speed camera re­
layers is sharp, the achievable density reduction is generally small
cordings. The high-speed (HS) camera is a commonly used tool for
(~30%) [1,21–23].
impact testing measurements to validate results and quantify maximum
The presented articles are limited mainly to the investigation of the
deformation [35]. For a similar research area, Hoohbor et al. [36]
morphology with the aim of assessing the applicability and reproduc­
effectively used HS camera–based digital image correlation to analyze
ibility of the production method. The limited number of experimental
the stress wave propagation of high-velocity impacts from pressure
data on the mechanical testing of functionally graded foams and the lack
Hopkinson bar tests with 3-layered rigid PU foams. They obtained the
of analysis on their deformation mechanism means that we can mostly
axial strains vs. time data to analyze the variation in the densification
conclude the material response from theoretical analysis. Uddin et al.
process of the structure as a function of the distance from the impact
[24] investigated virtual three-layered polyurea foam laminates sub­
side. However, there is currently no measurement method available for
jected to quasi-static uniaxial compression and showed that graded
the drop weight testing of polymeric foams that can monitor the
laminates could outperform single-density foams in terms of strength
deformed area’s shape, size, and variation over time by treating the
and energy absorption. Similar conclusions were found by Koohbor and
layers of different densities separately. While the current study focuses
Kidane [14] from the constitutive modeling of compression testing
only on cross-linked polyethylene (XPE) foams used in the sports
4-layer graded polyurethane foam structures. As the numerical
equipment industry, the method presented here can also be applied to

2
M. Tomin et al. Polymer Testing 122 (2023) 108014

the design of foams for packaging and automotive industrial use. Table 2
Layer order of the investigated foam structures.
2. Materials Sample Nominal Nominal Nominal Average
name density of the density of the density of the density
top layer [kg/ middle layer bottom layer [kg/m3]
We investigated discretely graded, three-layer, weakly cross-linked
m3] [kg/m3] [kg/m3]
polyethylene foam structures. The foams had similar average densities
but different densities in each layer. The cell-structural characteristics of 50-50-50 50 50 50 47.9 ± 0.5
30-50-70 30 50 70 48.7 ± 0.4
the individual layers are summarized in Table 1. Cell density shows the
40-70-40 40 70 40 51.6 ± 0.7
number of the cells in a given volume. In general, the higher the cell 70-30-70 70 30 70 54.4 ± 1.4
density, the better the mechanical properties of the foam. 70-50-30 70 50 30 48.7 ± 0.4
The foams were provided by Polifoam Ltd. (Budapest, Hungary). All
structures were produced on a flat film production line with an azodi­
Table 3
carbonamide foaming agent and a dicumyl peroxide cross-linking agent.
Test parameters of the impact tests.
The exact layer order of the functionally graded foam structures and
their average densities are summarized in Table 2. The average densities Property Value

were calculated from the specimens’ mass and dimensions. In each case, Impactor geometry Cylinder
the multilayer samples were welded together from three 10 mm thick Material of the impactor Steel
Impactor diameter 50 mm
layers by flame lamination. In the evaluation, all samples were
Drop height 400 mm
compared to the uniform-density “50-50-50′′ sample. Impact velocity 2.8 m/s
Impactor mass 2.514 kg 3.514 kg 4.514 kg 5.514 kg
3. Experimental Impact energy 9.862 J 13.784 J 17.707 J 21.630 J

3.1. Calculation of average density

The average densities of the structures were determined from the

mass and dimensions of the specimens. The mass was measured with an
Ohaus Explorer (Nänikon, Switzerland) balance (accuracy 0.0001 g),
while the volume of the foam blocks was determined with a GOM ATOS
Core 5 M (Gesellschaft für Optische Messtechnik GmbH, Germany) 3D
optical measuring system.

3.2. Evaluation of shock absorption

3.2.1. Drop weight impact tests

The dynamic testing of the foams was carried out at four different
impact energy levels with a Ceast 9350 impact tester using a 22 kN load
cell. We varied impact energy by placing extra weights in the dropping
frame. The exact measurement settings are summarized in Table 3.
The 100 × 100 × 30 mm size specimens were placed on a 40 mm
thick steel plate, which functioned as rigid support during the impacts.
When the cylindrical-shaped impactor was dropped down to the center
of the foam, the force versus time data were recorded, from which the
maximum reaction force, the maximum deformation, and the absorbed
Fig. 2. Schematics of the test setup of impact testing with a high-speed camera
energy were calculated. Data was acquired with a Ceast DAS 64k High-
(side view).
Speed Data Acquisition Unit at 1 MHz.
mass until the deformation of the samples fell within the range of
3.2.2. Deformation analysis with a high-speed camera 80–100%. The finalized measurement settings were set to 400 mm drop
A special arrangement was used to analyze the deformed area’s height and 2.514 kg mass, which resulted in an impact energy of 9.86 J
shape, size, and variation over time. The structural deformation of the and impact velocity of 2.8 m/s. We investigated the deformation
foam during impact testing was recorded with a Keyence VW-9000 high- mechanism of the 3-layer foam structures with a new evaluation method
speed camera. The schematics of the test setup are presented in Fig. 2. based on the image analysis of the HS-camera recordings.
We positioned the specimens with a dimension of 200 × 200 × 30 In this test, we prepared the samples in a special way, which made
mm in such a way that only half of the impactor hits them, so that we can the segmentation of layers easier during image processing. The middle
see the changes in the contours of the layers during deformation. The layer of the foams was masked with a tape, and an AESUB ASW102
schematics shows only that part of the foam sample which is affected white pigment spray was sprayed on the side of the foams. After the
during the impact. removal of the tape, a white-grey-white layer structure was formed (see
The measurement parameters were determined based on the results Fig. 3).
of preliminary tests, in which we varied the drop height and impactor

Table 1
Cell structural characteristics of the individual layers (based on our former study [13]).
Nominal density [kg/m3] 30 40 50 70

Average cell size [μm] 622 ± 151 567 ± 163 500 ± 138 464 ± 108
Cell density [cells/cm3] 2815 4194 4839 6947
Average cell wall thickness [μm] 6.7 8.7 9.1 12.3

3
M. Tomin et al. Polymer Testing 122 (2023) 108014

We recorded the deformation with the high-speed camera during

each impact at 2000 frames per second in 480 × 640 resolution, which
means we took photos of the process every 0.5 ms. After recording, we
extracted the process-related frames from the video files and investi­
gated them. The steps of image processing can be seen in Fig. 4.
The first frame is when the impactor just touches the surface of the
foam (where the Force–Deformation curves of the drop weight impact
tests start). The last frame is when the penetration of the impactor is
maximum (see Fig. 5/a). It is the same point where the Force­
–Deformation curve has a maximum.
The contour of the foam changes as a function of time during the
Fig. 3. Photo of the measurement layout with the white-grey-white layer impact and the deformation zone of the foam is not limited to the area
structured sample (front view taken from the position of the high- under the dart. As a first step in defining the shape and the size of the
speed camera). deformed zone, we identified the pixels which belong to the foam. We

Fig. 4. Image processing flowchart.

4
M. Tomin et al. Polymer Testing 122 (2023) 108014

Fig. 5. Visualization of the image processing steps for the 70-50-30 sample: (a) original image, (b) noise removal, (c) edge sharpening and histogram equalization,
(d) visualization of the label matrix, (e) labeling process of the edges.

used several image processing techniques to do that. We converted the It can be seen on the sharpened frames that the contours and the
extracted frames into grayscale images, where the color of the pixels is foam can be easily distinguished from the other part of the images. We
described as a number between 0 and 255, where the dark pixels have used the so-called “Flood fill” segmentation to identify the pixels that
low intensity (Feature extraction step 1). The recorded video and the belong to the foam. Flood fill segmentation requires a seed pixel, and it
extracted frames were noisy, so the next step was noise removal. To will investigate its neighborhood. Suppose the intensity of the neigh­
remove noise, we used motion filters, which blur the original image boring pixel does not differ significantly from that of the seed pixel. In
(Fig. 5/b) The advantage of using the motion filters is that we can that case, it will be identified as foam, otherwise it will be background.
control the direction of blurring on the image so we can more or less The next step will investigate the neighbors connected to the pixels
preserve the contours of the layer edges. identified as foam and so on until all the pixels are examined. We used
The effect of noise removal also worsens the possibility of detecting two seed points per layer on each side of the frame. The result of seg­
the edges on the blurry images, which is why we used other filters on the mentation is a label matrix, whose size is the same as the frames and only
frames to sharpen the edges and histogram equalization to increase contains zeros (not foam) and ones (foam). A separate label matrix be­
contrast (Fig. 5/c). longs to each different foam layers. The visualization of the label

5
M. Tomin et al. Polymer Testing 122 (2023) 108014

matrices can be seen in Fig. 5/d. The area of the layer cross-sections (A1, If the difference was bigger than 5 pixels, that part of the foam was
A2, A3) in pixels is the total number of pixels. deformed (ε>5). After we identified the deformed part of the edge, we
Based on the label matrices, we can extract the edges of the foam extracted the horizontal position for the starting and the end point of the
layers in every time step. We defined three important edges: top layer deformed part. The difference between these two is the diameter of the
edge (top edge of the top layer), middle layer edge (top edge of the deformed part (Equation (2)). As we know the DPI of the frames, we
middle layer), and bottom layer edge (top edge of the bottom layer). The were able to change the unit from pixels to mm. Each side of a pixel
process visualization of the labeling of the edges can be seen in Fig. 5/e. corresponded to a length of 0.2941 mm in our case. However, we must
For example, the position of the top layer edge in the frame is the highlight that this ratio can vary significantly depending on the mea­
indices (vertical and horizontal position) of the first nonzero element in surement setup, such as the distance between the camera and the object,
each column of the label matrix (Fig. 6). We defined the position of the and the resolution of the video recorder.
edges every 0.5 ms for each layer. The changing of the top layer edge as a
d(x, t) = f (x, 0) − f (x, t) (1)
function of time can be seen in Fig. 7 in the case of the “70-50-30” foam.
To compare the different foam structures, we calculated two pa­
D(t) = max (x|d(x, t) > ε) − min (x|d(x, t) > ε) (2)
rameters that can describe deformation behavior. One of the parameters
is the diameter of the deformed zone (Fig. 6) and the other is the where d(x,t) is the difference between the actual and starting vertical
decrease in the cross-sectional area of the foam layers. We calculated the edge positions as a function of time, f(x,0) is the vertical edge position at
diameter of the deformed zone from the vertical position of the edges. the first timestep and f(x,t) is the actual vertical edge position as a
The vertical position f(x,t) of the edges is a two-variable function whose function of time. D(t) is the diameter of the deformed zone as a function
value depends on the horizontal position and the time. For every time of time and ε is a threshold limit, which we defined as 5 pixels.
step, we calculated the difference (d(x,t)) between the actual vertical The decrease in the cross-sectional area of the top layer at any time
position and the starting position of the edges according to Equation (1). step can be calculated as the following (3):
( )
A1t
ΔAt = 1 − *100 (3)
A10

where ΔAt [%] is the decrease in the cross-sectional area, A10 [mm2] is
the area of the cross-section of the top layer at the first time step, A1t
[mm2] is the cross-sectional area at a given time step. The calculation
was the same for the middle and the bottom layers.
These two values can be used to compare the different foam struc­
tures and their deformation as a function of time.

4. Results and discussion

4.1. Drop weight impact tests

Fig. 8 shows the effect of increasing impact energy on the material

response of the “50-50-50′′ sample. All other samples had a similar trend
of force-deformation curves.
In contrast to a typical quasi-static compression test where the whole
specimen is compressed between two platens with a constant strain rate
(see Fig. 1), we performed a dynamic test, in which the diameter of the
impactor was only half of the specimen size. Therefore, the deformation
Fig. 6. The position of the edges at a given time. of the foam in our case is heterogeneous. In addition, the foam’s resis­
tance is negligible compared to the magnitude of the load, so there is no
significant initial elastic region. After an initial pre-loading stage, in the
so-called plateau zone, the force does not remain constant and increases

Fig. 8. Force–deformation curves of the “50-50-50′′ sample for different

Fig. 7. The changing of the top layer edge as a function of time. impact energies.

6
M. Tomin et al. Polymer Testing 122 (2023) 108014

linearly as the high strain rate contributes to the compression of the gas steadily as the drop weight was increased. The differently layered foam
enclosed in the cells. Furthermore, the resistance to the impactor in­ structures absorbed on average 79 ± 1% of the impact energy at the
creases with the indentation depth due to the shear and compression lowest impact energy level. With growing impact energy levels, this
deformation of the foam around the impactor head, which also increases efficiency gradually decreased to 73 ± 1%, 70 ± 1%, and finally 67 ±
the measured force. At larger strains, the nature of deformation changes, 1%. This is due to the fact that the energy absorption efficiency of the
and the slope of the curve increases steadily. At this point, similar to the foam decreases above a particular load since the rate of stress increase
densification zone, the foam resists deformation significantly due to air exceeds the rate of absorbed energy increase in the last part of the
compression and excessive compression of the cells [10,33]. force–displacement curves [9].
For the smallest impact energy (9.862 J), the degree of deformation Impact energy also significantly affected the maximum force gener­
is optimal, as there is no final densification stage, and the nature of the ated during the impact. Regardless of the impact energy, the “70-50-30′′
deformation remains linear throughout the entire measurement. In this and “70-30-70′′ layer orders showed the lowest maximum force. The
case, the samples stopped the downward movement of the dart before “30-50-70′′ specimen showed slightly higher maximum forces than the
the foam structure became excessively compacted. However, as the reference specimen (“50-50-50′′ ), while the “40-70-40′′ structure gave
impact mass—and hence the impact energy—increased, the last part of the worst results (highest force).
the curves (densification) became more significant. It is also visible that The maximum force results correlated well with the extent of me­
the increasing impact energy did not affect the slope of the curves at chanical damage in the samples as a result of impact. The samples that
lower strains since the impact velocity was kept constant (constant drop showed better shock absorption (”70-50-30” and “70-30-70”) suffered
height), and the impact energy was varied by modifying the weight of negligible residual deformation. In contrast, in the case of the samples
the impactor dropped onto the sample. with high maximum force results, the foam suffered not only irreversible
The results for absorbed energy (4), efficiency (5), maximum force, cellular, but external surface damages as well.
and maximum deformation derived from the curves are shown in Fig. 9. The maximum deformation of the tested samples also increased with
increasing impact energy and appeared to approach a threshold limit.
∫ε
This is caused by more excessive cellular compaction due to higher
Eabs = F(ε)dε (4)
impact energy. This assumes that at a given impact energy, the foam
structures are already fully compressed, so further increasing the drop
0

Eabs weight would not increase the maximum deformation but would cause
η= • 100 (5) an increase in the reaction force and produce more irreversible defor­
Eimp
mation in the cell structure.
where Eabs [J] is the absorbed energy, F [N] is the recorded force, ε [m] is Overall, the structures with a higher density foam layer at the side of
the deformation, η [%] is the efficiency, while Eimp [J] is the impact the load showed the best shock absorption capacity, presumably due to
energy. the stiffer and more resistant top layer. The comparison of the force­
Although the absorbed energy increased, the energy absorption ef­ –deformation curves of the samples at the highest impact energy also
ficiency (the ratio of absorbed energy to impact energy) decreased supports this (see Fig. 10.).

Fig. 9. Results of the impact tests for each sample: maximum force (a), absorbed energy (b) and maximum deformation (c).

7
M. Tomin et al. Polymer Testing 122 (2023) 108014

Fig. 10. Force–displacement curves of the samples at the impact energy of

Fig. 11. Total area where cells undergo deformation.
21.630 J.

4.2.1. Homogeneous - density sample (50-50-50)

It can be seen that the second phase of the curves of the “70-50-30′′ For the homogeneous-density “50-50-50” sample (Fig. 12.), the
and “70-30-70′′ samples (70 kg/m3 density top layer), which gave more progression of deformation is uniform, so the delay between the
favorable results, starts at a higher force due to the steeper initial pre- beginning of deformation in the top and middle and in the middle and
loading phase. Hence, they are able to absorb more energy before
densification starts, which results in a smaller increase in the reaction
force. For example, at 70% deformation, the “70-50-30′′ sample absor­
bed 20 J of energy compared to 15 J absorbed by the “30-50-70′′ sample.
We assume that due to the higher stiffness of the 70 kg/m3 top layer, the
impact deformation mechanisms are less concentrated in the area just
below the dart and the impact energy is distributed over a larger zone.
However, the demonstration of this necessitated the taking of high-
speed camera recordings and the development of a new evaluation
method. Since the difference between the shock absorption capacity of
the samples increased with increasing load, it can be concluded that the
positive effect of the layer order modification appears mainly in the
densification zone. For this reason, we set the measurement settings to
achieve a high degree of deformation for each sample (above 80%
strain) for high-speed camera recording.

4.2. Investigating deformation with a high-speed camera

By changing the variation of the layer order, the layers start to

deform at different times and in a different order, which fundamentally
affects the behavior of the foam. With image processing, we determined
the outer points of the deforming zone for the top, middle and bottom
edges. Plotting these points at the time of maximum deformation and
connecting them gives a good approximation of the total volume
involved in the deformation for the first two layers (see Fig. 11.). We
assume that all cells in the area between the boundary lines are
deformed, while no deformation has occurred outside the lines.
In addition, we determined the temporal variation of the decrease of
the cross-sectional area and the diameter of the deforming zone, which
properties have not been studied in the literature yet. In the following,
the deformation mechanisms of each foam type will be discussed in
detail. Fig. 12. Cross-sectional area decrease (a) and the diameter of the deforming
zone (b) for the “50-50-50” structure.

8
M. Tomin et al. Polymer Testing 122 (2023) 108014

bottom layers are identical. The magnitude of the cross-sectional area concentrated on the top layer and then shifted to the middle layer once
decrease differed noticeably among the top, bottom, and middle layers, the top layer is fully compacted. Afterward, the cells in the middle layer
although these differences were relatively small compared to the other also undergo full compaction, which leads to the load being concen­
samples with a non-uniform density distribution. Additionally, the trated on the densest bottom layer. Since the densification of the low-
shape of the curves was more similar in this sample. Their variation over density top layer takes less time than the denser middle layer, the
time is initially linear, then the slope gradually decreases, due to the delay between the beginning of deformation in the top and middle layers
compaction of the cell structure, and it finally reaches a threshold. The is smaller than the delay between the middle and bottom layers. The
saturation of the curves occurs simultaneously. The diameter of the diameter of the deformation zone also decreases from top to bottom, but
deformation zone gradually decreases downwards from the top surface the magnitude of this decrease is larger than in the uniform-density “50-
of the foam, with the number of cells involved in the deformation 50-50” sample.
decreasing from layer to layer. Since the deformation zone is narrower in
the bottom layer, the degree of compression in the cell structure is 4.2.3. 70-50-30 layered structure
locally higher, which increases the reaction force recorded during drop The sample “70-50-30” (Fig. 14.), which has the highest density on
weight testing. the impact side and then decreasing density downwards, showed a
deformation response opposite to the “30-50-70”. The stiffer top layer
4.2.2. 30-50-70 layered structure immediately transfers the load to the underlying lower density middle
The progression of deformation in the “30-50-70” layered system layer, which is less resistant to mechanical loads, then this layer trans­
(Fig. 13.), which showed worse results in terms of shock absorption mits it to the bottom least dense layer. Accordingly, all the layers start to
capacity, was different. In the first stage of the collision, the cross- be compressed at almost the same time without significant delay once
sectional decrease is clearly the most significant in the top layer. Its the deformation process begins. For this reason, the degree of cross-
variation over time was non-linear; the slope decreased steadily before sectional area reduction in the layers correlates with their density. The
reaching a threshold cross-sectional area reduction limit of around 27%. lowest-density bottom layer is compressed the most and the highest
The underlying mechanism for this phenomenon is that the top layer’s density top layer the least. A significant difference is that the diameter of
area under the impactor undergoes full compaction prior to significant the deformation zone is larger than in the “50-50-50” structure, and the
deformation of the lower layers. This results in the load being initially width of the zone does not decrease from top to bottom. It results in far
more cells being involved in energy absorption, so the degree of defor

Fig. 13. Cross-sectional area decrease (a) and diameter of the deforming zone
(b) for the “30-50-70” structure. Fig. 14. Cross-sectional area decrease (a) and diameter of the deforming zone
(b) for the “70-50-30” structure.

9
M. Tomin et al. Polymer Testing 122 (2023) 108014

mation in each cell is smaller. Due to the typical nature of the

compressive stress–strain curves of polymeric foams, this lower level of
deformation results in a smaller maximum reaction force.

4.2.4. 70-30-70 layered structure

In the “70-30-70” sample (see Fig. 15.), the top high-density layer
transmits the load to the less dense middle layer, causing both layers to
undergo compression simultaneously. The deformation of the bottom,
also dense layer, becomes more noticeable only after a relatively long
time delay. Due to its lower density, the rate of the cross-sectional area
decrease is clearly the greatest in the middle layer during the initial
phase of the collision. Still, the upper layer also gradually loses its cross-
sectional area. The variation in the cross-sectional area reduction over
time in these layers is also similar to a saturation curve, with the low-
density middle layer reaching a threshold value first (at about 8 ms),
followed by the complete compaction of the top layer (at about 9 ms).
The diameter of the deformation zone is the largest in the top layer, but
the width of the zone decreases strongly from the top downwards. As the
top and middle layers are deformed over a wide zone during the initial
phase of the impact, this results in a higher stress plateau, similar to the
“70-50-30” sample, which helps to absorb the impact energy at a rela­
tively low maximum reaction force.

4.2.5. 40-70-40 layered structure

In the “40-70-40” sample (Fig. 16.), the striker impacted on a low-
density layer with a denser layer located underneath. Thus, the top

Fig. 16. Cross-sectional area decrease (a) and diameter of the deforming zone
(b) for the “40-70-40” structure.

layer distributes most of the load to the middle layer only after the cell
structure is compacted, with a time delay of about 1.5 ms. As the middle
layer is compressed, it pushes down the bottom layer, resulting in
simultaneous but varying rates of deformation in these two layers.
Initially, the compression of the bottom layer is far more dominant, but
once it reaches a threshold value, deformation progresses through the
cellular compaction of the middle layer. The diameter of the deforma­
tion zone decreases from top to bottom at approximately the same rate
as in the uniform-density “50-50-50” sample, which explains the similar
magnitude of the maximum force results obtained in the regular drop
weight tests.

4.2.6. Summary of deformation characteristics

In summary, we can conclude that the variation of the layer order
significantly affected the deformation response of the foam structures
(see Table 4.). The use of layers with decreasing density from top to
bottom contributes to the simultaneous deformation of the layers below
each other. In contrast, if a low density layer is followed by a denser one,
most of the deformation is concentrated in the top layer, and the
deformation of the lower denser layers only becomes dominant after the
densification of the top one. A high-density layer on the side receiving
the impact also increases the diameter of the deformation zone. The time
order of layer compaction depends mainly on foam density. As low-
density layers have a less resistant structure (larger cells, thinner cell
Fig. 15. Cross-sectional area decrease (a) and diameter of the deforming zone
walls) to loads, full cellular compaction takes less time compared to
(b) for the “70-30-70” structure.
higher density layers with smaller cells.

10
M. Tomin et al. Polymer Testing 122 (2023) 108014

Table 4
Summary of the deformation characteristics of foam structures with different density variation.
Specimen Progression of Starting time of layer deformationa for top, Deforming zone diameter at the time of maximum Time order of layer
type deformation middle, and bottom layers [ms] deflection for top, middle and bottom layers [mm] compaction

50–50–50 Stepwise (uniform time 0 92.4 Simultaneous

steps) 1.5 86.8
3.0 77.4

70–50–30 All layers start to deform at 0 105.7 1. Bottom

the same time 0 112.0 2. Middle
0.5 106.5 3. Top

30–50–70 Stepwise (increasing time 0 89.9 1. Top

steps) 2.0 82.1 2. Middle
5.0 69.3 3. Bottom

70–30–70 Stepwise (increasing time 0 123.6 1. Middle

steps) 0.5 100.3 2. Top
4.5 67.4 3. Bottom

40–70–40 Stepwise (decreasing time 0 94.3 1. Top

steps) 1.5 91.6 2. Bottom
2.0 81.0 3. Middle
a
(Starting time of layer deformation is calculated based on the temporal variation of the diameter of the deforming zone).

5. Conclusions collapses irreversibly. The accuracy of the current measurement method

can be further improved by using a higher-resolution high-speed cam­
This study investigated functionally graded, three-layer, 30 mm era. Additionally, the use of pressure measurement films positioned
thick weakly cross-linked polyethylene foam structures with the same between the foam and support plate would allow us to also investigate
average density (50 kg/m3) but different densities in each layer. Based the energy transferred to the plate. With the implementation of these
on the results of drop weight impact tests performed with a 50 mm improvements, further studies can contribute to a more comprehensive
diameter cylindrical impactor in the impact energy range of ~10–22 J, understanding of the deformation response of functionally graded foam
we showed that increasing the impact energy from 9.862 J to 21.630 J structures.
decreases the energy absorption efficiency of the foam structures from
79 ± 1% to 67 ± 1%. We also showed that the shock absorption of the CRediT author statement
foams could be improved by modifying the density distribution along
the thickness. By applying a decreasing density variation from top to Márton Tomin: Conceptualization, Methodology, Software, Formal
bottom, the maximum reaction force of the impact tests decreased by analysis, Investigation, Validation, Writing - Original Draft, Supervision.
7.09% compared to the foam structure with uniform density. Dániel Török: Methodology, Software, Formal analysis, Investigation,
In order to analyze the deformation mechanism of each layer, we Data Curation, Writing - Review & Editing. Tamás Pászthy: Methodol­
developed a new evaluation method based on an image processing al­ ogy, Software, Data Curation, Visualization, Writing - Review & Editing.
gorithm, which can be used to quantify the size of the deformed area and Ákos Kmetty: Writing - Review & Editing, Supervision, Project admin­
its variation over time, and the number of cells involved in the defor­ istration, Funding acquisition.
mation, by analyzing high-speed camera recordings in a MATLAB
environment. Using the evaluation method, we demonstrated the posi­ Declaration of competing interest
tive effect of modifying the layer order on shock absorption capability.
Decreasing density from top to bottom causes the top layer to transfer The authors declare that they have no known competing financial
the load to the layer below, which transmits it to the bottom layer so that interests or personal relationships that could have appeared to influence
the cells in all three layers start to compress at the same time. In contrast, the work reported in this paper.
in the case of the uniform-density foam structure, there was a 1.5 ms
time delay between the beginning of successive layer deformation. The Data availability
high-density top layer is also more resistant to the impact and distributes
the load over a 27.1% larger area, so cellular deformation takes place in The authors do not have permission to share data.
a larger volume. The wider impact zone and the simultaneous defor­
mation of the layers significantly increase the energy absorption ca­ Acknowledgements
pacity in the initial stage of deformation, as the ratio of volume fraction
deforming under the densification strain is higher. As more cells absorb This research was supported by the Hungarian National Research,
the load, the structure becomes more resistant, and the plateau of the Development and Innovation Office (K 132462); and by United World
force–deformation curve is shifted upwards by approximately 10%. This Wrestling. The research reported in this paper is part of project no. BME-
will increase the amount of energy absorbed by the end of the plateau NVA-02, implemented with the support provided by the Ministry of
zone, and the dart will have a lower velocity at the start of the densifi­ Innovation and Technology of Hungary from the National Research,
cation zone. The downward movement of the dart stops sooner, thus Development and Innovation Fund, financed under the TKP2021 fund­
reducing the degree of critical cell compaction, which would cause a ing scheme. Á. Kmetty is thankful for the support of János Bolyai
significant increase in the reaction force. Research Scholarship of the Hungarian Academy of Sciences. The au­
As the test parameters in our study were limited to a narrow range of thors are thankful for the foam samples to Polifoam Ltd. (Budapest,
impact energies and the drop height was kept constant at 400 mm, Hungary).
further improvements may include extending the test parameters to
higher impact energies and velocities where the whole cellular structure

11
M. Tomin et al. Polymer Testing 122 (2023) 108014

References [19] J. Yu, L. Song, F. Chen, P. Fan, L. Sun, M. Zhong, J. Yang, Preparation of polymer
foams with a gradient of cell size: further exploring the nucleation effect of porous
inorganic materials in polymer foaming, Mater. Today Commun. 9 (2016) 1–6,
[1] N. Mills, Polymer Foams Handbook: Engineering and Biomechanics Applications
https://doi.org/10.1016/j.mtcomm.2016.08.006.
and Design Guide, Elsevier Science, Oxford, 2007.
[20] E. Cusson, A. Akbarzadeh, D. Therriault, D. Rodrigue, Density graded polyethylene
[2] C. Ge, Theory and practice of cushion curve: a supplementary discussion, Packag.
foams: effect of processing conditions on mechanical properties, Cell, Polym 38
Technol. Sci. 32 (2019) 185–197, https://doi.org/10.1002/pts.2427.
(2019) 3–14, https://doi.org/10.1177/0262489319839632.
[3] S. Liu, J. Duvigneau, G.J. Vancso, Nanocellular polymer foams as promising high
[21] C. Tovar-Cisneros, R. González-Núñez, D. Rodrigue, Effect of mold temperature on
performance thermal insulation materials, Eur. Polym. J. 65 (2015) 33–45, https://
morphology and mechanical properties of injection molded HDPE structural foams,
doi.org/10.1016/j.eurpolymj.2015.01.039.
J. Cell. Plast. 44 (2008) 223–237, https://doi.org/10.1177/0021955x07088044.
[4] S.A. Baghban, M. Khorasani, G.M.M. Sadeghi, Acoustic damping flexible
[22] A.N.J. Spörrer, V. Altstädt, Controlling morphology of injection molded structural
polyurethane foams: effect of isocyanate index and water content on the
foams by mold design and processing parameters, J. Cell. Plast. 43 (2007)
soundproofing, J. Appl. Polym. Sci. 136 (2019), 47363, https://doi.org/10.1002/
313–330, https://doi.org/10.1177/0021955x07079043.
app.47363.
[23] J. Gómez-Monterde, M. Schulte, S. Ilijevic, J. Hain, D. Arencón, M. Sánchez-Soto,
[5] J.C. Viana, Polymeric materials for impact and energy dissipation, Plast., Rubber
M.L. Maspoch, Morphology and mechanical characterization of ABS foamed by
Compos. 35 (2006) 260–267, https://doi.org/10.1179/174328906X146522.
microcellular injection molding, Procedia Eng. 132 (2015) 15–22, https://doi.org/
[6] Y. Chen, R. Das, A review on manufacture of polymeric foam cores for sandwich
10.1016/j.proeng.2015.12.462.
structures of complex shape in automotive applications, J. Sandw. Struct. Mater. 24
[24] K.Z. Uddin, G. Youssef, M. Trkov, H. Seyyedhosseinzadeh, B. Koohbor, Gradient
(2022) 789–819, https://doi.org/10.1177/10996362211030564.
optimization of multi-layered density-graded foam laminates for footwear material
[7] Z.X. Zhang, Y.M. Wang, Y. Zhao, X. Zhang, A.D. Phule, A new TPE-based foam
design, J. Biomech. 109 (2020), 109950, https://doi.org/10.1016/j.
material from EPDM/PPB blends, as a potential buffer energy-absorbing material,
jbiomech.2020.109950.
Express Polym. Lett. 15 (2021) 89–103, https://doi.org/10.3144/
[25] M. Tomin, A. Kossa, S. Berezvai, Á. Kmetty, Investigating the impact behavior of
expresspolymlett.2021.10.
wrestling mats via finite element simulation and falling weight impact tests, Polym.
[8] M. Tomin, Á. Kmetty, Polymer foams as advanced energy absorbing materials for
Test. 108 (2022), 107521, https://doi.org/10.1016/j.
sports applications—a review, J. Appl. Polym. Sci. 139 (2022), 51714, https://doi.
polymertesting.2022.107521.
org/10.1002/app.51714.
[26] L. Cui, S. Kiernan, M.D. Gilchrist, Designing the energy absorption capacity of
[9] M. Avalle, G. Belingardi, R. Montanini, Characterization of polymeric structural
functionally graded foam materials, Mater. Sci. Eng., A 507 (2009) 215–225,
foams under compressive impact loading by means of energy-absorption diagram,
https://doi.org/10.1016/j.msea.2008.12.011.
Int. J. Impact Eng. 25 (2001) 455–472, https://doi.org/10.1016/S0734-743X(00)
[27] S.T. Lee, Foam Extrusion: Principles and Practice, CRC Press, London, 2000.
00060-9.
[28] E. Shim, Bonding requirements in coating and laminating of textiles, in: I. Jones, G.
[10] L. Di Landro, G. Sala, D. Olivieri, Deformation mechanisms and energy absorption
K. Stylios (Eds.), Joining Textiles, Woodhead Publishing, Cambridge, 2013,
of polystyrene foams for protective helmets, Polym. Test. 21 (2002) 217–228,
pp. 309–351.
https://doi.org/10.1016/S0142-9418(01)00073-3.
[29] Y. Shimazaki, S. Nozu, T. Inoue, Shock-absorption properties of functionally
[11] P.C.C. Pinto, V.R. da Silva, M.I. Yoshida, M.A.L. Oliveira, Synthesis of flexible
graded EVA laminates for footwear design, Polym. Test. 54 (2016) 98–103,
polyurethane foams by the partial substitution of polyol by steatite, Polímeros 28
https://doi.org/10.1016/j.polymertesting.2016.04.024.
(2018) 323–331, https://doi.org/10.1590/0104-1428.10417.
[30] M.C. Saha, H. Mahfuz, U.K. Chakravarty, M. Uddin, M.E. Kabir, S. Jeelani, Effect of
[12] J.V. Mane, S. Chandra, S. Sharma, H. Ali, V.M. Chavan, B.S. Manjunath, R.J. Patel,
density, microstructure, and strain rate on compression behavior of polymeric
Mechanical property evaluation of polyurethane foam under quasi-static and
foams, Mater. Sci. Eng., A 406 (2005) 328–336, https://doi.org/10.1016/j.
dynamic strain rates- an experimental study, Procedia Eng. 173 (2017) 726–731,
msea.2005.07.006.
https://doi.org/10.1016/j.proeng.2016.12.160.
[31] G. Lyn, N.J. Mills, Design of foam crash mats for head impact protection, Sports
[13] M. Tomin, Á. Kmetty, Evaluating the cell structure-impact damping relation of
Eng. 4 (2001) 153–163, https://doi.org/10.1046/j.1460-2687.2001.00081.x.
cross-linked polyethylene foams by falling weight impact tests, J. Appl. Polym. Sci.
[32] A. Kmetty, M. Tomin, T. Barany, T. Czigany, Static and dynamic mechanical
138 (2021), 49999, https://doi.org/10.1002/app.49999.
characterization of cross-linked polyethylene foams: the effect of density, Express
[14] B. Koohbor, A. Kidane, Design optimization of continuously and discretely graded
Polym. Lett. 14 (2020) 503–509, https://doi.org/10.3144/
foam materials for efficient energy absorption, Mater. Des. 102 (2016) 151–161,
expresspolymlett.2020.40.
https://doi.org/10.1016/j.matdes.2016.04.031.
[33] S. Ouellet, D. Cronin, M. Worswick, Compressive response of polymeric foams
[15] N. Gupta, E. Woldesenbet, Microballoon wall thickness effects on properties of
under quasi-static, medium and high strain rate conditions, Polym. Test. 25 (2006)
syntactic foams, J. Cell. Plast. 40 (2004) 461–480, https://doi.org/10.1177/
731–743, https://doi.org/10.1016/j.polymertesting.2006.05.005.
0021955x04048421.
[34] B. Song, W. Chen, X. Jiang, Split Hopkinson pressure bar experiments on polymeric
[16] N. Gupta, A functionally graded syntactic foam material for high energy absorption
foams, Int. J. Veh. Des. 37 (2005) 185–198, https://doi.org/10.1504/
under compression, Mater. Lett. 61 (2007) 979–982, https://doi.org/10.1016/j.
ijvd.2005.006656.
matlet.2006.06.033.
[35] B.F. Berencsi, A. Kossa, Analyzing the effect of temperature on squash ball impacts
[17] M. Higuchi, T. Adachi, Y. Yokochi, K. Fujimoto, Controlling of distribution of
using high-speed camera recordings, Period. Polytech. - Mech. Eng. 65 (2021)
mechanical properties in functionally-graded syntactic foams for impact energy
354–362, https://doi.org/10.3311/PPme.18381.
absorption, Mater. Sci. Forum 706–709 (2012) 729–734. https://doi.org/10.40
[36] B. Koohbor, S. Ravindran, A. Kidane, In situ deformation characterization of
28/www.scientific.net/MSF.706-709.729.
density-graded foams in quasi-static and impact loading conditions, Int. J. Impact
[18] A. Mohyeddin, A. Fereidoon, A semi-empirical model for density gradient in
Eng. 150 (2021), 103820, https://doi.org/10.1016/j.ijimpeng.2021.103820.
microcellular thermoplastic foams, J. Cell. Plast. 47 (2011) 413–428, https://doi.
org/10.1177/0021955X11406003.

12