Numerical and experimental investigations on sandwich panels made with eco-friendly components under low-velocity impact

Abstract

A low-velocity impact characterisation of a sustainable sandwich panel based on upcycled bottle caps as circular honeycomb is conducted. The recycled core aims to develop an alternative route of reusing waste bottle caps disposed in landfills. Ecological alternatives to skin (recycled PET foil) and adhesive (bio-polyurethane) are also compared with classic components (aluminium skin and epoxy polymer). A low-cost reinforcement (cement particles) is also proposed to enhance the mechanical strength of the panel. The samples are tested at several levels of impact energy, according to the type of skin, to observe their effect on mechanical behaviour. Metal skins achieve higher impact loads and energy absorption compared to PET foil. The bio-adhesive leads to a similar or enhanced maximum impact load and energy absorption compared to the epoxy adhesive. Specific properties highlight the promising performance of the bio-based adhesive with aluminium skins, reaching increments of up to 378%. The cement increases the maximum load and reduces the duration of the impact event, leading to lower energy absorption. The unreinforced epoxy polymer shows a visible adhesive peeling off from aluminium skin, while particle inclusions lead to reduced overall delamination. Biopolymer exhibits marginal adhesive debonding and stable deformation, revealing a progressive failure. In general, PET samples show core shear failure due to rupture of the skin. Crack propagation in PET samples made with biopolymer adhesive is reduced at lower energy levels. The results evidence the promising application of bottle caps in a more sustainable honeycomb core to build eco-friendly structures.

Keywords

Sandwich panel recyclability bio-polymer low-velocity impact design of experiment FEA

Introduction

Sandwich panels are structures based on the bonding of two thin skins to a lightweight and thick core. The skins provide bending rigidity, while thick core is designed to increase shear rigidity [1]. Sandwich structures are widely applied due to their high strength- and stiffness-to-weight ratio, besides exhibiting high energy absorption under compression. However, they present limited performance under common impact loads, such as tool drops, runway debris and ballistic impact [2]. These events reduce the residual properties of panels and induce invisible defects [3].

Sustainability and high mechanical performance are two some of the most demanding – and also contradicting – targets in the design of sandwich structures. Fuel consumption limits and governmental regulations demand the development of greener solutions for infrastructure and mobility [4]. Alternative materials for skins and core of sandwich panels are considered a feasible route to obtain improved eco-friendly products with adequate mechanical properties and reduced environmental impacts. Some solutions are the use of natural fibre composites as core [5] and skins [6], as well as wood core agglomerates [7,8]. The incorporation of disposed materials in composites is another option to obtain sustainable engineering components. Composites based on upcycled materials present satisfactory mechanical performance under quasi-static and impact loads [9]. These structures can enhance specific properties, improve the panel’s end-of-life treatment, as well as ensuring high energy recovery and reduced manufacturing costs.

A discarded material with significant ecological impact and potential for future upcycling is the bottle cap, which achieved high mechanical strength as a honeycomb core under quasi-static bending [10 –12]. The use of bottle caps as a core aims to offer a feasible route for their upcycling, since they are disposed of in landfills in a rate of 320,000 tonnes/year [13] and are a source of pollution in marine ecosystems [14]. Their use into structures subject to impact loads is promising, since thermoplastic cores exhibit higher energy absorption and lower deformation than wooden cores in these conditions [15].

The components used in sandwich panels significantly affect their response under dynamic loads. Damghani and Gonabadi [16] reported that facings contribute to increasing the energy absorption ratio of honeycombs by 34%, which was also found by Shitta-Bey et al. [17]. Zhang, Fei and Zhang [18] found that the outer skin absorbs more than 40% of the total energy dissipated during the impact test. Furthermore, higher impactor velocities affect panel damage more significantly than mass increase [16], exhibiting deeper indentation in a limited area [18]. Despite its minor participation, the modifications in the core (e.g. material and geometry) significantly affect the impact response [19]. Mocian et al. [20] determined that a rigid polyurethane core with aluminium skins reduced overall panel deformation and increased peak loads during drop-tower impact tests compared to a soft polystyrene foam. However, serious damage was induced by the brittle core. Core geometry also affects impact performance. Auxetic lattices [21] and double layer honeycomb [22] have achieved an improved impact performance, the latter providing greater energy absorption and controllable crushing. Different honeycomb cell geometries, such as circular cell honeycombs, have also promising impact responses. Hu et al. [23,24] compared cubic and hexagonal packing systems (Figure 1). More energy is dissipated during deformation of the hexagonally packed core, due to the greater number of constraints and the reduced folding size of the cell walls [23]. However, imperfections of the cell, such as wall inclinations, reduced its efficiency [11].

The impact resistance is also increased by reducing core shear stiffness, thus increasing panel toughness [25,26]. One approach is the inclusion of soft particles, e.g. rubber particles, into brittle polymers [27,28]. Few studies have investigated particle inclusions in adhesives [29 –32]. Ramakrishnan et al. [32] tested a sandwich panel based on polymethacrylimide (PMI) foam core and aramid composite skins bonded by epoxy reinforced with block copolymers under low-velocity impact. Higher peak load, reduced upper skin deformation and higher stress propagation are achieved with particle inclusions. Low cost fillers have been added to bio-polyurethane (bio-PU) adhesive, achieving high mechanical strength and stiffness. An increment in impact resistance (i.e. the energy required to break a sample) is also found out due to chemical reactions between bio-PU and particles [31].

The use of upcycled plastic waste, recycled skins, and bio-based adhesive in sandwich panels can reduce significantly the ecological footprint of structures, while reaching satisfactory impact properties. A design for bottle cap panels based on alternated caps in a cubic packing has been developed in previous studies with satisfactory quasi-static properties (Figure 2) [12]. However, a limited amount of work has examined the performance of eco-friendly sandwich panels made of upcycled parts under out-of-plane impact loads [33]. Therefore, this study aims to characterise experimental and numerically the upcycled caps core under low-velocity impact. A full factorial design compares sustainable alternatives for skins (recycled PET foil) and adhesive (bio-sourced polyurethane) to classic components (aluminium skin and epoxy polymer), as well the effect of particles incorporation (cement particles) on the impact responses of sandwich panels.

Materials and methods

Materials

Bottle caps from a single brand (Coca-Cola™, ∼ Ø30.3 mm) are used as sustainable core cells. After collection, the caps are carefully washed and dried during 24 hours before the sandwich panel is manufactured. The aluminium skin type AW-5754 is tested in two thickness levels (1.0 and 1.5 mm). Sustainable skins consist of 1.3 mm thick PET foils from recycled bottles supplied by Armacell Benelux SCS (Belgium). The epoxy resin Renlan M and hardener HY956 supplied by Huntsman (Brazil) are mixed in a 5:1 weight ratio, respectively. A bio-sourced polyurethane (AGT 1315), produced by Imperveg (Brazil), is based on a prepolymer (component A) and a polyol (component B) mixed in a 1:1.2 weight proportion. Cement particles (type ASTM III-Holcim/Lafarge Brazil) are used as received as adhesive filler. The properties of skins, bottle cap, and adhesives are shown in Tables 1 and 2 [31].

Table 1.

Average properties and standard deviations (SDs) of the skin and core components.

Property	Aluminium AW-5754		PET foil		HDPE bottle cap
Property	Mean Value	SD	Mean Value	SD	Mean Value	SD
Young’s modulus [GPa]	70.6	3.5	0.5	0.03	1.01	0.08
Tensile strength [MPa]	246.6	3.2	6.9	0.2	16.7	1.5
Density [kg/m³]	2700	–	300	–	970	–

Table 2.

Average and standard deviations (in parentheses) of the main properties of tested adhesives.

Type of polymer	Type of particle	Amount of particle (%wt.)	Young’s modulus (GPa)	Tensile strength (MPa)	Impact resistance (kJ/m²)	Single lap strength (MPa)	Apparent density (g/cm³)
Epoxy	–	–	1.9 (0.2)	30.8 (2.5)	15.8 (1.0)	3.9 (0.3)	1.2 (0.0)
Epoxy	Cement	3	2.6 (0.2)	38.0 (3.2)	12.3 (0.9)	3.7 (0.4)	1.2 (0.0)
Biopolymer	–	–	0.4 (0.0)	11.1 (0.3)	8.5 (1.0)	3.8 (0.4)	1.1 (0.0)
Biopolymer	Cement	3	1.3 (0.1)	18.0 (0.8)	18.5 (1.3)	4.7 (0.4)	0.9 (0.0)

Design of experiment

The experimental setup consisted of two steps. The first independent experiment verifies the effect of the outer skin thickness on the impact response of the panel (preliminary conditions C1^p and C2^p). Two thickness levels of aluminium skins are tested: 1.0 and 1.5 mm. The data are statistically analysed by the Minitab v18 software using the techniques of Analysis of Variance (ANOVA) and Design of Experiment (DoE) [34,35]. A full factorial design type 2³ (Table 3) is performed as second step to investigate sustainable components. This technique consists of investigating factors (the parameters of interest) and the combination of their respective levels. Three factors are investigated: type of adhesive, type of skin, and particle reinforcements. Two types of adhesive (epoxy polymer and biopolymer) and two types of panel skin (aluminium sheets and recycled PET foil) are tested to provide a sustainable alternative to the proposed classic design in previous researches [10,11], as well as particle inclusions (with and without cement). The1.0 mm thick aluminium skin and cement particles are considered based on previous works [12,31]. Additional factorial designs are carried out to analyse the effect of impact energy on the energy absorption capacity of the samples. Panels made of aluminium skins are tested at impact energy levels of 100 J (as recommended by ASTM D7136 [36]) and 50 J, while PET-based samples are tested at four energy impact levels (50/25/12.5/6.25J), until reduced sample damage is achieved. Table 4 presents the additional full factorial designs for the assessment of the energy level. Three (3) samples are produced for each condition considering two (2) replicates, resulting in six (6) samples per condition.

Table 3.

Project planning matrix, full factorial design 2³, compared at 50 J energy level.

Condition	Type of adhesive	Particle reinforcement	Type of skin
C1	Epoxy	Without particles	Aluminium
C2	Epoxy	Without particles	PET foil
C3	Epoxy	Cement	Aluminium
C4	Epoxy	Cement	PET foil
C5	Biopolymer	Without particles	Aluminium
C6	Biopolymer	Without particles	PET foil
C7	Biopolymer	Cement	Aluminium
C8	Biopolymer	Cement	PET foil

Table 4.

Project planning matrix, full factorial design 2³ for aluminium skins and 2²4¹ for PET skins.

Condition	Type of skin	Type of adhesive	Particle reinforcement	Energy level
C1’ – C8’	Aluminium	Epoxy	Without particles	100/50 J
	Aluminium	Epoxy	Cement
	Aluminium	Biopolymer	Without particles
	Aluminium	Biopolymer	Cement
C1* – C12*	PET foil	Epoxy	Without particles	50/25/12.5/6.25J
	PET foil	Epoxy	Cement
	PET foil	Biopolymer	Without particles
	PET foil	Biopolymer	Cement

The full factorial design has some constant factors, such as the thickness of the skins (1.0 mm for aluminium skin and 1.3 mm for PET foil), the thickness of the adhesive layer (∼1.5 mm adhesive layer [11]), the type of bottle cap (Coca-Cola™ bottle caps), the cap orientation and packing system (alternated caps packed in cubic configuration [10,11]), polymer mixing ratio (1:1.2 for biopolymer and 5:1 for epoxy polymer), the amount of particles (3 wt% [31]), the mixing time (2 minutes) and the curing conditions (28 days at room temperature − ∼24°C and ∼55% relative humidity).

Panel manufacturing and testing

The manufacturing process for impact samples is similar to the previous work [12]. Both skins are cut to the dimensions required for the test (150 × 150 mm², adapted from ASTM D7136 and D7766 [36,37]). The metallic skins are cleaned with liquid degreaser and sanded under running water in ±45° direction (Figure 3(a)) to remove the oxide layer, improving the surface wettability to the polymeric adhesive [38]. The skins receive a final cleaning with acetone to remove all debris from the sandpaper. PET foil skins receive basic cleaning with light abrasion. After cleaning, the skins are covered laterally with a plastic film and inserted in the moulds fitted with release clothing (Figure 3(b)). The film protects the mould against adhesive leakage. The adhesive is then prepared and spread on the surface. The plastic caps are placed on the surface in cubic packing and alternated directions (Figure 3(c)), until the mould is filled. Different coloured caps with similar composition and geometry (diameter and height) are obtained from varied Coca-Cola™ bottle types. The mould is closed with a lid under constant pressure (∼3 kPa) for 24 hours, when the partial sample is removed (Figure 3(d)). The process for bonding the second skin is similar to the first step. The total curing time at room temperature (∼24 °C, RH 55%) is 28 days.

The impact test is carried out on a drop tower machine in Fraunhofer EMI (Freiburg), as shown in Figure 4. The equipment consists of an impactor of ∼10 kg with a 50 mm diameter semi-spherical tip coupled to a fixed vertical rail. The test parameters (energy level and impactor size) are adapted from ASTM D7136 [36] and D7766 [37]. All samples are tested at different energy levels, according to the experiments described in Tables 3 and 4, with an unsupported square area of 125 × 125 mm². The test is recorded using a high-speed camera (FASTCAM SA-X type 324 K-M2 – time resolution of 0.05 ms, Figure 4) to measure the velocity and displacement of the impactor during the impact event through image correlation software. In addition, a laser measurement system is applied using a mirror under the sample to detect the displacement of the lower skin. The measurement is based on the movements of laser patterns projected on the lower skin during deformation and captured by a second high speed camera (FASTCAM-APX RS model 250 K – time resolution of 0.3 ms). Fractured samples are inspected to analyse the extent of damage to the sample surface and the crushing effect on the impact area. Damage extension and indentation depth are measured following the guidelines of ASTM 7136 [36].

The maximum impact load and energy absorption responses are measured from the force vs. displacement curve obtained in the drop-tower tests. The equivalent density of the samples is measured following the procedures of ASTM C20 [39]. The specific properties are determined based on the ratio between absolute properties and density. Finally, the efficiency of the structures at different impact velocities is determined by the ratio between the energy absorption and the nominal impact energy during the drop-tower test. This property is the only response analysed by the additional full factorial designs, shown in Table 4, to compare the effect of different energy levels.

Finite element (FE) model

Finite element (FE) models are developed for comparison with experimental results. The model is designed in Hypermesh and simulated using the LS-Dyna software. The representative 3D model is shown in Figure 5. The skin and core are defined as elastoplastic materials, type 024 in LS-Dyna, considering the approximate properties of each component, as shown in Tables 1 and 2 [31]. The skin is modelled as a Belytschko-Tsay shell element with three (3) points of integration and a 7.5 mm element size. The bottle cap and adhesive core are modelled as solid elements (∼4.5 mm long) of constant stress. The mesh size is determined based on a convergence analysis to ensure efficient calculations and accuracy. The number of elements in the sample is 3100 (800 shell elements on the skins and 2300 solid elements on the core). The impactor and support are modelled as rigid 3D solid materials and relevant boundary conditions are applied (Figure 5). All displacement and rotation are constrained on the square support, while the Ø 50 mm impactor can only move in the local z-direction. An initial velocity corresponding to the initial energy level is applied to the impactor of a mass equal to the real impactor. The adhesion between the core and the skin is ensured using a tied contact linked with an offset between the nodes of upper/lower core surface and the skin shells. The failure of the skin/core contact is based on the failure of the elements, obtained by their mechanical properties. The remaining contacts (skin to impactor or support) are modelled as automatic contact between surfaces with reduced friction (µ = 0.1).

Results

Impact test results

Evaluation of skin thickness effect

The results of the first independent experiment with samples submitted to 100 J impact energy are shown in Table 5. P-values less than 0.05, which are underlined in Table 6, evidence that all properties are affected by the “Skin thickness” factor, considering 95% confidence interval. R² (adj) values vary from 91.6 to 96.6%, indicating high predictability of the models.

Table 5.

Average and standard deviations (in parenthesis) of ANOVA test for aluminium thickness.

	Absolute properties			Specific properties
	P_max [N]	E_abs [J]	ρ [kg/m³]	P_max [N.m³/kg]	E_abs [10⁻³N.m⁴/kg]
C1^p_(1.5 mm)	12,915.0 (129.2)	98.3 (0.98)	819.3 (1.70)	15.8 (0.16)	120.0 (1.20)
C2^p_(1.0 mm)	10,607.9 (212.2)	99.1 (0.50)	721.9 (2.26)	14.7 (0.07)	137.3 (2.75)

Table 6.

ANOVA results for absolute and specific properties.

Factor	Absolute properties			Weight-specific properties
Factor	P_max	E_abs	E_ratio	P_max	E_abs
Skinthickness (P-value ≤ 0.05)	0.011	0.025	0.025	0.026	0.028
R² (adj) (%)	96.6	92.5	92.5	92.4	91.6

The thicker aluminium skin (1.5 mm) contributes to increase the maximum impact load and its specific property by 21.7% and 7.5%, respectively (Table 5). Despite having reduced impact loads, panels with 1.0 mm aluminium skins lead to greater energy absorption (increments of 0.8%), which is still considered significant. A 14.4% increase in specific energy absorption is achieved by panels made with thinner skins (Table 5). Figure 6 shows the load vs time curves for both conditions. The longest deformation is observed in the panels made with 1.0 mm aluminium skin. The peaks in the plot are associated with the shear crack between the adjacent bottle caps. These results show a major efficiency obtained by 1.0 mm panels, especially when lightweight requirements are demanded. For this reason, this configuration is considered in the remaining steps.

Full factorial design – 2³

The results of the impact test (absolute and specific properties) and the equivalent density for the full factorial design (Table 3) are shown in Table 7 for the 50 J energy impact level. The force vs. time plots for all conditions are shown in Figure 7 for the highest energy level tested (100 J for aluminium based samples and 50 J for panels made of PET skin). In general, higher loads are achieved for epoxy-based panels when cement particles are added to the adhesive (Figure 7(a)). However, the duration of the impact event is significantly reduced, especially for samples with PET skin (Figure 7(b)). The curves show several peaks for panels based on aluminium skin, which correspond to the shear failure of the bonding between the adjacent caps, leading to core-skin debonding after impactor rebound. PET skin-based samples, on the other hand, exhibit rupture of the skin with subsequent sample fracture and complete penetration of the impactor, leading to a reduced impact load. Different failure modes limit the comparison between the two types of skin. Biopolymer samples exhibit the opposite behaviour of samples made with epoxy polymer, in which smoother curves are obtained due to reduced core failure and marginal debonding (Figure 7(c)). The maximum load is slightly reduced by particle inclusions, while the duration of the impact event is similar for reinforced and unreinforced biopolymer conditions. Higher loads are observed for samples based on biopolymers compared to epoxy polymer for panels made of aluminium skins, while similar impact resistance is achieved for samples based on PET skins due to the complete penetration of impactor. The duration of the impact event, however, is longer for the biopolymer adhesive with PET skin (Figure 7(d)). In general, a more flexible adhesive, such as the biopolymer adhesive, leads to greater relative displacements between the constituents and an enhanced toughness of the panels.

Table 7.

Average results and standard deviations (in parenthesis) of equivalent panel density and impact results for full factorial design (2³) at 50 J energy level.

Setup	Absolute properties				Specific properties
Setup	ρ [kg/m³]	P_max [N]	E_abs [J]	E_ratio [%]	P_max[N.m³/kg]	E_abs[10⁻³N.m⁴/kg]
C1: E, N, A	721.9 (2.3)	8,373.4 (167.5)	44.8 (0.9)	89.7 (1.8)	10.6 (0.2)	62.1 (1.2)
C2: E, N, P	398.2 (0.6)	1,131.3 (22.6)	11.3 (0.2)	22.6 (0.5)	2.8 (0.1)	28.3 (0.6)
C3: E, C, A	710.8 (3.5)	8,825.0 (88.3)	43.8 (0.4)	87.6 (0.9)	12.4 (0.1)	61.6 (0.6)
C4: E, C, P	390.8 (1.8)	1,218.8 (12.2)	8.1 (0.1)	16.2 (0.2)	3.1 (0.0)	20.7 (0.2)
C5: B, N, A	658.1 (4.5)	8,643.8 (43.2)	42.7 (0.2)	85.4 (0.4)	13.1 (0.1)	64.9 (0.3)
C6: B, N, P	342.9 (5.5)	1,012.5 (5.1)	11.7 (0.1)	23.3 (0.1)	3.0 (0.0)	34.0 (0.2)
C7: B, C, A	651.1 (3.4)	8,587.5 (257.6)	43.6 (1.3)	87.1 (2.6)	13.2 (0.4)	66.9 (2.0)
C8: B, C, P	343.3 (4.8)	875.0 (26.3)	10.4 (0.3)	20.7 (0.6)	2.5 (0.1)	30.2 (0.9)

E: epoxy; B: biopolymer; N: no particle; C: cement; A: aluminium; P: PET foil.

Table 8 shows the results of ANOVA for absolute and specific responses. P-Values less than or equal to 0.05 are shown in bold, indicating that the main factor or interaction is significantly affected by the response. An interaction indicates that the effect of one factor depends on the level of another factor. The high predictability of the models is highlighted by the R²(adj) ranging from 99.49 to 99.83%. The DoE/ANOVA results are validated by the Anderson-darling test, which indicates that the data follow a normal distribution when the P-Value is greater than or equal to 0.05. The P-Values of the highest-order interactions are underlined in Table 8. Most properties exhibit significant second-order interactions as highest-order interactions, while the specific maximum load is affected by a third-order interaction and the equivalent density is affected independently by the main factors. The underlined interactions are analysed in the main effect or interaction plots (Figures 8 to 12). Energy absorption ratio (E_ratio) will be presented in ‘Analysis of energy level effect’ section to compare with the additional experiments.

Table 8.

DoE results for full factorial design (2³).

Factors		Absolute properties				Weight-specific properties
Factors		ρ	P_max	E_abs	E_ratio	P_max-specific	E_specific
P-Value ≤0.05	Type of adhesive (TA)	0.000	0.222	0.881	0.881	0.000	0.000
	Particle reinforcement (PR)	0.040	0.319	0.026	0.026	0.006	0.006
	Type of skin (TS)	0.000	0.000	0.000	0.000	0.000	0.000
	TA*PR	0.280	0.050	0.056	0.056	0.001	0.046
	TA*TS	0.078	0.166	0.018	0.018	0.000	0.030
	TS*PR	0.307	0.207	0.035	0.035	0.003	0.001
	TATSPR	0.730	0.409	0.987	0.987	0.050	0.624
R² (adj) (%)		99.91	99.83	99.76	99.76	99.76	99.49
AD (P-Value ≥ 0.05)		0.315	0.123	0.505	0.505	0.377	0.989

Equivalent density is affected only by the main factors, as shown in Figure 8. The use of biopolymer contributes to reducing the panel density by 10.1%, which is attributed to its lower apparent density compared to the epoxy polymer, in addition to its discrete expansion during curing process (increase in average thickness from 14.6 mm for epoxy-based panels to 15.1 mm for biopolymer-based panels). A limited density reduction of about 1% is achieved when cement particles are added. The main contribution towards panel density is obtained from the “Type of Skin” factor, which induces an increase in density by almost 86% when aluminium skins are used in comparison to PET foil skin.

The maximum impact load is significantly affected by the main factor “Type of skin” and the interaction between “Type of adhesive” and “Particle reinforcement”, as shown in Figure 9. The aluminium skin leads to an 8-fold increase in impact load compared to the PET foil (Figure 9(a)) due to low strength of PET skin and its brittle failure, with the presence of impactor penetration and the propagation of cracks in the sample. The remaining factors affect the impact load up to 6.1%, especially when cement particles are added to the epoxy adhesive (Figure 9(b)). Fisher's test, for comparison of means by grouping, is carried out with ANOVA/DoE results. This test identifies the means in the interaction plots that are statistically distinct to each other, within a 95% confidence interval, attributing different letters to means that are not statistically equivalent [34]. The means with the letter B, shown in Figure 9(b), reveal that biopolymer and epoxy adhesives have similar impact load when they are not reinforced with particles. The cement particles do not provide a significant change when combined with biopolymers (Group B), while an increase of 5.7% (Group A) is observed for the reinforced epoxy polymer. The incorporation of cement particles in both polymers increases their stiffness and strength, as shown in Table 2, affecting the maximum load achieved in the impact tests. Microparticles are also responsible for delaying the propagation of cracks, with a consequent increase in strength [32], increasing the maximum load of the panels, as shown in Figure 9(b).

Figure 10 shows the interaction effect plot for the specific maximum load response. The presence of cement affects only the specific response when the epoxy polymer is used (Figure 10(i)). An increase of 19.8% in specific values is obtained with the incorporation of cement. The use of aluminium skins is highly effective when combined with biopolymer, as shown in Figure 10(ii) (Group A), reaching an increase of 378.5% compared to PET skins (Group C). This level also shows a 23% higher load than panels made of epoxy polymer and aluminium skins (Group B). Panels consisted of PET skins are not affected by cement particles (Figure 10(iii), Group C), while those made with aluminium skins and cement inclusions (Group A) achieve specific loads 352% higher compared to PET-based samples (Group C). Unreinforced samples with aluminium skins lead to a 9.4% reduction in specific loads compared to reinforced samples (Group B).

The energy absorption response has two significant second-order interactions between the factors “Type of Adhesive” and “Type of Skin” (Figure 11(i)) and “Type of Skin” and “Particle Reinforcement” (Figure 11(ii)). In general, aluminium panels offer up to a 374% increase in energy absorption compared to PET-skin panels (Figure 11). This behaviour is attributed to the greater toughness and strength of aluminium skins in relation to PET foil, which obtained less resistance to crack propagation. The epoxy polymer adhesive exhibits significant increase through the use of aluminium skin compared to the biopolymer (Figure 11(i)); while the inclusion of cement reduced the response of PET-based panels, according to Fisher’s test (Figure 11(ii)). Although the PET foil is made of thermoplastic material, its cellular structure formed during the recycling process makes it very fragile under bending loads, allowing the propagation of cracks in the structure, as shown in ‘Impact failure analysis’ section.

Specific energy absorption has three significant second-order interactions, shown in Figure 12. It is noteworthy that particle inclusions do not affect the performance of biopolymer panels (Figure 12(i), Group A). Biopolymer-based panels, especially without particles, achieve enhanced specific energy absorption (Figure 12(i), Group A) compared to epoxy-based panels, revealing an increase of 8.5% with the added particles (Group B) and 20.1% (Group C) without particles. Aluminium skins enhance the specific energy absorption of the panels by up to 152.7%, with a significant contribution from the type of polymer used (Figure 12(ii) and (iii)), as detected by Fisher’s test. Unlike absolute energy absorption, panels with aluminium skins bonded by biopolymer present the highest average results (Figure 12(ii), Group A). The increment in relation to the PET foil panel (Group C) is 105.3%, while aluminium panels bonded with epoxy polymer show a marginal reduction of 6.1% (Group B). Similar behaviour is found for PET skin-based samples, in which the epoxy polymer reduces specific energy absorption by 23.6% (Group D). Likewise, for the absolute response, the inclusion of particles does not significantly affect the specific energy absorption of the aluminium-based samples, sharing the letter A (Figure 12(iii)). However, the specific energy absorption shows a great decrease for panels with PET foil and cement inclusions (Group C), with a reduction of 12.0% in relation to unreinforced panels (Figure 12(iii), Group B) due to reduction of panel toughness with particle reinforcement.

Analysis of energy level effect

The energy absorption ratio for the total impact energy is shown in Table 9, considering the two additional factorial designs described in Table 4. The first experiment (2³) analyses the efficiency of aluminium-based panels under impact loads of 100 and 50 J, while the second experiment (2¹4¹) analyses the PET-based panels under four impact energy loads of 50, 25, 12.5 and 6.25 J. Table 10 shows the results of ANOVA (DoE) for the two experiments. In both cases, the energy impact level (EL) affects the absorbed energy ratio response (P-Value ≤ 0.05). Aluminium-based panels are also affected by a second-order interaction effect (TA*PR). PET-based panels are significantly affected by a third-order interaction (TA*TS*EL). All relevant P-Values are underlined in Table 8. The predictability of the models is satisfactory, with R² (adj) values of 86.14 and 99.80%. P-values greater than 0.05 presented by the Anderson-Darling (AD) test validate ANOVA. The interaction and main effect plots for the energy absorption ratio for each skin (Table 4) are shown in Figures 13 and 14.

Table 9.

Average results and standard deviations (in parenthesis) of energy absorption ratio for aluminium skin (2³ DoE) and for PET skin (2²4¹ DoE) full factorial designs.

	E_ratio [%]
2³ DoE	C1’		C2’		C3’		C4’		C5’		C6’		C7’		C8’
2³ DoE	99.1 (0.5)		89.7 (1.3)		96.9 (1.9)		87.6 (0.9)		92.7 (1.9)		85.4 (1.7)		96.1 (1.0)		87.1 (0.9)
2²4¹DoE	C1*	C2*	C3*	C4*	C5*	C6*	C7*	C8*	C9*	C10*	C11*	C12*	C13*	C14*	C15*	C16*
2²4¹DoE	22.6 (0.5)	19.9 (0.2)	46.4 (0.9)	79.0 (0.8)	16.2 (0.5)	25.6 (0.8)	38.8 (0.4)	64.0 (0.3)	23.3 (0.9)	36.9 (1.1)	98.6 (1.0)	97.8 (1.0)	20.7 (0.2)	38.1 (0.8)	76.7 (1.2)	81.5 (2.0)

Table 10.

DoE results for full factorial design (2³ DoE) for aluminium skin and for (2²4¹ DoE) for PET skin.

Factors		Absolute properties
Factors		Aluminium (100/50J)	PET (50/25/12.5/6.25J)
P-Value ≤0.05	Type of adhesive (TA)	0.014	0.000
	Particle reinforcement (PR)	0.857	0.000
	Energy level (EL)	0.000	0.000
	TA*PR	0.040	0.000
	TA*EL	0.544	0.000
	PR*EL	0.725	0.000
	TATSEL	0.645	0.000
R² (adj) (%)		86.14	99.80
AD (P-Value ≥ 0.05)		0.119	0.180

The energy absorption ratio depends directly on the energy absorption response. Therefore, the behaviour of the energy absorption ratio at 50 J corresponds to that seen in Figure 11, in which the unreinforced epoxy adhesive and the aluminium skins lead to an increase in the energy absorption capacity of the samples compared to biopolymer and PET skin, with a great contribution of the skin type. The failure mode of PET skin-based samples with complete perforation of the core and impactor penetration hinders the absorption of energy during impact. The type of polymer also exhibits a significant influence when considering different energy levels and skin types.

The influence of different energy levels on aluminium-based samples is shown in Figure 13. The use of the epoxy adhesive increases the energy absorption ratio of aluminium-based samples by 6% compared to the biopolymer adhesive in pristine condition (without particles), as shown in Figure 13(a) (Groups A and C). The incorporation of cement particles is beneficial only for panels based on biopolymers, which show similar behaviour to reinforced epoxy-based panels (Groups AB and B). Figure 13(b) shows that aluminium-based panels lead to a 10% higher absorption ratio when subjected to the highest impact energy level (100 J) above the level of 50 J. The amount of energy not absorbed by the panels is converted into kinetic energy, rebounding the impactor after the event. The core and skin are preserved during the test, with no visible rupture, as described in ‘Impact failure analysis’ section 3.2.

In contrast, the opposite behaviour is observed for PET-based panels, as shown in Figure 14. In general, biopolymer adhesives provide an approximately 50% increase in the energy absorption ratio compared to the epoxy polymer due to its ductile behaviour. In addition, the inclusion of cement reduces the energy absorption ratio by about 13.9% (Figure 14(i)) for both polymers due to the increase in adhesive stiffness by cement particles (Table 2), which reduces, in general, the energy absorption capacity [20]. The impact energy levels (6.25–50 J) substantially affect the energy absorption ratio, which shows percentage variations of up to 307% (Figure 14(ii)). PET-based panels lead to higher energy absorption ratios, especially when subjected to lower impact levels (6.25 J and 12.5 J), which induces less crack propagation in the sample and allows longer contact between impactor and the sample, enhancing energy absorption by ductile deformation. In opposite, at higher impact levels, the efficiency of the biopolymer panel is largely reduced due to the rapid propagation of cracks. This behaviour demonstrates that PET-based panels are not appropriated for high impact load applications. Cement-reinforced adhesives degrade the overall impact capacity by up to 21.5%, especially for 6.25 and 12.5 J (Figure 14(iii)). The increase in the energy level reduces the energy absorption efficiency by 74 and 74.6% for unreinforced and reinforced panels, respectively. A more rigid core and a low strength skin induce a complete perforation of the samples based on PET skin and limited absorption of energy at higher energy levels. The impactor rebound is not observed for PET-based panels.

Figure 15 shows typical plots of force vs. time for different energy levels for both skin types. PET-based panels reveal an increase in the event time as a function of the energy level due to the reduced impact velocity. In addition, the maximum load increases when lower energy levels are considered, attributed to the longer contact between the sample and the impactor before the sample breaks (Figure 15(a)), which also allows energy dissipation. Both aspects lead to increased energy absorption by PET foil samples, which agrees with the results discussed in Figure 14. In contrast, changing the energy level from 50 to 100 J leads to an increase in the maximum load supported by aluminium-based panels, followed by a slight reduction in the duration of the impact event, indicating its enhanced energy absorption capacity with higher impact loads (Figure 15(b)).

Impact failure analysis

The main features of the fractured samples are shown in Figures 16 and 17. PET-based panels exhibit extensive damage when subjected to higher impact loads and reduced rupture at the lower levels, as shown in Figures 16 and 17(a) and (c). Biopolymer-based samples with PET skin tested at a lower energy level (6.25 J) do not show significant rupture (Figure 17(a) and (c)). The use of biopolymer leads to a reduction in visual damage and less rupture of the skin and core, which may justify the greater amount of energy absorption observed for this condition. The biopolymer adhesive provides enhanced toughness and therefore greater energy absorption, also for aluminium based samples. Epoxy-based samples, on the other hand, exhibit substantial damage to the PET skin, leading to a failure of the core bonding between adjacent caps, without visible crushing of plastic bottle caps under dynamic loads. Cracks propagate through the skin and core along the edges of the caps to the borderlines of the sample (Figure 16(a) and (c)). The incorporation of particles does not visually affect failure modes.

Samples based on aluminium skins exhibit different failure modes, depending on the type of adhesive. The pristine epoxy polymer provides extended skin delamination at 100 J, with a substantial reduction at 50 J. Major cracks within the core are observed between adjacent cells in the central area of the sample (Figure 16(b)). The central cap shows the major deformation during impact, being partially crushed, with crack propagation along the adhesive layer to the adjacent cells in epoxy-based samples (Figure 16(b) and (d)). The inclusion of particles leads to significantly less delamination in the epoxy-based panels, limiting the debonding from the edge of the samples and crack propagation (Figure 16(d)). The onset of delamination is the loss of adhesion around the open side of the caps, spreading to the other caps located at the edges of the sample. Biopolymer-based panels achieve significantly reduced delamination compared to those based on epoxy adhesive for both reinforced and unreinforced formulations (Figure 17(b) and (d)). The central cap shows moderate deformation and no visible adhesive debonding, especially for the reinforced biopolymer, in a cohesive debonding failure.

The incorporation of particles reduces the overall damage of the structures based on the biopolymer adhesive. The indentation depth and the apparent damage diameter of the upper skin (ASTM 7136 [36]) are obtained only for aluminium-based panels, due to their specific type of damage, as shown in Table 11. The Fisher test is used to compare and verify the effect of the polymer type and particle reinforcement on the extent of the damage. In addition, the deformation of the lower skin at the maximum impact load is determined by the projection of the laser system. Deformations are represented in the contour map plots shown in Figure 18. Larger areas of deformation and reduced indentation for unreinforced panels made with biopolymer are verified, while the epoxy adhesive presents the largest indentation among the investigated conditions, besides inducing localised damage, mainly when associated with particles (Figure 18(b)). The changes are considered significant according to the Fisher test (Table 11). The dissimilar deformation of the upper skin and lower skin in Table 11 indicates a reduction in the core thickness by crushing. The enhanced crushing capacity of the reinforced biopolymer samples is observed, achieving an average reduction in core thickness by 41% compared to the core crushing of 37% for epoxy reinforced cores. In contrast, the opposite behaviour is found for pristine adhesives. In general, an increment in stiffness of particle reinforced polymers promotes a reduction in impact resistance and, consequently, in the crushing capacity, as shown for the epoxy polymers (Table 2 [31]). However, this effect is not evident for biopolymers (Table 2 [31]), which can be attributed to chemical interaction with cement particles, affecting the crushing effect of biopolymers. The results indicate greater toughness provided by the biopolymer adhesive, with the evident recovery of thickness and less penetration. Additionally, the extended damage diameter at the upper skin indicates that the energy is more effectively dissipated in the core, reducing the lower skin deformation, as seen in the contour maps of Figure 18, especially for biopolymer-based panels without particles. This important feature allows the use of the bottle cap panel in applications subject to low or moderate energy impact since this material can preserve the survival space in collisions by the propagation of energy and crushing of the core.

Table 11.

Apparent damage parameters result and Fisher test.

Conditions	Indentation depth						Damage diameter [mm]	Fisher
Conditions	Upper skin [mm]	Fisher	Lower Skin [mm]	Fisher	Core crushing ratio [%]	Fisher	Damage diameter [mm]	Fisher
C1 (Aluminium + pristine epoxy)	18.3	A	12.5	A	40.1%	A	27.6	A
C2 (Aluminium + cement reinforced epoxy)	15.6	B	10.2	BC	37.2%	B	28.1	B
C5 (Aluminium + pristine biopolymer)	14.9	B	9.4	C	36.6%	B	31.0	C
C6 (Aluminium + cement reinforced biopolymer)	17.1	C	11.0	B	40.3%	A	30.2	C
R² (adj)	97.91%		95.06%		86.92%		99.38%

FE results

The results of the simulation models are shown in Figures 19 and 20 for the samples based on aluminium and PET skins, respectively. The material properties used to obtain the responses are shown in Table 12, which are based on the properties of the individual components shown in Tables 1 and 2 [31], especially the properties of the polymeric adhesive on the core, since the failure of the core occurred in the adhesive bond to the skin and the core. The results indicate a good correlation between experimental and numerical behaviours. The only exception is the stiffness of samples based on biopolymers, which is most affected by the bottle cap modulus, shown in Table 1 [10], since the pristine biopolymer has less stiffness compared to the estimated modulus of HDPE caps, resulting in an intermediate elastic modulus for the core. Aluminium-based samples exhibit an increase in panel load with similar linear behaviour, and especially for epoxy-based samples, load drops occur at a point associated with the failure of the adhesive element. This failure mode is not observed in biopolymer-based samples. The decrease in the load reaction after the full penetration of the impactor is faster in the simulation models, indicating that the real samples exhibit longer recovery periods, which are not completely reproduced in the model. The finite element models of panels with PET skins exhibit behaviour similar to the experiment with less impact event due to rupture of the PET skin and consequent failure of the core. The loads after the maximum peak tend to be slightly higher in the models than in the experiments (Figure 20(b) and (c)). This effect can be attributed to the longer contact perforation experienced between the impactor and the sample in the model due to the geometry adopted (Ø50 mm spherical impactor and mass equivalent to the impactor construction), while the real impactor has limited displacement due to the extension of the impactor support. The behaviour observed in experiments is also predicted in the failure mode of the resulting models. The failure of the models is shown in Figure 21. The sandwich panels with aluminium skins exhibit localised central damage and failure of the core elements associated with the debonding of adhesive for epoxy-based panels (Figure 21(a)). The fragile failure of PET skin is also observed in the simulation due to the failure of the skin elements in a pattern similar to that as observed in Figures 16 and 17. The results indicate good predictability of the FE models based on the convergence of the analyses.

Table 12.

Modelling parameters for the panel's components.

Components		MAT-024 main parameters
Components		Density [kg/m³]	Young’s modulus [GPa]	Poisson’s ratio	Yield stress [MPa]	Tangent Modulus [GPa]	Failure strain
Aluminium skin		2720	70.5	0.3	143	2	0.16
PET skin		300	0.45	0.4	8.5	0	0.02
Bottle caps core	Unreinforced epoxy	1200	1.9	0.4	20	0.35	0.06
	Epoxy + cement	1200	2.5	0.4	20	0.1	0.04
	Unreinforced biopolymer	1100	0.9	0.4	9.8	0.2	0.12
	Biopolymer + cement	1100	1.2	0.4	15	0.1	0.14

Conclusions

The low-velocity impact response of environmentally friendly sandwich panels made from a recycled honeycomb core based on discarded bottle caps and recycled PET or aluminium skins is studied experimentally and numerically. The use of bio-sourced polyurethane is also compared with classic epoxy polymer as adhesive. The main conclusions are described below:

The use of thicker aluminium skins increases weight-specific panel maximum load. However, greater specific energy absorption and a more efficient structure is achieved by lightweight construction based on thinner metallic skins (1.0 mm).

The aluminium skin contributes to significantly increase the maximum impact load compared to recycled PET skin, achieving higher efficiency. The use of particle inclusions in the adhesive connecting the core and the skin leads to a marginal increment of the panel strength. Higher toughness of biopolymer contributes towards enhanced panel strength and reduced delamination, especially when combined with aluminium skins and cement particles.

The energy absorption capability of sandwich panels made from aluminium skins is similar for both adhesives and reinforcement conditions. PET skin based panels show lower potential for energy absorption. Here, higher efficiency is found for unreinforced biopolymer due to enhanced toughness and damping capacity.

Metallic skins present enhanced energy absorption ratio under higher energy levels due to higher ductility. The fragility of the PET skins leads to better performance only under lower impact loads, especially with core with higher toughness (with biopolymer). The inclusion of particles decreases the absorption ratio in most of the tested energy levels due to the brittle adhesive in the core.

The failure mode is characterised by rupture of the skin followed by the propagation of cracks along the cells bonded to the PET skins, which are not adequate under impact loads. Cement reinforcement reduced the apparent skin debonding in aluminium-based samples, especially with biopolymer, reducing lower skin deformation and enhancing core crushing.

FE models investigated in LS-Dyna show a good correlation with the experimental results in terms of failure mode and force-time-response. The simulation approach can therefore be chosen to conduct further parameter studies of the eco-friendly sandwich panels.

The conclusions indicate the applicability of using recycled plastic disposals in the construction of structural solutions. The characterisation under low-velocity impact test indicates the promising use of sustainable core and bio-based adhesive in conjunction with classic aluminium skin. A variety of fields, such as lightweight constructions, automotive parts, advertising walls, etc., can benefit from lightweight and low-cost sustainable materials. The low strength of the proposed sustainable skin, however, requires further research into sustainable high strength alternatives, such as laminates based on renewable fibres, to be used as sandwich panel skins.

Footnotes

Acknowledgements

The authors thank the support of Dr. Max Gulde (Fraunhofer EMI) in the setup of the laser system and acknowledge the donations from Armacell Benelux SCS (Belgium) and Imperveg Ltda. (Brazil).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this work was supported by CNPq-Brazil (grant numbers GDE 290224/2017–9; PQ 309885/2019–1).

ORCID iD

Tulio Panzera

References

Moongkhamklang

Deshpande

Wadley

The compressive and shear response of titanium matrix composite lattice structures. Acta Materialia 2010; 58: 2822–2835.

Wang

Waas

Wang

Experimental and numerical study on the low-velocity impact behavior of foam-core sandwich panels. Compos Struct 2013; 96: 298–311.

Bernard

Lagace

PA.

Impact resistance of composite sandwich plates. J Reinf Plast Compos 1989; 8: 432–445.

May

Hesebeck

Assessment of experimental methods for calibrating rate-dependent cohesive zone models for predicting failure in adhesively bonded metallic structures. Eng Fail Anal 2015; 56: 441–453.

Riccio

Raimondo

Saputo

, et al. A numerical study on the impact behaviour of natural fibres made honeycomb cores. Compos Struct 2018; 202: 909–916.

Susainathan

Eyma

Luycker

, et al. Experimental investigation of impact behavior of wood-based sandwich structures. Compos Part A: Appl Sci Manuf 2018; 109: 10–19.

Walsh

Kim

H-I

Suhr

Low velocity impact resistance and energy absorption of environmentally friendly expanded Cork core-carbon fiber sandwich composites. Compos Part A: Appl Sci Manuf 2017; 101: 290–296.

Sarasini

Tirillò

Lampani

, et al. Dynamic response of green sandwich structures. Proc Eng 2016; 167: 237–244.

Filho

SLMR

Oliveira

Panzera

, et al. Impact of hybrid composites based on rubber tyres particles and sugarcane bagasse fibres. Compos Part B: Eng 2019; 159: 157–164.

10.

Oliveira

Bonaccorsi

AMS

Panzera

, et al. Sustainable sandwich composite structures made from aluminium sheets and disposed bottle caps. Thin-Walled Struct 2017; 120: 38–45.

11.

Oliveira

Panzera

Freire

, et al. Sustainable sandwich structures made from bottle caps core and aluminium skins: a statistical approach. Thin-Walled Struct 2018; 130: 362–371.

12.

Oliveira

May

Panzera

, et al. Improved sustainable sandwich panels based on bottle caps core. Compos Part B: Eng 2020; 199: 108165.

13.

Recycling

NJ.

Bottle cap recycling (plastic and metal) in New Jersey, www.recyclingnj.com/recycle/caps.html (2016, accessed 7 June 2019).

14.

Boonstra

Van Hest

The findings of the first survey into plastic bottle cap pollution on beaches in the Netherlands. The North Sea Foundation - Netherlands, www.noordzee.nl/. project/userfiles/SDN_Doppenrapport_EN_2017_DEF_small.pdf (2017, accessed 7 June 2019).

15.

Wang

Ramakrishnan

Shankar

Experimental study of the medium velocity impact response of sandwich panels with different cores. Mater Des 2016; 99: 68–82.

16.

Damghani

Gonabadi

AM.

and Numerical Study of energy absorption in aluminum foam sandwich panel structures using drop hammer test. J Sandw Struct Mater 21: 3–18.

17.

Shitta-Bey

Carruthers

Soutis

, et al. The localized low-velocity impact response of aluminium honeycombs and sandwich panels for occupant head protection: experimental characterization and analytical modelling. Int J Crashworthiness 2007; 12: 549–558.

18.

Zhang

Fei

Zhang

Drop-weight impact behavior of honeycomb sandwich panels under a spherical impactor. Compos Struct 2017; 168: 633–645.

19.

Ozdemir

Karakuzu

Al-Shamary

AKJ.

Core-thickness effect on the impact response of sandwich composites with poly(vinyl chloride) and poly(ethylene terephthalate) foam cores. J Compos Mater 2015; 49: 1315–1329.

20.

Mocian

Constantinescu

Sandu

, et al. Impact response of polyurethane and polystyrene sandwich panels. Proc Struct Integrity 2017; 5: 653–658.

21.

Beharic

Egui

Yang

Drop-weight impact characteristics of additively manufactured sandwich structures with different cellular designs. Mater Des 2018; 145: 122–134.

22.

Palomba

Epasto

Crupi

, et al. Single and double-layer honeycomb sandwich panels under impact loading. Int J Impact Eng 2018; 121: 77–90.

23.

, et al. Dynamic crushing of the circular-celled honeycombs under out-of-plane impact. Int J Impact Eng 2015; 75: 150–161.

24.

Cai

, et al. Influence of internal pressure on the out-of-plane dynamic behavior of circular-celled honeycombs. Int J Impact Eng 2017; 104: 64–74.

25.

Nilsson

Wave propagation in and sound transmission through sandwich plates. J Sound Vibr 1990; 138: 73–94.

26.

Sargianis

Suhr

Core material effect on wave number and vibrational damping characteristics in carbon fiber sandwich composites. Compos Sci Technol 2012; 72: 1493–1499.

27.

Bagheri

Marouf

Pearson

RA.

Rubber-toughened epoxies: a critical review. Polym Rev 2009; 49: 201–225.

28.

Garg

Mai

Y-W.

Failure mechanisms in toughened epoxy resins – a review. Compos Sci Technol 1988; 31: 179–223.

29.

Malik

Kaur

Mechanical and thermal properties of castor Oil-Based polyurethane adhesive: effect of TiO2 filler. Adv Polym Technol 2018; 37: 24–30.

30.

Panchireddy

Grignard

Thomassin

J-M

, et al. Bio-based poly(hydroxyurethane) glues for metal substrates. Polym Chem 2018; 9: 2650–2659.

31.

Oliveira

May

Panzera

, et al. Reinforced biobased adhesive for eco-friendly sandwich panels. Int J Adhes Adhes 2020; 98: 102550.

32.

Ramakrishnan

Guérard

Viot

, et al. Effect of block copolymer nano-reinforcements on the low velocity impact response of sandwich structures. Compos Struct 2014; 110: 174–182.

33.

Oliveira PR,

Kilchert S,

May

, et al. Eco-friendly sandwich panel based on bottle caps core and sustainable components: Static and dynamic characterisation. Composites Part C: Open Access 2020; 3: 100069DOI: 10.1016/j.jcomc.2020.100069.

34.

Montgomery

DC.

Design and analysis of experiments. Hoboken, NJ: Wiley, 2020.

35.

Minitab 18 Statistical Software. State College, PA, 2017.

36.

ASTM D7136 2015. Standard test method for measuring the damage resistance of a fiber-reinforced polymer matrix composite to a drop-weight impact event.

37.

ASTM D77662016. Standard practice for damage resistance testing of sandwich constructions.

38.

Borsellino

Bella

Ruisi

Adhesive joining of aluminium AA6082: the effects of resin and surface treatment. Int J Adhes Adhes 2009; 29: 36–44.

39.

ASTM C20 2015. Standard test methods for apparent porosity, water absorption, apparent specific gravity, and bulk density of burned refractory brick and shapes by boiling water.

Numerical and experimental investigations on sandwich panels made with eco-friendly components under low-velocity impact

Abstract

Keywords

Introduction

Materials and methods

Materials

Design of experiment

Panel manufacturing and testing

Finite element (FE) model

Results

Impact test results

Evaluation of skin thickness effect

Full factorial design – 23

Analysis of energy level effect

Impact failure analysis

FE results

Conclusions

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

ORCID iD

References

Full factorial design – 2³