Integrated Analytical and Finite Element-Based Modelling, Manufacturing, and Characterisation of Vacuum-Infused Thermoplastic Composite Laminates Cured at Room Temperature

Abstract

Due to their improved recyclability, thermoplastic composites (TPCs) are increasing their application across industries. The current work deals with the dimensioning, manufacturing, and characterisation of vacuum-infused TPCs cured at RT and made of non-crimp glass fabric and the liquid acrylic-based resin Elium©. Laminates with 10 and 12 layers achieved a fibre weight content of 73% measured by the burn-off process, which corresponds to a fibre volume content of 55%. Three-point bending tests revealed a bending strength of 636.17 ± 25.70 MPa and a bending modulus of 24,600 ± 400 MPa for the 12 layer laminate. Using micro-mechanical models, unidirectional elastic constants are calculated and applied in classical laminate theory (CLT) for optimising composite lay-ups by maximising bending stiffness, whilst yielding a laminate thickness prediction error of −0.18% and a bending modulus prediction error of −1.99%. Additionally, FEA simulations predicted the bending modulus with a −4.47% error and illustrated, with the aid of the Tsai–Hill criterion, the relationship between the onset of layer failure and discrepancies between experimental results and simulations. This investigation demonstrates the effective application of analytical and numerical methods in the dimensioning and performance prediction of TPCs.

1. Introduction

The demand for recycling approaches for composite materials is growing significantly, driven by environmental concerns from a societal and regulatory perspective, as well as the need to reduce waste in industries such as aerospace, automotive, and construction [1,2]. Composite materials, known for their high strength/stiffness-to-weight ratios, are often used in applications requiring lightweight, durability, and performance [3,4]. However, recycling these materials, especially thermoset composites, presents significant challenges [5,6]. Despite presenting high mechanical performance, thermoset composites (TSCs) cannot be remelted or reshaped due to their cross-linked molecular structure, which makes traditional recycling methods less effective. As a result, innovative recycling techniques are being explored to address these limitations and support a more sustainable lifecycle for composite materials in the context of a circular economy [7,8].

Thermoplastic composites (TPCs), on the other hand, present several advantages in terms of recyclability, such as meltability, reshapeability, and reusability [8,9]. This makes them ideal for applications involving repair [10,11]. However, some of the limitations hindering a widespread application of TPCs are the lower Tg, the high viscosity of resins, which complicates the impregnation of fibres [12,13], and the lower strength/stiffness compared to traditional TSCs such as epoxy. In this regard, in recent years, recyclable thermoplastic resins were introduced, which addressed the aforementioned issues.

Among them, Arkema’s Elium© is an acrylic-based liquid thermoplastic resin, which can be welded [14], adhesively bonded [15], mechanically joined [15], and recycled [16,17]. Regarding mechanical performance, studies have revealed that, compared to epoxy, in some conditions, composites made with Elium© have similar to higher properties related to impact [18,19], as well as tensile strength/modulus, flexural strength/modulus, and interlaminar shear strength [20,21]. Moreover, Elium©-based composites are compatible with several fibre systems such as carbon [22], glass [21], and even natural fibre (e.g., flax) [17,20].

In the context of processing, Elium© can be manufactured using methods such as vacuum infusion [19,23], pultrusion [24], and RTM [25]. On this front, vacuum infusion (VI) is advantageous in composite manufacturing for its ability to produce high-quality and low-void parts [21,26]. The process uses vacuum pressure to draw resin into fibre reinforcement, reducing the need for high-pressure devices and other energy-intensive equipment [27,28]. By compaction and air evacuation, vacuum infusion also enhances structural integrity, creating lighter, stronger parts with excellent fibre-to-resin ratios. Additionally, it reduces resin waste and VOC exposure, making it a cleaner, safer, and more environmentally friendly alternative to traditional open moulding techniques [29,30].

In terms of composite part design, the ability to predict the properties of laminate composites, both geometrical and mechanical, is essential for dimensioning structures with tailored performance. By accurately modelling laminate geometry, such as layer thickness, fibre orientation, and stacking sequence, it is possible to forecast mechanical properties like stiffness as well as failure behaviour. Analytical approaches such as classical laminate theory (CLT), micromechanical modelling [31,32], and advanced computational methods, e.g., finite element analysis (FEA) [33], allow for precise simulations of how laminates will behave under various loads and conditions. This predictive capability reduces the need for extensive physical testing, enabling more efficient material selection and design optimisation [34].

Regarding this scenario, the current work aims to investigate how analytical models can be employed along with FEA for the dimensioning and prediction of composite laminates made of glass fibre and Elium©. For this purpose, vacuum-infused laminates were manufactured and characterised in terms of fibre weight content, geometry, bending strength, and bending modulus. This investigation addressed the following questions:

▪

Is it possible to obtain the elastic constants of the UD composites for FEA simulation based solely on micro-mechanical models and the properties of fibre and matrix?

▪

Is it possible to predict the elastic constants and thickness of multi-layered composites based solely on micro-mechanical models, classical laminate theory, and the properties of fibre and matrix?

▪

Is it possible to assess the failure mechanisms of multi-layered composites using FEA simulations?

2. Materials and Methods

2.1. Materials

The matrix system consisted of a mixture of 49% Elium© 150, 49% Elium© 188, and 2% peroxide hardener Metox 50W. Both Elium© formulations were optimised for the vacuum infusion process. Elium© is an amorphous thermoplastic resin of methyl methacrylate (MMA). Three non-crimp fabrics (NCFs) of E-glass were employed:

▪

Saertex U-E-640 g/m²-1260 mm—0° (areal weight A_F = 640 g/m²);

▪

Saertex B-E-625 g/m²-1270 mm—0°-90 (areal weight A_F = 625 g/m²);

▪

Saertex X-E-610 g/m²-1270 mm—±45° (areal weight A_F = 610 g/m²);

The properties of the matrix made of Elium© and glass fibre are given in

Table 1. Material parameters of the matrix and fibre.

${\mathit{E}}_{\mathit{M}}$ [MPa]	${\mathit{\nu }}_{\mathit{M}}$ [-]	${\mathit{\rho }}_{\mathit{M}}$ [g/cm³]	${\mathit{E}}_{\mathit{F}}$ [MPa]	${\mathit{\nu }}_{\mathit{F}}$ [-]	${\mathit{\rho }}_{\mathit{F}}$ [g/cm³]
2700 [18]	0.37 [35]	1.178 [36]	73,000 [37]	0.2 [37]	2.56 [37]

.

2.2. Manufacturing Using Vacuum Infusion

The set-up for vacuum infusion, which was mounted over an aluminium plate, is shown in Figure 1. The following sequence was stacked from bottom to top:

▪

Release film of ETFE: WL 5200R;

▪

Peel-ply: Release Ply 960;

▪

Stacked glass fabric (10 or 12 layers);

▪

Peel-ply: Release Ply 960;

▪

Mesh flow;

▪

Breather cloth;

▪

Vacuum film: Securlon^® L1000.

Additional elements include inlet and outlet hoses for infusion. After the vacuum bag was sealed, pressure of ca. 10 mbar was applied to remove entrapped air and to impregnate the fibres with resin. Due the exothermic nature of the curing of the matrix, although the curing took place at RT, a peak temperature of 46 °C (see Figure 1 right) was reached, which was similar to the temperature of 49.5 °C reported by Han et al. [13].

After vacuum infusion, following the datasheet’s recommendation, the composite laminates were post-cured at 80 °C for 2 h. The samples for 3-point-bending and fibre weight determination were cut into final geometry from a larger plate using water pressure cutting.

2.3. Characterisation of Composite Laminates

2.3.1. Morphological Analysis Through Optical Microscopy

The morphological analysis of the laminate cross-section was carried out using optical microscopy in a VHX-7000 microscope (Supplier: Keyence, Neu-Isenburg, Germany). The surface treatment before image making was carried out with a grade 1200 sandpaper, followed by a grade 2500 one, and finished by polishing with corundum.

2.3.2. Fibre Weight Content Through Burn-Off Process

The fibre weight content was determined according to the process A of the DIN EN ISO 1172 [38] using samples with a nominal geometry of 30 × 20 × thickness mm³. In this test, the matrix was burnt until only the fibre remained. The sample preparation consisted of drying in a convection oven Nabertherm© L9/S11 (Supplier: Nabertherm, Lilienthal, Germany) at 105 ± 2 °C. Weighting before and after burning was carried out using the scale MSE 125 P-100-DU (Supplier: Sartorius, Göttingen, Germany). A total of 3 repetitions per laminate type were carried out. Afterwards, the fibre weight content ($W_F$) was converted to fibre volume content ($V_F$) based on the densities of the fibre ${\rho }_F$ and of the matrix ${\rho }_M$, as follows:

$$V_F={\displaystyle \frac{W_F}{W_F+\left(1-W_F\right)\frac{{\rho }_F}{{\rho }_M}}}$$

(1)

2.3.3. Mechanical Characterisation Through 3-Point-Bending Test

The bending modulus and bending strength were obtained through the 3-point-bending test according to the DIN EN ISO 14125 [39] (see Figure 2). The following parameters were employed for the tests:

• Repetitions = 6;
• Nominal sample geometry = 100 × 25 × thickness mm³;
• Radius of the loading pin = 5 mm;
• Radius of the supporting pin = 5 mm;
• Span length ($L$) = 84.8 mm.

Edge fibre strain (${ϵ}_B)$, bending stress (${\sigma }_B),$and the bending modulus ($E_B$) were calculated between 0.05 and 0.25% of strain as follows.

$$E_{B,EXP}={\displaystyle \frac{L^3}{4w{t}_T³}}\ast {\displaystyle \frac{∆P}{∆s}}$$

(2)

$${\sigma }_{B,EXP}={\displaystyle \frac{3PL}{2w{t}_T²}}$$

(3)

$${ϵ}_{B,EXP}={\displaystyle \frac{6t_{T}s}{L²}}$$

(4)

where $P$ is the applied bending load, $s$ is the deflection, $w$ is the sample width, $t_T$ is the sample thickness, and $L$ is the span length.

2.4. Analytical Calculation of Elastic Constants Through Classical Laminate Theory

A detailed description of the Equations from the CLT are given in Appendix A. The first step in the analytical determination of elastic constants is related to the calculation of the composite properties for each layer by means of a micromechanical model using the fibre and matrix properties. In the current work, the rule of mixtures from Voigt [32] and Reuss [31] were employed. The shear modulus $G$ of both fibre and matrix can be obtained from the tensile modulus $E$and Poisson’s ratio $\nu$:

$$G_{ForM}={\displaystyle \frac{E_{ForM}}{2\left(1-{\nu }_{ForM}\right)}}$$

(5)

Considering the rule of mixture from Voigt [32], using the fibre volume content $V_F=1-V_M$, it is possible to obtain the tensile modulus in fibre direction $E_{L,X}$, the Poisson’s ratio ${\nu }_{C,XY}$, and the density ${\rho }_C$ of the composite:

$${\mathrm{E}}_{L,\mathrm{X}}={\mathrm{E}}_{\mathrm{F}}{V}_F+E_M(1-V_F)$$

(6)

$${\mathsf{\nu }}_{\mathrm{L},\mathrm{X}}={\mathsf{\nu }}_{\mathrm{F}}{V}_F+{\nu }_M(1-V_F)$$

(7)

$${\rho }_{\mathrm{L}}={\rho }_F{V}_F+{\rho }_M{V}_M$$

(8)

For the tensile modulus in transverse fibre direction $E_{C,Y}$ and the shear modulus $G_{C,XY}$, the rule of mixture according to Reuss [31] can be employed:

$$E_{L,Y}={\displaystyle \frac{1}{\frac{{\nu }_F}{E_F}+\frac{{\nu }_M}{E_M}}}$$

(9)

$$G_{L,Y}={\displaystyle \frac{1}{\frac{{\nu }_F}{G_F}+\frac{{\nu }_M}{G_M}}}$$

(10)

The thickness of a single layer can be calculated based on the fibre area weight ${\mathrm{A}}_{\mathrm{F}}$ being in [g/cm²], as well as the fibre density ${\rho }_F$ and fibre volume content $V_F$:

$${\mathrm{t}}_{\mathrm{L}}={\displaystyle \frac{{\mathrm{A}}_{\mathrm{F}}}{{\rho }_F{V}_F}}$$

(11)

2.5. Finite Element Analysis

A finite element analysis (FEA) was conducted using the commercial software Abaqus© 2023 to simulate the 3-point-bending test on the composite laminate to assess the failure mechanisms and validate material parameters, particularly focusing on bending stiffness. The material was modelled as *LAMINA, assuming equal shear moduli for G_XY, G_XZ, and G_YZ. The elastic constants, whose values are given in

Table 2. Elastic constants for the *LAMINA material for the FEA model.

${\mathit{E}}_{\mathit{X}}$ [MPa]	${\mathit{E}}_{\mathit{Y}}$ [MPa]	${\mathit{\nu }}_{\mathit{X}\mathit{Y}}$ [-]	${\mathit{G}}_{\mathit{X}\mathit{Y}}={\mathit{G}}_{\mathit{Y}\mathit{Z}}={\mathit{G}}_{\mathit{X}\mathit{Z}}$ [MPa]
41,365	5740	0.276	2106

, were calculated using Equations (5)–(10).

The composite laminate was represented using 3D-shell elements (S4R) with an element size of 1 mm (Figure 3a), while 0–90° and ±45° plies were modelled as the half-layer thickness of each orientation (Figure 3b). The support and loading pins were modelled as 3D discrete rigid bodies, with the load applied via a reference point located at the bottom of the loading pin. The support pins were encastered, meaning that no degrees of freedom were allowed. The interaction between the pins and the composite laminate was defined using general contact for tangential behaviour, incorporating a penalty friction formulation with a default friction coefficient of 0.1. Normal behaviour was modelled as hard contact.

Failure analysis was conducted using the in-built Tsai–Hill criterion [40] to predict failure initiation in the laminate:

$${\displaystyle \frac{{\mathsf{\sigma }}_{11}^2}{X²}}-{\displaystyle \frac{{\sigma }_{11}{\mathsf{\sigma }}_{22}}{X^2}}+{\displaystyle \frac{{\mathsf{\sigma }}_{22}^2}{Y^2}}+{\displaystyle \frac{{\mathsf{\sigma }}_{12}^2}{S^2}}<1$$

(12)

where ${\sigma }_{11}$, ${\sigma }_{22}$, and ${\sigma }_{12}$ are principal stresses. If ${\sigma }_{11}>0$, then $X=X_{+}$ as the tensile strength in the fibre direction; otherwise, $X=X_{-}$ as the compressive strength in the fibre direction. Similarly, if ${\sigma }_{22}>0$, then $Y=Y_{+}$ as the tensile strength in the transverse fibre direction; otherwise, $Y=Y_{-}$ as the compressive strength in the transverse fibre direction. Finally, $S$ is the shear strength. The strength parameters for the failure analysis are given in

Table 3. Strength parameters for the Tsai–Hill failure criterion and respective sources.

${\mathit{X}}_{+}$ [MPa]	${\mathit{X}}_{-}$ [MPa]	${\mathit{Y}}_{+}$ [MPa]	${\mathit{Y}}_{-}$ [MPa]	$\mathit{S}$ [MPa]
304.5 [18]	420 [18]	76 [35]	130 [35]	70 [18]

.

3. Results and Discussion: Dimensioning, Manufacturing, and Characterisation

3.1. Dimensioning: Selection of Laminate Lay-Up

The dimensioning and selection of optimal laminate lay-up were carried considering the following requirements:

▪

Manufacturing process of vacuum infusion;

▪

Maximum bending stiffness in the fibre direction;

▪

Laminate optimised for adhesive bonding → neither 0° nor 90° as outermost layer [41];

▪

Symmetrical lay-up;

▪

Available NCF-textile: 0°, 0–90°, ±45°;

▪

Minimal thickness of 5 mm.

At the dimensioning stage, i.e., prior to manufacturing, there was no available information about the fibre volume content, which is a parameter that impacts the layer thickness, as well as mechanical constants. Reports in the literature for the manufacturing of vacuum-infused laminates with Elium© varied between 46% [42] and 58% [43], so that to ensure the 5 mm minimal thickness, laminate lay-ups were considered with 10 and 12 layers.

Two non-quasi-isotropic and two quasi-isotropic laminate lay-ups were considered for 10 and 12 layers. Calculations were carried out assuming a fibre volume content of $V_{F,G}$ = 46%, the lower bound of reference values in the literature. Based on Equation (11), the estimated layer thickness was $t_{L,G}=$ 0.52 mm.

The properties of the UD composite were obtained using Equations (5)–(10). The composite lay-ups and respective predicted elastic constants (marked with the underscore Pr), obtained with the CLT, are given in

Table 4. Composite lay-ups and respective predicted elastic constants for an assumed fibre volume content V_F,G = 46% according to the CLT.

# Layers	ID	Lay-Up	EB,X,Pr [MPa]	EB,Y,Pr [MPa]	GB,XY,Pr [MPa]
10	10L-NQI-A	[±45/0/±45/0–90/0]s	16,415	8411	6595
10	10L-NQI-B	[±45/0–90/±45/0–90/0]s	13,561	11,832	6600
10	10L-QI-A	[±45/0/±45/90–0/90]s	15,899	9126	6597
10	10L-QI-B	[±45/0–90/±45/90/0]s	12,632	12,785	6600
12	12L-NQI-A	[±45/0/0/±45/0–90/0]s	20,394	7934	5624
12	12L-NQI-B	[±45/0/0–90/±45/0–90/0]s	18,750	10,053	5627
12	12L-QI-A	[±45/0/90–0/±45/90–0/90]s	17,782	11,177	5627
12	12L-QI-B	[±45/0/0–90/0–90/±45/90/0]s	14,951	14,192	5628

# means number, i.e., Number of Layers.

, with the sample IDs 10 L and 12 L meaning 10 and 12 layers, NQI meaning non-quasi-isotropic, and QI meaning quasi-isotropic. The highest bending stiffness was obtained for the lay-ups 10L-NQI-A and 12L-NQI-A. Based on these results, the next step was the manufacturing of composite laminates, which will be discussed in the next section.

3.2. Manufacturing: Morphological, Fibre Content, and Geometrical Analysis

The burn-off process was carried out in the composite laminate so that the matrix was incinerated until only the glass fabric remained. This characterisation was performed considering three repetitions for samples with 10 and 12 layers with optimal lay-up from Table 4 (10L- and 12-L-NQI-A), with BO meaning burn-off. As seen in

Table 5. Results of the burn-off process.

10-Layer Sample	${\mathit{l}}_{\mathit{C}}$ [mm]	${\mathit{w}}_{\mathit{C}}$ [mm]	${\mathit{t}}_{\mathit{T}}$ [mm]	${\mathit{W}}_{\mathit{F}}$ [%]	12-Layer Sample	${\mathit{l}}_{\mathit{C}}$ [mm]	${\mathit{w}}_{\mathit{C}}$ [mm]	${\mathit{t}}_{\mathit{T}}$ [mm]	${\mathit{W}}_{\mathit{F}}$ [%]
10L-NQI-A-BO-1	29.86	19.77	4.46	72.97	12L-NQI-A-BO-1	30.12	19.92	5.36	73.66
10L-NQI-A-BO-2	29.79	19.86	4.48	72.93	12L-NQI-A-BO-2	30.08	19.95	5.34	73.76
10L-NQI-A-BO-3	29.82	19.76	4.48	73.25	12L-NQI-A-BO-3	30.13	19.96	5.33	73.46
Average	29.82	19.80	4.47	73.05	Average	30.11	19.95	5.34	73.62
St. Dev	0.04	0.06	0.01	0.17	St. Dev	0.02	0.02	0.01	0.15

, the obtained fibre weight contents for the samples with 10 and 12 layers were $W_{F,10L}$ = 73.05 ± 0.17% and $W_{F,12L}$ = 73.62 ± 0.15%, respectively. These results indicate a suitable and reproducible manufacturing process, as a similar fibre weight content was obtained for 10- and 12-layered laminates. Using Equation (1), it is possible to obtain the respective fibre volume content, based on the fibre weight content of $W_F$ ≈ 73%, which leads to an actual value of $V_{F,A}$ ≈ 55%. Compared to benchmark values of fibre volume content found in the literature, as given in

Table 6. Benchmark values for the fibre volume content of GFRP laminates made with Elium©.

Author	$\mathbf{Fibre}\mathbf{Volume}\mathbf{Content}{\mathit{V}}_{\mathit{F}}$ [%]
Mamalis et al. [44]	49.0
Obande et al. [21]	49.5
Chilali et al. [20]	52.4
Devine et al. [45]	53.4–56.8
Han et al. [13]	55
Boissin et al. [43]	51–58

, the current results are located in the upper bound, which again corroborates the conclusion of a suitable manufacturing process.

The laminate thickness for the samples with 10 and 12 layers was $t_{T,10L}$ = 4.47 ± 0.01 mm and $t_{T,12L}$ = 5.34 ± 0.01 mm, respectively, so that the actual average layer thickness was $t_{L,A}$= 0.445 mm, less than the estimated value of 0.52 mm. This could be related to the higher fibre volume content of $V_{F,A}$ ≈ 55% obtained with the actual composite laminate, compared to the estimated value of $V_{F,G}$ = 46%. At the same time, by taking Equation (11) and the actual fibre volume content of 55%, a layer thickness of 0.44 mm was calculated, which matched exactly the value obtained experimentally. As one of the requirements of the composite laminates was a minimum thickness of at least 5 mm, the remaining part of the characterisation was carried out only for the composite with 12 layers.

An optical microscopy image of the composite with 12 layers, in which each layer is identified, is shown in Figure 4. The 90° plies had a dark colour, as they allowed the passing of light, and the microscope background was black. No visible voids were observable with a magnification of 200×.

3.3. Mechanical Characterisation: Three-Point-Bending-Test

The results of the three-point-bending-test, considering six repetitions of the sample with 12 layers, are given in the form of deflection as a function of the bending force in Figure 5. Inside the plot, the six samples can be seen, with the bending line slightly lighter right in the middle of the samples. The bending force increased monotonically until the fracture of the samples.

As seen in

Table 7. Results of the three-point-bending test.

Sample	${\mathit{t}}_{\mathit{T}}$ [mm]	${\mathit{w}}_{\mathit{C}}$ [mm]	${\mathit{\sigma }}_{\mathit{B},\mathit{m}\mathit{a}\mathit{x}}$ [MPa]	${\mathit{E}}_{\mathit{B},\mathit{X},\mathit{E}\mathit{X}\mathit{P}}$ [MPa]	$\mathit{s}@{\mathit{\sigma }}_{\mathit{B},\mathit{m}\mathit{a}\mathit{x}}$ [MPa]
12L-NQI-A-3PB-1	5.32	25.13	634	24,568	6.68
12L-NQI-A-3PB-2	5.31	25.13	630	23,851	6.77
12L-NQI-A-3PB-3	5.42	25.11	594	24,920	6.50
12L-NQI-A-3PB-4	5.31	25.12	639	24,579	6.92
12L-NQI-A-3PB-5	5.35	25.11	673	24,939	7.29
12L-NQI-A-3PB-6	5.30	25.02	647	24,742	6.96
Average	5.34	25.10	636.17	24,600	6.85
Std. Dev.	0.05	0.04	25.70	400	0.27

, based on the bending force $P$ and the deflection $s$, the bending strength ${\sigma }_{B,max}$, its respective deflection and maximum strength $s@{\sigma }_{B,max}$, and the bending modulus $E_{B,X,EXP}$ can be calculated with Equations (2) and (3). The bending strength obtained was ${\sigma }_{B,max}$ = 636.17 ± 25.70 MPa, whereas the bending modulus was $E_{B,X,EXP}$ = 24,600 ± 400 MPa.

4. FEA Simulation: Failure Assessment and Bending Modulus

The results of the FEA simulation (solid blue line) of the three-point-bending test, considering the modelling parameters described in Section 2.5, are plotted along with experimental curves (dashed lines) in Figure 6. The bending stress and edge fibre strain were calculated from Equations (3) and (4), respectively. Based on the FEA results, it was possible to calculate the bending modulus by taking the slope of the bending stress and edge fibre strain between strain values of 0.05 and 0.25%, similarly to what was done experimentally. A value E_B,X,FEA = 23,499 MPa for the bending modulus was obtained.

By comparing the bending modulus obtained with different approaches, the following results were achieved:

▪

Experiment, three-point-bending:

⚬

E_B,X,EXP = 24,600 ± 400 MPa (Table 7).

▪

Classical laminate theory, $V_{F,A}$ = 55%:

⚬

E_B,X,CLT = 24,110 MPa (error = −1.99%).

▪

Finite element analysis:

⚬

E_B,X,FEA = 23,499 MPa (error = −4.47%).

This result underscores the effectiveness of analytical approaches such as micromechanical models and classical laminate theory, as both prediction with CLT and FEA simulations (using micromechanical material parameters) accurately estimated the laminate’s bending modulus, with errors of −1.99% and −4.47%, respectively.

As seen in Figure 6, up to an edge fibre strain of around 1%, the correlation between experiment and FEA simulation was very good. However, for a larger strain value, there was a distinction between them. Using the Tsai–Hill failure criterion and strength parameters from Table 3, a simulation was carried to obtain the bending stress leading to the failure of the composite laminate. These results can be seen in Figure 7a, which indicates that the first layer to fail was L11 (Figure 3a) under tensile stress (Figure 7b). The respective bending stress was ${\sigma }_B$ = 220 MPa, matching the position in which the FEA simulation and experiments started to diverge. As soon as the failure started to occur, a reduced number of layers was able to bear the load, so that the stiffness of the sample decayed. However, since no progressive damage, i.e., stiffness degradation, was implemented in the FE model, this behaviour could not be reproduced.

According to the FEA simulation, the sequence of layer failure was the following:

▪

L11 0° (symL2), tension → ${\sigma }_B$ = 220 MPa;

▪

L10 0° (symL3), tension and L2 0°, compression → ${\sigma }_B$ = 324 MPa;

▪

L12 − 45° (symL1), tension → ${\sigma }_B$ = 359 MPa;

▪

L12 + 45° (symL1), tension → ${\sigma }_B$ = 393 MPa;

▪

L1 − 45°, compression and L3 0°, compression → ${\sigma }_B$ = 429 MPa;

▪

L1 + 45°, compression → ${\sigma }_B$ = 463 MPa.

In the three-point-bending-test, half of the layer was under compressive stress and the other half under tensile stress. By looking at Figure 8, it is possible to correlate that the bottom fibres (under tension) failed much more strongly that the fibres under compression. This could be seen by the stress whitening on the cross-section, which can be correlated to the micro-cracking of the matrix [46]. Moreover, the fibres in the centre apparently did not fail, as corroborated by FEA simulations, which indicated failure at the top and bottom three layers of the composite laminate.

5. Conclusions

The current investigation dealt with the dimensioning, manufacturing, and characterisation of vacuum-infused thermoplastic composites made of non-crimp glass fabric and the liquid acrylic-based resin Elium©. Composite laminates were manufactured with 10 and 12 layers, achieving an actual fibre weight content of W_F,A = 73%. From the three-point-bending-testing, a bending strength of 636.17 ± 25.70 MPa and a bending modulus of 24,600 ± 400 MPa were obtained for the laminates with 12 layers.

Based on micro-mechanical models, UD elastic constants were calculated, which were employed for classical laminate theory (CLT) and finite element analysis (FEA). By having the actual fibre volume content of V_F,A = 55% and applying the CLT, it was possible to dimension and select composite lay-ups with the maximum bending stiffness and to predict the laminate thickness (with an error of −0.18%) and the bending modulus (with an error of −1.99%).

FEA simulations were able to predict the bending modulus with an error of −4.47%, as well as to demonstrate, by using the Tsai–Hill criterion, the correlation between the failure of specific layers of the composite laminate with variations between experiments and simulations.

As for future works, further research could focus on extending the methodology to more complex composite geometries beyond flat laminates, as well as considering aspects of joining (e.g., adhesive bonding, welding, and mechanical fastening) of such laminates.