Polymer composites with short recycled glass fibres manufactured by vacuum infusion

Recycling glass fibre reinforced thermoset composites remains a challenge despite years of research in the field. Among the barriers for a successful recycling, the properties of the recycled materials and their application in new products represent the main ones. With a pyrolysis recycling process, clean glass fibres can be extracted and separated from the polymer matrix. In this paper, recycled glass fibres are incorporated in a new polymer composite structure using both virgin and recycled glass fibres. Recycled glass fibres have a Young’s modulus comparable to virgin glass fibres. The objective is to identify a composite lay-up using virgin and recycled glass fibres, and where the Young’s modulus of recycled fibres is beneficial for the resulting composites mechanical properties. To do so, a sandwich structure composite with virgin fibres on the outer layer and recycled fibres in the inside is designed and manufactured. In order to produce flat panels despite the short and random fibres, a manufacturing procedure ensuring a good dispersion of the fibres and controlling the thickness of the panel is developed. Panels are produced incorporating two different weight fraction of recycled glass fibres. The quality of the panels are characterized using microstructure observations, fibre porosity analysis and mechanical testing. Results shows that the sandwich composites have a low porosity content and achieves higher bending stiffness than a composite only made of unidirectional virgin glass fibres.


Introduction
The recycling of wind turbine blades and of glass fibre reinforced thermoset polymer composites has been studied for many years, but remains challenging.A wind turbine blade is a one-piece component, which cannot easily be separated.The hollow structures made of two aerodynamic shells and a beam are composed of different composite materials.The type of material and the diversity of materials used in blades is a challenge from a recycling point of view.The aerodynamic shells are made of thin glass fibre reinforced laminate separated by a core materials made of balsa wood, PET or PVC foam.The long and slender beam is a combination of unidirectional fibre reinforced polymer composites and multiaxial fibre sandwich composite named shear webs.The fibres used are glass fibres, carbon fibres or a combination of both.The polymer resin may be epoxy, polyester or vinylester.Some blades may be manufactured as one single piece, while other blades will be assembled using adhesive joints, which cannot be easily released [1].
At end-of-life, due to the cross-linked nature of thermoset polymers, glass fibre reinforced polymer composite, as used in wind turbine blades, cannot be soften, reshaped or easily separated into fibres and matrix material.Nevertheless, recycling processes have been developed and wind turbine blades can be transformed into different types of recycled materials [2,3].These processes can involve cutting, shredding, heat treatment and/or chemical treatment.The resulting recycled materials can be sections of blades, shredded composites or fibres; with dimension varying from a few meters down to few 1293 (2023) 012036 IOP Publishing doi:10.1088/1757-899X/1293/1/012036 2 millimetres or microns.The applications for these recycled materials are diverse.For larger blade sections, footbridges or outdoor furniture may be considered [4][5][6].With a pyrolysis recycling process, clean glass fibres can be extracted and separated from the polymer matrix [7][8][9].An obvious application for recycled glass fibres is the replacement of virgin glass fibres in existing products.This option is challenged by the properties of recycled glass fibres, which are shorter, randomly oriented and with lower strength properties compared to virgin glass fibres.For thermal recycling temperatures reaching 450 o C or more, the loss in fibre strength is of at least 40% [10].To overcome this challenge, part of the research has explored restoring the strength properties of recycled glass fibres [11].Another option is to investigate applications for recycled glass fibres, where the recycled fibres do not replace virgin fibres, but are used for their own characteristics and properties.The advantage of recycled glass fibres is that compared to other recycled materials such as shredded composite obtained with mechanical recycling, they have a clean surface.It was shown in previous publication that the adhesion between shredded composite and new polymer composite was poor [12].Glass fibres obtained from pyrolysis are likely to have better fibre/matrix bonding.
In this paper, recycled glass fibres obtained from pyrolysis are incorporated in a new polymer composite structure using both virgin and recycled glass fibres.The idea is to take advantage of the high stiffness properties of recycled glass fibres and combine them to virgin glass fibre to provide high strength properties to the final composite.To do so, a sandwich structure composite with virgin fibres on the outer layer and recycled fibres in the inside is proposed and investigated.The objective of this paper is twofold.First, this paper aims at establishing a manufacturing procedure, which results in high quality composite materials.Second, this paper studies the impact of using recycled glass fibres on the mechanical properties of the composites produced.Some of the challenges to overcome when manufacturing composites with short randomly oriented recycled glass fibres and a vacuum infusion process are to achieve: (i) a good dispersion and distribution of the fibres in the composite, (ii) an even thickness and a flat surface and (iii) a high fibre content.To study the impact of incorporating recycled glass fibres, the glass fibres virgin and recycled are thoroughly characterized and composites are produced incorporating two different weight fraction of recycled glass fibres.The quality of the composites produced are characterized using microstructure observations, fibre porosity analysis and mechanical testing.

Fibres and matrix properties
In this paper, the virgin fibres used are E-glass fibres in a unidirectional fibre fabric from Saertex (Textile structure 7004118) with an areal weight of 640 g/m 2 .The epoxy used is the PRO SET INF 114 epoxy with the PRO SET INF 211 epoxy hardener.The recycled glass fibres are obtained from a pyrolysis process performed by the company Makeen Energy [13].The origin of the composite processed with pyrolysis is unknown.

2.1.1
Density.To measure the density of the glass fibres, a gas pycnometer Quantachrome Ultrapyc 1200e is used, with nitrogen as the displacement medium.For the virgin fibres, the fibres are winded onto a spool and placed in the pycnometer.For the recycled glass fibres, the fibres are ground with a mortar and placed in a container.Before the measurements, the fibre samples (7.2 g of virgin fibres and 13.5 g of recycled fibres) are dried for a minimum of 12 hours.For both types of fibres, 3 repeated density measurements are performed.To measure the density of the epoxy resin, samples are collected from the inlet tube during the manufacturing of the composites.The dimension of the specimen are 2 cm long and the diameter of the tube.The samples are weighed in air, and in water and the density of the material is calculated using Archimedes law.

2.1.2
Mechanical properties.Single fibre tensile tests were performed using a semi-automated single fibre testing machine, "Favimat+", combined with Robot 2. Initially, a vibroscope test is performed to determine the linear density of the fibre.Based on the linear density ( [kg/m]) and the measured fibre density ( [kg/m 3 ]) values, the cross-sectional area ( = /) of the fibre is calculated.The fibres were tested at a gauge length of 30 mm.For both types of fibres, at least 100 single fibres are tested.The fibre Young's modulus is calculated according to the composite materials standards, between 0.05% and 0.25% strain.The mechanical properties of the epoxy resin are obtained from the literature.

Manufacturing of composites
The composites are manufactured using a vacuum infusion process.To enable a good dispersion and distribution of the recycled fibres, achieve an even thickness, a flat surface and reach high fibre content, the manufacturing procedure is made of four steps: fibre dispersion, fibre distribution, pressing and vacuum infusion.Figure 1 shows a schematic representation of the steps.In the following sections, the steps are described in more details.

2.2.1
Fibre dispersion.To improve the distribution and the resin impregnation of the short recycled fibres, the fibres are dispersed using compressed air in a closed bucket.The procedure lasts for approximately 1 min.The objective of the procedure is to separate the fibre bundle and create a woollen structure of individual fibres.

2.2.2
Fibre distribution and lay-up.The laminate dimensions are 300 x 300mm.The lay-up consists of three layers in total, two outer layers of unidirectional virgin fibres and an inner layer of recycled fibres.Two laminates are prepared where the mass of recycled glass fibres is varied.These composites are named Composite #1 and Composite #2.The mass of recycled glass fibre is set to be half (Composite #1) or equal (Composite #2) to the weight of virgin fibre as illustrated in figure 2. In any case, the portion of recycled fibres is divided in two and distributed manually on the two outer layers of virgin fibre fabrics.In order to close the sandwich structure and avoid the fibres to move around, an adhesive spray is used (INFUTAC Green from Diatex).The adhesive is sprayed on the fabric and the recycled fibres are pressed against the fabric.Then, the two outer layers of unidirectional virgin fibres with the recycled glass fibres attached are assembled to close the structure into a sandwich structure containing the recycled fibres as an inner layer.

2.2.3
Pressing.Once the sandwich lay-up is closed, it is placed into a press facility in order to compress it.The objective of this step is to reduce the thickness of the dry lay-up before the vacuum infusion.The applied force improves the compaction of the dry fibres.The pressing procedure is as follow.The fibre lay-up is pressed for 1 min at 70 kN.Considering the dimension of the lay-up, the equivalent pressure is 0.8 MPa.The procedure is repeated until the height of the lay-up could not be further decreased.After pressing, the lay-up is packed for vacuum infusion.An aluminium plate with dimension 240mm x 240mm x 3.5 mm is introduced in the vacuum bag.It is placed on top of the laminate above the release film and peel ply and below the vacuum bag.The vacuum bag encloses the whole system.The aluminium plate plays the role of a second-sided mould during the vacuum infusion.The plate helps to avoid a wavy surface for the composite.

2.2.4
Vacuum level and curing profile.The composite is infused with a vacuum level of -999 mbar.The curing cycle selected is 24 hours at 25 o C, followed by 8 hours at 82 o C.

2.2.5
Additional composite materials.In order to estimate the volumetric composition of the individual layers of the sandwich composites manufactured, a unidirectional fibre composite was prepared with vacuum infusion and using the same fabric and epoxy resin as in the sandwich composites.This composite is named Reference UD.To compare the mechanical properties of Composites #1 and #2, two additional reference composites are modelled, but not manufactured.These two reference composites are sandwich composites with the same fibre volume fraction and thickness of individual layers, as Composites #1 and #2, however only assuming virgin glass fibre properties.These composites are named Reference #1 and Reference #2.Reference #1 uses the characteristics of Composite #1 and Reference #2 uses the characteristics of Composite #2.

Characterization of composites 2.3.1
Microstructure characterization.To characterize the microstructure, samples with dimensions 20 mm x thickness of the plate are cut from each composite.The samples are casted in epoxy, and the sample cross-section in the plane normal to the unidirectional fibre direction is grinded and polished.Observations are made using an optical microscope (Leica, IM5000).

2.3.2
Volumetric composition.To determine the fibre, matrix and porosity content, five rectangular samples of dimensions 15 x 20 mm x thickness are cut from each composite.The samples are roughly polished, dried overnight, and weighed.The samples are then sealed with paraffin wax to include the correct porosity volume in the determinations.The sealed samples are weighed in air, and in water in order to determine the composite density.Then, the epoxy matrix in the samples is burnt off overnight at 565°C.The weight of the remaining glass fibres is recorded in order to determine the fibre weight content.Using standard equations for composite materials and based on the measured values of fibre weight content, composite density, fibre density, and matrix density, the volume content of fibres, matrix and porosity in the composites is calculated [14].

2.3.3
Static tensile test and Young's modulus.A static tensile test is performed to determine the Young's modulus of the composite.The test follows the ISO 527-5:2009(E) standard.The length of the tensile specimens is however adjusted to the dimensions of the composite manufactured.4 tensile specimens are tested for Composite #1 and Composite #2.The cross-sectional area of the specimens is measured at 3 different locations, and the averaged cross-sectional area is used for the calculation of stress.Tensile tests are performed with an Instron tensile test machine (Instron, Buckinghamshire, United Kingdom,) having a load cell of 500 kN and a cross-head speed of 2 mm/min.The Young's modulus is determined in the strain range from 0.05 to 0.25%.The Young's modulus of the Reference UD plate is calculated using a rule of mixture, see equation (1), where ERef is the Young's modulus of the Reference UD composite, Vf is the fibre volume fraction, Ef and Em the Young's modulus of the fibre and matrix respectively. (1)

2.3.4
Flexural properties.To determine the impact of the sandwich structure on the flexural properties of the composite, 3-point bending tests are performed according to the ISO 14125:1999 standard.The width of the specimens is 15 mm and the length equals the thickness multiplied with 20. 10 specimens are tested for Composite #1 and Composite #2.
The bending stiffness [EI] of the composites is calculated using equation ( 2), where Ecore is the Young's modulus of the inner section of the composite made of recycled fibres.Ecore is calculated using equation ( 4).b is equal to 15 mm for all composites and corresponds to the width of the 3-point bending test specimens, hcore is the thickness of the inner section of the composite.ERef is the Young's modulus of the Reference UD composite, hskin is the thickness of one of the skin layers made of UD virgin glass fibres.hcore and hskin are roughly measured from micrographs.d is the distance to the neutral axis of the composite beam and is calculated using equation (3). Figure 3 shows a schematic representation of the parameters used in equation ( 2) and (3).The Young's modulus of the core section of Composite #1 and #2 is estimated using the modified rule of mixture in equation ( 4), where ηo is an orientation efficiency factor defined by Krenchel [15].It is assumed that the fibres are randomly oriented in 2D, therefore the factor is set equal to 0.375.η1 is the fibre length efficiency factor as defined by Cox [16].In the present study, the fibres have a length ranging from few mm to few cm.There are therefore considered to be long fibres and η1 is set equal to 1.
=   .  .  .  +   (1 −   ) The deflection of the specimens during the 3-point bending test is calculated using the Timoshenkobeam theory in equation ( 5), where P is the load applied, L is the length of the 3-point bending test specimen, [EI] is the bending stiffness, hcore is the thickness of the inner section of the composite and Gcore is the shear properties of the core section, which is calculated using equation ( 6).

𝛿 = 𝑃𝐿 3 48[𝐸𝐼]
+ ℎ  16   2  (5) The shear properties of the core section of the sandwich structured composite, Gcore is calculated using equation ( 6), where Ecore is the Young's modulus of the core section, and υ is the Poisson's ratio, which is set equal to 0.4.

Fibres and matrix properties
Figure 4 shows the recycled glass fibres.The fibres have different length, ranging from a few cm up to 10 cm and are randomly oriented.The fibres are generally white, however some fibre bundles are found to have a light yellow colour and few fibre bundles are grey.The variation in length is due to the shredding process taking place before the pyrolysis process.The properties of the fibres recycled and virgin are presented in table 1.In addition, the stress strain curves of single fibre test are presented in figure 5.

3.2.1
Fibre dispersion.The fibre dispersion procedure is applied for approximately 1 min.The procedure is illustrated in figure 6.In this figure, the same bucket is used in all 3 images.The fibres before and after the dispersion are randomly oriented, however, before dispersion, they show a bundle morphology, whereas the fibres after dispersion show a woolly or a candy floss like structure, where fibres are separated from each other.

3.2.2
Fibre distribution and lay-up.Figure 7 shows some of the main steps in the distribution of the fibres and the fibre lay-up.Figure 7 (a) shows how the adhesive is centred in the middle of the fabric.Figure 7 (b) shows the resulting recycled glass fibres distributed on a virgin glass fibre fabric.Figure 7 (c) shows how the adhesive is effective in keeping the fibres in place.The measured mass of recycled glass fibres used for Composite #1 and Composite #2 is detailed in table 2. The weight fractions of recycled fibres, with respect to the total weight of fibres is calculated and presented in table 2.

3.2.3
Pressing.Figure 6 shows the impact of the pressing procedure.The thickness of the fibre layup is measured.Given the uneven surface of the lay-up, the thickness is approximately measured using a ruler placed next to the fibre lay-up.The thickness decreased from above 20 mm in figure 6 (a) down to 10 mm after two rounds of pressing in figure 6 (b).After two rounds of pressing, the thickness of the lay-up could not be further decreased.During pressing, fibres cracking could be heard.

3.2.4
Vacuum infusion and curing.The composites produced are shown in figure 9, where images (a) and (b) show Composite #1 and Composite #2 respectively.As the recycled glass fibres are placed in the centre of the unidirectional fibre fabric, the edges of the composites are thinner than in the centre of the composite.To exclude this part from the characterization, 25 mm of the edges of the panel are cut out, as illustrated in figure 9 (b). Figure 9 also shows that the colour of the plate is not uniform.The black and brown areas indicate areas where recycled fibres are concentrated.Finally, figure 9 show the samples and specimens cut out for the characterization.The thickness of the composite is measured in the gauge section of the tensile test specimens and the bending test specimens.The measurements are reported in table 3 and show that Composite #1 is nearly 2 mm thinner than Composite #2.The variation in thickness is rather small for the Reference UD and Composite #1.For Composite #2 the variation in thickness is of about 10%, which is significant.This could be due to the larger amount of recycled glass fibres which needed to be distributed manually.Nevertheless, these measurements tend to indicate that the manufacturing process is quite efficient in producing uniform and flat panels.The remaining of the properties presented in table 3 are discussed in section 3.3.10 and 11 show the microstructure of Composite #1 and #2, respectively.The overview shows the sandwich construction of the composites, where the outer layers are made of densely packed unidirectional fibre bundles, see figure 10 (c), and the inner section of the composite is made of randomly oriented fibres with lower fibre volume fractions, see figure 11 (c).There are no large porosities visible and composite #1 and #2 are found to be of good quality.The approximate thickness of the different layers measured on the micrographs is reported in table 4. The outer layer made of one layer of unidirectional fibre fabric is referred to as skin.Figure 10 and 11 also show details of the microstructure of the two composites.In both composites, small porosities located in between the unidirectional glass fibres are observed, see figure 10 (c) and figure 11 (b).The size and the location of these porosities suggest dry and not wetted fibres or a poor impregnation of resin.These small porosities are consistently found at the interface between the unidirectional fibre bundles and the randomly oriented recycled fibres.This could indicate that the adhesive used in order to prepare the fibre mats also prevents resin to properly impregnate the fibres.The use of adhesive is an advantage for the distribution of the recycled fibres onto the virgin fibre fabric.However, in order to determine the impact of the adhesive on the composite performance, additional experiments should be conducted without the adhesive.Another observation from the detailed microstructure images are broken fibres with sharp edges, highlighting the very brittle nature of the recycled glass fibres, see figure 10 (b) and figure 11 (c).Finally, when adding the thickness measured on the microstructure images for the skin and the core, the total thickness is approximately matching the thickness measured on the composites.In the next section, the fibre volume fraction of the individual layer is calculated.

3.3.2
Volumetric composition.The fibre volume fraction and the porosity content of all composites is reported in table 3. The porosity content for all composites is low and this result is in good agreement with the microstructure observations.The fibre volume fraction is higher for the Reference UD composite than for the composite including recycled glass fibres.This is expected as the Reference UD composite only contains unidirectional fibres which can be packed to a higher degree than 2D randomly oriented fibres.The volume fraction of recycled fibres in the core section of the composite, is estimated with a rule of mixture using the thickness of the different layers as measured on the micrographs and the total fibre volume fraction measured on the three composites.The fibre volume fraction in the core section of the composite is estimated to 15.9 % for Composite #1 and 18.2% for Composite #2.The results are reported in Table 4. Composite #2 achieved a slightly higher fibre volume fraction in the core section of the composite, this indicates that the fibres may have packed to a higher degree than in Composite #1.However, given the standard deviation on the fibre volume fraction measurement and the heterogeneity of the material, these results may be considered similar.To increase the fibre volume fraction in the core section, future work may include to manufacture panels with different fibre length.Longer fibres may be easier to pack in a 2D fashion.Future work may also consider how to distribute short recycled fibres to facilitate their packing.

3.3.3
Mechanical properties.Table 3 also summarizes the experimental results obtained for the mechanical testing of the composites.The Young's modulus for the composite made with recycled fibres is obtained experimentally while the Young's modulus of the Reference UD composite is estimated theoretically and presented in table 4. The Young's modulus on the Reference UD plate is calculated using the material properties reported in table 1 and equation (1).The Young's modulus on the Reference UD plate is higher than the ones measured on Composite #1 and #2.This result is expected since Composite #1 and #2 are thicker, with a core section made of a lower fibre volume fraction and fibres randomly oriented.A theoretical estimation of the Young's modulus of the skin and the core section of Composite #1 and #2 is calculated using the fibre volume fraction and the fibre properties reported in table 1 and 4. With these Young's modulus, the bending stiffness and the deflection at max load and at 1000N are calculated.The results are presented in table 5.A first observation is that the calculated deflections at maximum load are smaller than the ones measured experimentally.An example of load displacement curve for a sample tested in Composite #1 is presented in figure 12. From this curve, one can see that around the maximum load the curve shows serrations probably indicating that local damages are initiating before the maximum load is reached.The theoretical estimations of deflection do not take these aspects into account.In future work, fitting the experimental curve with a second order polynomial fitting as described in [18] could provide the deflection at maximum load assuming no serration.Another observation is that the deflection of Composite #2 at maximum load is larger than the one of Composite #1.The theoretical deflection calculated indicate that it should be opposite.At maximum load, in theory, the thicker sandwich composite should have a smaller deflection.To further analyse the reasons for this observation, the use of strain gauges on the specimens would be needed to obtain more precise displacement measurements.Finally, when comparing Composite #1 to Reference #1 and Composite #2 to Reference #2, the use of recycled fibres in the core section does not impact the resulting bending stiffness and deflections.The results between the manufactured and the modelled reference composites are similar.This results could be interesting for composite applications, where bending stiffness is required in order to avoid large deflection, as for example in nacelle covers.However, future work is needed to check the requirements for nacelle covers and validate Composite #1 and #2 as potential candidates for this application.

Conclusions
Glass fibres recovered from thermoset polymer composites with a thermal recycling process are known to be brittle.Despite the brittleness of these fibres, the objective of the present study is to identify composite structures that can take advantage of their high stiffness properties.Recycled glass fibres have Young's modulus comparable to virgin glass fibres.Sandwich structured composite were identified as a potential candidate.Sandwich composite with virgin glass fibres on the outer layer and short randomly oriented recycled fibres in the inner section were manufactured.The experimental work focused on the production of flat panels despite the short length and the random orientation of the recycled fibres.A method was developed to ensure a good dispersion of the fibres and a uniform thickness of the panels and composite panels were manufactured with vacuum infusion.Results show that the composites are of good quality, with a low porosity content.Future work, should include the manufacturing of panels without any fibre preparation in order to investigate the impact of these preparation steps on the resulting composites.The mechanical testing results show that the sandwich composites manufactured with recycled glass fibres and composites modelled with virgin fibres achieve 1293 (2023) 012036 IOP Publishing doi:10.1088/1757-899X/1293/1/01203613 the same flexural properties.This result is expected since the flexural properties are mainly governed by geometrical parameters.Nevertheless, this demonstration is interesting from an application point as many products are designed according to the stiffness properties of the material.Since there is no impact on the flexural properties of the composites, using recycled glass fibres instead of virgin glass fibres becomes an attractive option from a circularity perspective.Future work is needed in several aspects.From a manufacturing point of view, the adhesive used in the production of the panels resulted in regions with improper wetted fibres, which may lead to a reduction in composite performance.Experimental work performed without the use of adhesive needs to be realized in order to determine the effect of it.Future work also need to look into how to achieve higher fibre volume fraction in the inner section.This could be done with the use of fibres with a different length or with the placement of fibres in specific direction.With a higher fibre volume content in the inner section higher stiffness properties may be achieved.From a characterization point of view, the inner section made of recycled fibres requires additional characterization work to determine more precisely the fibre orientation and fibre volume fraction.In addition, the characterization of the impact of varying fibre volume fraction could also be interesting.Finally in terms of application, future work should consider the upscaling feasibility of the manufacturing process developed in this study, as well as the cost and environmental impacts of such material.

Figure 1 .
Figure 1.Schematic representation of the manufacturing steps

Figure 2 .
Figure 2. Schematic representation of the composites manufactured and modelled

Figure 3 .
Figure 3. Schematic representation of the thickness parameters used in equations

Figure 4 .
Figure 4. Recycled glass fibres obtained from pyrolysis

Figure 5 .
Figure 5. Stress-strain curves obtained from single fibre testing of virgin and recycled glass fibres.

Figure 6 .
Figure 6.Illustration of the dispersion procedure (a) Recycled glass fibres before dispersion, (b) Compressed air is introduced in the bucket using a hole in the lid and (c) recycled glass fibres after dispersion

Figure 7 .
Figure 7. Illustration of the fibre distribution process (a) The adhesive is distributed on the virgin glass fibre fabric; (b) The recycled glass fibres are pressed onto the fabric with the adhesive; (c) An example of how the adhesive keeps the recycled fibres on the fabric layer

Figure 8 .
Figure 8. Illustration of the impact of pressing on the fibre lay-up for Composite #2 (a) Fibre layup before pressing, (b) After two pressing.

Figure 10 .Figure 11 .
Figure 10.Microstructure of Composite #1 with (a) Overview (b) Details of the recycled fibres, and (c) Details of the region between UD virgin glass fibres and recycled fibres

Figure 12 .
Figure 12.Example of load displacement curve obtained with the 3-point bending test

Table 1 .
Fibres and matrix properties

Table 2 .
Mass of virgin and recycled glass fibres used in the composites

Table 4 .
Properties of individual layers in sandwich composite