Additive manufacturing of continuous fibre reinforced thermoplastic composites using fused deposition modelling: Effect of process parameters on mechanical properties

Composites Science and Technology 181 (2019) 107688

Contents lists available at ScienceDirect

Composites Science and Technology journal homepage: www.elsevier.com/locate/compscitech

Additive manufacturing of continuous fibre reinforced thermoplastic composites using fused deposition modelling: Effect of process parameters on mechanical properties

T

J.M. Chacóna, M.A. Caminerob,∗, P.J. Núñezb, E. García-Plazab, I. García-Morenob, J.M. Revertea a b

Escuela Técnica Superior de Ingenieros Industriales, IMACI, Universidad de Castilla-La Mancha. Avda, Camilo José Cela s/n, 13005, Ciudad Real, Spain Escuela Técnica Superior de Ingenieros Industriales, INEI, Universidad de Castilla-La Mancha. Avda, Camilo José Cela s/n, 13005, Ciudad Real, Spain

A R T I C LE I N FO

A B S T R A C T

Keywords: 3D printing Fused deposition modelling Continuous fibre reinforced thermoplastic composites Mechanical characterization Process parameters Failure analysis

Continuous Fibre Reinforced Thermoplastic Composites (CFRTPCs) are becoming alternative materials to replace the conventional thermosetting polymers and metals due to excellent mechanical performance, recycling and potential used in lightweight structures. Fused deposition modelling (FDM) is a promising additive manufacturing technology and an alternative of conventional processes for the fabrication of CFRTPCs due to its ability to build functional parts having complex geometries. The mechanical properties of a built part depend on several process parameters. The aim of this study is to characterize the effect of build orientation, layer thickness and fibre volume content on the mechanical performance of 3D printed continuous fibre reinforced composites components manufactured by a desktop 3D printer. Tensile and three-point bending tests are carried out to determine the mechanical response of the printed specimens. SEM images of fractured surfaces are evaluated to determine the effects of process parameters on failure modes. It is observed that the effect of layer thickness of nylon samples on the mechanical performance is marginally significant. In addition, continuous fibre reinforced samples show higher strength and stiffness values than unreinforced ones. The results show that carbon fibre reinforced composites exhibit the best mechanical performance with higher stiffness and flat samples exhibit higher values of strength and stiffness than on-edge samples. Additionally, the results show that strength and stiffness increase as fibre volume content increases in most cases but, conversely, the level of increment in mechanical performance is moderate with continued rise in fibre content, particularly in the case of Kevlar® and glass fibres, due to weak bonding between the fibre/nylon layers as well as the presence of increased levels of defects. Finally, the practicality of the results is assessed by testing an evaluation structure.

1. Introduction Additive manufacturing (AM) is one of the most promising areas in the fabrication of components from prototypes to functional structures with complex geometries [1–11]. Compared to conventional methods, AM technologies can shorten the design manufacturing cycle, reduce production costs and increase competitiveness [8–10]. AM technology is a very broad term encompassing numerous methods such as Stereolithography (STL) of a photopolymer liquid, Fused Deposition Modelling (FDM) from polymer filaments, Laminated Object Manufacturing (LOM) from plastic laminations, and Selective Laser Sintering (SLS) from plastic or metal powders [3,12]. However, the FDM technique is of particular interest due to its relative low cost, low material wastage and ease of use [7]. FDM forms a 3D geometry through the deposition of

∗

successive layers of extruded thermoplastic filament, such as acrylonitrile butadiene styrene (ABS), polylactic acid (PLA), polypropylene (PP) or polyethylene (PE). In addition, engineering thermoplastics with improved mechanical performance, such as polyamide (PA or Nylon), polycarbonate (PC), polyetheretherketone (PEEK), polyetherimide (PEI), polyethersulfone (PES) or polyphenylene sulfide (PPS) are also available [13]. Due to this process, delamination of the component layers can occur resulting in premature failure. Additionally, FDM printed parts typically have lower elastic properties than injectionmoulded components of the same thermoplastic [3]. FDM has started to move beyond its initial role as prototyping technology to a process that can build finished parts. However, most of 3D printed polymer products are still used as conceptual prototypes rather than functional components, since pure polymer products built

Corresponding author. E-mail address: [email protected] (M.A. Caminero).

https://doi.org/10.1016/j.compscitech.2019.107688 Received 1 June 2018; Received in revised form 20 May 2019; Accepted 13 June 2019 Available online 16 June 2019 0266-3538/ © 2019 Elsevier Ltd. All rights reserved.

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

have been analysed in the literature: continuous carbon, glass or Kevlar® fibre reinforcements embedded in PLA, ABS or nylon thermoplastic composites [8–10]. The results of the previous studies have shown an increase in mechanical strength of continuous fibre reinforced 3D printed composite structures compared to pure thermoplastic 3D printed structures. However, the level of increase in strength was moderated upon continued increase in fibre content due to weak bonding between the fibre/matrix layers as well as the presence of increased levels of air voids (porosity) with fibre content. Since mechanical properties are crucial for functional parts, it is essential to examine the influence of process parameters on mechanical performance [9,30]. The aim of this research was to gain an inside into a process-property relationship that can help the designers and production engineers in selecting the optimum parameters for the manufacturing of 3D printed continuous fibre-reinforced thermoplastic composites following their requirements. In addition, due to the anisotropic characteristic of additive manufacturing based on FDM, 3D printed parts may present different mechanical behaviour under different orientations. Thus, further research is required to determine printer parameters such as build orientation, layer thickness or fibre volume fraction, since the literature on the mechanical properties of CFRTPCs parts processed by FDM technique is somewhat scarce [9,14,22]. In this study, the characterization and assessment of the effect of build orientation, layer thickness and fibre volume content on the mechanical properties of 3D printed continuous fibre reinforced composites components manufactured by FDM technique are examined. Additionally, manufacturing costs as a function of printing time are evaluated. The performance of the MarkForged® Mark Two system is also evaluated for the fabrication of composites with continuous fibre reinforcements. This study involves the fabrication with Carbon, Kevlar and Glass fibres, provided by MarkForged®, and the mechanical performance of all three composite types are compared. To date, Markforged materials have been assessed individually, and under different testing conditions and standards. This study presents a comprehensive tensile and flexural characterization of these three fibres commercially available for the MarkForged® system under uniform conditions, facilitating their direct comparison. Partially and fully reinforced coupons made of Carbon, Glass and Kevlar fibre filaments manufactured using FDM are characterized. The fibre filaments consisted on a mixture of a bundle of long fibres and resin, forming a preimpregnated-like material. Furthermore, the potential and limitations of 3D printing of continuous fibre-reinforced thermoplastic composites using MarkForged® System are also addressed. In addition, SEM images of fractured surfaces of tensile samples are evaluated to determine the effects of process parameters on failure modes. Finally, the practicality of the previous results is assessed further by testing a functionally static load-bearing assembly as a case example. The rest of the paper is organized as follows. First, the experimental methodology carried out in this study is briefly summarized with particular emphasis on specimen preparation, 3D printing process and the experimental set-up. Thereafter, the key results of the investigation are summarized, and the effects of the different process parameters are highlighted. Finally, conclusions and extensions of this work are outlined.

by FDM are lack of strength and functionality as fully functional and load-bearing parts [14–17]. Furthermore, mechanical properties of parts fabricated by conventional FDM are inherently poor because of the thermoplastic resin used, although the optimization of processing parameters, such as build orientation, layer thickness or feed rate, has been investigated for improving the mechanical properties of thermoplastic parts [7,11,18,19]. Such drawbacks restrict the wide industrial application of 3D printed thermoplastic polymers, leaving prototyping as the primary application [14]. However, there has been an increasing interest in introducing these technologies in the manufacturing of primary structural parts. 3D printing of polymer composites with enhanced mechanical properties solves the previous limitations by the addition of reinforcements, such as particles, fibres or nanomaterials, into thermoplastic polymers permitting the fabrication of polymer matrix composites, which are characterized by high performance and excellent functionality [8,10,20]. However, these composites show poor mechanical properties compared to composites manufactured by conventional methods, because composites reinforced with short fibres or particles are mechanically lower to composites reinforced with continuous fibres. The possibility of employing continuous fibre reinforced thermoplastic composites may lead to product with much higher mechanical performance, which are potentially useful for advanced applications [21]. FDM is a promising alternative of conventional processes for the fabrication of CFRTPCs, such as vacuum forming, filament winding, pultrusion, bladder-assisted moulding or compression, that require expensive facilities and equipment, such as autoclaves or complex rigid moulds for out-of-autoclave processes, hindering the wide application of composites [13,22,23]. The use of FDM for the manufacturing of CFRTPCs has not been extensively investigated in literature [3,9,10,14,16,20,24–26]. Most studies have been focused on the development and characterization of 3D printed composites with short carbon fibre reinforcements [27–35] and only a few with continuous fibre reinforcements [8–10]. However, this technology is still in its infancy. Two possibilities for embedding the continuous fibres into thermoplastic filament have been considered in the literature: embedding the continuous fibre in the injector in a “coextrusion” process [15,20–23,36–42] or embedding the continuous fibre directly in the component with a dual extrusion method. This last method has been implemented by MarkForged® in a successful way [3,14,16,24,43–48]. Co-extrusion (pre-impregnated continuous fibre) does not give the flexibility to change fibre volume fraction, but it eliminates the problem of poor fibre-matrix interface. This is because good impregnation can be performed with proper monitoring and quality control during fabrication of pre-impregnated fibres. On the other hand, remarkable improvement in mechanical performances of 3D printed composites are claimed by MarkForged®, which employs a method involving the dual extrusion of a continuous fibre reinforced composite filament. Matsuzaki et al. [20] and Dickson et al. [14] highlighted that the maximum effective fibre content for FDM composites was about 40–50%. This conclusion was supported by the results of previous works [16,20,36,44,45,47]. In addition, detailed evaluations of the mechanical properties of 3D printed continuous fibre reinforced composites (using Carbon, Glass or Kevlar fibres as reinforcements) are very limited to the authors’ knowledge. These studies are focused and limited in some parameters of this technology, such as material and fibre-laydown pattern. However, there are more parameters to be tested and reported to extend this technology characterization and design input parameters for functional parts. Hence, there is a need to characterize the mechanical properties and interlaminar bonding performance of additively manufactured continuous thermoplastic composites to give product designers a detailed understanding on their characteristics. FDM is a complex process with a large number of parameters that influence product quality and material properties, and the combination of these parameters is often difficult to understand [7,11,49,50]. Different combinations of fibre reinforcements and thermoplastic matrix

2. Experimental methodology 2.1. Materials, 3D printer and specimen preparation The goal of this study is to analyse the mechanical performance of 3D printed CFRTPCs samples. Nylon filament (matrix) was supplied by Markforged® with a diameter of 1.75 mm. Prior to use, this thermoplastic polymer was store in a Pelican® 1430 modified dry box to minimize moisture absorption in the same way as the manufactured specimens prior to testing [7]. The reinforced fibres, such as glass, 2

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 1. SEM images showing details of the individual fibres (5500x). (a) Carbon fibre. (b) Kevlar fibre. (c) Glass fibre.

carbon and Kevlar®, were also supplied by Markforged®. The fibre filaments consisted on a mixture of a bundle of long fibres and a polymer resin (nylon), forming a preimpregnated-like material. The fibre volume fraction of the fibre bundles has been studied and it turned out to be an average value of 34.5%. This value is in good agreement with the results of Klift et at [24] and Justo et al. [16]. The blinder is a polymer resin (nylon) that melts in the print head and ensures that the fibres will adhere to the previous layers. Even though this concept allows for adherence of the fibre bundle, it creates lower mechanical strength than fibre bundles with higher fibre volume fraction. FDM parts have lower fibre volume fractions than the pre-pregs materials used as reference (59.2% in carbon fibre reinforced composites [51] and 65% in glass fibre reinforced composites [16]). The average diameter and density of the carbon fibres bundles were measured to be 350 μm in diameter and a density of 1.4 gr/cm3. In addition, glass and Kevlar fibre bundles were approximately 300 μm in diameter and the density of 1.5 gr/cm3 and 1.2 gr/cm3, respectively. The average individual fibre diameters within these bundles were as follows: Carbon fibres 6.5 μm, Kevlar fibres 9.3 μm and glass fibres 8.5 μm as depicted in SEM images of Fig. 1. Based on the SEM images, the carbon bundles appeared to consist of a higher amount of sizing agent (nylon), thus giving them their larger diameter. The number of fibre per bundle was around 1000 fibres. It is noticed that resin accumulation appears around fibres but porosity was very low. This fact leads to conclude that porosity appears on the printed laminates as a consequence of a non-adequate bonding between filaments. The basic stiffness and strength properties of nylon and continuous fibres used in this study are depicted in Table 1 [52]. In addition, typical ranges of mechanical properties for PLA and ABS materials manufactured by FDM technology are also included for comparative purposes [7,12,53]. 3D printed CFRTPCs samples were manufactured using a

MarkForged® MarkTwo desktop 3D printer. This system can print two kind of materials independently and, hence, it has two extruders and two print heads (Fig. 2). One of the print heads is used to extrude nylon (matrix) and the other one is used to print fibre reinforcement. The nylon and fibre layers are printed with a hot end temperature of 273 °C and 232 °C, respectively, on a non-heated print bed. Carbon fibre was printed in layers of 0.125 mm and Kevlar® and glass fibres were printed in layers of 0.1 mm. The design of the 3D printer allows continuous fibre reinforcement to be placed as required. It allows specifying the fibre orientation on the layer-by-layer deposition process. MarkTwo uses its own software designated as Eiger®. This software allows to import. STL and. OBJ models. There are no standard test methods for determining tensile and flexural properties of CFRTPCs parts manufactured using FDM. In this study, the ASTM D3039 [54] and D790 [55] methods were applied for testing tensile and flexural specimens, respectively. The geometry of the 3D printed specimens were modelled using SolidWorks®, exported as an STL file and imported to the 3D printing software. Two types of fibre pattern can be selected in the MarkForged® MarkTwo desktop 3D printer: concentric and isotropic. In this study, isotropic fibre patterns were analysed. The term “isotropic” does not define the mechanical properties of the test specimens. Isotropic fill is in this case resulted in an unidirectional anisotropic specimen. Furthermore, the isotropic pattern was observed to yield higher tensile strength and stiffness than concentric pattern [14]. The main dimensions and details of layer thickness and build orientation (Flat and On-edge) of the 3D printed specimens are shown in Fig. 3. 2.2. Process parameters The mechanical properties of parts fabricated using FDM technologies depend on the selection of process parameters [7,10,56]. Two

Table 1 Stiffness and strength properties of nylon matrix and continuous fibres provided by the manufacturer MarkForged® [52]. Typical ranges of mechanical properties of PLA and ABS materials fabricated by FDM technology are also included for comparative purposes [7,12,53]. Properties

Nylon

Carbon

Glass

Kevlar®

PLA

ABS

Tensile Strength (MPa) Tensile Modulus (MPa) Tensile strain at Break (%) Flexural Strength (MPa) Flexural Modulus (MPa) Flexural strain at Break (%) Compressive Strength (MPa) Compressive Modulus (MPa) Compressive strain at Break (%) Izod Impact notched (J/m) Density (g/cm3)

54 940 260 32 840 – – – – 1015 1.1

700 54000 1.5 470 51000 1.2 320 54000 0.7 958 1.4

590 21000 3.8 210 22000 1.1 140 21000 – 2603 1.5

610 27000 2.7 190 26000 2.1 97 28000 1.5 1873 1.2

15.5–72.2 2020–3550 0.5–9.2 52–115.1 2392–4930 – – – – 27–192 –

36–71.6 99.8–2413 3–20 48–110 1917–2507 – – – – 47–174 –

3

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 2. Mark Two composite 3D printer system. (a) Details of the printing process. (b) General view of the MarkForged® Mark Two desktop 3D printer.

and fully reinforced specimens are labelled as Type C. A schematic representation of the cross-sectional internal structure of the 3D printed fibre reinforced composite specimen as a function of the build orientation (Flat and On-edge orientations) is shown in Fig. 5. It can be seen several distinct regions exist within the test samples: wall layers from the external structure of the test specimen where the extruded nylon is oriented along the longitudinal axis of the test specimen, roof/ floor layers which consist of closed layers of nylon and alternate between ± 45° from the longitudinal axis, infill layers which consist of sparse layers of nylon and fibre layers which consist on bundles of fibres with a sizing agent (nylon) oriented along the longitudinal axis of the test sample. Table 2 depicts the geometrical data of the 3D printed fibre reinforced samples and values of different parameters used in the calculations of the fibre and nylon volume contents. The fibre volume content (%Vfibre ) of each configuration (Flat and On-edge build orientations) of the test specimen as a function of the number of fibre layers was determined from the geometry of the sample

types of specimens have been considered: unreinforced and continuous reinforced nylon specimens. Two build orientations were assessed: Flat and On-edge, where the fused filament deposition and reinforcements were positioned in the direction of the longest dimension of the 3D printed specimen (Fig. 4a and b). Fig. 4c depicts SEM images showing details of the interfaces between fibre bundles and matrix (nylon). In addition, three different layer thickness (Lt) were assessed for the unreinforced nylon specimens: Lt = {0.1, 0.125, 0.2} mm. Two different layer thickness were used for continuous reinforced specimens: Lt = 0.125 mm for carbon fibres and Lt = 0.1 mm for glass and Kevlar® fibres. The usual minimum layer thickness found in the literature was Lt = 0.1 mm [3]. Additionally, manufacturing costs as a function of printing time were evaluated for both build orientations. Furthermore, different fibre volume contents were considered: partially and fully reinforced specimens. At least two nylon layers must be added in each specimen (roof and floor layers). For ease of reference purposes, partially reinforced specimens are labelled as Types A and B

Fig. 3. Standard specimen for mechanical testing. (a) Tensile specimen. (b) 3-point bending specimen. c) Details of layer thickness (Lt) and build orientations (Flat and On-edge). Dimension are in mm [54,55]. 4

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 4. Process parameters of unreinforced and continuous reinforced nylon specimens: Layer thickness, build orientations and fibre volume content (Type A, B and C). a) Flat specimens with different fibre volume content. b) On-edge specimens with different fibre volume content. c) SEM image showing details of the fibre bundles.

Fig. 5. Schematic representation of the cross-sectional internal structure of the 3D printed fibre reinforced composite specimens as a function of the build orientation (Flat and On-edge orientations). Dimensions are in mm.

5

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Table 2 Geometrical data and fibre filament bundle content of 3D printed composite samples used in this work. Parameter

Value

Parameter

Value

Length (L) Width (W) Thickness (t) Number of wall layers (nwall) Number of roof/floor layers (nroof/floor) Width of wall layer (Wwall)

127 mm 12.7 mm 4 mm 4 4 0.4 mm

Length of wall layer (Lwall) Flat/On-edge Nylon layer thickness (tnylon) Carbon fibre layer thickness (tcarbon) Glass/Kevlar layer thickness (tglass/Kevlar) Width of fibre layer Wlayer (Flat/On edge)

4/12.7 mm 0.1, 0.125 0.125 0.1 11.1/2.4 mm 0.345

Average fibre filament bundle content (%Fibrebundle )

failure mechanism for the failure of the part, with nylon residue remaining on the exposed fibres, suggesting insufficient adhesion between the fibres and matrix. Kevlar fibres presented a visually cleaner fracture surface with minimal residue on the fibre interface, suggesting a weaker bond to the nylon, when compared to carbon and glass fibres. These imperfect fibre-matrix bonds could results in lower strength and stiffness than the expected ones. In addition, there were voids in several regions of the extruded filament. This could be improper impregnation of nylon matrix during the fabrication of fibre composite filament. Another reason could be the lack of consolidation process to enhance the fibre-matrix bonding and to reduce porosity. These results are in agreement with previous works [3,16,20,47]. Certain parameters, such as temperature, feed rate, fill density and pattern of the thermoplastic matrix (nylon) and fibre angle or fibre fill type (isotropic) were fixed for all the samples in order to focus on the influence of the previous parameters (Table 4).

as follows,

VTOTAL = L × W × t Vwall layers = n wall × L wall × Wwall × L Vroof / floor layers = nroof / floor × t fibre layer × Wlayer × L Vinner nylon layers = nnylon layers × tnylon layer × Wlayer × L Vnylon = Vinner nylon layers + Vwall layers + Vroof / floor layer Vfibre = nfibre layers × t fibre layer × Wlayer × %Fibrebundle × L %Vfibre =

Vfibre VTOTAL

where the values of the different parameters used in the previous equations are depicted in Table 2. Table 3 outlines the fibre volume content of partially reinforced samples (Type A and Type B) and fully reinforced samples (Type C) of the different type of fibre reinforcements (Carbon, Kevlar and Glass fibres) as a function of the build orientations (Flat and On-edge orientations). Types A and B were selected with similar fibre volume content in both build orientations (Flat and Onedge) for comparative purposes. Type C depicts different fibre volume content for Flat and On-edge orientations due to different internal structure of the 3D printed nylon specimens as a function of the build orientation. Flat and On-edge samples show different fixed nylon volume content because of their different internal structure, and, hence, different maximum fibre volume content could be achieved. (Fig. 5). These fibre volume contents helped to assess the effect of the fibre content (fibre layers) on the mechanical performance and gained an insight into the failure modes bonding response of fibre-fibre and fibrematrix interfaces. As indicated by the manufacturer, reinforced fibres are composed of multiple strands of carbon, glass of Kevlar fibres coated with nylon matrix. The presence of nylon as binder in fibre filament may help adhesion between consecutive fibre-fibre layers and between fibrenylon layers, however, fibre pull-out and a small amount of fibre breakage was observed when they were subjected to tensile loading. Fig. 6 supports the proposition that fibre pull-out may have a primary

2.3. Experimental set-up Each sample set consisted of five specimens for a given group of process parameters. Average strength and stiffness values of the mechanical test were taken as the results. Since the physical properties of many materials (especially thermoplastics) can vary depending on ambient temperature, tests were carried out according to the standards for room temperature. The uniaxial tensile tests were performed following the standard ASTM D3039 [34]. A 50 kN universal electro-mechanical testing machine with 5 kN and 50 kN load cells at a fixed loading rate of 2 mm/ min was used for both the tensile and 3-point bending tests. This displacement rate was within the proposed ASTM test speed range: D3039 and D790 propose a test speed range of 1–5 mm/min. The selection of this displacement rate was in agreement with the displacement rate used in other studies [1,5–7,53]. Strain was measured using a MTS 634.14 high-performance axial extensometer. The experimental data were processed following the recommendations of the previous standard, for the determination of the maximum tensile strength (σt) and the tensile Young's modulus (Et). Young's modulus was determined considering the linear part of the stress-strain curve and the slope was

Table 3 Printing parameters and their levels used in this work. Parameter

Value

Layer thickness of unreinforced samples (mm) Layer thickness of fibre reinforced samples (mm) Fibre volume content (%) (nfibre layers/nlayers)

Lt = {0.1, 0.125, 0.2} LtCARBON = 0.125, LtGLASS = LtKEVLAR = 0.1 Fibre volume content type Build Orientation Type A Carbon fibre Flat 1.88 (2/32) On-edge 2.03 (28/102) ® Kevlar fibres Flat 3.01 (4/40) On-edge 2.93 (18/127) Glass fibre Flat 3.01 (4/40) On-edge 2.93 (18/127)

6

Type B

Type C

15.07 (16/32) 15.06 (74/102)

26.38 (28/32) 19.96 (98/102)

15.07 (20/40) 14.99 (94/127)

27.13 (36/40) 20.04 (123/127)

15.07 (20/40) 14.99 (94/127)

27.13 (36/40) 20.04 (123/127)

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 6. SEM images showing details of the fibre-matrix interface of different bundles of fibres. (a) Carbon fibre. (b) Kevlar fibre. (c) Glass fibre.

and for reinforced nylon samples in Tables 6 and 7. In order to clarify the results for a better understand of the correlation between the different ranges of process parameters on mechanical performance, the graphical representation of these results is shown in Figs. 7 and 8. These figures show the average maximum tensile and flexural strengths as a function of layer thickness and fibre volume content, respectively. In addition, Figs. 9 and 10 report some representative stress-strain curves for the tensile and flexural samples with different process parameters in order to characterize and assess different types of damage observed. Finally, Table 8 depicts the printing time of tensile samples as a function of the process parameters. Printing time is a key material process parameter because it is related to manufacturing cost. The main effects of the process parameters on the mechanical characteristics of 3D printed samples are summarized in the following sections.

Table 4 Fixed parameters.

Unreinforced samples

Reinforced samples

Print parameters

Value

Fill density (%) Fill pattern Roof and floor layers Fibre fill type Fibre angle

100 Rectangular 2 Isotropic 0°

estimated by a linear fit. σt was calculated as a ratio between the maximum load reached during the test and the cross-sectional area. According to the previous standard, the specimens must include some tabs at their ending zones in order to avoid any damage that may be produced by the gripping system. In addition, the results of 3D printed composites tests of previous works [16,24] reported some problems with discontinuities (start and end points of the fibre reinforcement). The use of tabs overcomes this issue and the specimens were not modified between printing and testing. The 3-point bending tests were performed following the ASTM D790-10 procedure [55] using a three point bending test fixture. The radius of the loading nose and the radii of the support noses of the 3point bending specimen test fixture were 3 mm. The flexural modulus of elasticity (Ef) was determined following the previous standard, based on the Classical Beam Theory, supposing that shear effects are negligible. We can define the maximum normal stress σf in the three-point bending test as

3.1. Effects of build orientation and layer thickness on the mechanical response of unreinforced nylon specimens A first glance at the results of Table 5 and Fig. 7 reveal that build orientation and layer thickness marginally affected the mechanical resistance of the unreinforced nylon samples. Layer thickness is directly related to the number of layers needed to print a part and hence to printing time. Thus, manufacturing costs decrease as layer thickness increases. The variations of maximum tensile and flexural strength and stiffness in the range of layer thickness Lt = {0.1, 0.125, 0.2} mm were marginally significant. Flat orientation showed the highest values for maximum tensile and flexural strength and stiffness in most cases. For example, the maximum tensile strength and stiffness for the flat orientation were about 8.4% and 4.5% higher than on-edge orientation. In this case, the specimens were pulled perpendicular to the layer deposition direction and hence layers were pulled parallel to the loading direction, resulting in trans-layer failure [7]. Individual layers withstood most of the applied load and layer breakage was observed [6,37]. These results have confirmed the observations of previous studies [4,5,7]. Fig. 9 show the average stressstrain behaviour for the unreinforced samples as a function of layer thickness and build orientation under tensile and flexural loading. In general, the results highlighted a ductile behaviour for both orientations, with significant plastic deformation. In the case of the tensile response of unreinforced samples, the results showed high tensile strain to failure for most cases, where the variations of maximum tensile to failure with layer thickness and build orientation were marginally significant. On the other hand, in the case of the flexural response of unreinforced samples, the results showed that flexural strain to failure was lower that tensile strain to failure, particularly in the case of low values of layer thickness. In this case, a larger number of nylon layers were needed for a given total thickness, therefore the number of layer bonds was increased, and a reduction of mechanical performance was expected. The effect of layer thickness and build orientation in the case of flexural response were also slightly significant, except for maximum

3PL 2wt 2

σf =

where P is the fracture force, L is the support span, w is the width of the specimen, t is the thickness of the specimen, and the maximum strain ε of the outer surface at mid-span, which was calculated as

ε=

6δt L²

where δ is the mid-span deflection. The flexural modulus of elasticity Ef is the ratio of stress to the corresponding strain at a given point on the stress-strain curve. Hence, it can be calculated as

Ef =

L³m 4wt ³

where m is the slope of the secant of the load-displacement curve. The slope m was measured between the 25% and 75% of the maximum load. It is also recommended to prescribe a strain range instead of a load range. If the ultimate strain of the material is higher than 1%, a 0.3–0.5% strain range or even higher can be used [57]. 3. Results and discussion Average and standard deviation of the test results of the maximum strength (σt, σf) and stiffness (Et, Ef) for the 3D printed composites samples are tabulated for the unreinforced nylon samples in Table 5, 7

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Table 5 Average tensile and 3-point bending test results of unreinforced nylon samples and process parameter ranges. Standard deviation is depicted in brackets. Average tensile test results of nylon samples Layer thickness

Build orientation

Lt = 0.1 mm σt (MPa)

Flat 30.9 (0.9) On-edge 28.2 (0.2) Average 3-point bending test results of nylon samples

Et (MPa)

Lt = 0.125 mm σt (MPa)

Et (MPa)

Lt = 0.2 mm σt (MPa)

Et (MPa)

412.3 (27.6) 453.1 (20.6)

29.1 (0.1) 27.2 (0.1)

453.2 (6.4) 425.1 (6.3)

28.4 (0.2) 28.5 (0.2)

473.3 (22.4) 444.1 (25.7)

Build orientation

σf (MPa)

Ef (MPa)

σf (MPa)

Ef (MPa)

σf (MPa)

Ef (MPa)

Flat On-edge

20.9 (0.3) 19.8 (0.1)

512.3 (19.6) 433.2 (22.7)

21.3 (0.2) 20.2 (0.3)

521.4 (7.8) 486.3 (5.2)

20.8 (0.1) 20.5 (0.4)

524.6 (14.5) 478.7 (11.4)

3.2. Effects of type of fibre reinforcement, build orientation and fibre volume fraction on the mechanical response of reinforced nylon specimens

flexural strain to failure, where an increase was observed with the increase of layer thickness. This effect requires a more detailed study in further research work. If we focus on the tensile stress-strain curves of nylon samples, there are not significant differences between tensile samples as a function of layer thickness and build orientation in terms of ductility. Fig. 11 depict SEM images showing details of the tensile fractured surfaces of unreinforced nylon samples at different build orientations and layer thickness. It cannot be claimed that there are any differences of fracture behaviour. Flat oriented samples showed better mechanical performance in terms of tensile strength and stiffness. In the case of the flexural samples, the variation of flexural strength with the process parameters was marginally significant. However, the maximum flexural stiffness for the flat orientation was about 7.9% higher than on-edge orientation. In addition, on-edge and flat orientation showed lower values of flexural strengths than tensile ones, compared with the flexural response of the nylon matrix. The different mechanical performance can be explained by considering the internal structure of the 3D printed nylon specimen as a function of the build orientation (Flat and On-edge). In the case of flat samples, fewer layers were needed for a given total thickness, therefore the number of layer bonds was reduced, and strength increased. In addition, the results show higher strength for the tensile test than the flexural test in most cases. This may be explained as an effect of the 3D printed nylon samples response under tensile and flexural loads. In the case of flexural loading, there is a combination of tensile and compressive response. Compressive strength of unreinforced 3D printed materials is expected to be lower than tensile strength [4,7,12]. Hence, compressive failure is expected to occur before tensile failure and, therefore flexural strength is expected to be lower than tensile one.

The bonding performance of the interface between fibres and matrix is a main factor for the mechanical performance of 3D printed composite structures. The maximum achievable strength of 3D printed fibre reinforced composites is limited by the fibre-matrix interaction in porous areas [10,14,23,25]. The role of type of fibre reinforcement (carbon, glass or Kevlar®), build orientation and fibre volume fraction on the mechanical strength and stiffness of 3D printed composite samples were analysed. It is important to note that, during the printing procedure, no pressure was applied after a layer was laid up. It is well known that pressure plays a fundamental role on the manufacturing of laminated parts from both thermoset and thermoplastic-based composites. The absence of pressure in highly viscous matrices is directly related to the presence of defects (pores or matrix-dominated areas) [14,16], [25,26,47]. Tables 6 and 7 and Fig. 8 depict the average and standard deviation of the maximum strength and stiffness for the 3D printed reinforced nylon samples. The results reveal that the type of fibre reinforcement, build orientation and fibre volume content significantly affected the mechanical performance of the 3D printed composite samples. The differences on mechanical properties of 3D printed composite samples due to the type of fibre reinforcement can be explained by the differences on mechanical properties of the individual continuous fibres. Table 1 depicts the mechanical performance of the nylon matrix and the individual continuous fibres. More specifically, it is observed that carbon fibre showed the best performance in terms of tensile and flexural strength and stiffness, while glass and Kevlar® fibres resulted in lower ones with similar values. Carbon fibres are known for their high stiffness-to-weight ratio but are expensive and, hence they are mainly used in industries such as automotive or aerospace industry. On the other hand, glass and Kevlar® fibres are relatively inexpensive, and

Table 6 Average tensile test results of reinforced nylon samples and process parameters ranges. Standard deviation is depicted in brackets. Average tensile test results of reinforced nylon samples Build orientation

Carbon fibre Flat On-edge Kevlar® fibre Flat On-edge Glass fibre Flat On-edge

Type A

Type B

Type C

σt (MPa)

Et (GPa)

σt (MPa)

Et (GPa)

σt (MPa)

Et (GPa)

96.6 (1.3) 63.9 (0.5)

7.6 (0.1) 5.2 (0.1)

239.8 (14.9) 177.2 (7.9)

25.3 (0.8) 17.6 (0.5)

436.7 (6.2) 341.2 (10.2)

51.7 (0.6) 31.6 (1.4)

97.1 (2.1) 55.8 (1.9)

5.2 (0.1) 2.3 (0.4)

211.7 (4.9) 127.5 (7.1)

15.1 (0.1) 8.5 (0.5)

305.2 (36.2) 235.6 (2.1)

25.5 (0.7) 15.9 (0.2)

113.4 (2.5) 66.8 (2.3)

3.7 (0.1) 1.9 (0.1)

235.9 (11.0) 146.2 (8.1)

10.3 (0.2) 5.7 (0.1)

381.2 (28.1) 295.2 (23.5)

19.6 (0.6) 11.8 (0.4)

8

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Table 7 Average 3-point bending test results of reinforced nylon samples and process parameters ranges. Standard deviation is depicted in brackets. Average 3-point bending test results of reinforced nylon samples Build orientation

Carbon fibre Flat On-edge Kevlar® fibre Flat On-edge Glass fibre Flat On-edge

Type A

Type B

Type C

σf (MPa)

Ef (GPa)

σf (MPa)

Ef (GPa)

σf (MPa)

Ef (GPa)

80.7 (6.7) 38.1 (0.4)

6.1 (0.1) 1.6 (0.1)

355.6 (35.1) 122.3 (8.4)

31.1 (0.5) 6.9 (0.1)

423.5 (10.3) 157.1 (5.5)

39.2 (0.3) 12.1 (0.2)

85.3 (11.3) 37.9 (1.6)

3.7 (0.1) 0.9 (0.2)

162.1 (3.2) 61.7 (1.2)

12.4 (0.3) 2.1 (0.2)

189.8 (2.5) 73.8 (2.1)

14.1 (0.5) 2.8 (0.2)

121.1 (21.5) 37.5 (3.4)

8.2 (0.2) 1.1 (0.1)

170.7 (23.8) 71.2 (2.6)

15.1 (0.3) 2.4 (0.1)

205.1 (6.4) 97.9 (4.6)

16.2 (0.1) 3.2 (0.3)

Fig. 7. Graphical comparison of average maximum (a) tensile and (b) flexural strength of unreinforced nylon specimens. Effects of build orientation and layer thickness.

Fig. 8. Graphical comparison of average maximum (a) tensile and (b) flexural strength of reinforced nylon specimens. Effects of fibre reinforcement, build orientation and fibre volume content.

accordance with the results of 3D printed reinforced samples depicted in Tables 6 and 7 The advantage of using carbon fibre to reinforce a 3D printed part was further evident from the flexural results, as the stiffness of the samples reinforced with carbon fibre are twice that of

exhibit fairly good mechanical properties and they are suitable for parts that are less stringent on weight and strength, so that parts can be manufactured at lower cost. In addition, Kevlar® fibres are well known for its enhanced impact properties. These observations were in good 9

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 9. Average stress-strain curves for the unreinforced tensile and flexural samples as a function of layer thickness and build orientation.

Kevlar® and three times that achieved using glass fibre. For example, an increase in tensile and flexural strength was observed from nylon samples to Type C flat composite samples for the three fibre reinforcements, increasing by 1313% and 1888% for carbon fibre, 887% and 791% for Kevlar® fibre and 1133% and 863% for glass fibre samples. These improvements in mechanical performance are even higher

in terms of stiffness (Tables 6 and 7). Tensile and flexural testing demonstrated strengths of up to 436 MPa for tensile and 423.5 MPa for flexural loading with Type C, which are higher that the reported results of some non-ferrous metal alloys, such as Aluminium 6061-T6 [14]. Fig. 10 shows the average stress-strain behaviour for 3D printed reinforced samples under tensile and flexural loading. The stress-strain 10

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 10. Average stress-strain curves for the reinforced tensile and flexural samples as a function of fibre reinforcement, build orientation and fibre volume content.

quantity of fibre reinforcement increased. Moreover, in the case of onedge oriented samples, a larger number of fibre layers were needed for a given fibre volume content, therefore a reduction of mechanical performance was expected. This fact was heightened in the case of flexural performance. These improvements in mechanical performance with fibre volume content were expected because fibres have much greater elastic modulus and ultimate strength than nylon. Therefore, the addition of fibres resulted in an increase in the effective properties of the 3D printed reinforced samples. In addition, the stress-strain curves depicted non-linear behaviour for low stress and strain values. This non-linear behaviour of the 3D printed reinforced specimens could be caused by the initial adjustment between the gripping system and

plots demonstrate the effect of fibre volume content and build orientation on the mechanical behaviour on 3D printed composites. We must consider jointly both process parameters. Without taking into account the former relationship, a strict relationship between mechanical performance and fibre volume content cannot be claimed. In we focus on the performance of carbon fibre reinforced samples, in the case of partially reinforced samples (Type A and Type B), the results showed that flat oriented samples were better in terms of mechanical performance than on-edge samples. Furthermore, in the case of fully reinforced samples (Type C), flat oriented samples were able to achieve higher fibre volume content due to their internal structure. In this situation, tensile and flexural strength and stiffness increased as the 11

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

specimens (at the centre and close to the tabs) perpendicular to the loading direction, generating cracks in the direction parallel to the fibres. These observations were in accordance with previous studies [16,47]. In addition, Fig. 12 shows SEM images with the details of the tensile fractured surfaces of flat oriented carbon fibre reinforced samples as a function of the type of the fibre volume content. For example, an increase of fibre content from Type A flat samples to Type C flat samples tended to promote higher tensile and flexural strength (352% and 424.7%, respectively). In the case of Kevlar® and glass fibres, increasing fibre content from Type A to Type C depicted an increasing trend of tensile and flexural strength, but to a lesser extent than carbon fibre reinforced samples (214% and 122% for Kevlar® and 236% and 125% for glass fibres samples). In addition, Type A glass fibre reinforced samples showed a more efficient mechanical performance than carbon and Kevlar® reinforced samples, with higher strength and stiffness. However, the level of increase in strength and stiffness for the different fibre reinforcements was moderated from Type B to Type C upon continued increase in fibre volume content. This effect was more remarkable in the case of Kevlar® and glass fibre reinforced samples. The addition of fibres had two opposite effects. On one hand, it could prevent crack propagation efficiently, which contributed to the improvement of the strength and stiffness of the reinforced samples. On the other hand, the increase of fibre content could increase the defect density due to weak bonding between the fibre/matrix layers as well as the presence of increased levels of air voids (porosity) with fibre content. In addition, fibre reinforced samples showed poor wettability of bundles by the nylon and, hence, weaker bonding between fibres and matrix (nylon). These results were in agreement with previous findings [14,47]. In addition, Fig. 13 depicts SEM images of Type C flat samples showing some details of the fractured surfaces as a function of fibre reinforcement. It is worth noting that carbon fibre reinforcement presented the highest mechanical performance, leading to a more brittle behaviour (Table 1). Extensively fibre pull-out and minor fibre breakage was observed. The extensive pull-out indicated improper coating of nylon matrix onto the fibre surface during the manufacturing of composite filament and poor interfacial bonding between matrix and fibres. The internal fibres were hardly infiltrated by the thermoplastic matrix and a limited number of external fibres could adhere to the matrix (nylon), suggesting a weaker bond to the nylon. These results were in agreement with previous studies [14,47]. Fig. 14 shows SEM images showing details of the fractured surfaces of Type C carbon fibre reinforced samples as a function of the type of build orientation. As a general remark, flat oriented samples depicted higher tensile strength than on-edge ones for the same fibre volume

Table 8 Printing time of tensile reinforced samples as a function of the process parameters. Printing time (min) Build orientation Carbon fibre Flat On-edge Kevlar® fibre Flat On-edge Glass fibre Flat On-edge

Type A

Type B

Type C

61 73

71 85

74 87

76 92

88 108

94 115

76 91

88 102

94 115

the samples and, additionally, the manufacturing process. Fibres were embedded into the test specimen in specific regions of the 3D printed part. However, during the fibre embedding process, tension was not applied to the fibre strands, as a result fibre waviness could occur in the fibre strands. Fibre waviness could affected the mechanical properties of the 3D printed parts as the embedded fibres were not entirely aligned with the loading axis of the test samples. In addition, it was observed that the stress increased almost linearly with the strain before breaking abruptly, indicating brittle nature of the composite. The results showed that the tensile performance of reinforced samples was better than their flexural performance in terms of strength and stiffness. This may be explained as an effect of the 3D printed composite response under tensile loads, in which the continuous fibres support the tensile stress. In this case, tensile behaviour is a fibre-dominated property, and the brittle nature with low tensile strain to failure of fibre reinforcements was observed in the experimental results of Fig. 10. On the other hand, under flexural loads, there is a combination of tensile and compressive response. The nylon matrix avoided fibre buckling effects and supported the compressive stress. In this case, compressive strength is also a matrix-dominated failure mode. Compressive strength of 3D printer continuous reinforced composites was expect to be lower than tensile strength. Hence, flexural strength was expected to be lower than tensile one, with higher strain to failure due to the ductile nature of polymer matrix that support compressive loads. For tensile coupons of this kind, an energetic failure could be expected, due to alignment of the fibre with the loading direction. The failure occurred with a break almost perpendicular to the loading direction in the case of carbon fibre coupons. It may be explained by the presence of internal defects in the specimens. In the case of glass and Kevlar® fibres, the failure mechanisms were different. Failure was observed at several places of the

Fig. 11. SEM images showing details of the tensile fractured surfaces of unreinforced nylon samples at different build orientations and layer thickness (Lt) (x140): (a) Lt = 0.1 mm, (b) Lt = 0.125 mm, (c) Lt = 0.2 mm. 12

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 12. SEM images showing details of the fractured surfaces of tensile composite samples (carbon fibre, flat orientation) as a function of the fibre volume content of 140x: (a) Type A, (b) Type B, (c) Type C.

content. Moreover, flat oriented samples showed the highest flexural strength for almost all fibre volume contents, except Type A carbon fibre flat samples. The different mechanical performance can be explained by considering the internal structure of the 3D printed nylon specimen as a function of the build orientation (Flat and On-edge) for the same fibre volume content. In the case of flat samples, fewer layers were needed for a given total thickness, therefore the number of layer bonds was reduced, and strength increased. For example, flat Type C reinforced samples depicted an averaged increase in tensile strength and stiffness of 28.8% and 63.4%, respectively, compared with on-edge ones. In the case of flexural performance, a further increase in average flexural strength and stiffness between flat and on-edge orientations was observed, increasing by 169.5% and 224% for carbon fibre, 157% and 404% for Kevlar® fibre and 109% and 406% for glass fibre samples. Finally, in order to study the use of 3D printed composite materials in functional applications, Fig. 15 shows a graphical comparison of the tensile strength and stiffness of continuous fibre reinforced nylon composites with other common 3D printed thermoplastic and prepreg composites with similar fibre volume content (thermoset-matrix composite) manufactured with vacuum bag and autoclave [3,7,11,14,16,20,21,24,29,47,51,58]. The prepreg material was M21/ IMA carbon fibre/epoxy composite with 59.8% fibre volume fraction, an unidirectional prepreg used in Airbus A350XWB primary structures (wing spars and wing covers, fuselage sections, keel beam and central wing box) [38–40]. Note that this is a qualitative comparison, since the polymeric matrix and the fibres are different for 3D printing (thermoplastic) and prepreg manufactured composites (thermoset). The comparison shows that the properties of 3D printed carbon fibre continuous reinforced composites are poor compared with conventional prepreg materials. The reasons are the differences of the raw materials (matrix and fibres) compared with the epoxy matrix and carbon fibres of prepreg composites and the manufacturing process itself, without any compaction stage between filaments laid-up and even between layers (porosity). However, the use of carbon, glass or Kevlar® fibres as reinforcements greatly increased the mechanical performance of 3D printed reinforced composites and, in some cases substantially higher than conventional metal alloys [14]. Despite the poor properties of the 3D printed carbon fibre continuous reinforced composite material with respect to traditional

prepreg composites, the 3D printed continuous reinforced composites greatly improves the usual mechanical performance of traditional 3D printing parts or reinforced with short fibres and particles and in some cases over conventional metal materials. Moreover, another advantage of continuous fibre reinforced 3D printing compared to the unreinforced thermoplastic printing is the stability that the fibre gives to the parts, reducing and even avoiding the shape distortions that usually appear in 3D printed parts [16]. 3.3. Effects of process parameters on printing time of reinforced nylon specimens Table 8 depicts printing time as a function of build orientation, fibre reinforcement and fibre volume content. As discussed above, printing time was directly related to manufacturing costs and hence it is crucial to bear this in mind. It is also clear from the previous results that printing time was directly related to the build orientation. Reinforced on-edge samples had longer printing time than flat samples. The reason was the number of fibre/nylon layers for a given thickness. In the case of reinforced on-edge samples, more fibre/nylon layers are required for the same thickness, increasing the printing time. Hence, a logical recommendation is to print flat in order to save time and money, with improved mechanical performance. 4. Analysis of a functional structural assembly A two-part arbitrary functional structural assembly was manufactured to demonstrate the practicality of the previous results. This assembly consists on a two-part quick release buckle [59] and it was selected as a functional static load-bearing structural component. Fig. 16a depicts the 3D geometry of the proposed structural assembly and its main dimensions. Tensile tests were performed to evaluate the functional structural assembly printed with the settings mentioned in Table 9. Two types of structural assemblies were considered: unreinforced and continuous reinforced nylon structural assemblies with carbon, glass and Kevlar® fibres as reinforcements. Fig. 16b depicts the test set-up for determining the maximum strength of the assembly. Owing to the complex geometry of the selected functional structure, the maximum load at failure was considered

Fig. 13. SEM images showing details of the fractured surfaces of tensile composite samples (flat orientation, Type C) as a function of the type of fibre reinforcement of 140x: (a) Carbon fibre, (b) Kevlar® fibre and (c) Glass fibre. 13

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 14. SEM images showing details of the fractured surfaces of tensile composite samples (carbon fibre, Type C) as a function of the type of build orientation of 140x: (a) Flat orientation, (b) On-edge orientation.

printed with nylon. Several nylon layers can protect the inner fibres in the part. Because of layer-by-layer FDM technology, build orientation should be considered during the design stage. Finally, another limitation of the FDM process was caused by slicing software Eiger®. In short, the main shortcomings of the current MarkForged® dual extrusion method are presented below:

for strength comparison. Each structural assembly set tested consisted of five samples for a given group of process parameters. Table 10 shows the average and standard deviation of the maximum load at failure for the selected combinations of build orientation, layer thickness and fibre reinforcement. As a general remark, the assembly tensile performance was in agreement with the coupons results and followed the general trends observed for the unreinforced samples, i.e., flat samples depicted higher strength and minor differences in the assembly tensile performance as a function of layer thickness was observed. In the case of the reinforced samples, the addition of fibres reinforcement contributed to the improvement of the strength of the reinforced samples compared to unreinforced ones. In addition, flat carbon fibre reinforced samples depicted the best mechanical performance. These results were in accordance with the coupon ones. Nevertheless, the assembly tensile performance could be affected by the complex geometry of the functional assembly and the combination of different stress states induced during the test. Hence, the combination of build orientation, layer thickness, and fibre reinforcement had a significant effect on the final strength (Fmax) of the functional assembly, showing variations in terms of average values between different process parameter sets in the range of 813.7–2825.1 N, i.e., an increase in strength of 247.2%.

1 Low bonding strength between fibres and matrix layers. 2 Porosity (air inclusions) due to the absence of pressure during the filament deposition. 3 Low properties of the raw materials of the matrix and fibre filaments and their low fibre volume fractions compared with pre-pregs conventional materials. Nylon has lower mechanical properties than the epoxy resin and the fibres used in FDM are low modulus and low strength fibres [47]. MarkForged supplies fibre bundles specifically created for this system, but they have shortcomings. The fibre volume fraction of the fibre bundles had an average value of 34.5%. The blinder is a polymer resin (nylon) that melts in the print head and ensures that the fibres will adhere to the previous layers. Even though this concept allows for adherence of the fibre bundle, it creates lower mechanical strength than fibre bundles with higher fibre volume fraction. 4 The Eiger® Software: a Eiger does not allow a print to start with a fibre layer, nor to end with one. The outer layers need to be of nylon. b Layer thickness of Glass and Kevlar fibres is fixed at 0.1 mm, while the layer thickness of Carbon fibre is fixed at 0.125 mm. c The maximum and minimum size of the reinforced samples. There is a minimum area that can be reinforced with fibres (A > 100 mm2). d The outer layers of the part need to be printed with nylon (wall layers). e The temperature and print head velocity are predetermined and cannot be customized. The average temperatures are 273 °C for

5. Main shortcomings of the current MarkForged® dual extrusion technology In the design of 3D printed continuous fibre reinforced thermoplastic composite parts, some guidelines should be followed. Firstly, the bonding between fibres layers and matrix layers should be considered during the design process. The bonding strength was relatively weak compared to the strength of fibres [25]. The failure was more likely caused by debonding between layers and the formation of cracks between fibres and polymer matrix [60]. The second issue was the manufacturing process itself. The outer layers of the part needed to be 14

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 15. Graphical comparison of the (a) tensile strength and (b) stiffness of different 3D printed thermoplastic materials: Unreinforced samples1, Continuous reinforced samples2 and prepreg thermoset-matrix samples3.

the nylon filament and 232 °C for fibre filament in order to melt the fibre filament resin, which acts as a binder between the fibres and the matrix. Hence, there is no temperature control of the print head, making it impossible to heal to higher temperatures than 273 °C. That means that the printer would be unable to use matrix

material with melting temperature higher than this value, as for example PEEK (343 °C). In addition, polymer materials with a lower melting temperature will cause problems, limiting the ranges of materials for this system. f The toolpath of fibres cannot be customized. 15

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

Fig. 16. Functional load-bearing assembly [41]. (a) Details of the main dimensions. (b) Test set-up. Dimensions are in mm.

3 Include a compaction stage after the deposition of the filaments, in order to reduce void content. 4 Use a heated print bed. By increasing the print bed temperature, the part can be annealed during the printing, reducing the residual stress and void content. 5 Annealing parts in an oven after printing to improve the mechanical properties due to reducing the residual stresses. 6 Further research focused on impact performance would give a greater insight in the possibilities of 3D printed continuous fibre reinforced composite materials [26].

Table 9 Printing settings used in the functional load-bearing structural assembly. Print parameter

Value

Build orientation Layer thickness of nylon samples (mm) Layer thickness of reinforced samples (mm) Fibre reinforcement

Flat, On-edge Lt = {0.1, 0.125, 0.2} LtCARBON = 0.125, LtGLASS = LtKEVLAR = 0.1 Carbon, Kevlar®, Glass

Table 10 Average maximum load of tensile test results of the unreinforced and reinforced functional assembly with different process parameter ranges. Standard deviation is depicted in brackets.

6. Conclusions The mechanical performance of 3D printed continuous fibre reinforced thermoplastic composites was studied. Continuous glass, carbon and Kevlar® reinforced nylon composites were manufactured by FDM technology. Tensile and three-point bending tests were carried out to determine the mechanical response of the printed specimens following the ASTM standard recommendations. Different ranges of three main process parameters were analysed: layer thickness of unreinforced samples (Lt = {0.1, 0.125, 0.2} mm), build orientation (flat and onedge) and fibre volume content (Types A, B and C). The practicality of the results was assessed by testing an evaluation structure. It has shown that the effect of layer thickness on the mechanical performance of nylon specimens was marginally significant. It was observed that flat orientation showed the highest values for maximum tensile and flexural strength and stiffness in most cases. Moreover, the effects of build orientation, the type of reinforcement and fibre volume content on the mechanical performance of reinforced nylon specimens were of particular significance. Flat samples exhibited higher values of strength and stiffness than on-edge samples. Furthermore, carbon fibre reinforced composites exhibited the best tensile and flexural strength performance with higher stiffness. On the other hand, Kevlar® fibre reinforced composites had the lowest mechanical performance. By reason of the anisotropic nature of Kevlar® fibre, the compressive strength was much lower than that of glass and carbon fibres. Thus, its composite easily failed due to the failure of fibres under compression (and hence flexural performance was also lower). Extensively fibre pull-out and minor fibre breakage was observed for the fibre reinforcement. The extensive pull-out indicated

Nylon samples Layer thickness Build orientation

Lt = 0.1 mm Fmax (N)

Lt = 0.125 mm Fmax (N)

Lt = 0.2 mm Fmax (N)

Flat On-edge Reinforced samples

1204.7 (14.2) 884.2 (15.2)

1181.4 (9.6) 837.8 (13.2)

1140.7 (12.1) 843.7 (16.1)

Build orientation

Kevlar® Fmax (N)

Glass Fmax (N)

Carbon Fmax (N)

Flat On-edge

2145.7 (18.6) 1263.1 (20.1)

2456.1 (24.6) 1489.2 (27.2)

2825.1 (21.3) 1745.7 (16.8)

g Some discontinuities in the printed fibre In view of the obtained results, some ideas that will help in the improvement of the FDM composite mechanical performance are proposed: 1 Improving the software: temperature and velocity control, so that more types of matrix with better properties (PEEK, PEI, PEK, …) and fibre materials can be used by the printer. Use different printing velocities and extruding temperatures to analyse the void content. 2 Produce nozzles and filament to achieve a higher fibre content. 16

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

improper coating of nylon matrix onto the fibre surface during the manufacturing of composite filament and poor interfacial bonding between matrix and fibres. Finally, the results have shown that strength and stiffness increased as fibre volume content increased, depending on build orientation. Nevertheless, the level of increase in mechanical performance was moderate with continued increase in fibre content, particularly in the case of Kevlar® and glass fibres. This was in part due to weak bonding between the fibre/nylon layers as well as the presence of increased levels of defects. Hence, in the light of the previous conclusions, a logical recommendation was to print flat in order to save time and money, with improved mechanical performance. The results have shown that it is still a challenge to increase the mechanical performance of 3D printed continuous fibre reinforced composite materials with respect to the autoclave manufacturing of common prepreg materials. A compaction stage after the deposition of the filament would be desirable to reduce porosity. FDM composite parts have lower fibre volume fractions than the pre-pregs materials used as reference [16,51]. These were the main causes for the differences found in tensile and flexural properties. Nevertheless, the properties obtained by 3D printed continuous reinforced composites by FDM are significantly higher than the usual 3D FDM thermoplastics. In conclusion, 3D printing of continuous fibre-reinforced parts complements the existing composite manufacturing processes by taking advantage of the complexity that FDM enables in the design of composite parts. Hence, it seems to be a very promising procedure for the manufacturing of 3D printed continuous fibre-reinforced composite parts once some of the current disadvantages are overcome. It is a relatively new technique and there is a lack of experimental data on the mechanical performance of structures manufactured by this process, underscoring the need for further research to improve our understanding of the mechanical behaviour of 3D printed composites. In addition, further work is required for the characterization of complex parts made with this technology and the study of a treatment to reduce the porosity and low interlayer bonding strength of the parts and maybe including a compaction stage.

[7]

[8]

[9] [10] [11] [12] [13]

[14]

[15] [16] [17] [18]

[19]

[20]

[21] [22]

[23]

[24]

[25]

Acknowledgements

[26]

This research was supported by the Spanish Ministerio de Economía y Competitividad (Plan Nacional de I+D+i), under research grants DPI2016-77715-R and DPI2015-65472-R, co-financed by the ERDF and Grants no. GI20163590 and GI20174156 financed by University of Castilla-la Mancha. I García-Moreno would like to acknowledge the financial support of the Castilla-La Mancha Government (JCCM) and the ERDF (SBPLY/16/180501/000041). J.M. Reverte also would like to acknowledge the financial support of Spanish Ministerio de Economía y Competitividad (Plan Nacional de I+D+i) under research grant BES2016-076639. We also would like to acknowledge the referees for their suggestions, which improved the quality of this article.

[27]

[28]

[29]

[30]

[31]

References [32]

[1] B.M. Tymrak, M. Kreiger, J.M. Pearce, Mechanical properties of components fabricated with open-source 3D printers under realistic environmental conditions, Mater. Des. 58 (2014) 242–246. [2] M. Sugavaneswaran, G. Arumaikkannu, Analytical and experimental investigation on elastic modulus or reinforced additive manufactured structure, Mater. Des. 66 (2015) 29–36. [3] G.W. Melenka, B.K.O. Cheung, J.S. Schofield, M.R. Dawson, J.P. Carey, Evaluation and prediction of the tensile properties of continuous fiber-reinforced 3D printed structures, Compos. Struct. 153 (2016) 866–875. [4] M. Domingo, J.M. Puigriol, A.A. Garcia, J. Lluma, S. Borros, G. Reyes, Mechanical property characterization and simulation of fused deposition modeling polycarbonate parts, Mater. Des. 83 (2015) 670–677. [5] C. Casavola, A. Cazzato, V. Moramarco, C. Pappalettere, Orthotropic mechanical properties of fused deposition modelling parts described by classical laminate theory, Mater. Des. 90 (2016) 453–458. [6] B. Rankouhi, S. Javadpour, F. Delfanian, T. Letcher, Failure analysis and

[33]

[34]

[35]

[36]

[37]

17

mechanical characterization of 3D printed ABS respect to later thickness and orientation, J. Fail. Anal. Prev. 16 (2016) 467–481. J.M. Chacón, M.A. Caminero, E. García-Plaza, P.J. Nuñez, Additive manufacturing of PLA structures using fused deposition modelling: effect of process parameters on mechanical properties and their optimal selection, Mater. Des. 124 (2017) 143–157. T. Hofstatter, D.B. Pedersen, G. Tosello, H.N. Hansen, State-of-the-art of fiber-reinforced polymers in additive manufacturing technologies, J. Reinf. Plast. Compos. 36 (2017) 1061–1073. P. Parandoush, D. Lin, A review on additive manufacturing of polymer-fiber composites, Compos. Struct. 182 (2017) 36–53. X. Wang, M. Jiang, Z. Zhou, J. Gou, D. Hui, 3D printing of polymer matrix: a review and prospective, Composites: Part B 110 (2017) 442–458. Y. Song, Y. Li, W. Song, K.Y. Lee, V.L. Tagarielli, Measurements of the mechanical response of unidirectional 3D-printed PLA, Mater. Des. 123 (2017) 154–164. J. Kotlinski, Mechanical properties of commercial rapid prototyping materials, Rapid Prototyp. J. 20 (2014) 499–510. S.S. Yao, F.L. Jin, K.Y. Rhee, D. Hui, S.J. Park, Recent advances in carbon fiberreinforced thermoplastic composites: a review, Composites: Part B 142 (2018) 241–250. A.N. Dickson, J.N. Barry, K.A. McDonnell, D.P. Dowling, Fabrication of continuous carbon, glass and kevlar fibre reinforced polymer composites using additive manufacturing, Addit. Manuf. 16 (2017) 146–152. Y. Nakagawa, K.I. Mori, T. Maeno, 3D printing of carbon fibre-reinforced plastic parts, Int. J. Adv. Manuf. Technol. 91 (2017) 2811–2817. J. Justo, K. Tavara, L. Garcia-Guzman, F. Paris, Characterization of 3D printed long fibre reinforced composites, Compos. Struct. 185 (2018) 537–548. H.A. Albadi, H.T. Thai, V. Paton-Cole, V.I. Patel, Elastic properties of 3D printed fibre reinforced structures, Compos. Struct. 193 (2018) 8–18. N. Aliheidari, R. Tripuranenia, A. Ameli, S. Nadimpalli, Fracture resistance measurement of fused deposition modeling 3D printed polymers, Polym. Test. 60 (2017) 94–101. W. Gao, Y. Zhang, D. Ramanujan, K. Ramani, Y. Chen, C.B. Williams, C.C.L. Wang, Y.C. Shin, S. Zhang, P.D. Zavattieri, The status, challenges, and future of additive manufacturing in engineering, Comput. Aided Des. 69 (2015) 65–89. R. Matsuzaki, M. Ueda, M. Namiki, K. Joeong, H. Asahara, K. Horiguchi, T. Nakamura, A. Todoroki, Y. Hirano, Three dimensional printing of continuous fibre composites by in-nozzle impregnation, Sci. Rep. 6 (2016) 23058. P. Bettini, G. Alitta, G. Sala, L.D. Landro, Fused deposition technique for continuous fiber reinforced thermoplastic, J. Mater. Eng. Perform. 26 (2017) 843–848. X. Tian, T. Liu, C. Yang, Q. Wang, D. Li, Interface and performance of 3D printed continuous carbon fiber reinforce of PLA composites, Composites: Part A 88 (2016) 198–205. C. Yang, X. Tian, T. Liu, Y. Cao, D. Li, 3D printing for continuous fiber reinforced thermoplastic composites: mechanism and performance, Rapid Prototyp. J. 23 (2017) 209–215. F. van der Klift, Y. Koga, A. Todoroki, M. Ueda, Y. Hirano, R. Matsuzaki, 3D printing of continuous carbon fibre reinforced thermo-plastic CFRTP tensile test specimens, Open J. Compos. Mater. 6 (2016) 18–27. M.A. Caminero, J.M. Chacón, I. Garcia-Moreno, J.M. Reverte, Interlaminar bonding performance of 3D printed continuous fibre reinforced thermoplastic composites using fused deposition modelling, Polym. Test. 68 (2018) 415–423. M.A. Caminero, J.M. Chacón, I. Garcia-Moreno, G.P. Rodriguez, Impact damage resistance of 3D printed continuous fibre reinforced thermoplastic composites using fused deposition modelling, Composites Part B 148 (2018) 93–103. D. Young, N. Wetmore, M. Czabaj, Interlayer fracture toughness of additively manufactured unreinforced and carbon-fiber-reinforced acrylonitrile butadiene styrene, Addit. Manuf. 22 (2018) 508–515. W. Zhang, C. Cotton, J. Sun, D. Heider, B. Gu, B. Sun, T.-W. Chou, Interfacial bonding strength of short carbon fiber/acrylonitrile-butadiene-styrene composites fabricated by fused deposition modeling, Compos. B Eng. 137 (2018) 51–59. G. Liao, Z. Li, Y. Cheng, D. Xu, D. Zhu, S. Jiang, J. Guo, X. Chen, G. Xu, Y. Zhu, Properties of oriented carbon fiber/polyamide 12 composite parts fabricated by fused deposition modeling, Mater. Des. 139 (2018) 283–292. D.-A. Türk, F. Brenni, M. Zogg, M. Meboldt, Mechanical characterization of 3D printed polymers for fiber reinforced polymers processing, Mater. Des. 118 (2017) 256–265. H.L. Tekinalp, V. Kunc, G.M. Velez-Garcia, C.E. Duty, L.J. Love, A.K. Naskar, C.A. Blue, S. Ozcan, Highly oriented carbon fiber–polymer composites via additive manufacturing, Compos. Sci. Technol. 105 (2014) 144–150. R.T.L. Ferreira, I.C. Amatte, T.A. Dutra, D. Bürger, Experimental characterization and micrography of 3D printed PLA and PLA reinforced with short carbon fibers, Compos. B Eng. 124 (2017) 88–100. F. Ning, W. Cong, J. Qiu, J. Wei, S. Wang, Additive manufacturing of carbon fiber reinforced thermoplastic composites using fused deposition modeling, Compos. B Eng. 80 (2015) 369–378. F. Ning, W. Cong, Y. Hu, H. Wang, Additive manufacturing of carbon fiber-reinforced plastic composites using fused deposition modeling: effects of process parameters on tensile properties, J. Compos. Mater. 51 (2017) 451–462. M. Ivey, G.W. Melenka, J.P. Carey, C. Ayranci, Characterizing short-fiber-reinforced composites produced using additive manufacturing, Adv. Manuf. Polym. Compos. Sci. 3 (2017) 81–91. B. Akhoundi, A.H. Behravesh, A. Bagheri Saed, Improving mechanical properties of continuous fiber-reinforced thermoplastic composites produced by FDM 3D printer, J. Reinf. Plast. Compos. (2018) 073168441880730. Y. Sano, R. Matsuzaki, M. Ueda, A. Todoroki, Y. Hirano, 3D printing of discontinuous and continuous fibre composites using stereolithography, Addit. Manuf.

Composites Science and Technology 181 (2019) 107688

J.M. Chacón, et al.

[48] H. Al Abadi, H.-T. Thai, V. Paton-Cole, V.I. Patel, Elastic properties of 3D printed fibre-reinforced structures, Compos. Struct. 193 (2018) 8–18. [49] R.J. Zaldivar, D.B. Witkin, T. McLouth, D.N. Patel, K. Schmitt, J.P. Nokes, Influence of processing and orientation print effects on the mechanical and thermal behaviour of 3D-printed ULTEM 9085 material, Addit. Manuf. 13 (2017) 71–80. [50] J.R.C. Dizon, A.H. Espera, Q. Chen, R.C. Advincula, Mechanical characterization of 3D-printed polymers, Addit. Manuf. 20 (2018) 44–67. [51] M.A. Caminero, G.P. Rodriguez, V. Muñoz, Effect of stacking sequence on Charpy impact and flexural damage behaviour of composite laminates, Compos. Struct. 136 (2016) 345–357. [52] MarkForged, MarkForged MarkTwo, https://Markforged.Com/, (2018). [53] A. Lanzotti, M. Grasso, G. Staiano, M. Martorelli, The impact of process parameters on mechanical properties of parts fabricated in PLA with an open-source 3-D printer, Rapid Prototyp. J. 21 (2015) 604–617. [54] D3039/D3039M, ASTM standard test method for tensile properties of polymer matrix composite materials, Annu. Book ASTM Stand. 15 (2012) 1–10. [55] D790, ASTM standard test method for flexural properties of unreinforced and reinforced plastics and electrical insulating materials, Annu. Book ASTM Stand. 10 (2010) 1–11. [56] M. Fernandez-Vicente, W. Calle, S. Ferrandiz, A. Conejero, Effect of infill parameters on tensile mechanical behaviour in desktop 3D printing, 3D Print. Addit. Manuf. 3 (2016) 183–192. [57] F. Mujika, On the effect of shear and local deformation in three-point bending test, Polym. Test. 26 (2017) 869–877. [58] Z. Li, A. Haigh, C. Soutis, A. Gibson, R. Sloan, N. Kariman, Detection and evaluation of damage in aircraft composites using electromagnetically coupled inductors, Compos. Struct. (2016) 252–261. [59] Quick Release Buckle, Quechua forclaz 30L, https://www.thingiverse.com, (2018). [60] H. Brooks, D. Tyas, S. Molony, Tensile and fatigue failure of 3D printed parts with continuous fibre reinforcement, Int. J. Rapid Manuf. 6 (2017) 97.

24 (2018) 521–527. [38] Z. Hou, X. Tian, J. Zhang, D. Li, 3D printed continuous fibre reinforced composite corrugated structure, Compos. Struct. 184 (2018) 1005–1010. [39] W. Hao, Y. Liu, H. Zhou, H. Chen, D. Fang, Preparation and characterization of 3D printed continuous carbon fiber reinforced thermosetting composites, Polym. Test. 65 (2018) 29–34. [40] K. Sugiyama, R. Matsuzaki, M. Ueda, A. Todoroki, Y. Hirano, 3D printing of composite sandwich structures using continuous carbon fiber and fiber tension, Compos. Part A Appl. Sci. Manuf. 113 (2018) 114–121. [41] T. Liu, X. Tian, M. Zhang, D. Abliz, D. Li, G. Ziegmann, Interfacial performance and fracture patterns of 3D printed continuous carbon fiber with sizing reinforced PA6 composites, Compos. Part A Appl. Sci. Manuf. 114 (2018) 368–376. [42] N. Li, Y. Li, S. Liu, Rapid prototyping of continuous carbon fiber reinforced polylactic acid composites by 3D printing, J. Mater. Process. Technol. 238 (2016) 218–225. [43] G. Dong, Y. Tang, D. Li, Y.F. Zhao, Mechanical properties of continuous kevlar fiber reinforced composites fabricated by fused deposition modeling process, Procedia Manuf. 26 (2018) 774–781. [44] L.G. Blok, M.L. Longana, H. Yu, B.K.S. Woods, An investigation into 3D printing of fibre reinforced thermoplastic composites, Addit. Manuf. 22 (2018) 176–186. [45] K. Agarwal, S.K. Kuchipudi, B. Girard, M. Houser, Mechanical properties of fiber reinforced polymer composites: a comparative study of conventional and additive manufacturing methods, J. Compos. Mater. 52 (2018) 3173–3181. [46] M. Araya-Calvo, I. López-Gómez, N. Chamberlain-Simon, J.L. León-Salazar, T. Guillén-Girón, J.S. Corrales-Cordero, O. Sánchez-Brenes, Evaluation of compressive and flexural properties of continuous fiber fabrication additive manufacturing technology, Addit. Manuf. 22 (2018) 157–164. [47] G.D. Goh, V. Dikshit, A.P. Nagalingam, G.L. Goh, S. Agarwala, S.L. Sing, J. Wei, W.Y. Yeong, Characterization of mechanical properties and fracture mode of additively manufactured carbon fiber and glass fiber reinforced thermoplastics, Mater. Des. 137 (2018) 79–89.

18