Geometric Accuracy and Mechanical Behavior of Polymer-Based Composite Curved Tubes produced by Fused Filament Fabrication (FFF)

The present study aims to assess and characterize the effect of processing parameters including inll pattern and reinforcement type on the dimensional accuracy of products manufactured by Fused Filament Fabrication (FFF) process as well as on the mechanical properties of the printed components. The reinforcements used were carbon, Kevlar and glass bers supplied by MarkForged®; they were utilized to manufacture the PA6 matrix composite. The mechanical properties of the stated composites were compared. Finally, the results obtained conrmed that the selection of the appropriate type of the reinforcements and inll patterns among the several available types during the printing process is effective in improving the mechanical properties and also in providing a better geometrical quality of the surfaces and the consequent dimensional precision improvement of the parts printed by FFF process.


Introduction
Additive manufacturing (AM) [1][2], 3D printing and prototyping [3] [4], or solid freeform fabrication [5], are all instances of a layer-by-layer manufacturing process in which parts are manufactured from designed geometries [6]. Given modern technological progress, AM is gaining popularity, which can make it a common manufacturing process [7]. This approach has provided many bene ts; it can improve exibility and convenience, reduce manufacturing costs and reduce turnout time for multiple manufacturing applications [8,9].
One of the main and effective advantages of AM is the geometric exibility, which allows the assembly of the complex parts without increasing the manufacturing costs [8][9][10]. The successful implementation and expansion of AM requires the improvements in the surface quality, shear and durability, accuracy and precision of these processes [11]. In these issues, surface quality, accuracy and precision are the main obstacles, which don't let AM to be presented and considered as a primary production process [10,12].
In AM we distinguish several methods to manufacture parts such as, selective laser melting, laserengineered net shaping, 3D welding, laminated object manufacturing. Various materials have been developed in polymer printing, which are speci c to the used AM techniques [11]. For instance, for selective laser sintering we have powder materials [12], and for polyjet and stereolithography we have liquid photopolymers [13,14], thermoplastic polymer are developed for Fused Deposition Modeling (FDM) process [15].
Fused Filament Fabrication (FFF) technology is based on manufacturing the parts from computer aided design (CAD) data by fusing a superposition of thin coils of lament through a heated nozzle, in form of layer-by-layer [16]. The lament is deposited layer by layer until the whole desired component is formed [17]. The used raw material in the form of lament could be thermoplastic polymers such as polypropylene (PP), acrylonitrile butadiene styrene (ABS), polyethylene (PE) or polylactic acid (PLA). polycarbonate (PC), polyamide (PA or Nylon), polyetherimide (PEI), polyetheretherketone (PEEK) and etc [18].
Most of the industries are looking for stronger and lighter materials. Therefore, polymer matrix composites (PMCs) have been studied by many researchers [19], as they can achieve the suitable thermomechanical properties when the appropriate reinforcement is used to reinforce the polymer matrix [20]. For more than a decade study and research on the improvement and development of composite materials for AM processes has been ongoing [21]. However, as some polymers cannot reach the required mechanical properties through AM, FDM manufacturing process of polymer matrix composites (PMCs) are developed to obtain composite-based structural components with satisfactory mechanical properties for speci c applications [22,23]. This method was successfully implemented by MarkForged®. In fact, the continuous ber reinforcement has been introduced into the 3D geometry via the double extrusion method [24][25][26][27][28][29][30].
Lately composite materials have been designed and manufactured for several applications and have been introduced as common type of engineering materials. It is understood that the composite tubes have the potential to replace even the metal products on many applications. High attention is directed to producing the composite tubes and the consequent characterizations [31].
The in uence of the in ll pattern on mechanical properties of FDM manufactured components has been studied by many researchers due to the existence of many suggested and available patterns. In fact, it has been proven that selecting the suitable in ll pattern is an important step in producing 3D geometries.
The effects of percentage in ll and in ll patterns on the tensile strength of printed ABS parts were investigated by Fernandez-Vicente et al. [32] : the maximum tensile strength was reported to be around 36.6 MPa related to an in ll percentage of 100% in a rectilinear con guration. The effects of in ll patterns on the cost and strength of the printed components have been studied by Baich et al. [33]; many types of in ll patterns, including solid, high, low, and double dense were considered. The solid-samples presented the maximum tensile, exural, and compressive strengths and also modulus. Akhoundi et al. [34] investigated experimentally the effect of in ll pattern on tensile and exural strength and modulus of parts printed via FFF. Their selected conditions were in ll percentages of 20, 50, and 100% and also the different types of in ll patterns including rectilinear, concentric, hilbert curve and honeycomb. According to the results obtained the concentric pattern presented the required exural and tensile behaviours at all the stated in ll percentages [35].
FFF printed parts have presented poor mechanical property issues which arise from weak bonding and adhesion between the printed layers, minor discontinuities in the extrusion of lament, also the existence of shrinkage which is uncontrolled during the cooling process, etc. [36][37][38].
It is stated that, the control of the required dimensional accuracy is a signi cant issue for the application of FDM process in direct manufacturing [39][40]. Multiple variation sources can cause shape deviation and inaccuracy of AM components in comparison with the desired and designed shapes. Several research studies on optimization of the required geometric accuracy of the manufactured parts via FDM processes have been conducted. According to the Bochmann et al. [41] investigation, it is stated that the magnitude of the errors signi cantly varied in the x, y and z directions in FFF process, which can in uence the accuracy, precision, and quality of nal surface. El-Katatny et al. [42] measured and analyzed the error obtained in geometric characteristics of determined sections on anatomical parts which have been manufactured by FDM method. A methodology of spectral graph theory was used by Tootooni et al. [43,44] and Rao et al. [45] in order to quantify and evaluate the geometric precision of FFF parts using the deviations of the 3D point cloud coordinate measurements from the speci cations of the design. It was clari ed that the proposed indicator did not propose a relationship or correlation between the geometric precision and the process parameters, but only facilitated the comparison of the geometric precision of the parts. Statistical analysis of dimensional accuracy based on the Taguchi method and arti cial neural network (ANN) Sood et al. [46]optimized processing parameters including layer thickness, part orientation and raster angle in FFF. Saqib et al. [47] reported that the geometry of an object affects the accuracy more than processing parameters in FFF process. Also, the perpendicularity and atness features of geometries could in uence the accuracy of the printed components. Chang et al. [48] found that pro le errors and extruding apertures are two essential quality factors which need to be taken into account via FFF process. Also the accuracy depends on transmission machinery and lament diameter.
The mechanical properties of the manufactured parts can be signi cantly improved by the suitable adjustment of the process parameters obtained from the conducted research. It is understood that there is a clear relationship between the selected and excerpted parameters and the obtained mechanical properties of the manufactured part. Optimization of process parameters has signi cantly attracted the attention of different researchers, such as lling velocity (Ning et al. 2016) [49], diameter of the nozzle  [58][59]. All these parameters should be controlled to achieve a suitable part quality with satisfactory mechanical properties.
This study assesses and characterizes the in uence of the process parameters including material type and the selected in ll pattern on the dimensional accuracy, as well as on the mechanical properties of the parts manufactured by FFF process. A compression test was applied on the PA6 reinforced with carbon, Kevlar and glass bers composites under uniform conditions and the mechanical performance of all three composite types are compared. The material used as raw material was polyamide 6 (PA 6). It is introduced as one of the newest matrix materials for fabricating the composite parts with Markforged 3D printers. The Isotropic ber ll type made of Carbon, Glass and Kevlar was chosen as the reinforcement printing type. The printed model for this study was a tube with height, thickness, and external diameter of 50 mm, 4 mm, and 40 mm, respectively (Fig. 1).

3D printer device
One of the Markforged desktop printers, is Mark Two Printer which was used in this study (Fig. 2). This printer was used to print parts from Nylon 6 supplied by Markforged. It alows reinforcing parts with continuous carbon, glass or Kevlar bers.
The nylon 6 was printed with a temperature of 273°C and ber layers were printed with a temperature of 232°C, on a non-heated printer bed platform. The carbon ber was printed in layer height value of 0.125 mm, and the Kevlar and glass bers used were printed with a layer height of 0.1 mm. The dual extrusion system allows continuous ber reinforcement to be placed as the required and determined layers. Also this possibility is provided to specify the ber orientation in the component during the deposition process. Eiger® is the designated software for MarkTwo, which makes it possible to import OBJ and STL models.
Mark Two has ability to produce different structures at different percentages. According to the related printer software, it is possible to choose three main types of in ll pattern, which are rectangular, triangular and hexagonal. Also, during our study, we considered the solid in ll status as another structure or in ll pattern. In fact, in the solid ll pattern, the raster orientations of the layers were + 45 , -45 . Concentric and isotropic are the ber patterns that could be selected in the Markforged Mark Two desktop 3D printer.
Moreover, two types of specimens were considered: unreinforced and continuous reinforced nylon specimens. The printing conditions for the polymer(polyamide-6) samples were: 37% of ll density for the triangular and rectangular, hexagonal ll patterns, 4 roof and layers (the number of layers of solid plastic are used on the top and bottom of the part) and 2 wall layers (the thick of the walls of the part). More walls will make a pure plastic part stronger, but will also reduce the area that ber will be able to t into). For the solid ll pattern the density was 100% with 2 wall layers. For Nylon full reinforced (Carbon, Glass, and Kevlar) specimens were printed with solid ll pattern ,100% ll density and 4 roof and layers, 2 wall layers .The total ber layers was 490 (the total number of layers lled with bers) with a concentric ber ll type (the ber ll type determines the algorithms which control how ber will be used to reinforce the part). All the walls were reinforced (inner holes and outer shell), and 2 concentric ber rings (the number of rings of concentric ber ll added per layer).

Geometric accuracy measurement
A desktop 3D laser scanner (Solutionix D500) was used to scan the geometry and obtain point-by-point coordinate measurements of the component, referred to as a 3D point cloud. The laser scanner records re ected light from the part surface as a point in the 3D space, with a maximum volumetric deviation. Solutionix D500 is powered by Solutionix ezScan. The program is used to calibrate devices, as well as process scan data stitch images taken from different sides at different angles. The desktop rotates scanned objects from different angles. A ray of blue light bounces off objects and enters camera lenses.
For dimensional and quality control a professional 3D Geomagic® Control X ™ software was used, which captures and processes data from 3D scanners. It makes possible calculation of geometric deviations by comparing the data from the point cloud with the original computer aided design (CAD model). The calculation procedure consists of several steps. The alignment of the measured scan to the CAD requires a careful part alignment procedure to achieve consistent results. The alignment step requires matching at least four points of the raw point cloud data to the CAD model and subsequent analysis, each of which has its own literature [36,37]

Quasi-static compression test
The sample used was according to Fig. 1. Quasi-static compression experiments were achieved with the INSTRON 5966 machine, the loading cell of 50 kN, and the loading speed used was 5 mm/min. The special jaws were designed to perform the compression tests and the tubes were positioned between two jaws as sketched in Fig. 3. In order to ensure reproducibility of the results, at least three samples were created in the compression test study.

Dimensional Accuracy And Mechanical Properties Relation
The tubes were printed using the nylon 6 lament and continuous bers, under the main stated ll patterns. Then they were analyzed for the geometric accuracy. Finally the tubes were tested in compression loading. The comparison with the compression strength of the different patterns and also with the solid pattern, for which ll percentage was 100% was carried out. So, the compression strength was considered as the criterion to make the comparison.

Effect of in ll patterns
This part treats three important parameters; the rst parameter is the choice of materials; we used a polymer (Polyamide 6) and also composites; nylon (PA 6) reinforced with three types of bers which are: Carbon ber, Glass ber and Kevlar. The second parameter is in ll pattern: triangular, and rectangular, hexagonal, solid. These parameters were varied to investigate which parameter affects the dimensional accuracy the most and what are the more appropriate for a better print quality and high precision accuracy.
Geometric accuracy results for nylon with different ll patterns; triangular, rectangular, hexagonal, and solid are presented in Figure. 4. One can see that the parts (specimens) are deformed inwards which is known as the shrinkage phenomenon and with different amplitudes (Figs. 5 and 6). The rectangular ll pattern is less deformed comparing with triangular, hexagonal and solid ones. One can note that from 10 mm to 40 mm of height of tube for all ll patterns, there is homogenous deformation.At the initial time of printing (from 0 to 10 mm of height) the First layers can exchange temperature in all directions, and the Page 7/23 ending time of printing (from 40 to 50 mm of height) last layers which can also exchange with air but also receives heat exchange from previous layers.
Moreover, an increasing gap with retraction of each layer and also the retraction of the lower layer which will add up is signi cant (Fig. 7).
The compression test results for the different in ll pattern samples, which were made of nylon 6 are presented in Fig. 8. According to comparison between the effect of the different in ll patterns in the case of compression strength, the solid in ll pattern had the highest strength which was about 52.37 ± 3.5 MPa. Then the compression strength was decreased by changing the in ll pattern from the solid in ll to Hexagonal, Triangular, and Rectangular. The related compression strengths of the printed specimens with the in ll patterns of Hexagonal, Triangular, and Rectangular were 51.02 ± 5 MPa, 28.73 ± 0.5 MPa, and 23.42 ± 2.3 MPa, respectively (Fig. 9). In fact, by changing the in ll pattern from solid in ll to Hexagonal in ll, the compression strength was decreased by about 2.58%. But, in the case of the in ll pattern change from Solid to Triangular and Solid to Rectangular, the compression strength was decreased by about 45.14% and 55.28% respectively.  One can notice that the rectangular in ll pattern has minimum compression properties; however after geometric accuracy results it presents minimum deformation. The latter can be explained by the relation time of the polymer used. Cooling speed of the lament for different in lls can be another reason.
The temperature selection is highly dependent on the viscosity of the polymer and should be adjusted with the right printing speed; too high temperature may cause a reduction in the polymer viscosity and the melt will become too uid and highly owable which result in a lot of plastic leaking out from the hot end (nozzle) during printing, and reducing the dimensional accuracy. Otherwise when the temperature is too low, the new layer will simply not stick to the previous layer and the surface of the thread could be a bit rough [50].

Effect of reinforcement type
The geometric accuracy results for composite specimens (Fig. 10) shows different deformations. The printed PA 6 reinforced with carbon ber composite is deformed outwards (dilatation) contrary to Kevlar and glass ber which are deformed inwards (shrinkage). Also, Kevlar is more deformed in the rst layers. As the graph shows the deformation of the reinforced tube with Kevlar and glass ber is almost the same. The results showed different part behaviors after the FFF process, and the deformation of measured parts (3D models) changed with the variation of reinforcement. It is important to take into account the reinforcement lament thickness: carbon ber was printed with a layer height value of 0.125 mm, and the Kevlar and glass bers were printed with a layer height of 0.1 mm, so it is logical to obtain this signi cant difference in deformation.  One can note that the nylon reinforced with carbon ber has minimum compression properties in comparison to kevlar and glass bers.

Macroscopic observation
One can observe from Fig. 13, the shape of the specimens after compression test which were produced by solid in ll pattern. The rectangular was found to be damaged more signi cantly in macroscopic observations.
The cooling rate of the specimens through the printing process is affected by the movement of the extrusion head temperature (which is higher than the envelope temperature), as a consequence it will in uence the adhesion and bonding between the adjacent deposited lament [49] The use of various materials in a dedicated and optimized system may change its standard melt rheological behavior requirement, thereby in uencing the melt processes. Therefore, many parameters need to be adjusted in order to obtain the best quality for the nal product.

Conclusion
In this study, the assessment of the in uence of the process parameters including material and in ll pattern on the dimensional accuracy and precision, as well as on the mechanical properties of components manufactured by FDM process, were examined. This involves the manufacturing of polymer matrix composites (PMCs) with carbon, Kevlar and glass bers reinforcements, provided by MarkForged®. Then the compression performance of the manufactured composites were evaluated and compared. The reinforcements and the in ll patterns make it possible to improve the mechanical behavior while also obtaining a better geometrical quality and precision of the FDM manufactured parts.

Availability of data and materials
The authors declare that the data and the materials of this study are available within the article.

Declarations
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Consent to participate: Informed consent was obtained from all individual participants included in the study. The dimensions of the printed tube Mark Two printer (a) and printer during the printing of the required specimens (b) Figure 3 Experimental setup of compression test  Shirinkage during printing of the last layer   Compression results for nylon with different in ll patterns Figure 9 Compression strength of the different in ll patterns Compression results for nylon with different reinforcements Figure 12 The effect of the different reinforcements on the compression strength (reinforcing, perpendicular to the stress direction) Figure 13 Macroscopic observation of tubes after compression tests