Fused filament fabrication of carbon fiber-reinforced polymer composite: Effect of process parameters on flexural properties

Abstract

Short carbon fiber-reinforced polymer composites are desirable for many additive manufacturing (AM) applications as they are light and structurally strong. The process parameters in fused filament fabrication (FFF) significantly affect the mechanical properties of SFRP parts. In this work, three-point bending tests are carried out to investigate the flexural behavior of 3D-printed polyamide 12 carbon fiber (PA12-CF) specimens. An L18 Taguchi design of experiments with Gray relational analysis is applied to optimize the FFF parameters. It is shown that build orientation has the most influence on flexural properties and the distribution of short fibers has an additional effect. For a rectilinear infill pattern, the maximum flexural strength of the part was realized at the printing speed, layer thickness, and extrusion temperature of 30 mm/s, 0.15 mm, and 270°C, respectively. With these parameters, the flexural strength of the part is 119.9 MPa and the flexural modulus is 3038 MPa. For a concentric infill pattern, the flexural strength of the part is 15.8% higher at 138.8 MPa. The flexural modulus is also higher at 3692 MPa. This study's results contribute toward optimizing FFF parameters to suit the specific flexural loading requirements of an AM part.

INTRODUCTION

Additive manufacturing (AM) is one of the priority areas of research related to manufacturing today. The advancement of AM technologies can be attributed to their applicability in the fabrication of parts with complex geometries. As opposed to conventional manufacturing methods, AM technologies do not require product-specific tools, and the production lead time is low. In addition, material wastage is minimal.^1–5 Fused filament fabrication (FFF) is the most established and widely used AM process. In FFF, a filament feedstock is melted and deposited by an extrusion head. The extrusion head moves in both horizontal and vertical directions, using a computer-controlled mechanism. The extruder nozzle follows a tool path generated from computer-aided manufacturing software, and the part is built layer-by-layer from the bottom to the top.⁶ FFF is widely used mainly because of its cost-effectiveness, reliability, ease of material variation, dimensional accuracy, and ability to produce parts with good resolution.⁷

The FFF technique is mostly used in the manufacture of parts using thermoplastic polymers. In recent years, 3D printers have become less costly and more widely available, which in turn has expanded the opportunities to print functional parts for a variety of applications.⁸ This is in addition to the more prevalent application in printing prototypes. According to a systematic survey by Dey and Yodo,⁹ acrylonitrile–butadiene styrene (ABS) and polylactic acid (PLA), which are thermoplastic polymers, are the most widely used and studied materials for FFF. Composite materials offer an alternative to conventional materials as designers have greater flexibility to optimize the structure or component to meet functional requirements. Some of the composites used within the scope of FFF are polymer matrix composites, metal matrix composites, ceramic composites, and fiber-reinforced composites. Fibers and fillers are often added to polymers to improve their mechanical properties, especially strength, stiffness, and thermal stability.¹⁰ As more materials are made that are compatible with the FFF process, the scope of parts that can be manufactured is increased.¹¹ Composites offer great potential for reducing the weight of parts while still maintaining the required strength and stiffness. For this reason, composites have found many applications in engineering such as the manufacture of parts for aerospace, automotive,¹² sound-absorbing structures,¹³ unique lightweight structures,¹⁴ and sporting, tool and mold,¹⁵ as well as for prosthetics,¹⁶ and orthotics industries.¹⁷ Carbon fiber-reinforced polymer (CFRP) composites consist of short or continuous carbon fibers for reinforcement, embedded into a polymer matrix. The coordinated sizing and arrangement of reinforcement fibers is a current topic of research.¹⁸ The advantageous properties of CFRP composites include high strength, lightweight, corrosion resistance, and good fatigue resistance. There is great potential in the use of CFRPs as materials for 3D printing, not only prototypes but also functional parts. The main challenge with CFRP composites is their high cost, but they are gradually becoming more viable for many applications, as equipment to manufacture parts from CFRC becomes accessible and technology development advances.

Parts made using the FFF technique are anisotropic and their mechanical properties largely depend on the process parameters used.¹⁹ There remain a need for accurate prediction of the properties and failure modes for parts made by AM.^11,20 Usually, operators choose the parameters through their intuition and experience, but they do not have enough information on how to select the best manufacturing parameters. In addition, the default selection of printing process parameters provided by manufacturers of filaments and printers cannot guarantee the quality of printed products.²¹ It is, therefore, necessary to establish the effect of process parameters on mechanical properties such as flexural strength to determine the best process parameters for enhanced functional life of the fabricated parts.²² With the knowledge of the effect of the parameters, the design and prediction of the performance of FRAM parts will be more accurate and reliable. Many studies have been carried out where properties of 3D printed parts from conventional materials are evaluated with respect to the printing parameters. In a study by Ahn et al.,²³ the effects of raster orientation on tensile and compressive strengths were investigated experimentally for parts made of ABS P400 material, and compared with injection-molded FFF ABS P400 material. The parameters studied were air gap, raster orientation, bead width, and temperature. An important outcome of this study was the formulation of build rules based on the experimental results. Application of the build rules formulated can lead to improved strength and quality of FFF-manufactured parts. Chacon et al.²⁴ conducted a study to characterize the effect of FFF parameters (layer thickness, build orientation, and feed rate) on the mechanical performance of parts that were 3D printed using PLA. They carried out tensile and flexural tests to determine the mechanical performance. Analysis of variance (ANOVA) is a reliable method to study the influence of parameters on multiple responses after a Taguchi analysis as seen in Reference 25 and an ANOVA was done to determine the significance of the three factors. It was found that the build orientation had the most influence specifically on ductility and failure behavior. Flat orientations gave the best tensile and flexural strength. The authors, however, did not consider the effect of the printing pattern, which could also be significant.

Continuous fiber-reinforced composites (CFRC) have also been applied in AM, and they are suitable for high-strength applications.²⁶ In a study by Chacón,²⁷ continuous glass, Kevlar®, and carbon-reinforced nylon composites were manufactured using the FFF technique. In the study, samples were printed using a dual extruder printer with one print head depositing nylon while the other deposited the reinforcement. The results showed a moderate increase in tensile strength, with an increase in fiber content. Due to porosity occasioned by the weak bonding between the reinforcing fiber and thermoplastic layers, a compaction stage would be necessary after the deposition. Nevertheless, the strength properties obtained in the samples were still significantly higher than those of the usual 3D-printed thermoplastics. The challenge faced in the processing of CFRCs is in the placement of continuous fibers,²⁸ making the use of chopped carbon fiber reinforcement an easier and less costly option. A study by Goh et al. realized high strength properties of FFF-fabricated continuous carbon fiber and glass fiber-reinforced nylon composite. A challenge in low deposition rate was reported for this technique, which would be addressed by the use of multi-nozzle AM systems.²⁹ The use of short or chopped fibers to reinforce thermoplastics enhances the strength properties of 3D-printed CFRC parts. The strength attained, however, does not reach the strength level of continuous fiber composite parts.^30,31 There is a wide range of short fiber-reinforced polymer (SFRP) composites available for application in the FFF process. As more materials become available for FFF, it is necessary to establish how their properties are affected by the printing parameters used. The knowledge of flexural properties of additively manufactured short fiber composites is crucial as they have found many applications where a certain degree of flexibility or stiffness is required including weather seals, diving equipment, medical catheters, orthosis, and prostheses. The objective of this study was to investigate the relationship between the process parameters and the mechanical properties of short fiber-reinforced polymer parts produced using the FFF technique. This is part of an ongoing study whose objective is to optimize the flexural properties of an additively manufactured ankle-foot orthosis (AFO). Currently, a patient's pathological foot profile can be obtained through 3D scanning. The digital data from this method is not applicable with conventional manufacturing method for AFOs using thermoformed polymer sheets. Additive manufacturing thus provides the opportunity for a patient-specific customization, with minimal possibility of error. With AM, it is also possible to use advanced materials such as reinforced polymer composites and enable more complex designs. The potential for such devices has been demonstrated using polymers and different AM techniques.³² Carbon fiber-reinforced polymer composites have a high strength-to-weight ratio, which would result in thinner and lighter AFOs. While in use, an AFO experiences different loadings including bending and low-frequency cyclic flexural loads. Understanding the flexural behavior is therefore key to optimizing the AFO. In this study, the effects of four process parameters, namely, layer thickness, extrusion temperature, print speed, and build orientation, on the flexural properties of short carbon fiber-reinforced nylon parts produced by the FFF method were investigated. Taguchi optimization technique was used to characterize the flexural properties and to establish the optimum FFF process parameters.

MATERIALS AND METHODS

The material used in this study is PolyMide 12 Carbon Fiber (PA12-CF) filament of diameter 1.75 mm with 10% fiber content by weight. The average aspect ratio of the carbon fibers was 80 and the diameter was about 5 μm. The filament is produced by a melt compounding process, where the polymer matrix and the preprocessed chopped fibers are mixed and then extruded into a 1.75 mm diameter filament. The flow through the extruder die results in the imperfect alignment of short fibers. As per the manufacturer's datasheet, the filament has a flexural modulus of 3535 ± 239 MPa. The mechanical properties of the PA12-CF filament are given in Table 1.

TABLE 1 Mechanical properties of the PA 12-CF filament.³³

Property	Typical value
Density	1.06 g/cm3 at 23°C
Melt index	25 g/10 min
Glass transition temperature	108°C
Young's modulus (X–Y)	3304 ± 145 MPa
Young's modulus (Z)	1801.1 ± 80 MPa
Tensile strength (X–Y)	71.6 ± 1.7 MPa
Tensile strength (Z)	43.3 ± 3.3 MPa
Bending strength (X–Y)	109.9 ± 1.4 MPa
Bending modulus (X–Y)	3535 ± 239 MPa

This study investigation focuses on the following parameters:

1. Layer thickness: the height of each deposited layer along the z (vertical) axis
2. Print speed: the distance traveled by the print head per unit of time during deposition
3. Extrusion temperature: the temperature to which the material is heated during printing
4. Build orientation: how a part is oriented on the build platform with respect to the x-y-z axes

The four parameters were selected based on literature findings showing their significance in flexural strength properties of FFF parts printed from polymers such as PLA and ABS.

Experimental procedure

The experiments were planned using Taguchi design of experiments (DoE) whereby an L18 orthogonal array was used to study all of the above parameters using a minimum number of experiments. This method is reliable when carrying out experiments for all combinations of factors is not possible.²¹ The different combinations of parameter values gave a broad range of testing conditions. The effect of each parameter on flexural properties was thus investigated. The parameters that were kept constant for all the tests are given in Table 2. The variable parameter levels for each test run are given in Table 3. Rectilinear build orientation was used in all 18 experiments. In addition, a further study was carried out, using a concentric infill pattern and both flat and on-edge build orientations. The variation of build orientation and the infill pattern is illustrated in Figure 1, and the Taguchi DoE for the additional tests is given in Table 4.

TABLE 2 Fixed printing parameters.

Parameter	Description
3D Printer	Delta Wasp 2040
Material	Nylon 12 CF, 1.75 mm diameter
Infill percentage	100%
Fill pattern	Rectilinear
Raster angle	+45°/−45°
Number of perimeters	3
Perimeter/Infill overlap	10%
Bed temperature	80°C
Cooling	Fan off
Nozzle diameter	0.4 mm

TABLE 3 Experimental factors and their levels.

Factor	Level 1	Level 2	Level 3
Print speed, mm/s	30	40	50
Layer thickness, mm	0.1	0.15	0.2
Build orientation	Flat	On-edge
Extrusion temperature, °C	270	280	290

TABLE 4 Layer thickness, print speed, and extrusion temperature values for all tests in A, B, and C setups.

Test run	Layer thickness (mm)	Print speed (mm/s)	Extrusion temperature (°C)
1	0.1	30	270
2	0.1	40	280
3	0.1	50	290
4	0.15	30	280
5	0.15	40	290
6	0.15	50	270
7	0.2	30	290
8	0.2	40	270
9	0.2	50	280

Samples were modeled in CAD software and exported as STL files onto slicing software, PrusaSlicer©. A screenshot of the slicing is shown in Figure 2. The slicing process generated G-code files, which contain information on all the FFF parameter values. The samples were 3D printed using the setup shown in Figure 3A. The recommended print speed ranges from 30 to 60 mm/s, and therefore a range of 30–50 mm/s was used in the study. The recommended extrusion temperature by the material manufacturer is 260–300°C, and for this study, the range was selected between 270 and 290°C. The layer thickness was limited by the capability of the FFF machine used. The samples measured 76.2 mm in length, 12.7 mm in width, and 3.2 mm in thickness by ASTM D790 standard.³⁴ Samples were printed using a Delta Wasp 2040 3D printer which has a closed chamber and a heated circular bed of a diameter of 200 mm. Since nylon is hygroscopic, the filament was stored and supplied from a dry box to avoid deterioration in quality, and the humidity was maintained below 16%. There were 18 samples printed for each set of tests, and three samples for each experimental run. A photograph of the printed samples is shown in Figure 3B.

For static flexural testing, a three-point bending setup was used on a Universal Testing Machine model ME-8236. This setup is shown in Figure 3C. The samples were held on two supports, with the support span-to-depth ratio of 16, and the load was applied at the midpoint. The cross-head speed, R was determined using Equation 1 where: [Image Omitted. See PDF] where Z is the strain rate, L is the span length between supports, and t is the thickness of the sample.

Using a standard strain rate of 0.01, R was obtained as 1.4 mm/min for a span length of 50.8 mm and a thickness of 3.2 mm. The loading time, load, and deflection data were recorded simultaneously. The displacement of the crosshead was used to determine the deflection of the part. From the experimental values of load and deflection, the flexural strength (σ_f) for each sample was determined as³⁴: [Image Omitted. See PDF] where F is the maximum load applied (N), L is the support span (mm), b is the width of the sample (mm), t is the thickness/depth of the sample (mm), and the flexural strain (ε_f) as³⁴: [Image Omitted. See PDF] where D is the maximum deflection (mm) and t is the depth (mm).

The flexural modulus of elasticity (E) was also obtained as³⁴: [Image Omitted. See PDF] where m is the slope of the tangent to the initial straight-line portion of the load-deflection curve.

Optimization of process parameters

Using the Taguchi method, a loss function is employed to determine the optimal conditions required to achieve the desired response. The loss function is converted to an S/N ratio to assess the correlation between quality and variability.³⁵ The signal-to-noise (S/N) ratio was calculated and used as a performance characteristic of the deviation of responses from the desired values. The S/N ratio, a logarithmic function, was computed by assessing the proportion of the response signal to the noise. Thus, a high S/N ratio indicates better quality of the response. The optimal combination of process parameters is therefore given by the test run with the greatest S/N ratio. This technique is successfully applied in multi-response optimization of process parameters for the batch processing of a microcellular nanocomposite.³⁶ The S/N ratio required for both flexural strength and flexural modulus is the higher-the-better,³⁷ and was thus calculated using Equation 5,³⁸ where n is the number of experiments and y_i is the value of response for the ith experiment. [Image Omitted. See PDF]

To get the optimum parameters for maximum flexural strength and maximum flexural modulus, gray relational analysis (GRA) was used. Taguchi design with GRA has been successfully applied for the multi-objective optimization of process parameters.³⁹ To determine the optimal parameters that maximize both flexural strength and flexural modulus, the S/N ratio data of the responses were normalized in the range of 0 to 1 using Equation 6.⁴⁰ [Image Omitted. See PDF] where x_i(k) is the normalized sequence for the ith experiment and y_i(k) is the sequence of the mean of the responses.

The deviation sequence was then determined for each response and used to calculate the gray relational coefficients (GRC), ζ_i, for each run of tests, using Equation 7.⁴⁰ [Image Omitted. See PDF] where Δ_i(k) is the deviation of the normalized response for the sequence, and Δ_max and Δ_min are the maximum and minimum deviations of each response. The identification coefficient was set as 0.5, to allocate weight to each parameter. The average of the GRCs was used to compute the GRG, γ_i, using Equation 8.⁴⁰ The responses were ranked according to GRG, with the highest GRG being ranked first. [Image Omitted. See PDF]

RESULTS AND DISCUSSION Effect of process parameters on flexural strength and strain

The results of flexural strength, flexural strain, and flexural modulus averaged from three replicas of each experimental run are given in Table 5. From the results, the average flexural strength for the flat orientation was 56.30 MPa while that of the on-edge orientation was 98% greater at 111.3 MPa. The maximum flexural strength in the on-edge orientation was 119.9 MPa while the maximum flexural strength in the flat orientation was 62.30 MPa. According to the manufacturer's material card, PA12-CF has a flexural strength of 109.9 ± 1.4 MPa when printed. The difference in values between the experimental data and the manufacturer's data could be attributed to differences in printing parameters. The reinforcement of PA12 with carbon fibers is also seen to significantly increase the strength, as compared to neat PA12 which has a flexural strength of about 62–63 MPa when printed by FFF and a flexural modulus of 1265–1500 MPa.^41,42 In an un-reinforced polymer such as PLA, studies show a less significant influence of build orientation on flexural strength.

TABLE 5 Results of static flexural strength tests (PS-print speed, LT-layer thickness, BO-build orientation, ET-extrusion temperature).

Test run	PS (mm/s)	LT (mm)	BO	ET (°C)	Flexural yield strength (MPa)	Flexural strain	Flexural modulus (MPa)
1	30	0.1	Flat	270	62.30 ± 0.44	0.1109 ± 0.008	1500 ± 9.80
2	30	0.15	Flat	280	62.89 ± 0.28	0.1173 ± 0.014	1461 ± 53.4
3	30	0.2	Flat	290	56.95 ± 0.53	0.1150 ± 0.000	1308 ± 15.0
4	40	0.1	Flat	270	54.49 ± 0.96	0.1166 ± 0.003	1313 ± 4.27
5	40	0.15	Flat	280	57.55 ± 0.05	0.1100 ± 0.000	1346 ± 11.1
6	40	0.2	Flat	290	58.14 ± 0.96	0.1184 ± 0.000	1346 ± 11.2
7	50	0.15	Flat	270	59.92 ± 0.60	0.1156 ± 0.001	1346 ± 12.5
8	50	0.2	Flat	280	47.46 ± 0.56	0.1221 ± 0.004	1115 ± 4.29
9	50	0.1	Flat	290	57.55 ± 0.12	0.1140 ± 0.003	1307 ± 5.78
10	30	0.2	On-edge	270	100.9 ± 0.40	0.1186 ± 0.007	2807 ± 11.5
11	30	0.1	On-edge	280	119.9 ± 0.41	0.0864 ± 0.000	3038 ± 4.22
12	30	0.15	On-edge	290	115.1 ± 0.07	0.0920 ± 0.000	3038 ± 6.20
13	40	0.15	On-edge	270	113.3 ± 0.50	0.1201 ± 0.000	2999 ± 1.66
14	40	0.2	On-edge	280	106.8 ± 0.16	0.1075 ± 0.000	2807 ± 0.59
15	40	0.1	On-edge	290	115.1 ± 0.12	0.0748 ± 0.000	2846 ± 3.40
16	50	0.2	On-edge	270	104.4 ± 0.17	0.0779 ± 0.000	2769 ± 0.17
17	50	0.1	On-edge	280	118.1 ± 0.53	0.0944 ± 0.001	3038 ± 7.36
18	50	0.15	On-edge	290	108.6 ± 0.91	0.1615 ± 0.003	2730 ± 12.0

The build orientation of the samples in SFRP is observed to be additionally affected by the distribution of fibers. The distribution of fibers is a factor of lengths of short fibers in a material and the fabrication method. A study by Bardiya et al. shows a maximum difference of 41.5% for orientations between 0° and 60°.⁴³ The strength-bearing role of reinforcement fibers and their orientation in their layers results in a much more significant influence on flexural strength. In the study, there was also observed a decrease in flexural strength for build orientation above 60°, while in this study the maximum flexural strength was obtained at a build orientation of 90°.

A large difference in flexural yield strength was observed between the flat-oriented samples in Tests 1–9 and on-edge-oriented samples in Tests 10–18. This was attributed to the positioning of the contours and rasters' relative to the loading direction. Each of the samples was printed with three perimeters, which ran concentric to the rectangular profile. This meant that the on-edge build orientation had the perimeters, which run perpendicular to the loading direction, taking a large percentage of the load direction. The change in orientation of the part results in changing the orientation of the lines in a path. The anisotropy due to pores along these lines varies for the different orientations and is one cause of strength loss in a printed part. This anisotropy can be reduced by optimizing the print direction. Further, in the flat-oriented samples debonding failure mechanism between the layers is involved, and contributes to the lower strength.

The main effects plot is shown in Figure 4. The build orientation had the greatest influence on the flexural strength of the sample, followed by layer thickness, as shown in the response table for the signal-to-noise ratio in Table 6. The extrusion temperature had the second-largest influence on flexural stress. The flexural strength of the parts increased with increased temperature from level 1 to 2. This can be attributed to improved fusion between the layers. There was however a slight decrease at level 3 of extrusion temperature. This could be as a result of increased fluidity of the filament at higher temperatures, which resulted in improper material deposition and poorer interlayer bonding. In a study by Alafaghani et al., an increase in extrusion temperature beyond a certain limit was observed to result in increased dimensional errors.² From ANOVA analysis, build orientation and layer thickness had p-values less than 0.05, proving statistical significance in determining the flexural strength of the part. The print speed had the least influence followed by the layer thickness. The optimum parameters for maximum flexural strength were obtained as on-edge build orientation, a layer thickness of 0.1 mm, a print speed of 30 mm/s, and an extrusion temperature of 280°C.

TABLE 6 Response table for flexural stress signal-to-noise ratios.

Response table for signal to noise ratios
Larger is better
Level	Build orientation	Layer thickness	Print speed	Extrusion temperature
1	34.78	38.10	37.58	37.86
2	40.86	37.53	37.84	38.10
3		37.81	38.03	37.49
Delta	6.08	0.57	0.45	0.60
Rank	1	3	4	2

Another cause of strength loss in FFF parts is inter-layer debonding on the layer interface. Debonding was not observed in the on-edge oriented samples, as this interface was oriented perpendicular to the tensile and compressive forces that occur at the top and bottom of the sample during the test. The flat orientation on the other hand had the rectilinear infill taking a larger percentage, making the part more susceptible to failure. In the rectilinear infill, rasters run at 45° to the loading direction. The layer interface was thus oriented at 45° to the loading direction. The effect of having less fiber length to bear the load resulted in premature failure in these samples. A visual inspection was carried out using an Olympus SC50 microscope and Figure 5 shows that the highest percentage of carbon fibers are aligned along the print direction.

The failure pattern was observed macroscopically to identify the weak path where the crack occurs. The path of failure was observed on the point of application of load on the side of the samples that was under tension. None of the samples broke completely, but on some samples, a crack was observed to form and propagate. Figure 6A,B show the cracks formed on sample 1 as observed on an Olympus SC50 light microscope. The failure is mainly due to pulled matrix, fiber pullout and fiber breakage. Scanning electron microscopy was also done to obtain further imaging of the failure region as shown in Figure 7. For sample 1 which was built flat, there is less resistance to breakage, thus the crack is well formed with fiber on the surface under tension completely broken. On sample 11, the greater resistance to bending offered by the parallel orientation of fiber resulted in minimal failure predominantly in the form of pulled matrix.

In the subsequent study using the L9 Taguchi DoE given in Table 4, the flexural strength results were obtained and plotted as shown in Figure 8. Samples built on-edge and printed using a concentric infill pattern had the highest flexural strength, regardless of the print speed, extrusion temperature, and layer thickness levels. Samples that were printed using a rectilinear infill and built flat on the bed had the lowest flexural strength for all print speed, extrusion temperature, and layer thickness level combinations. The highest flexural strength was 138.8 MPa, obtained at a layer thickness of 0.1 mm, print speed of 50 mm/s, and extrusion temperature of 290°C using an on-edge build orientation and a concentric infill pattern. This was a 15.8% increase from the highest strength obtained in Table 5. For this sample, the concentric infill pattern with on-edge build orientation meant that the fibers were predominantly aligned parallel to the bending plane, and offered a higher resistance to flexure due to the longer fiber length over which the bending load could be distributed. Figure 9 shows the stress–strain curves for Test 3, whose factor levels are layer thickness of 0.1 mm, print speed of 50 mm/s, and extrusion temperature of 290°C. The infill pattern and the build orientation resulted in varied stress–strain curves for each combination, with on-edge build orientation and concentric infill having a higher flexural strength for all strain levels, and a flat build orientation with a rectilinear infill pattern having the lowest flexural strength. The parts underwent a significant amount of plastic deformation before yielding, which occurred at different points depending on the combination of parameters.

Effect of process parameters on flexural modulus

The build orientation had the highest influence on the flexural modulus as shown in the main effect plots in Figure 10. A maximum modulus of 3038 MPa was obtained in tests 11, 12, and 17. The response table for flexural modulus is given in Table 7, and it shows the build orientation having the greatest influence on the flexural modulus, with a delta value of 6.73. The extrusion temperature had the least influence with a delta value of 0.19. Notably, all of these test runs resulted in the three highest values of flexural strength as seen in Table 5. The similar values of modulus may be a result of low sensitivity in the measurement of the deflection of the sample. No deflection measurement instrument was used, instead, the movement of the crosshead was used to determine the deflection. The use of an extensometer may provide more accurate results for the measurement of strain. According to the manufacturer's material card, PA12-CF has a flexural modulus of 3535 ± 239 MPa when printed. The difference in the values obtained was attributed to the use of different printing parameters. The optimum parameters for maximum flexural modulus were obtained as an on-edge build orientation, a layer thickness of 0.1 mm, a print speed of 30 mm/s, and an extrusion temperature of 280°C.

TABLE 7 Response table for flexural modulus signal-to-noise ratios.

Response table for signal to noise ratios
Larger is better
Level	Build orientation	Layer thickness	Print speed	Extrusion temperature
1	62.50	66.10	66.24	65.97
2	69.23	66.06	65.85	65.85
3		65.44	65.51	65.78
Delta	6.73	0.66	0.72	0.19
Rank	1	3	2	4

Optimization of process parameters

Table 8 shows the normalized S/N ratios for flexural strength (SNRA1) and flexural modulus (SNRA2), gray relational coefficients GRC1 and GRC2, and GRG for all tests in the DoE given in Table 3. Test 11 was ranked first, meaning it offered the best combination of process parameters for maximum flexural strength and flexural modulus. The optimum parameters were obtained as an on-edge build orientation, a layer thickness of 0.1 mm, a print speed of 30 mm/s, and an extrusion temperature of 280°C. This optimal combination of parameters obtained for the multi-response optimization was part of the initial DoE, and it is highlighted in Table 8. For this combination of parameters, the flexural strength was obtained as 119.9 MPa while the flexural modulus was obtained as 3038 MPa.

TABLE 8 Normalized S/N ratios, calculated gray relational coefficients and overall gray relational grades.

Test run	SNRA1	SNRA2	Normalized SNRA1	Normalized SNRA2	GRC1	GRC2	GRG	Rank
1	35.89	0.294	63.52	0.296	0.414	0.415	0.415	10
2	35.97	0.304	63.30	0.270	0.418	0.406	0.412	11
3	35.11	0.197	62.33	0.159	0.384	0.373	0.378	16
4	34.73	0.149	62.36	0.163	0.370	0.374	0.372	17
5	35.20	0.208	62.58	0.188	0.387	0.381	0.384	14
6	35.29	0.219	62.58	0.188	0.390	0.381	0.386	13
7	35.55	0.252	62.58	0.188	0.401	0.381	0.391	12
8	33.53	0.000	60.95	0.000	0.333	0.333	0.333	18
9	35.20	0.208	62.33	0.159	0.387	0.373	0.380	15
10	40.07	0.814	68.96	0.921	0.729	0.864	0.796	9
11	41.57	1.000	69.65	1.000	1.000	1.000	1.000	1
12	41.22	0.956	69.65	1.000	0.920	1.000	0.960	3
13	41.09	0.940	69.54	0.987	0.892	0.975	0.934	4
14	40.57	0.875	68.97	0.921	0.801	0.864	0.832	6
15	41.22	0.956	69.08	0.935	0.920	0.885	0.902	5
16	40.38	0.851	68.85	0.907	0.771	0.844	0.807	8
17	41.44	0.984	69.65	1.000	0.969	1.000	0.984	2
18	40.71	0.893	68.72	0.893	0.824	0.824	0.824	7

Note: The bold entry highlights the test with the best rank.

CONCLUSION

This paper focused on the experimental study of flexural properties of carbon fiber-reinforced PA12 parts manufactured using FFF. An L18 Taguchi DoE was used to investigate and analyze the effect of FFF process parameters, namely, layer thickness, print speed, build orientation, and extrusion temperature, on the flexural strength of carbon fiber-reinforced PA12. All the parameters were significant to the responses. Based on the results of this study, the main conclusions are summarized as follows:

1. The optimal parameters for maximum flexural strength and maximum flexural modulus are a printing speed of 30 mm/s, layer thickness of 0.15 mm, extrusion temperature of 270°C, and built using the on-edge orientation.
2. The build orientation had the most influence on both the flexural strength and flexural modulus. The average flexural strength of the samples using a rectilinear infill in the on-edge build direction was 57.6% higher than samples built in a flat direction.
3. Based on the results, it was shown that the average flexural strength could be increased further by using a concentric infill pattern and an on-edge build orientation. A 36.5% increase in average flexural strength was realized, and a flexural strength of 138.8 MPa.
4. The results of this study help design parts such as a custom orthotic device with special mechanical requirements. The data demonstrated that consideration of build orientation and infill pattern would enhance the applicability and durability of a 3D-printed part

AUTHOR CONTRIBUTIONS

Lucy W. Kariuki: Conceptualization (equal); data curation (lead); formal analysis (lead); funding acquisition (equal); methodology (equal); project administration (equal); resources (equal); validation (equal); writing – original draft (lead); writing – review and editing (equal). Bernard W. Ikua: Conceptualization (equal); data curation (supporting); formal analysis (supporting); funding acquisition (equal); methodology (supporting); project administration (equal); resources (equal); supervision (lead); validation (equal); writing – review and editing (equal). Samuel K. Karanja: Conceptualization (equal); data curation (supporting); formal analysis (supporting); funding acquisition (equal); investigation (supporting); methodology (equal); project administration (equal); resources (equal); supervision (equal); validation (equal); writing – review and editing (equal). Stephen P. Ng'ang'a: Conceptualization (equal); data curation (supporting); formal analysis (supporting); funding acquisition (equal); investigation (supporting); methodology (equal); project administration (equal); resources (equal); supervision (equal); validation (equal); writing – review and editing (equal). Henning Zeidler: Data curation (supporting); formal analysis (supporting); investigation (supporting); methodology (supporting); resources (equal); validation (equal); writing – review and editing (equal).

ACKNOWLEDGMENTS

This study was conducted with the support of Jomo Kenyatta University of Agriculture and Technology (JKUAT), Kenya, and funded by the German Academic Exchange Service (DAAD). The experimental work was carried out at Technische Universität Bergakademie Freiberg (TUBAF), Germany.

FUNDING INFORMATION

DAAD In-Country/In-Region Scholarship Programme-Kenya, 2019. Grant/Award Number 57499637. Short-Term Research Scholarship for In-Country/In-Region Scholarship Holders, 2021. Grant/Award Number: 57560509.

CONFLICT OF INTEREST STATEMENT

The authors declare no potential conflict of interest.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.