Optimizing process parameters of fused deposition molding for 3D printing customized control of polyetheretherketone scaffolds

Abstract

Polyetheretherketone (PEEK), with good biocompatibility and similar mechanical properties to natural bone, is extensively employed in the manufacture of prostheses. However, the precision and mechanical properties of current implants are major challenges for clinical applications. In this study, the effect of pore size, raster angle and printing temperature were investigated on length, width, thickness, material consumption, compressive strength and Young’s modulus. Taguchi design of experiment method was used to reduce the number of experiments and optimize the printing process parameters. Finally, predictive analysis was exploited to give the optimal set of process parameters. Experimental results indicated that the approach applied in this work provided more accurate predictions and control of the response variables. The maximum compressive strength and compressive modulus of PEEK scaffolds reached 43.4 MPa and 253.3 MPa, respectively. Therefore, the methodology of present work has the potential to meet the demand of design precision and manufacture of customized bone substitutes.

Keywords

Rapid prototyping Taguchi’s method compressive properties dimensional precision customization

Introduction

Compared to SLS technologies, fused deposition modeling (FDM) has been widely used in diverse application fields, such as medicine, automotive, electronics, construction and other fields.^1–4 Simple operation, inexpensive machines, and durable parts⁵ result in FDM being the most accessible and extensively used rapid prototyping (RP) technology⁶ with its ability to build prototypes of complex and intricate shapes shorting production cycles and reducing cost.⁷ Polyether ether ketone (PEEK), with excellent biocompatibility and high temperature resistance, similar density and mechanical properties to natural bone tissue,^8,9 has been used as replacement for mental in some industrial fields.^10,11 However, the mechanical properties and dimensional accuracy of PEEK-printed parts to meet the requirements of orthopedic applications are still a hot topic.^12,13

Researchers have studied the impact of 3D printing parameters on FDM metrics in order to improve the condition of parts and guarantee reliable structural performance by maximizing mechanical properties.^14,15 Three-point bending of the specimens printed horizontally was examined.¹⁶ The effects of the printing path and printing speed on the mechanical properties of PEEK samples were evaluated.^17,18 The maximum strength of PEEK (74.2 MPa) was achieved with a printing speed of 30 mm/s and a printing raster angle of ±15°. Using the Taguchi method for experimental design, it was concluded that a decent Young's modulus could be obtained by setting the nozzle temperature, printing speed and bed temperature to high levels and setting the layer thickness and radiation temperature to low levels.¹⁹ Experiments were conducted on cylindrical parts fabricated by nanocomposite deposition to examine the influence of build orientation on compressive strength. The results indicated that the oblique specimens possessed the greatest compressive strength compared to the axial specimens.²⁰ Nozzle temperature and layer height were the process parameters that significantly influenced the resultant roughness as well as the elastic modulus measured by using the Taguchi method along with the analysis of variances.²¹ The new fractal structure improved the first cracking strength and residual bearing capacity in the three-point bending test.²² Different FDM process parameters: printing speed, layer thickness, printing temperature and extrusion width were carried out by Taguchi method and the tensile strength of the specimens was improved.^23–25 The best comprehensive mechanical properties were achieved for pure PEEK fabricated along the horizontal zero-degree direction.²⁶

Lot of studies have focused on the effect of printing speed, layer thickness, nozzle temperature, infill pattern, and infill density on the tensile or flexural properties of FDM-3D printed specimens using the Taguchi DOE method.^27–29 However, less attention is paid to the printed parts’ attributes (e.g., dimensional accuracy, weight and compressive properties), which results in printed parts that do not meet the needs of the clinical application. The mismatch in properties contributes to the loosening or dislodging of implants. To a certain extent, a solution to such a problem is urgently required. Customization to achieve the demands of the application is becoming a practical mainstream. The objective of the present study was to optimize the process parameters by employing the Taguchi DOE method to seek the best combination of each response variable to meet different needs. The process parameters were pore size, infill angle and printing temperature. Analysis of variance (ANOVA),^30,31 S/N ratio and means plots were used to help determine the optimal level of each process parameter. Predictive analysis was exploited to define a set of parameters.

Experimental

Preparation of Filaments and Scaffolds

Polyetheretherketone (PEEK 550G, Jilin Joinature Polymer Co., Ltd; China) was selected as the building material. Prior to preparing filaments, PEEK was dried at 120°C overnight to remove any moisture before the extrusion. The dried PEEK pellets were performed in a twin-screw extruder (SJZS-10B, Wuhan Ruiming Experimental Instrument Co., Ltd; Wuhan, China), from which printed filaments for fused filament fabrication (FFF) with a diameter of 1.75 ± 0.1 mm were extruded and wound with a micro traction winder (Wuhan Ruiming Experimental Instrument Co., Ltd; Wuhan, China). Before printing, the filaments were put into the oven at 120°C for 2 h to remove moisture. A 3D model of a cubic porous scaffold sample (Figure 1) of 10 mm in length, width and thickness was modeled using Solidworks software and exported in STL format. The import file was converted to the G-code by slicing software (IEMAI 3D EXPERT). Finally, the prepared PEEK filament was printed using a high-performance 3D printer (MAGIC-HT-M, Imai Intelligent Technology Co., Ltd; Dongguan, China) to produce the scaffolds at a print speed of 30 mm/s with 0.2 mm layer thickness.

Figure 1.

The schematic drawing of 3D printing process.

Variable printing parameters

Among the FDM printing parameters, a wide variety of printing parameters can significantly affect the mechanical properties of the printed parts. Taking into account most of the previous studies, pore size, raster angle and printing temperature were chosen as the main FDM printing parameters to be explored in this paper. Each of the printing parameters considered was assigned three levels to be investigated as shown in Table 1.

Table 1.

Variable printing parameters and their levels.

Parameters	Symbol	Level				Unit
Parameters	Symbol	1	2	3	4	Unit
Pore size	A	0.3	0.5	0.7	0.9	Millimeters, mm
Raster angle	B	0, 90	15, −75	30, −60	45, −45	Degree, °
Printing temperature	C	410	420	430	440	Degrees Celsius, °C

Mechanical properties and measurement accuracy

For the purpose of evaluating the effects of pore size, raster angle, and printing temperature on the mechanical properties and material consumption of the scaffolds, a caliper with an accuracy of 0.01 mm was used to estimate the length (L), width (W) and thickness (T) of the scaffolds (n = 5). Three measurements were taken for each of the dimensions as shown in Figure 2(a), and they were averaged. In addition, dimensional measurement accuracy was calculated using equation (1).³²

∆ M = | M_{m} - M_{0} |

(1)

where

∆ M

is the deviation between the dimension of 3D model (

M_{0}

) and the measured estimation of the printed sample size (

M_{m}

). The closer

∆ M

is close to zero, the higher the accuracy of the printed sample is.

Figure 2.

3D model of porous scaffolds and testing method of mechanical properties: measurement positions of the scaffolds (a) and scaffolds compressed along X or Z-axis direction (b).

The material consumption of the scaffolds (n = 5) was quantified by the weight of the scaffolds and weighed by electronic balance (FA-2104, Shanghai Sunny Hengping Scientific Instrument Co., Ltd; Shanghai, China) with an accuracy of 0.1 mg. Besides, the scaffolds (n = 5) were compressed parallel along the X-axis printing direction and vertically along the Z-axis printing direction (Figure 2(b)), respectively, with a compression speed of 1 mm/min. The compressive test of the samples employed an electronic universal testing machine (UTM-4204, Shenzhen Sansi Zongheng Technology Co., Ltd; Shenzhen, China). The data obtained from the test were pre-processed using the Excel software to calculate the values of compressive strength and modulus for each scaffold.

Design of experiments

To analyze the effect of printing parameters on the response variables, the L16 orthogonal array was adopted. For each unique combination of the 16 trials included in the design, five replications were performed in a randomized order. The S/N ratio has three types, including nominal is the best, larger the better, and smaller the better. According to the principle that the variability decreases when the S/N ratio reaches a maximum, the weight of the scaffold and the dimensional accuracy were analyzed using the criterion of the S/N ratio smaller the better. For the compressive strength and modulus, the tendency was to select the S/N ratio larger the better, because these variables should be maximized for enhancing the mechanical performance of the part. The S/N ratio “larger the better” was obtained using equation (2).³³ “Smaller the better” was calculated through equation (3).³³

S / N = - 10 \log [(\sum (1 / {y_{i}}^{2}) / n)]

(2)

S / N = - 10 \log [(1 / n) (\sum {y_{i}}^{2})]

(3)

Where

y_{i}

is called the observed value, and

y_{0}

is a target value.

ANOVA was used to estimate the significance of each parameter, determine the percentage contribution of each process parameter and determine the percentage contribution of the error. In the analysis process, 0.05, the significant level, was used. The equations (4)–(6)³³ are required in the ANOVA calculation as follows.

S S A = \sum_{i = 1}^{r} \sum_{j = 1}^{m} {(\bar{y_{i \cdot}} - y)}^{2}

(4)

S S E = \sum_{i = 1}^{r} \sum_{j = 1}^{m} {(y_{i j} - \bar{y_{i \cdot}})}^{2}

(5)

where

S S A

and

S S E

are the sum of the square between groups and the sum of square error, respectively.

\bar{y_{i \cdot}}

is the mean of

i

th group.

y

is the overall mean. The value of the

j

th level of the

i

th factor is described by

y_{i j}

. The

r

represents the number of factors, whose level is denoted by m.

F (r - 1, n - r) = \frac{S S A / r - 1}{S S E / n - r}

(6)

F

is a statistic that is used to judge whether the test is significant or not.

r - 1

and

n - r

are the degrees of freedom of

S S A

and

S S E

Results and discussion

DOE plays a critical role in the whole quality control process. The Taguchi method is the most common quality engineering technique,³⁴ which has been widely used in various process optimization and product design studies.³⁵ The method involves the use of an orthogonal array for performing experimental trials, a signal-to-noise ratio plot for reducing process variability, and a means graph for tuning the process to the desired values, which chooses the control level combination of the parameters. In this case, the Taguchi method was selected to reduce the number of experiments required as well as to obtain the significance and sensitivity of the parameters.³⁶

Table 2 presented the average values of the response variables corresponding to each experimental group of the L16 orthogonal array. As the objective of the current work was to optimize the process parameters, it did not focus too much on explaining the impacts of process parameters on mechanical properties. Therefore, by analyzing the main effects plots for means and signal-to-noise ratios, the optimal combination of process parameters was ascertained.

Table 2.

Experimental results of Taguchi’s L16 orthogonal array.

No.	ΔL (mm)	ΔW (mm)	ΔT (mm)	Materials consumption (g)	Compressive strength-Z (MPa)	Young’s modulus-Z (MPa)	Compressive strength-X (MPa)	Young’s modulus-X (MPa)
1	0.33	0.32	0.13	0.7058	46.557	241.373	12.822	207.298
2	0.28	0.31	0.09	0.6751	45.563	211.623	12.003	177.039
3	0.42	0.38	0.15	0.6920	40.248	293.451	5.322	128.034
4	0.57	0.28	0.19	0.6952	42.710	266.516	4.837	108.615
5	0.08	0.08	0.14	0.5315	29.207	214.535	9.094	140.831
6	0.33	0.32	0.14	0.5413	28.765	225.841	8.886	96.488
7	0.36	0.41	0.05	0.5444	20.901	252.843	3.286	89.902
8	0.42	0.46	0.08	0.5441	29.107	247.177	4.219	101.132
9	0.09	0.11	0.06	0.4361	17.601	241.793	5.626	144.977
10	0.26	0.34	0.10	0.4470	16.729	189.472	2.795	45.974
11	0.41	0.41	0.04	0.4574	21.704	219.610	4.119	56.860
12	0.60	0.55	0.06	0.4423	16.306	215.812	3.904	55.959
13	0.24	0.29	0.08	0.3399	11.596	169.417	2.426	53.518
14	0.31	0.32	0.08	0.3767	10.721	223.579	1.688	44.853
15	0.43	0.45	0.05	0.3807	11.258	158.918	2.145	23.756
16	0.50	0.48	0.08	0.3832	12.385	171.504	1.987	22.026

To eliminate the impact of the scaffolds variation on experimental results, five scaffolds were printed for each trail to repeat the experiment. Hence, Figure 3 presented only one of stress-strain curves from the 16 experiments listed in Table 2. As observed from Figure 3 and Table 2, sample one shown the maximum compressive strength when compressed along the Z-axis and X-axis, respectively. The maximum modulus for compression along Z-axis was sample 3, while the largest modulus could be obtained for X-axis compression by sample 1. The sample 14 also exhibited minimal compressive strength whether along the Z direction or X direction. With compression along Z-axis, the minimal modulus was found from sample 15, whereas sample 16 shown the minimum modulus for compressing along X-axis. And the failure process could be acquired in Figure 3.

Figure 3.

Stress versus strain curve. (a) Compression along Z-axis. (b) Compression along X-axis.

Mian effect plots of means and S/N ratio

The means analysis and the S/N ratios graphs for each level of each variable correspond to the Figures 4 and 6 and Figures 7 and 8, respectively.

Figure 4.

Mean effect plot. (a) $∆ L$ . (b) $∆ W$ . (c) $∆ T$ . (d) Materials consumption.

At the printing temperature of 420°C, the smallest deviation in dimensional accuracy about $∆ L$ (Figure 4(a)) was achieved when the second level of pore size and infill angle of 0°.^13,17 Meanwhile, the value of the process parameters that increased $∆ W$ (Figure 4(b)) was the same as the ones that enhanced $∆ L$ . When the infill angle increased, the missing printing of the scaffolds, as shown in black circles and arrows in Figure 5, would become increasingly obvious, which led to a decrease in dimensional accuracy. The missing printing always occurred at the corners of printing specimens. The filament, extruded from the nozzle and supposedly deposited on the corners, could not adhere well to the previously deposited layer or the printing nozzle leaves the deposited position before the filament was extruded. The pore size of 0.5 mm was exactly a multiple of the scaffold size, which explained why the variables had the lowest variance values when the pore size was at its second level. As for the printing temperature, when it rises, the viscosity of the PEEK decreases further and the flowability increases.²⁵ When the printing temperature was at its fourth level, it was less viscous and more fluid than other levels,³⁷ resulting in more likely non-programmed extrusion of the material during the printing process, and the high printing temperature made the molten deposition material solidify slowly and stretch to both sides,³⁸ all of which made the printed parts accuracy most severely degraded.

Figure 5.

Schematic diagram of infill angle.

The dimensional deviation corresponding to $∆ T$ (Figure 4(c)) decreased when the pore size was 0.7 mm applying at 0° as the infill angle, with the third level of printing temperature. Similar to the results for $∆ L$ and $∆ W$ , the difference in thickness was minimal at the infill angle of 0°, which meant that the reduction in printing deficiency could improve the printing accuracy. Unlike the optimal printing temperature for $∆ L$ and $∆ W$ in this research, the optimal printing temperature for $∆ T$ was 430°C, probably due to the high temperature, which caused more material to be extruded non-programmatically, thus increasing the layer thickness and making the scaffolds less inaccurate. As could be seen from the means plot (Figure 4(d)), as the pore size increases from 0.3 to 0.9 mm, the material consumption per scaffold decreased from about 0.7 g to about 0.37 g. Cause the larger the pore size, the less the actual volume.

The effect diagrams of compressive strength and the modulus were shown in Figure 6. On the compressive strength of the scaffolds compressed along the Z and X axes respectively, the compressive strength was maximum at the lowest values of pore size and infill angle, with the printing temperature of 440°C. The compressive strength compressed along Z and X direction decreased from 43.4 MPa to 11.5 MPa and from 8.8 MPa to 2.1 MPa as the pore size enlarged from 0.3 mm to 0.9 mm. Young’s modulus along the Z-axis was significantly higher than that of scaffolds along the X-axis with the same pore size. The Young’s modulus of the scaffold ranges from 253.3 MPa to 36.1 MPa. To some extent, the printing temperature would affect the compressive strength. Compared to the X-axis, the increase in compressive strength in the Z-axis was smaller as the printing temperature increases. The mechanical properties obtained for the scaffolds were similar to those of natural bone at all factor levels upon compression along the x-axis, unless the mechanical properties of the scaffolds with pore sizes of 0.7 mm and 0.9 mm and infill angles of 30° and 45° were less satisfactory.

Figure 6.

Mean effect plot. (a) Compressive strength-Z. (b) Young’s modulus-Z. (c) Compressive strength-X. (d) Young’s modulus-X.

Figures 7 and 8 showed the main effect plots of all the different levels of the three parameters on the S/N ratios of the response variables. The degree of process parameters of the optimization variables was equivalent to the reduction of process variations and enhancement of mechanical properties values, which applied to the materials consumption, compressive strength and modulus, dimensional accuracy of the scaffold’s length, width and thickness. From Figure 8, the trends of compressive strength were and Young’s modulus along the Z and X axes were approximately the same.

Figure 7.

S/N ratio plot. (a) $∆ L$ . (b) $∆ W$ . (c) $∆ T$ . (d) Materials consumption.

Figure 8.

S/N ratio plot. (a) Compressive strength-Z. (b) Young’s modulus-Z. (c) Compressive strength-X. (d) Young’s modulus-X.

Results of analysis of variance

To investigate whether different levels of the factors have significant effect on the response variables, ANOVA, a significance test method, was carried out. In the test, the significance level, α, was set as 0.05.

Tables 3–10 were the outcomes of ANOVA for each one of the response variables. The dimensional variation of the porous scaffold’s length and width was influenced by the infill angle as could be seen in Tables 3 and 4. Although the results given in Table 5 pointed out that pore size, infill angle and printing temperature did not have a significant effect on the thickness, it could be assumed from the mean plots and SN plots that the third level of pore size and infill angle had some effect on the thickness. Table 6 presented the ANOVA result of the variables on the deviation of the thickness. From the table, the P-value per variable on the thickness deviation was greater than 0.05, which gives the conclusion that the variables on the thickness deviation were considered insignificant. However, the results of ANOVA were used to provide decision makers with a better understanding of the relationship between the independent and dependent variables for reference, instead of using the ANOVA results as decisions. The results of ANOVA in Table 6 indicate that the effect of the factors on the response variables was not significant at the 0.05 level of significance. But considering the mean effect and the S/N ratio plot of

∆ T

, it could be concluded that the pore size was a significant factor. The quantity of material consumption was affected by pore size. The pore size was the most significant factor for compressive strength-Z, Young’s modulus-Z, compressive strength-X and Young’s modulus-X.^26,39 And printing temperature affected not only Young’s modulus-Z, but also compressive strength-X. The infill angle has a significant effect on the mechanical properties of compression along the X-axis.

Table 3.

Results of ANOVA for $∆ L$ .

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	0.011839	0.011839	0.003946	0.38	.772
Infill angle	3	0.144604	0.144604	0.048201	4.63	.050
Printing temperature	3	0.009110	0.009110	0.003037	0.29	.830
Residual error	6	0.062445	0.062445	0.010408
Total	15	0.227998

Table 4.

Results of ANOVA for $∆ W$ .

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	0.02365	0.02365	0.007884	1.47	.315
Infill angle	3	0.25149	0.25149	0.083830	15.61	.003
Printing temperature	3	0.01498	0.01498	0.004995	0.93	.482
Residual error	6	0.03222	0.03222	0.005370
Total	15	0.32235

Table 5.

Results of ANOVA for $∆ T$ .

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	0.015225	0.015225	0.005075	2.86	.126
Infill angle	3	0.002538	0.002538	0.000846	0.48	.710
Printing temperature	3	0.000877	0.000877	0.000292	0.16	.916
Residual error	6	0.010636	0.010636	0.001773
Total	15	0.029276

Table 6.

Results of ANOVA for material consumption.

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	0.230933	0.230933	0.076978	502.45	.000
Infill angle	3	0.000562	0.000562	0.000187	1.22	.380
Printing temperature	3	0.000593	0.000593	0.000198	1.29	.360
Residual error	6	0.000919	0.000919	0.000153
Total	15	0.233008

Table 7.

Results of ANOVA for compressive strength-Z.

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	2294.26	2294.26	764.753	115.02	.000
Infill angle	3	12.69	12.69	4.230	0.64	.619
Printing temperature	3	34.99	34.99	11.664	1.75	.255
Residual error	6	39.89	39.89	6.649
Total	15	2381.83

Table 8.

Results of ANOVA for Young’s modulus-Z.

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	11,471.2	11,471.2	3823.7	11.00	.007
Infill angle	3	836.9	836.9	279	0.80	.536
Printing temperature	3	5617.4	5617.4	1872.5	5.39	.039
Residual error	6	2086.1	2086.1	347.7
Total	15	20,011.7

Table 9.

Results of ANOVA for compressive strength-X.

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	99.695	99.695	33.2316	62.20	.000
Infill angle	3	43.351	43.351	14.4502	27.05	.001
Printing temperature	3	39.914	39.914	13.3046	24.90	.001
Residual error	6	3.205	3.205	0.5342
Total	15	186.165

Table 10.

Results of ANOVA for Young’s modulus-X.

Factor	DoF	SS	Adj SS	Adj MS	F	p
Pore size	3	30,430	30,430	10,143.2	30.82	.000
Infill angle	3	10,757	10,757	3585.5	10.89	.008
Printing temperature	3	2108	2108	702.5	2.13	.197
Residual error	6	1975	1975	329.1
Total	15	45,269

Prediction and verification analysis

Enhancing one of the response variables by adjusting the process parameters might yield less benefits for the others. Hence, the best level for the factors that have a major influence on the experiment would be used, while for the minor factors, the appropriate level could be selected according to the established conditions. From the mean effect and S/N ratio plots of the response variables in this study, the optimal treatment combination corresponding to each response variable was summarized in Table 11.

Table 11.

The optimal prediction combination corresponding to each response variable.

Response variables	Combination	A	B	C
$∆ L$ (mm)	A₂B₁C₂	0.5	0	420
$∆ W$ (mm)	A₂B₁C₂	0.5	0	420
$∆ T$ (mm)	A₃B₃C₃	0.7	30	430
Material consumption (g)	A₄B₁C₁	0.9	0	410
Compressive strength-Z (MPa)	A₁B₁C₄	0.3	0	440
Young’s modulus-Z (MPa)	A₁B₃C₂	0.3	30	420
Compressive strength-X (MPa)	A₁B₂C₃	0.3	15	430
Young’s modulus-X (MPa)	A₁B₁C₂	0.3	0	420

In order to verify the best combinations predicted by Taguchi’s method, a validation test was conducted for each combination, with the average values of the five replications taken for each combination. Table 12 lists the trials values and the differences between the predicted and experimental values. The values of some response variables were 20% higher than that of the predicted values, which included Young’s modulus-Z, compressive strength-X and Young’s modulus-X. The values of

∆ T

, material consumptions and compressive strength-Z were pretty close to the predicted values, within a 10% difference.

Table 12.

Comparison between prediction and verification.

Response variables	Predicted result	Experimental result	Difference, %
$∆ L$ (mm)	0.148	0.044	236.36
$∆ W$ (mm)	0.084	0.032	162.50
$∆ T$ (mm)	0.031	0.033	−6.06
Material consumption (g)	0.356	0.335	6.27
Compressive strength-Z (MPa)	46.315	45.922	0.86
Young’s modulus-Z (MPa)	293.012	301.017	−2.66
Compressive strength-X (MPa)	11.231	17.268	−34.96
Young’s modulus-X (MPa)	209.494	279.322	−25.00

The experimental values of $∆ L$ and $∆ W$ deviated far from the prediction, exceeding 200% and 100% difference, respectively, but they were what we expected to obtain. The reason of that was that the predicted values were obtained by combining different levels of the factors, while the experimental values were derived by selecting the specific level of each factor. Pore size and infill angle had a greater effect on $∆ L$ and $∆ W$ . Than the printing temperature. The pore size of 0.5 mm and infill angle of 0° were the best levels of main factors in the best predicted combination of $∆ L$ and $∆ W$ , similar to the level setting of trial five of the L16 orthogonal array. The values of $∆ L$ and $∆ W$ in trial five were smaller than other tests, which explained why it was still reasonable and desirable for the experimental results to be so much smaller than the predicted ones.

Conclusion

In this study, three process parameters of pore size, infill angle and printing temperature were chosen for they were the parameters that deserve to pay attention to. Several response variables regarding length, width, thickness, material consumption, compressive strength along the Z and X directions, and Young’s modulus along the Z and X directions of the parts of PEEK were optimized using Taguchi method. The optimal process parameters for length and width were A₂B₁C₂. The deviation of thickness achieved a minimum in A₃B₃C₃. Compared to other settings, fabricating with a pore size of 0.9 mm, infill angle of 0°and 410°C printing temperature results in minimal material consumption. According to the “the larger, the better” criterion, the optimum conditions for compressive strength along the Z and X axes and the Young’s modulus along the Z direction could follow the parameters setting of trials 1, two and three in L16 orthogonal array respectively. Young’s modulus along the X direction could be obtained as the maximal value using A₁B₁C₂. The predictive analysis in Taguchi's method predicts other experiments not listed in the orthogonal array with high accuracy. The mechanical strength and Young’s modulus of the porous scaffolds obtained from the L16 orthogonal array and verification array in this paper were similar to natural bones. The printing parameters were optimized to achieve the application demands on cost, dimensional precision, weight or mechanical properties. Therefore, this research had the potential to be extended to other fields as a reference such as customized bone repair.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was financially supported by the Guangzhou Science and Technology Plan Project (SL2022B03J01173), Nanning Local Scientific Research and Technology Development Plan Project (20213122), IER Foundation 2021 (IERF202103), the University Teachers Innovation Research Project (2021SWYY04) are gratefully acknowledged.

ORCID iD

Wuyi Zhou

References

Parandoush

Lin

. A review on additive manufacturing of polymer-fiber composites. Compos Struct 2017; 182: 36–53.

Guo

Zou

Huang

, et al. Study on microstructure, mechanical properties and machinability of efficiently additive manufactured AISI 316L stainless steel by high-power direct laser deposition. J Mater Process Technol 2017; 240: 12–22.

Sun

Qin

Wang

, et al. Fabrication of novel printable artificial bone composites used as cartilage scaffolds by an additive manufacturing process. J Appl Polym Sci 2023; 140(43).

Xing

Zou

Wang

, et al. Fabrication and characterization of SiC whiskers toughened Al2O3 paste for stereolithography 3D printing applications. J Alloys Compd 2020; 828: 154347.

Sood

Ohdar

Mahapatra

. Parametric appraisal of fused deposition modelling process using the grey Taguchi method. Proc IME B J Eng Manufact 2010; 224(1): 135–145.

Bellini

Güçeri

. Mechanical characterization of parts fabricated using fused deposition modeling. Rapid Prototyp J 2003; 9(4): 252–264.

Ngo

Kashani

Imbalzano

, et al. Additive manufacturing (3D printing): a review of materials, methods, applications and challenges. Compos B Eng 2018; 143: 172–196.

Challa

Gummadi

Elhattab

, et al. In-house processing of 3D printable polyetheretherketone (PEEK) filaments and the effect of fused deposition modeling parameters on 3D-printed PEEK structures. Int J Adv Manuf Technol 2022; 121(3-4): 1675–1688.

Saini

Elhattab

Gummadi

, et al. Fused filament fabrication-3D printing of poly-ether-ether-ketone (PEEK) spinal fusion cages. Mater Lett 2022; 328: 133206.

10.

Kang

Zheng

Hui

, et al. Mechanical properties of 3D-printed PEEK/HA composite filaments. Polymers 2022; 14(20): 4293.

11.

Arif

Kumar

Varadarajan

, et al. Performance of biocompatible PEEK processed by fused deposition additive manufacturing. Mater Des 2018; 146: 249–259.

12.

Brenken

Barocio

Favaloro

, et al. Fused filament fabrication of fiber-reinforced polymers: a review. Addit Manuf 2018; 21: 1–16.

13.

Zhou

Chen

, et al. Experimental investigation and numerical simulations of temperature and morphology in material extrusion additive manufacturing. Int J Adv Manuf Technol 2022; 119(7-8): 4863–4876.

14.

Veeman

Palaniyappan

. Process optimisation on the compressive strength property for the 3D printing of PLA/almond shell composite. J Thermoplast Compos Mater 2023; 36(6): 2435–2458.

15.

Maguluri

Suresh

Rao

. Assessing the effect of FDM processing parameters on mechanical properties of PLA parts using Taguchi method. J Thermoplast Compos Mater 2023; 36(4): 1472–1488.

16.

Chacón

Caminero

García-Plaza

, et al. Additive manufacturing of PLA structures using fused deposition modelling: effect of process parameters on mechanical properties and their optimal selection. Mater Des 2017; 124: 143–157.

17.

El Magri

El Mabrouk

Vaudreuil

, et al. Optimization of printing parameters for improvement of mechanical and thermal performances of 3D printed poly(ether ether ketone) parts. J Appl Polym Sci 2020; 137(37): 49087.

18.

Akhoundi

Nabipour

Kordi

, et al. Calculating printing speed in order to correctly print PLA/continuous glass fiber composites via fused filament fabrication 3D printer. J Thermoplast Compos Mater 2023; 36(1): 162–181.

19.

Mohamed

Barhoumi

Lamnawar

, et al. Optimization of fused deposition modeling process parameters using the Taguchi method to improve the tensile properties of 3D-printed polyether ether ketone. Proc Inst Mech Eng Part L 2021; 235(11): 2565–2573.

20.

Lee

Abdullah

Khan

. Optimization of rapid prototyping parameters for production of flexible ABS object. J Mater Process Technol 2005; 169(1): 54–61.

21.

Pulipaka

Gide

Beheshti

, et al. Effect of 3D printing process parameters on surface and mechanical properties of FFF-printed PEEK. J Manuf Process 2023; 85: 368–386.

22.

Farina

Goodall

Hernández-Nava

, et al. Design, microstructure and mechanical characterization of Ti6Al4V reinforcing elements for cement composites with fractal architecture. Mater Des 2019; 172: 107758.

23.

Jiang

Cheng

Lin

, et al. Optimization of FDM 3D printing parameters for high strength PEEK using the Taguchi method and experimental validation. Rapid Prototyp J 2022; 28(7): 1260–1271.

24.

Palaniyappan

Veeman

Narain Kumar

, et al. Effect of printing characteristics for the incorporation of hexagonal-shaped lattice structure on the PLA polymeric material. J Thermoplast Compos Mater 2023; 36(5): 2009–2030.

25.

Wang

Zou

Ding

, et al. Effects of FDM-3D printing parameters on mechanical properties and microstructure of CF/PEEK and GF/PEEK. Chin J Aeronaut 2021; 34(9): 236–246.

26.

Zheng

Kang

Sun

, et al.. Effects of printing path and material components on mechanical properties of 3D-printed polyether-ether-ketone/hydroxyapatite composites. J Mech Behav Biomed Mater 2021; 118: 104475.

27.

Heidari-Rarani

Ezati

Sadeghi

, et al. Optimization of FDM process parameters for tensile properties of polylactic acid specimens using Taguchi design of experiment method. J Thermoplast Compos Mater 2022; 35(12): 2435–2452.

28.

Mohammadian

Rezaei

Azdast

. Microstructure, physical, and mechanical properties of LDPE/UHMWPE blend foams. J Thermoplast Compos Mater 2016; 29(9): 1229–1260.

29.

Singh

Ahuja

IPS

, et al. Metal matrix composite from recycled materials by using additive manufacturing assisted investment casting. Compos Struct 2019; 207: 129–135.

30.

Gaito

Firth

. Procedures for estimating magnitude of effects. J Psychol 1973; 83(1): 151–161.

31.

Lee

Kim

, et al. Measurement of anisotropic compressive strength of rapid prototyping parts. J Mater Process Technol 2007; 187–188: 627–630.

32.

Camposeco-Negrete

. Optimization of printing parameters in fused deposition modeling for improving part quality and process sustainability. Int J Adv Manuf Technol 2020; 108(7–8): 2131–2147.

33.

Sood

Ohdar

Mahapatra

. Improving dimensional accuracy of fused deposition modelling processed part using grey Taguchi method. Mater Des 2009; 30(10): 4243–4252.

34.

Burdick

Borror

Montgomery

. A review of methods for measurement systems capability analysis. J Qual Technol 2003; 35(4): 342–354.

35.

Mahmood

Qureshi

Talamona

. Taguchi based process optimization for dimension and tolerance control for fused deposition modelling. Addit Manuf 2018; 21: 183–190.

36.

Kacker

Lagergren

Filliben

. Taguchi's orthogonal arrays are classical designs of experiments. J Res Natl Inst Stand Technol 1991; 96(5): 577–591.

37.

Garg

Singh

. Investigations for melt flow index of Nylon6-Fe composite based hybrid FDM filament. Rapid Prototyp J 2016; 22(2): 338–343.

38.

Wang

Zou

Xiao

, et al. Effects of printing parameters of fused deposition modeling on mechanical properties, surface quality, and microstructure of PEEK. J Mater Process Technol 2019; 271: 62–74.

39.

Zheng

Zhao

Dong

, et al. Additively-manufactured PEEK/HA porous scaffolds with highly-controllable mechanical properties and excellent biocompatibility. Mater Sci Eng C 2021; 128: 112333.