J Mater Sci COMPOSITES Composites

Manufacturing and 3D printing of continuous carbon fiber prepreg filament Qingxi Hu1,3,4, Yongchao Duan1,2, Haiguang Zhang1,3,4,*, Dali Liu2,*, Biao Yan2, and Fujun Peng2 1

Rapid Manufacturing Engineering Center, School of Mechatronical Engineering and Automation, Shanghai University, Shanghai 200444, China 2 Shanghai Key Laboratory of Spacecraft Mechanism, Aerospace System Engineering Shanghai, Shanghai 201108, China 3 National Demonstration Center for Experimental Engineering Training Education, Shanghai University, Shanghai 200444, China 4 Shanghai Key Laboratory of Intelligent Manufacturing and Robotics, Shanghai University, Shanghai 200444, China

Received: 3 June 2017

ABSTRACT

Accepted: 20 September 2017

The current research proposes a novel method for printing continuous carbon fiber composite parts. At first, continuous carbon fiber prepreg filament for Fused Deposition Modeling 3D printing was manufactured, followed by modification of extruder head of 3D printers to print the filament. Thereafter, threepoint flexural test and Response Surface Methodology were adopted to study the mechanical properties of the composite parts printed with the filament. After testing, a mathematical model was developed to describe and analyze the relationship between the printing parameters (printing temperature, printing speed, and layer thickness) and the flexural strength of printed composite parts. We discovered that the flexural strength and flexural modulus of printed composites significantly improved with the proposed method with specified printing parameters, and while of all the parameters, the layer thickness had the greatest contribution towards the final flexural strength. The results indicate that the discussed method could be a promising approach to print CCF composites.

Ó

Springer Science+Business

Media, LLC 2017

Introduction The 3D printing technology, also known as additive manufacturing, has had a worldwide spread lately owing to its rapid modeling capability [1]. Fused Deposition Modeling (FDM) is the most widely used 3D printing technology due to its employment of inexpensive devices and ease of use [1–3] to build complex parts with thermoplastic filaments, such as PLA, ABS, or Nylon. However, the parts printed with

thermoplastic filaments suffer from a major drawback of poor mechanical properties which restricts their applications [4, 5]. In recent years, researchers have enhanced the mechanical properties of thermoplastics by combining them with reinforced materials [6–8]. Owing to its outstanding mechanical performance and lightweight characteristics, carbon fiber plays a crucial role in the field of composites. Short Carbon Fiber (SCF) is a noteworthy reinforced material because of

Address correspondence to E-mail: [email protected]; [email protected]

DOI 10.1007/s10853-017-1624-2

J Mater Sci

the relatively simple manufacturing procedure of SCF reinforced thermoplastics [8]. In general, the SCF and thermoplastic pellets are mixed in a blender for uniform distribution and the resultant well-distributed compound is extruded through a screw extruder to form short carbon fiber-reinforced plastic (SCFRP) filament [9, 10]. Ning et al. studied the mechanical performance of printed composite part with SCFRP filament while Tekinalp et al. compared the samples printed by SCF reinforced ABS filament with the samples prepared by compression molding from the same material [11, 12]. It was found that although the parts printed with SCFRP could improve the mechanical properties, these properties were only slightly better than those of pure plastic while additional porosity and poor bonding could be detected because of the existence of SCF [13, 14]. The ideal solution to achieve significant strength improvement is printing with Continuous Carbon Fibers (CCF) [13]. CCF has less specific surface area than SCF, which reduces the possibility of pore formation. Furthermore, former shows much better mechanical properties than the latter [15]. According to the majority of the reported research work until now, there are two primary methods to manufacture parts of continuous carbon fiber by FDM. As shown in Fig. 1a, b, the methods include real-time 3D printing of CCF and embedding of CCF after 3D printing with former grabbing most of the attention. The method is realized by redesigning the extrusion head to simultaneously get the continuous carbon fibers and pure thermoplastic filament while CCF can be impregnated in the extruder head in the process of printing [16–18]. As a result, the continuous carbon fiber reinforced plastic (CCFRP) parts can be printed in a single step and the content of carbon fiber can be adjusted by changing extruding speed of pure thermoplastic filament or CCF at the same time. Figure 1 Current methods for manufacturing parts reinforced with continuous carbon fiber by FDM; a real-time 3D printing with CCF; and b embedding of CCF after 3D printing.

Matsuzaki et al. printed CCF reinforced PLA tensile test specimens by this method and concluded that there was a great enhancement in mechanical properties [19]. Tian et al. studied the influence of printing parameters (temperature of liquefier, layer thickness, feed rate of filament, etc.) on the flexural strength and modulus of printed specimens by this method. Embedding of CCF after 3D printing seems like an indirect way to use CCF as reinforcement in 3D printing [20]. Mori et al. studied this method and first manufactured a lower plate and upper plate by FDM; thereafter, CCF was sandwiched between both plates; and finally, the three parts were bonded by thermal treatment. The results showed that the strength increased to almost double of the previous values using this method [21]. As suggested via the introduction of the two methods discussed above, the former method might be a quicker way to print composite parts with CCF but it is not an appropriate technique to achieve decent bonding between the matrix and the reinforcement. This is because the extrusion head cannot produce sufficient pressure to squeeze the melting resin into CCF; besides, the short time of impregnation of CCF is another weakness. As regards the latter, the method is much complicated, and also cannot make sure that the CCF obtains good impregnation. Taking above discussed factors into consideration, this paper presents a novel method to print CCF. At first, the CCF prepreg filament suitable for FDM printer was manufactured and thereafter the filament was printed using a common FDM printer with a simply modified extrusion head. In the first step, the CCF filaments have a high probability of good impregnation when manufactured with professional devices while the second step of 3D printing can achieve the second impregnation for CCF reinforcement.

J Mater Sci

Design and methods Device design for manufacturing of CCF prepreg filament As shown in Fig. 2, the single screw extruder and coaxial extrusion moulds were employed to manufacture the CCF prepreg filament. The single screw extruder causes the molten resin to have higher pressure, which could squeeze the molten resin into the CCF filaments in the nozzle. The coaxial extrusion mould, especially its nozzle, apart from ensuring that the resin easily squeezes into the filaments, also produces resin membrane to protect the inner CCF filament. Before it entered into the coaxial extrusion mould, the CCF was heated in heating pipe to keep it dry and warm and to ensure that the filaments are easily impregnated with the molten resin. After impregnation, the CCF prepreg filament entered into the cooling water for solidification while in the next step, a blower was used to keep the filament dry. The dragger machine provided the power to maintain the whole manufacturing process at a constant speed.

Extruder head design for printing CCF prepreg filament The extruder head was simplified as shown in Fig. 3. Since CCF could not be broken by melting and could be dragged down by printed CCF prepreg filament bonded with the base plate, there was no need for extruder motor to give the filament. The PTEF (Polytetrafluoroethylene) tube should go through the extruder head from the entrance of guide pipe until the end of the nozzle to make sure that the filament would get uniformly heated and also reduce friction from the metal guide pipe and nozzle. Additionally, the nozzle was also replaced with the one with a Figure 2 Schematic for manufacturing 3Dprintable CCF prepreg filament.

Figure 3 Schematic of 3D printing extruder head with 3Dprintable CCF prepreg filament.

bigger diameter to prevent clogging and fracture of filament while printing.

Experiment method The mechanical properties of printed composite parts were investigated to estimate the potential of the method of printing CCF composites. In this study, we chose flexural strength and flexural modulus as a measure of the mechanical properties of printed composite parts. According to the standard of GB/T 1449:2005, the three-point flexural tests were performed in the microcomputer-controlled electronic testing machine (SongDun WDW-1). The flexural strength (rt ) was calculated using Eq. (1), rf ¼

3P  l 2b  h2

ð1Þ

where P is the failure load (N), l is the distance between the two supports while testing, b is the width of specimen, and h is the thickness of the specimen.

J Mater Sci

The flexural modulus (Ef ) was also calculated in this section using Eq. (2): Ef ¼

l3  DP 4b  h3  DS

ð2Þ

where l, b, and h have the same meaning as the variables in Eq. (1); DP is the increment of load in the initial linear part of the load & deflection curve; and DS is the increment of deflection corresponding to DP at the midpoint of l. It is interesting to notice that the printing process can further impregnate the CCF filament which can additionally enhance the mechanical properties of printed composite parts. Hence, the relationship between the printing parameters and the flexural strength was explored using response surface methodology (RSM). RSM is a design of experiment (DOE) method which is widely used to explore the relationships between explanatory variables and response variables. This method can obtain reasonable results through a small number of experiments, which can improve the experiment efficiency and reduce the experimental cost and time [22, 23]. At the end of the experiment, the mathematical model is developed to describe the relationship between the printing parameters and the flexural strength. In this study, the quadratic regression model was used to fit the experimental data of factors and response and the model employed can be expressed via Eq. (3), Y ¼ a0 þ

k X i¼1

ai X i þ

k X i¼1

aii Xi2 þ

k X

aij Xi Xj

ð3Þ

i\j

where Y represents the fitted response; Xi and Xj are the variables; ai, aj, and aij are the coefficients of every term and k is the total number of variables.

Materials and experiment CCF prepreg filament manufacturing The continuous carbon fiber bundle made of 1000 single carbon fiber filaments shown in Fig. 4a (T300–1000, Toray) was used as the reinforcement of composite. As shown in Fig. 4b, the thermoplastic raw materials used are Polylactic Acid (PLA) pellets (Nature Works 4043D) while Fig. 4c, d shows that the CCF prepreg filament was manufactured via coaxial

extrusion mould. Figure 4e shows the CCF prepreg filament with diameter of 1.2 mm which was manufactured by a 1 mm-nozzle in mould while Fig. 4f shows the cross-section of CCF prepreg filament where it is noticeable that only a tiny amount of CCFs are impregnated with PLA resin and most of the resin adheres to the surface of bundle.

CCF prepreg filament 3D printing In the 3D printing process, the Mendel type of FDM 3D printer, which was assembled by open-source modules, was employed (see Fig. 7a). The common FDM printer nozzle in extruder head was replaced by a larger 1.5-mm diameter one for unrestrained printing of the filament. This is because that a small nozzle can cause clogging and fracture of the filament in the small-diameter nozzle. The opensource 3D printing software Repetier-Host was adopted to control the movement of FDM 3D printer while the 3D printer was installed with the Marlin firmware. Thereafter, the impact of printing speed and layer thickness was explored to understand the relationship with impregnation of printed prepreg filament. The printing speed can influence the residence time of prepreg filament in the extruder head and further affect the melting degree of PLA resin. Hence, the printing speed was replaced by the residence time to observe the desired relationship. From Fig. 5a, it can be observed that longer the original prepreg filament stayed in the extruder, which was a result of lowering the printing speed, better the bonding between the PLA and CCF is. Especially, at the resident time of 200 s, the number of pulled-out CCF filaments is extremely less. It is noteworthy that there was no pressure on CCF in the test and impregnation could be the result of the adsorption of micro-pores formed by carbon fiber bundles. The layer thickness also significantly affects the interlaminar bonding and impregnation of printed prepreg filament [3, 5, 24]. Hence, we also performed the experiment to explore the relationship between the layer thickness and impregnation of printed prepreg filament. As shown in Fig. 5b, from the number of pulled-out CCF filaments it could be concluded that the bonding has a negative relationship with layer thickness, and 0.3 mm-layer thickness resulted in complete impregnation.

J Mater Sci

Figure 4 The raw materials and production of 3D-printable CCF prepreg filament; a 1 k continuous carbon fiber bundle, b polylactic acid (PLA) pellets, c device for manufacturing CCF prepreg

Figure 5 Influence of the a printing speed and b layer thickness on the printed prepreg filament.

RSM experiment In this experiment, three printing parameters, namely printing temperature, printing speed, and layer

filament, d CCF prepreg filament pulled out from the impregnation mould, e production of 3D-printable CCF prepreg filament, and f cross-section schematic of 3D-printable CCF prepreg filament.

thickness were taken into consideration to study the relationship with flexural strength of printed composite parts. In the case of printing temperature, since molten PLA resin has good flowability when printing temperature is between 200 and 230 °C and it thermally degrades at a temperature higher than 230 °C [25, 26], 200–230 °C was used as the printing temperature range in RSM experiment. Also, in the case of printing speed experiments, it could be found that the bonding would be better when the residence time was over 200 s in the previous experiment. However, it would seriously affect the efficiency. This is due to the fact that the speed needs to be below 0.15 mm/s considering the length of the hot path of extruder head in the experiment was only 30 mm. Therefore, the final three levels of printing speed chosen were 60, 90, and 120 mm min-1. Finally, in the case of the layer thickness, although the 0.5 mm layer thickness filament could be printed, the success rate was low. This is because the molten resin could overflow in the PTEF tube due to the small layer thickness resulting in less molten resin printing, and the overflowed molten resin would solidify in the entrance of extruder head, further causing the fracture of CCF.

J Mater Sci

Hence, the three levels of layer thickness chosen were 0.6, 0.9, and 1.2 mm in this study. The three levels of three factors are listed in Table 1. The 3D-printed flexural specimens were designed to 60 mm in length and 11 mm in width and since height is an explanatory variable in the current study, all the specimens were printed three layers in height. The printing path of flexural specimens was designed according to the anisotropy of carbon fibers to enhance the mechanical properties (see Fig. 6b). The printing parameters of 15 experiments were designed and the flexural specimens were printed according to the specified design dimensions and printing parameters (see Fig. 6a). Finally, the threepoint flexural test was conducted on all specimens (Fig. 7). The printing parameters and final flexural strength are listed in Table 2.

Results and discussion Flexural property analysis In Fig. 8, stress–strain curves of five flexural specimens are shown in detail. The specimens are selected according to flexural strengths of around 100, 200, 300, 400, and 500 MPa, respectively. It can be observed from the figure that stress–strain curves are divided into four stages. In the beginning, the curve is almost a straight line and the printed composite specimens exhibit a linear stress–strain relationship. This is because both the reinforcement (carbon fiber) and the matrix (PLA resin) showed elastic properties in the beginning of the flexural test. At this stage, the slope of the linear part of the curve corresponds to Modulus of Elasticity. As the flexural strain increased further, at a sufficiently large strength, the slope of the curve began to reduce, for instance, in the curves

Table 1 Three-factor and three-level central composite design used in present study Factors

A: Printing temperature (°C) B: Printing speed (mm min-1) C: Layer thickness (mm)

Coded levels -1

0

1

200 60 0.6

215 90 0.9

230 120 1.2

for 4th, 7th or 15th specimens, the stage could be noticed clearly and it shortened with an increase in strength. However, in the cases of 5th and 12th specimens, there was no distinctly observable phenomenon. At this stage, some of the CCF filaments gradually pulled out from the matrix, which indicates that these filaments were not well impregnated and could no longer work as reinforcement; hence it might result in a decrease of the change rate of flexural stress. The flexural stress still increased further to a maximum value, although the rate of increase became lower than the last stage. In the next stage, after reaching the maximum flexural stress, the CCF filament which could bear most of the load cracked in a short amount of time. This is suggested from the curve where the flexural stress drops rapidly for a minuscule amount of change in flexural strain. Thereafter, in the last stage, the only remaining matrix resin bears the subsequent load and the fractured reinforcement CCF filaments do not value anymore. There are certain differences between 12th, 5th, and other three specimens in the last stage; the plausible reason for the differences could be that the resin in the 4th, 7th, and 15th specimens did not impregnate CCF well. Therefore, these parts shows better plastic properties in the last stage where the curves did not decline, whereas the resin of the 5th and 12th specimens had lots of remaining holes after the CCF filaments pulled out, hence, in these parts stress descended continuously with the increase in strain. Additionally, the flexural specimens with pure PLA which had 100% infill were printed and tested and the flexural strength was measured as 74.161 MPa. The curve could be clearly observed as different from the other curve with filament reinforcement that does not have a fracture region and the flexural modulus is as low as 2.3 GPa. Figure 8 also compares the flexural strength obtained in this study with the ones for other specimens that were prepared using different methods and materials. It can be observed that the maximum flexural strength obtained in this research comes to 541.65 Mpa, which is more than double the strength of the specimens prepared by the method of Solution casting and hot press method with 25% carbon-fiber reinforced PLA composites [27]. Also, the maximum flexural strength improved by 61.7% as compared to the value of 335 MPa obtained by the first method (real-time 3D printing of CCF) and exceeds flexural

J Mater Sci

Figure 6 The flexural specimen. a The 3D printed flexural specimen with CCF prepreg filament. b The size and 3D printing path (black lines) of designed flexural specimen.

Figure 7 The experimental devices. a FDM printer for printing the CCF prepreg filament and the controlling software. b The flexural specimen in the printing process. c The flexural test

process of flexural specimen in the microcomputer-controlled electronic testing machine.

strength of 3D printing part printed with Markforged 3D printer and its own carbon CFF materials [20, 28]. Also, the flexural modulus of PLA or PLA matrix composites manufactured via different methods are listed in Table 3. In the current research, the highest flexural modulus reaches to 40.13 GPa, which is more than ten times than that of pure PLA filament [29].

Also, the highest flexural modulus increased by 14.3% as compared to the flexural modulus of material prepared by solution casting and hot press method with 25% short carbon fiber-reinforced PLA composites [27], and it is 33.8% higher than the strength of parts manufactured using the first method [20].

J Mater Sci

Table 2 Factors and responses of every run in the flexural testing with RSM method Run

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Factors

Response

A (°C)

B (mm min-1)

C (mm)

Y (MPa)

215 230 200 230 200 200 215 230 215 215 200 215 230 215 215

120 120 60 60 90 90 90 90 90 90 120 60 90 120 60

1.2 0.9 0.9 0.9 0.6 1.2 0.9 1.2 0.9 0.9 0.9 0.6 0.6 0.6 1.2

142.953 214.649 282.112 295.457 451.886 129.73 185.193 147.109 159.339 178.238 248.752 541.647 463.398 304.747 115.918

Figure 8 Stress–strain curves of five of the flexural specimens’ and comparison of flexural stress with other specimens prepared using different methods.

Mathematical model and verification The specific mathematical model of relationship between the printing parameters and flexural strength is calculated according to Eq. (3) and Table 3, and described in Eq. (4). Y ¼12869:55  100:77  A  8:71  B  2639:21  C  0:03AB þ 0:33AC þ 7:33BC 2

2

ð4Þ

2

þ 0:24A þ 0:04B þ 76:94C

The fit indexes of the mathematical model based on the analysis of variance (ANOVA) are presented in the Table 4. The Model F-value is 64.48, which implies that the quadratic model chosen for this study is significant. Also, the Model P value less than 0.05 also proves that the model is significant. The Lack of fit test diagnoses how well the model fits the experimental data. The lack of fit F-value of 3.31 and the lack of fit P value of 0.2407 which is greater than 0.10, both indicate that the lack of fit of this model is not significant. The R-Squared, Adj R-Squared, and Pred R-Square are estimated to be extremely close to 1.0 and in addition, the Pred R-square and the Adj RSquared are within 0.20 of each other. The Adeq Precision, which measures the signal-to-noise ratio, is [ 4, which suggests that the model has decent noise discrimination. All the evaluated results indicate that the quadratic model chosen for this experiment fits well with the experimental data. Finally, verification test was performed to check the correctness of the mathematic model of this experiment. The same size specimen as the one used in the previous three-point flexural test was printed using the parameters of 230 °C of printing temperature, 60 mm min-1 of printing speed, and 0.6 mm of layer thickness. The three-point flexural test was performed under the same conditions as the other flexural tests and the results showed that the flexural strength was estimated to be 610.092 MPa, which is

Table 3 Flexural modulus of PLA or PLA matrix composites prepared via different methods Methods

FDM with CCF prepreg filament

Real-time 3D printing of CCF

FDM with pure PLA filament

Solution casting and hot press method with 25% SCFR PLA composites

Injection molding with pure PLA

Flexural modulus (GPa)

40.13

30

2.485

35

2.253

J Mater Sci

Table 4 Model P value, F-value, lack of fit, R-squared, Adj R-squared, and Pred R-square of the RSM experiment Model F-value Model P value Lack of fit F-value Lack of fit P value Rsquared

Adj R-squared Pred R-square Adeq Precision

64.48

0.9764

0.001

3.31

0.2407

0.9915

0.8830

25.987

very close to 601 MPa predicted by the mathematic model. The verification test indicates that the mathematical model has a definite creditability from one aspect. As a result, Eq. (4) can provide the relationship between the printing parameters and the flexural strength of the parts printed with CCF filament.

Response surface analysis Figure 9 displays the 3D surface plots based on Eq. (4) to understand the relationship between the printing parameters and the flexural strength. As observed in Fig. 9a, the printing temperature has a minor influence on the strength. This might be because the melting flow index of pure PLA resin used in current study changes only slightly in the temperature range of 200–230 °C [25]. Also, it can be noticed that the flexural strength increases by only a small amount with the decrease in printing speed. Although the longer residence time in the extruder head would result in better impregnation in the above test, the longest residence time could be 30 s given that the hot path is 30 mm in length. By taking printing efficiency into account, the printing speed is relatively faster than speed converted by the printing time in the RSM experiment. At least, it was found that higher temperature in the range of 200–230 °C and lower printing speed would result in better bonding between PLA and CCF bundles and consequently obtaining better mechanical properties. As shown in Fig. 9b, c, it can be observed that the layer thickness has a huge impact on the final flexural strength. The layer thickness is the vertical distance between the nozzle of extruder head and the printing baseplate or the last printed layer and it determines the pressure between the nozzle and the printed filament. Therefore, the smaller layer thickness leads to more pressure on the molten CCF prepreg filament, which causes more molten PLA resin to be squeezed into the gaps and pores within the CCF bundles. In addition, smaller layer thickness leads to less air-gap between adjacent layers causing a decrease in air-gap to material ratio. This phenomenon results in higher

Figure 9 3D surface graphs of relationship between flexural strength and printing parameters (printing temperature, printing speed, and layer thickness); The graphs are shown when layer thickness is 0.9 mm (a), printing speed is 90 mm min-1 (b), and printing temperature is 215 °C (c).

strength for specimens as verified by Rankouhi et al. [24]. Shubham et al. similarly concluded that decreasing layer thickness has a positive effect on

J Mater Sci

Figure 10 SEM micrograph of cracked surface of printed flexural specimens with the continuous carbon fiber prepreg filament after three-point flexural testing. (The picture shows the fracture surface

of the specimen printed with a higher printing speed and bigger layer thickness and b lower printing speed and smaller layer thickness).

mechanical properties of printed samples with ABS [30]. Moreover, decreasing layer thickness dramatically improves the CCF content of the specimens, a significant contributor towards the enhancement of flexural strength. The trend of change in the strength of printed specimens with the change in layer thickness is same irrespective if the printing temperature is high or low (see Fig. 9b). However, as shown in Fig. 9c, at a low printing speed, reducing the layer thickness has a significant influence on the flexural strength of the printed specimens. In the valid range of layer thickness, a large layer thickness would at least slightly improve the flexural strength, irrespective of the printing speed. Therefore, the largest flexural strength occurs at the point where the printing speed is lowest and the layer thickness is the smallest in the range set in this experiment. The microstructures in the fracture of flexural specimens were additionally observed through the Scanning Electron Microscope (HITACHI SU-1500) picture. Figure 10a is the specimen fracture printed at the printing temperature of 215 °C, the printing speed of 90 mm min-1, and the layer thickness of 0.9 mm while corresponding values for Fig. 10b are 215 °C, 60 mm min-1 and 0.6 mm, respectively. A comparison of these two images reveals that there is more amount of CCF filament adhered to PLA resin after printing at lower printing speed and layer thickness. Especially in Fig. 10b, it can be seen that the PLA resin not only exists around CCF bundles but also appears within the CCF filaments. This indicates that high pressure due to lower thickness squeezes more molten PLA resin into the CCF

bundles, resulting in better bonding between PLA matrix and CCF reinforcement than the other.

Conclusion To improve the mechanical properties of printed parts by FDM printer, the CCF prepreg filament suitable for the printer was manufactured. The printing method of this composite filament was explored and the flexural properties of parts printed with the filament were studied. It was found that layer thickness has a significant influence on the final strength and modulus, while the printing temperature and speed have a minor influence. By decreasing the layer thickness and printing speed, increasing the printing temperature, the flexural strength of final printed parts can reach 610.092 MPa, and the flexural modulus can reach 40.13 GPa. Both achieved a great improvement compared to parts printed with pure PLA and by other methods mentioned in fourth part. It can be concluded that the method we proposed is a promising technique to print CCFRP composite parts. In addition, we found that the impregnation quality also plays an important role to final parts’ mechanical properties, which suggests that future investigations can focus on the impregnation of the prepreg filament.

Acknowledgements This work was supported by the http://dx.doi.org/ 10.13039/501100001809 National Nature Science

J Mater Sci

Foundation of China (NSFC) under Grant No. 51405305 and the Shanghai Key Laboratory of Intelligent Manufacturing and Robotics under Grant No. ZK1304. The authors would like to thank the colleges of Aerospace System Engineering Shanghai for their assistance.

References [1]

Snyder TJ, Andrews M, Weislogel M, Moeck P, StoneSundberg J, Birkes D, Friedman S (2014) 3D systems’ technology overview and new applications in manufacturing engineering science and education. 3D Print Addit Manuf 1(3):169–176 [2] Horn TJ, Harrysson OL (2012) Overview of current additive manufacturing technologies and selected applications. Sci Prog 95(3):255–282 [3] Turner NB, Strong R, Gold SA (2014) A review of melt extrusion additive manufacturing processes: I. Process design and modeling. Rapid Prototyp J 20(3):192–204 [4] Bellini A, Gu¨c¸eri S (2003) Mechanical characterization of parts fabricated using fused deposition modeling. Rapid Prototyp J 9(4):252–264 [5] Wang J, Xie H, Weng Z, Senthil T, Wu L (2016) A novel approach to improve mechanical properties of parts fabricated by fused deposition modeling. Mater Des 105:152–159 [6] Dul S, Fambri L, Pegoretti A (2016) Fused deposition modelling with ABS–graphene nanocomposites. Compos A Appl Sci Manuf 85:181–191 [7] Prashantha K, Roger F (2017) Multifunctional properties of 3D printed poly (lactic acid)/graphene nanocomposites by fused deposition modeling. J Macromol Sci A 54(1):24–29 [8] Bettini P, Alitta G, Sala G, Di Landro L (2017) Fused deposition technique for continuous fiber reinforced thermoplastic. J Mater Eng Perform 26(2):843–848 [9] Ashori A, Menbari S, Bahrami R (2016) Mechanical and thermo-mechanical properties of short carbon fiber reinforced polypropylene composites using exfoliated graphene nanoplatelets coating. J Ind Eng Chem 38:37–42 [10] Hull E, Grove W, Zhang M, Song X, Pei ZJ, Cong W (2015) Effects of process variables on extrusion of carbon fiber reinforced ABS filament for additive manufacturing. In: ASME 2015 international manufacturing science and engineering conference, American Society of Mechanical Engineers. pp V001T02A065–V001T02A065 [11] Ning F, Cong W, Qiu J, Wei J, Wang S (2015) Additive manufacturing of carbon fiber reinforced thermoplastic composites using fused deposition modeling. Compos B Eng 80:369–378

[12] Tekinalp HL, Kunc V, Velez-Garcia GM, Duty CE, Love LJ, Naskar AK, Ozcan S (2014) Highly oriented carbon fiber– polymer composites via additive manufacturing. Compos Sci Technol 105:144–150 [13] Quan Z, Larimore Z, Wu A, Yu J, Qin X, Mirotznik M, Chou TW (2016) Microstructural design and additive manufacturing and characterization of 3D orthogonal short carbon fiber/acrylonitrile-butadiene-styrene preform and composite. Compos Sci Technol 126:139–148 [14] Li ZH, Rong RY, Li YX, Li J (2011) Effect of fiber length on mechanical properties of short carbon fiber reinforced PTFE composites. In: Advanced materials research, Trans Tech Publications, vol 311. pp 193–196 ´ , Borda J, Dea´k G, Zsuga M, Marossy K, [15] Ra´thy I, Kuki A Ke´ki S (2012) Preparation and characterization of poly (vinyl chloride)–continuous carbon fiber composites. J Appl Polym Sci 124(1):190–194 [16] Fischer A, Rommel S, Bauernhansl T (2013) New fiber matrix process with 3D fiber printer—a strategic in-process integration of endless fibers using fused deposition modeling (FDM). In: NEW PROLAMAT. pp 167–175 [17] Pru¨ß H, Vietor T (2015) Design for fiber-reinforced additive manufacturing. J Mech Des 137(11):111409-1–111409-7 [18] Li N, Li Y, Liu S (2016) Rapid prototyping of continuous carbon fiber reinforced polylactic acid composites by 3D printing. J Mater Process Technol 238:218–225 [19] Matsuzaki R, Ueda M, Namiki M, Jeong TK, Asahara H, Horiguchi K, Hirano Y (2016) Three-dimensional printing of continuous-fiber composites by in-nozzle impregnation. Sci Rep 6:23058-1–23058-7 [20] Tian X, Liu T, Yang C, Wang Q, Li D (2016) Interface and performance of 3D printed continuous carbon fiber reinforced PLA composites. Compos A Appl Sci Manuf 88:198–205 [21] Mori KI, Maeno T, Nakagawa Y (2014) Dieless forming of carbon fibre reinforced plastic parts using 3D printer. Proced Eng 81:1595–1600 [22] Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA (2008) Response surface methodology (RSM) as a tool for optimization in analytical chemistry. Talanta 76(5):965–977 [23] Aslan N, Cebeci Y (2007) Application of Box–Behnken design and response surface methodology for modeling of some Turkish coals. Fuel 86(1):90–97 [24] Rankouhi B, Javadpour S, Delfanian F, Letcher T (2016) Failure analysis and mechanical characterization of 3D printed ABS with respect to layer thickness and orientation. J Fail Anal Prev 16(3):467–481 [25] NatureWorks Inc. (2013) Biopolymer 4043D technical data sheet. http://www.natureworksllc.com/*/media/Files/Nature Works/Technical-Documents/Technical-Data-heets/Technical DataSheet_4043D_films_pdf.pdf

J Mater Sci

[26] Aized T, Shirinzadeh B (2011) Robotic fiber placement process analysis and optimization using response surface method. Int J Adv Manuf Technol 55(1):393–404 [27] Gang ZF, Hong DX, Lin WY et al (2016) Mechanical performance of carbon fiber-reinforced polylactide matrix (C/ PLA) composites(I). J Mater Eng 5:16–18 [28] Markforged Inc. (2016) Mechanical properties of continuous fibers. https://markforged.com/ty/download-data-sheet/

[29] Afrose MF (2016) Mechanical and viscoelastic properties of polylactic acid (PLA) materials processed through fused deposition modelling (FDM). Doctoral dissertation, Swinburne University of Technology [30] Shubham P, Sikidar A, Chand T (2016) The influence of layer thickness on mechanical properties of the 3D printed ABS polymer by fused deposition modeling. In: Key engineering materials, Trans Tech Publications, vol 706. pp 63–67