Fused filament fabrication of fiber-reinforced polymers: A review

Additive Manufacturing 21 (2018) 1–16

Contents lists available at ScienceDirect

Additive Manufacturing journal homepage: www.elsevier.com/locate/addma

Review

Fused filament fabrication of fiber-reinforced polymers: A review a,b,⁎

a,c

Bastian Brenken , Eduardo Barocio , Anthony Favaloro R. Byron Pipesa,b,c

a,b

, Vlastimil Kunc

b,d

T

,

a

Composites Manufacturing and Simulation Center (CMSC), Indiana Manufacturing Institute, 1105 Challenger Avenue, Suite 100, West Lafayette, IN, 47906-1168, United States1 School of Aeronautics and Astronautics, Purdue University, 701 West Stadium Avenue, West Lafayette, IN, 47907-2045, United States c School of Materials Engineering, Purdue University, 701 West Stadium Avenue, West Lafayette, IN, 47907-2045, United States d Manufacturing Demonstration Facility, Oak Ridge National Laboratory, 2370 Cherahala Blvd, Knoxville, TN, 37932, United States b

A R T I C L E I N F O

A B S T R A C T

Keywords: Additive manufacturing Fused filament fabrication Fused deposition modeling Extrusion deposition Polymer composite materials

Recent advancements in the Additive Manufacturing (AM) Fused Filament Fabrication (FFF) approach are described with focus on the application to tooling and molds for composite materials and structures. A detailed summary of mechanical properties of printed parts for different composite material systems is presented and discussed. These material systems are comprised of discontinuous fiber-reinforced polymers characterized by fiber orientation dominantly parallel to the direction of the extrudate. An overview of the FFF process and its physical phenomena is given including the flow and resulting fiber orientation, the bond formation between adjacent beads and the thermomechanical solidification behavior of the deposited material. Based on reviewed research in these different phenomena, future research needs are discussed and desirable objectives are formulated.

1. Introduction Additive Manufacturing (AM) technologies offer the potential for significant cost savings due to reduced material waste and the capability for a tool-less production of intricate geometries. Therefore, they have gained considerable attention during the last decade. The Fused Filament Fabrication (FFF) process, which is also termed Fused Filament Fabrication (FDM), is one of the most popular AM methods. The recent development of large scale printers, which melt pelletized material deposited by a screw extruder, has yielded yet another name, Extrusion Deposition. In the extrusion based process, printing is achieved by controlled deposition of molten feed stock material. By following a predefined machine path, the 3D printer builds a geometry layer by layer on a printing bed. For complex geometries, support material can be utilized to enable the generation of geometric overhangs. Significant technical advancements have been made in FFF in the last five years. A special focus has been on the improvement of the maximum print size, the maximum material output per hour and the printing speed. The Big Area Additive Manufacturing (BAAM) system has been developed at Oak Ridge National Laboratory (ORNL) in

collaboration with Cincinnati Incorporated©. The system is capable of printing large scale geometries (6 m length × 2.5 m width × 1.8 m height) at material output rates up to 45 kg/h [1]. To demonstrate this capability, a full scale automobile was live-printed at the International Manufacturing Technology Show (IMTS) in Chicago in 2014 [2]. The BAAM system is commercially available from Cincinnati Incorporated© [3]. Fig. 1 illustrates the print of a full scale aerospace trim tool and gives an impression of the size of the system. Similarly, Thermwood© Incorporated developed the Large Scale Additive Manufacturing (LSAM) system which enables material output rates of up to 226 kg/hr (500 lbs/hr) and offers a maximum printing envelope of 30 m length × 3 m width × 1.5 m height. The system incorporates a printer head and a CNC router on the same gantry system and thus a part can be printed and then machined to size within the same system [4]. The extension of scale in the FFF process is enabled by utilizing discontinuous fiber-reinforced polymers. The fibers in the printing material significantly enhance mechanical performance, reduce warpage and add dimensional stability during the printing process [5]. Recently, the Composite Additive Manufacturing Research Instrument (CAMRI) was developed at Purdue University, as a medium size Extrusion Deposition printer [6]. While the dominant experience to

⁎ Corresponding author at: Composites Manufacturing and Simulation Center (CMSC), Indiana Manufacturing Institute, 1105 Challenger Avenue, Suite 100, West Lafayette, IN, 479061168, United States. E-mail address: [email protected] (B. Brenken). 1 www.purdue.edu/cmsc.

https://doi.org/10.1016/j.addma.2018.01.002 Received 3 October 2017; Accepted 22 January 2018 Available online 02 February 2018 2214-8604/ © 2018 Elsevier B.V. All rights reserved.

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

extruded material in order to fill the voids that otherwise form between the usually elliptically shaped extrudate cross sections. As a result, more intricate shapes can be printed and subsequently machined. The pin bracket compression molding tool illustrated in Fig. 3 was printed with the upgraded CAMRI system utilizing the same carbon fiber reinforced PPS material as for the previous molds. The pin bracket part depicted in the image was successfully compression molded (150 °C @ 250psi) with the printed tool using chopped discontinuous carbon fiber prepreg material. Another potential benefit of printed tooling is the tailorability of the coefficients of thermal expansion (CTE). During the printing process, the majority of fibers align with the flow in the direction of the extruded beads [11]. As a result of this induced alignment and the low CTE of carbon fiber, the CTE in the bead direction is much lower than the two in the directions transverse to it. By controlling the bead orientations with the printing trajectory, a tool can be specifically designed to possess expansions similar to the material molded during the manufacturing cycle. This enables precise control of the final part dimensions. Despite the significant technological advancements and the progress that has been made with the application to tooling, the FFF process itself remains largely empirically and must be calibrated. To date, the development of adequate simulation tools to model either the manufacturing process or the performance of printed parts has just begun. There is a great need to predict the process induced internal stresses and resulting deformations and how these affect the performance of the printed geometry. To fulfill this need, software companies like Digimat [12] or Dassault Systèmes [13] are developing simulation software and released first AM simulation capabilities already. These simulation tools now have to be utilized and combined with the right physics descriptions to correctly model the printing process. Based on a discussion of the current performance of printed parts with the FFF method in Section 2 and a detailed explanation of the process and the involved physical phenomena in Section 3, past relevant research work for the FFF process is summarized in this paper and future needs are discussed in Section 4. Finally, the main findings are concluded and research recommendations are given. Here, the primary focus is on FFF with fiber-reinforced materials. For a review on AM techniques in general [14,15], FFF and related extrusion processes for neat polymers [16,17] and a summary of existing common AM methods for composite materials [18,19], the cited references should be consulted.

Fig. 1. Aerospace trim tool being printed at ORNL’s BAAM system – 5.2 m long structure made of ABS with 20% carbon fiber.

date has been in the printing of intermediate temperature polymers like ABS, the focus in the development of CAMRI has been on the capability to process high temperature polymer systems. Fiber-reinforced polymers with high fiber loadings (50 wt.%) can be processed in CAMRI at temperatures of up to 400 °C. Furthermore, this system was primarily developed as a validation tool to support the simulation development for the Extrusion Deposition process needed to design products fabricated by the Extrusion Deposition process. A promising initial application of the FFF process is printing molds and tooling for traditional composite manufacturing methods like Vacuum Assisted Resin Transfer Molding (VARTM) [1], hand layup [7], compression molding and autoclave curing [8,9]. Utilizing fiber-reinforced polymers, the quoted papers demonstrate the applicability of FFF for printing tooling. By measuring the tool shapes before and after the manufacturing processes, deviations were found to be minimal for a printed VARTM tool [1] and an autoclave tool [8] indicating sufficient dimensional stability of the printed molds. Especially for the prototyping phase during the development of a new tool or for low volume production, the application of FFF can lead to significant cost savings replacing the traditional processes for manufacturing a tool [10]. In contrast to this traditional manufacturing method, a tooling master is not required for the FFF process. Instead, a modestly over-dimensioned version of a tool can be printed and then machined to size in a second process step. Optionally, a coating can be added afterwards to improve the surface quality and durability. In first tool manufacturing trials at Purdue University with the CAMRI system, the total manufacturing time for such an overall manufacturing process was less than one day. Both an autoclave and a compression molding tool were printed with a Polyphenylene sulfide (PPS) material reinforced with 50 wt.% of carbon fibers. Woven carbon fiber thermoset prepreg material was used to manufacture both a laid up part in the autoclave (180 °C @ 80psi) and a compression molded part (180 °C @ 300psi) utilizing the printed tools. Fig. 2 depicts the produced parts and the tools after being used in a manufacturing cycle. In order to improve the quality of printed tools and reduce the void content in printed geometries, most recently a mechanical tamper was added to the CAMRI system. It allows for an in-situ compaction of the

2. Performance of printed parts In order to give an impression of the current performance with respect to tensile modulus and strength for parts produced with fiberreinforced polymers using the FFF process, a large collection of mechanical data from tensile tests is summarized graphically in this section for various fiber-reinforced material systems. In addition, the data collection can also be found in table form in the Appendix A. The mechanical characterization of printed fiber-reinforced polymers with a comparison to the neat polymer material is the main focus of a majority of published research works in the field of FFF with composite materials. As fibrous reinforcements, short fibers [5,6,9,11,20–32], fibrils [33,34], nanofibers [35–37] and continuous fibers [38–47] have been utilized. The fibers were combined with thermoplastic matrix materials such as Nylon, PLA, ABS, PPS and PEI for most of the cited references. Three publications [24,28,32] developed a FFF process for a fiber-reinforced epoxy resin. When studying the reported data in this chapter, the reader should be aware of the fact that in addition to the utilized material system, the processing conditions and printing parameters play a vital role in determining the final properties of printed geometries. This is illustrated for example by the paper of Ning et al. [31]. As a consequence, the data presented may not follow simple volume fraction trends expected. 2

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 2. a) Printed tool after an autoclave curing cycle with a cured laminated composite part, b) Two-sided compression molding tool with molded Purdue P.

Fig. 3. Printed compression molding pin bracket tool with molded pin bracket part.

made with FFF. Due to the primarily elliptical cross sections of printed beads, voids form inside of the printed geometry. Furthermore, by assuming a rectangular cross section based on caliper measurements of the most outer points of the beads, the load carrying areas are overestimated. Therefore, due to this discrepancy, it is assumed that for most of the data presented herein, the actual mechanical properties are greater than the data reported in Fig. 3. Nevertheless, it is not expected that the effect of a correction with respect to the true area would change the conclusion drawn in terms of the comparison to aluminum. One reason for the limited mechanical properties is the short resulting fiber lengths. Usually, the extrusion process significantly damages the fibers and reduces their resulting fiber aspect ratios. In addition, included voids and printing defects are expected to act as stress concentrators causing the specimens to fail prematurely. A summary of mechanical properties is given in Fig. 5 for discontinuous fiber material systems loaded transverse to the printing direction. It is apparent that compared to the properties parallel to the printed beads, the performance is significantly reduced. A clear difference can be observed for most of the thermoplastic materials compared to the epoxy material systems. Especially the strength values of most thermoplastic composites, ranging between 5 and 15 MPa, are clearly less than the value of 45 MPa reported for both epoxy systems. Only one carbon-fiber reinforced PLA reached a comparable strength of 35.4 MPa. One explanation could be the lower viscosity of the epoxy. During the deposition process of the material, the surface of previously laid down beads can be wetted more easily by a low viscosity system,

Parameters such as extrusion temperature, bead size and the time between the depositions of adjacent layers have a significant effect on the resulting performance. However, the objective of this paper is to give a general overview of the current performance of the FFF process with fiber reinforced materials. Therefore, the detailed discussion of different processing conditions and details on the material systems used is beyond the scope of this text. For the investigation of specific differences in the results, the cited references should be examined. Fig. 4 provides an overview of tensile properties for manufactured samples that were tested parallel to the printed bead direction. Most of the fibers align in this direction as indicated in the sketch in the upper left of the figure. This was confirmed by optical microscopy for various samples. Consequently, this testing direction yields the highest mechanical properties of the anisotropic printed material. The greatest stiffness of 26.4 GPa was found for a 50 wt.% carbon fiber (CF) reinforced PPS printed with the CAMRI system at Purdue University [6]. The greatest strength of 125.3 MPa was provided by a PEI material reinforced with 4.7 wt.% carbon nanotubes (CNT) [36]. Most of the tested material systems are ABS based composites with carbon fiber reinforcements. The majority of the presented strength data ranges between 30 and 70 MPa, while the tensile modulus is largely between 2 and 15 GPa. Despite the fact that most of the fibers align in the loading direction and high fiber loadings up to 50 wt.% were investigated, none of the mechanical properties shown in Fig. 2 are comparable to aircraftgrade aluminum [48]. However, it has to be mentioned that it is quite difficult to determine the actual cross sections of tensile specimens 3

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 4. Summary of tensile properties for printed discontinuous fiber composites loaded parallel to the printing direction [5,6,11,21–26,28,29,35–37]. Data marked with a star (*) was extracted graphically.

mechanical performance in the transverse direction and it is controlled by process parameters like the extruded material temperature or the layer time. This underlines the importance of the process conditions, especially for the transverse properties of printed parts. In addition to discontinuous fiber-reinforced materials systems, the FFF process with continuous fiber systems is being advanced as well. MarkForged© developed the first continuous fiber printer and technology improvements are ongoing [49]. Printed samples generated with MarkForged© printers have been characterized [40,42,46]. In addition, several research groups developed their own printing heads to print continuous fiber materials and tested the resulting printed parts [38,41,44,45,47]. Therefore, a summary of mechanical properties for continuous fiber systems is included in this section as well and provided

resulting in a larger contact area and thus a stronger bond between adjacent beads. From the results presented in Fig. 5 for the most commonly used thermoplastic matrices, it can be concluded that the mechanical properties in the transverse direction are one of the key limiting factors for utilizing printed composite parts in structural applications. With tensile moduli ranging between 1 and 3 GPa, the printed samples have about the same or even a lower modulus than the neat polymers. It is interesting to notice that similar results can be found for very different material systems. A PPS material reinforced with 50 wt.% CF yielded similar mechanical properties than an ABS material with 20 wt.% CF. Thus, the polymer material itself does not seem to play a major role in determining the transverse properties. In fact, the bond formation between adjacent beads governs the

Fig. 5. Summary of mechanical properties for printed discontinuous fiber composites loaded transverse to the printing direction [5,6,22–25,29,30]. Data marked with a star (*) was extracted graphically.

4

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 6. Summary of mechanical properties for printed continuous fiber composites loaded parallel to the printing direction [38,40–43,45–47]. Data marked with a star (*) was extracted graphically. For the values labeled with two stars (**), the volume fraction was found from correspondence with the author.

3. Physical phenomena

in Fig. 6. Depending on the material systems utilized, a wide variety of mechanical values has been found in the literature. The data ranges from a tensile strength of 31 MPa with a corresponding modulus of 1.8 GPa for a Nylon material with 4 vol.% of Kevlar fibers to a strength of 464 MPa with a modulus of 35.7 GPa for a carbon fiber reinforced Nylon material with 18 wt.% of fibers. The tensile properties are dependent on the achieved fiber volume fractions and on the quality of fiber wetting during the printing process. In general, carbon-fiber reinforced materials yielded the highest strengths and moduli. However, the PLA material reinforced with a high fiber content of 34vol.% carbon fibers tested by Li et al. [41] yielded a strength of only 91 MPa. From studying the paper, this low value can be explained by a poor wetting of the fibers in the printed samples. The low modulus of the printed ABS/ CF of Yang et al. [47] is caused by fiber pullout due to poor wetting as well. On the contrary, the results reported by van der Klift et al. [40] show that printed continuous fiber parts can reach or even exceed the strength of aluminum, while the stiffness is still at about half of its value [48]. When studying the fiber volume fractions of the printed samples, it is apparent that to date only limited amounts of continuous fibers can be incorporated in the FFF process. Furthermore, the wetting of the fibers in the extrusion nozzle is a challenge. Improvements must focus on these two main aspects. Koga et al. [39] discuss the limitation of a low fiber volume fraction for the Mark One© printer by MarkForged© and they give solution approaches. In conclusion, the summaries of mechanical properties for printed composite parts manufactured with the FFF process illustrate that in general, printed parts are still inferior to aluminum [48]. Especially transverse to the printing direction, the mechanical properties are limited and no real improvement compared to the printed neat polymeric materials has been observed. No transverse data could be found for the continuous fiber systems, but it is expected that the same conclusions apply as for the discontinuous material systems. A consequence of these findings is that tooling was chosen as a primary target application for the FFF process due to reduced strength requirements. Since mechanical properties of printed parts evolve during the solidification process of the material, a fundamental understanding of the material behavior during the cooling process is required in order to simulate the performance of printed parts. An improvement of the mechanical properties can then be targeted based on process simulations.

In this section, the FFF process and related physical phenomena are addressed in depth. In order to simulate this manufacturing method for improving the process itself and the quality of printed parts, a thorough understanding is essential. Fig. 7 depicts a detailed view on the FFF process. The first important process that can be identified is the flow of the fiber-reinforced material through the die and the subsequent deposition turn onto the printing bed or previously laid down material. At this stage, the resulting fiber orientation is defined by the flow in the orifice and thereby governs the mechanical anisotropic behavior of the printed part. For the materials of interest with significant fiber volume fractions, a viscous fiber suspension flow is present. As a consequence, the flow and the fiber orientation are mutually dependent. The flow field determines the fiber orientation in the printed beads mostly through shear alignment and the converging zone within the die. In turn, the fiber orientation has a large influence on the primary viscosities of the material. For instance, the extensional viscosity is orders of magnitudes larger parallel to the fiber direction than transverse to it [50]. Therefore, the fiber orientation affects the resulting flow field as well. The deposition turn is also expected to have an important influence. Based on different turning radii on the bottom and the top of the bead, different shearing rates are applied to the material, which can affect the resulting fiber orientation. During the deposition, the hot material wets and reheats or re-melts previously laid down material beads. Both effects are critical for the bond formed between adjacent beads. The wetting process determines the contact area between the beads, while a longer exposure of the interface to elevated temperatures promotes the coalescence of adjacent beads and interdiffusion of polymer chains through the interface. The wetting process is governed by the viscosity transverse to the bead direction and the surface tension of the printing material, which are affected by the reinforcing fibers. A strong dependence of temperature on the viscosity of the polymer limits the range of temperatures for forming a bond between layers. High temperatures are needed for a diffusion based fusion of the adjacent beads after an interface has formed. At lower temperatures, the molecular mobility decreases and the diffusion process is hindered. Hence, the temperature history at the bead-to-bead interface is the crucial factor for the bond formation. 5

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 7. Detailed illustration of the FFF process and its related physical phenomena.

solidification behavior as well as deformation and residual stresses, which were introduced in the last section, promising work is presented in this section. In this regard, due to a limited amount of studies for reinforced materials, relevant work for neat materials is considered that could be extended to be applied for fiber reinforced systems. In addition, future needs are underlined which account for the complexity of the FFF process discussed in the last section.

Additionally, for semi-crystalline polymers, the viscosity increases rapidly as the material begins to crystallize and thus the bond formation process is interrupted. Finally, the deposited materials begin to cool and the temperature based solidification process is initiated. The transient thermal conditions are governed by convective and radiative heat losses on external surfaces of the bead. Furthermore, thermal contacts between the beads and between the material and the printing bed drive the conductive distribution of heat and thus the cooling process during printing. If the layer deposition times are small and a large amount of material is deposited without allowing the previously laid down beads to cool sufficiently, sagging of the material due to gravity effects can occur. If the sagging effect becomes extensive, the ongoing print cannot be achieved. As the part cools, the material transitions from a viscous fluid to a viscoelastic solid. During this process, the time and temperature dependent relaxation moduli evolve defining the viscoelastic constitutive relationship of the material. Based on the coefficients of thermal expansion (CTE), the material shrinks and internal stresses start to build up due to the evolving material stiffness upon solidification and the bonds between the beads constraining the material. A fraction of these stresses is released due to viscoelastic relaxation and deformations of the material. In the case of a semi-crystalline polymer, the crystallization reaction changes both mechanical and thermal properties of the polymer during cool down. In particular, additional strains are introduced that further promote internal stresses and part deformations. The combined behavior of these simultaneously occurring phenomena is anisotropic. Besides the well-known influence on the mechanical properties, the aligned fibers in the bead direction affect the thermal properties as well. They significantly increase the conductivity in this direction, compared to the transverse directions. The shrinkage, both due to thermomechanical and crystallization effects, is constrained by the fibers in the bead direction, but not to the same degree in the lateral directions. The previous discussion of the physical phenomena in the FFF process underlines the complexity of the problem. In order to simulate the process, the described physical phenomena and their interactions have to be understood, characterized and modeled.

4.1. Flow and fiber orientation Bellini [51] was one of the first to characterize and model the material flow during the FFF process. For the extrusion of ceramic materials, the flow of the material through the liquefier, the nozzle and the subsequent deposition including die swell, was discussed and modeled. The focus was on the pressure drop in the nozzle and the resulting shapes of the printed beads before and after deposition on the printing bed. In addition, the velocity and temperature fields of the material during deposition were simulated and compared to experiments. Although it was not implemented in the simulations, an observed orientation of incorporated particles in the material parallel to the direction of the extruded beads was discussed. Ramanath et al. [52] modeled the 2D flow behavior in an extrusion liquefier of a FFF printer. The flow and temperature fields, as well as, pressure gradients were simulated in the FEA program ANSYS© for the bio-polymer Polycaprolactone (PLC). The results of the pressure drop were compared to a mathematical model and a reasonable correlation was found. Although not presented in the paper, ongoing work to extend the study for a PCL/Hydroxyapatite bio-composite was mentioned. In a subsequent work, the authors extended this work to investigate the melt flow behavior for varying nozzle diameters and angles [53]. The flow of iron powder reinforced ABS material through a liquefier head was investigated and modeled by Nikzad et al. [54]. After a characterization of the novel composite ABS material, the authors built 2D and 3D models in ANSYS© to simulate the temperature and velocity fields and the pressure drop at the exit of the nozzle. One of the first studies that investigated the fiber orientation in the nozzle of a FFF printer was carried out by Nixon et al. [55]. For three different nozzle geometries (a convergent, a straight and a divergent nozzle), the resulting fiber orientations were modeled in the nozzles for different injection rates and filler volume fractions of a carbon fiberreinforced PEI material. The authors utilized the software Moldflow© with its implemented Folgar-Tucker model [56] to describe the flow induced fiber orientation. In the simulations, the material was modeled

4. Past research work and future needs One of the primary objectives of this paper is to summarize relevant research work in the field of FFF with fiber-reinforced materials. Based on the different areas flow and fiber orientation, bond formation, 6

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 8. Fiber orientation in the flow (z-) direction for two different diverging nozzle geometries, illustrating a reduction of the fiber alignment in this direction in the diverging zones of the nozzles [57]. Reproduced with permission.

to flow from the liquefier through the selected nozzle geometry into a large, open cavity. The flow was considered up to the end of the nozzle outlet and free boundary effects like die swell were not accounted for after the outlet. For the three nozzle designs, the authors found a significant fiber alignment with the flow before the entrance of the nozzles at the end of the liquefier. This alignment increased for higher fiber loadings. For the convergent nozzle, the greatest overall fiber alignment was observed, whereas for the divergent nozzle, it was the least. Unfortunately, the authors only reported fiber alignment with the material flow without explicitly illustrating the flow fields. Therefore, the actual fiber orientations with respect to the nozzle geometries can just be estimated from the paper. In a similar work, Garcia [57] analyzed different divergent nozzle geometries to investigate the amount of the fiber alignment at the nozzle exit. The goal of this study was to find a nozzle geometry that results in a 3D random fiber orientation to achieve isotropic material properties of the extruded material. As in the study of Nixon et al. [55], Moldflow© with the implemented Folgar-Tucker model [56] was utilized. The effect of die swell was not accounted for. Fig. 8 illustrates the modeled fiber orientation in the flow direction, where red regions indicate a high and blue regions a low fiber alignment in the z-direction. While a re-alignment from the flow (z-) direction in the y-direction could be observed for all diverging nozzle geometries in the diverging zones, no random fiber orientation and thus isotropic material characteristics were found. In a series of publications, Heller et al. [58–60] investigated the resulting fiber orientation in the FFF process with consideration of die swell effects after the nozzle exit for a reinforced ABS material. In an initial publication [58], COMSOL© was utilized with the assumption of a Newtonian fluid and the velocity field and velocity gradients were computed based on a single density and an isotropic viscosity. The flow in the nozzle was modeled as creeping flow in an axisymmetric simulation. Based on these results, the fiber orientation was computed with

the Advani & Tucker orientation tensor approach [61] and the isotropic rotary diffusion formulation from Folgar-Tucker [56] utilizing the Fast Exact Closure method [62]. A high fiber alignment parallel to the flow was observed for the end of the convergence zone of the nozzle. The subsequent die swell effect reduced the fiber alignment in regions where the material leaves the nozzle. The evolution of fiber alignment in the flow direction is illustrated for the modeled axisymmetric nozzle in Fig. 9. In a later work [59], the effect of the die swell on mechanical properties was characterized, based on the changing fiber orientations. It was shown that the die swell reduced the axial modulus by about 20%. In addition, geometrical nozzle parameters and their influence on the resulting fiber orientation and mechanical properties were investigated. Finally, the latest study [60] considered the deposition turn onto the printing bed as well. A 2D planar model was built in COMSOL© to model the flow of the material. It was shown that the material deposition results in a high fiber alignment in the printing direction. In a paper by Lewicki et al. [32], the flow of a 8vol.% carbon fiber reinforced epoxy through the tip of a nozzle was modeled. In these simulations, the carbon fibers were modeled as discrete particles, so fiber-fiber interactions and fiber-wall interactions were considered. Here, the fibers were treated as separate domains in which the fluid was modeled as an incompressible rigid body. Particle-particle interactions were implemented with frictional, inelastic contact forces. The interactions of the particles with the surrounding fluid were described with the fluid dynamics equations. Fig. 10 illustrates the orientation evolution of an initially randomly oriented arrangement of fibers. Starting from an assumed random orientation state during the simulation, the fibers align in the flow orientation, starting from the walls where the shear forcers are highest then and progressing inwards according to the parabolic velocity profile of the fluid. Due to the high computational cost of this simulation and the limitation to small volumes, a second, larger scale analysis was run using CFD software. However, the focus here was on the global flow dynamics which the corresponding velocity 7

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 9. Fiber alignment in the flow direction (z- or 3-direction) at different points of the modeled extrusion nozzle [59]. A decrease of fiber alignment in the flow direction can be observed after the nozzle exit due to die swell. Reproduced with permission.

predict the flow and fiber orientation during the FFF process with fiber composites in a fashion that simulates their anisotropic flow.

fields. Fiber orientations were not computed, but assumed based on the velocity information. The studies presented in this section show that the amount of research regarding the flow and fiber orientation in the FFF process still needs to be extended. Except for one paper modeling the fibers as discrete particles, the studies that modeled the fiber orientation only considered a Newtonian isotropic fluid. However, the material systems of interest with concentrated fiber volume fractions show a different behavior. These systems should be modeled as an anisotropic viscous fluid/solid. Modeling fibers as discrete particles yields a great amount of detail, but this approach is computationally very expensive, so only small domains can be simulated. As discussed in the last section, the flow and the resulting fiber orientation are mutually dependent. However, the reviewed work only shows uncoupled analyses. Current flow solvers do not account for anisotropic flow properties. Furthermore, the solvers are constrained to second-order orientation tensors, which do not provide an unambiguous description of the fiber orientation state. Consequently, there is a need for improved tools to

4.2. Bond formation The bond formation between beads is of great importance for the mechanical performance of printed parts. It governs the properties in the transverse and thus the minimum strength directions. The bond formation in the FFF process is similar to the welding process of polymer interfaces [63] with respect to the physical mechanisms involved. However in the classical FFF process without external additional energy sources, the newly deposited material also serves as the heat source to form the bond between beads, when no additional heat inputs are present. The importance of the bond formation to the FFF process was identified early and accounted for in research studies. Thomas and Rodriguez [64] were among the first authors to investigate and model the fracture strength between extruded beads. A 2D analytical heat

Fig. 10. Fiber orientation evolution with time at a constant pressure. The fibers were modeled as discrete particles [32]. Reproduced with permission.

8

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

predict the degree of healing (welding) and thus the last stage of the bond formation process. However, the results were found to be of limited value for predicting the bond strength due to a high sensitivity to temperature changes and varying estimates for the welding time required for the model. In a later work, Sun et al. [65] investigated the effects of changing processing conditions like extrusion temperature and build envelope temperature on the temperature history and the related bond strength, based on the sintering model. The authors found a strong correlation where conditions promoting a lower cooling history yielded greater strength values. The critical sintering temperature was confirmed to be significant. Gurrala and Regalla [70] extended the previously described sintering model initially derived for spherical particles to be applicable to a cylindrical geometry. They found a strong correlation between strength and neck growth. The strength was predicted based on the computed neck growth and a reasonable agreement with experimental data was observed. A couple of studies were found that measure Mode I fracture toughness and fracture resistance for printed samples made from both unreinforced and carbon fiber reinforced ABS. Aliheidari et al. [71] designed a DCB specimen geometry and printed test samples with neat ABS material at different temperatures to assess the influence of printing temperature on the fracture resistance. The authors used the Jintegral method with a finite element model to evaluate the fracture resistance of the printed specimens. They found that the fracture resistance increased with printing temperature and explained this with a greater formed interfacial area between the layers, which was confirmed by optical microscopy. For the samples printed at the largest nozzle temperature of 240 °C, a fracture resistance close to the bulk value of ABS was found, indicating large bond strength. Young et al. [72] investigated the Mode I fracture toughness for printed samples of both neat ABS and carbon fiber reinforced ABS. For crack initiation, they used an embedded Kapton film. Specimens were manufactured both with 3D printing and compression molding for both material systems and results regarding the fracture toughness were evaluated. Compared to the compression molded samples, the printed carbon fiber ABS samples showed a 10-fold reduction in fracture toughness. Furthermore, their fracture toughness was only about a fifth of the one of the printed neat ABS samples. Consequently, the authors observed a significant reduction in fracture toughness for the printed carbon fiber reinforced material. From investigating the fracture surfaces, the authors attributed a very poor adhesion between the layers of these samples as the main reason for this reduction. In order to improve the bond strength between adjacent beads, two recent experimental studies were found that investigate the effect of applying additional heat to re-melt the laid down material before deposition of the next bead. An investigation considering a fiber-reinforced ABS material was conducted by Kishore et al. [73]. Infrared preheating was applied to previously deposited material in order to determine the effect on the resulting bond strength. Based on the length of time of the reheated material above the glass transition temperature, a stronger bond was expected based on the thermally driven polymer interdiffusion. However, in DCB experiments, the authors found both increases and reductions in measured fracture energy. They concluded that while infrared preheating can help to increase the bond strength, a degradation of the material could cause its reduction as well. In another study, Ravi et al. [74] investigated an in-process laser preheating technique to enhance the bond strength between extruded beads. Before deposition of new elevated temperature material, the previous layer was reheated with a near-IR laser. For similar printing parameters, an increase in inter-layer bond strength of 50% was observed for laser heated neat ABS samples compared to those which were printed without preheating. This increase was attributed to a promoted fusion due to the elevated interface temperature. While several studies exist that address the bond formation in the FFF process, no modeling work could be found that incorporates the

Fig. 11. Illustration of the bond formation process between adjacent filaments. After two beads get in contact (1), a neck grows (2) and finally molecular diffusion happens at the interface and the polymer chains randomize (3)[65]. Reproduced with permission.

transfer model was developed to predict local thermal histories at the interface between rectangular beads. This information was utilized to model the fracture strength between the beads, based on reptation theory describing the interdiffusion of long polymer molecules across the bead-to-bead interfaces. According to the model, most of the fracture strength develops during the initial wetting stage at elevated temperatures. As anticipated, lower cooling rates of the material promoted greater strength values. An illustration of the bond formation process for actual, round shaped extruded beads is provided in Fig. 11. When two beads get in contact (1), a neck between these beads starts to form (2). Once a connection is established, polymer chains diffuse through the interface and a randomization takes place concluding the process. Bellehumeur et al. [66] applied a Newtonian polymer sintering model developed by Pokluda et al. [67] to model the neck growth part of the bond formation process between adjacent beads during FFF with an ABS material. A 1D lumped heat transfer model developed by Li et al. [68] was used for the temperature predictions. In sintering experiments, the dynamics of neck formation for the ABS material were evaluated to determine the surface energy of the polymer. This neck formation process is depicted in Fig. 12. Since the sintering model utilized was limited to describing the wetting between beads, it was only indicative of fracture strength. The authors found that most of the necking process between the beads took place above a critical sintering temperature. Due to the rapid cooling during the FFF process, they concluded that no complete bond between filaments could form and consequently, the transverse tensile properties of the printed ABS material were lowered compared to the feedstock material. In a related work, Sun [69] also employed a non-isothermal diffusion model to

Fig. 12. Neck formation process for ABS polymer, observed in a sintering experiment [66]. Reproduced with permission.

9

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

effect of the fibers on bond strength. In spite of the fact that most fibers align parallel to the bead direction and greatly increase the extensional viscosity in this direction, it is expected that fibers will have an influence on the transverse viscosities that drive the necking process as well. In micrographs of the surface of a fiber-reinforced printed bead taken at Purdue University, a large number of fibers could be observed directly at the bead surface. Therefore, the fibers are expected to significantly affect the surface tension that governs the necking behavior for the previously described sintering model [66]. Additionally, since it is not possible that polymer chains penetrate the fibers, additional constraints are imposed by the fibers hindering the diffusion of polymer chains as described by the reptation theory. Furthermore, the fibers affect the thermal properties of the material significantly and thus they change the thermal history, which has been shown to be of paramount importance for the bond formation by most of the summarized work in this section. In the case of a semi-crystalline polymer, the initiated crystallization reaction suppresses a further necking and interdiffusion process due to the largely increased viscosity of the material during crystallization. At this point, the statements made in this paragraph with respect to the effect of fibers are based on assumptions. Fundamental research and understanding is required in order to characterize and model the bond formation behavior of fiber reinforced polymers during the FFF process.

thermal histories for an actual print of a part. Beads were modeled as filaments of constant cross section and length. Based on the deposition time, the temperature was allowed to vary in the longitudinal axis of the filaments. It was assumed to be constant across the bead cross section, so filaments were considered to be thermally thin. Thermal contacts to adjacent beads and the printing bed were accounted for. With a Matlab© model these contacts were shown to be significant for the temperature evolution. More details on the implementation and the model structure can be found in a second publication [78] and more detailed results were presented there as well. By implementing a stepwise activation of the filaments and a discretization of the geometry, the authors were able to model the transient 3D temperature history for simple parts by applying the 1D solution to each of the bead elements. For more complex analyses, the problem must be discretized and numerical tools need to be applied. One of the first to investigate the cool down during FFF with FEA was Rodriguez [79]. In his PhD thesis work, he modeled five elliptical cross-sections in a 2D model, considering constant heat convection and a negligible thermal resistance between the beads. In a work presented at the SAMPE conference in 2016 by Brenken et al. [80], a 2D heat transfer model was employed as well. It was strongly coupled with a non-isothermal dual crystallization kinetics model in COMSOL© to predict the crystallization behavior for a semi-crystalline, fiber-reinforced polymer. By depicting the printing process through a stepwise activation of bead cross sections, thermal histories and crystallization states could be predicted for various square packed geometries. Due to the two dimensional character of the model, temperature gradients in the printing direction were assumed to be small. Fig. 13 illustrates an example result of the 2D model. Both a temperature and distribution is shown for a simple 2 by 2 bead cross section. Here, bead 4 is still inactive, which means that it is not yet considered as deposited in the analysis. Ultimately, 3D analyses are required to model a realistic behavior of printed materials. Zhou et al. [81] modeled a three-dimensional bead on a support platform for a thermal analysis in ANSYS©. The latent heat of phase change was included, as well as, temperature dependent heat conduction and heat capacity. The printing process was modeled by a stepwise activation of the bead and temperature profiles were shown for different times during the print. Costa and coauthors [82] modeled simple 3D extruded beads as well in order to evaluate different heat transfer analysis parameters regarding their significance. They investigated both convection and radiation with the environment, effects of different thermal resistances between the beads themselves as well as between the beads and the printing bed. Furthermore, radiation between adjacent filaments and convection with entrapped air was considered. The printing process was depicted by activating boundary conditions for different short rod structures in a stepwise manner. Based on the results, the varying thermal resistances and the convection conditions were found to be most significant for the heat transfer analysis. Pooladvand and Furlong [83] derived a transient enthalpy formulation to predict temperature profiles for a printed cylinder. The authors developed a 3D discretization of the model and used alternating direction implicit methods in order to solve the set of equations at each deposition step using MATLAB. Convection and radiation heat losses were considered with a total heat transfer coefficient, which was determined in 1D calibration experiments. In subsequent printing experiments, the temperature profiles were recorded with a thermal camera and compared to the simulated ones and a reasonable agreement was found. In a recent paper, Brenken et al. [84] extended the 2D work presented above. The crystallization kinetics model and a melting model to describe re-melting of deposited beads were implemented in a UMATHT user subroutine to model these phenomena in Abaqus. A 3D model was build and the printing process was modeled by activation of boundary conditions of rod shaped elements representing the printed part. Fig. 14 illustrates a temperature and crystallinity distribution at a

4.3. Solidification behavior This section summarizes past work on the solidification analysis of materials made from FFF. The combined material behavior is governed by the thermal history of the material. Therefore, much of the initial work presented focuses on describing the cool down during FFF. Herein, the majority of papers concentrates on unreinforced polymers. Yardimci and Güceri [75] published one of the first papers to underline the importance of the heat transfer analysis for the FFF process. They described a 1D heat transfer formulation considering 1D beads modeled as grid blocks. Interactions with the environment and other beads were approximated by including evolving heat sink terms in the formulation. Convection boundary conditions were added for the end surfaces of the beads. Based on different ambient cooling conditions, the bonding potential of adjacent beads was estimated. In a later work [76], the two authors extended their model to a 2D quasi-steady state heat transfer analysis. In order to describe the process as a steady state analysis, the moving deposition head was chosen as reference frame. With a non-dimensionalized form of the equations, effects of different Peclet and Biot numbers on the temperature distribution of a three layer geometry made from two different materials were investigated. As already mentioned in the last subsection, analytical heat transfer models were developed with the focus of predicting the bond strength between extruded beads. For their temperature predictions, Thomas and Rodriguez [64] developed an analytical 2D heat transfer model assuming rectangular cross sections of the extruded beads in a single bead wall. Free convection boundary conditions were imposed for the outer surfaces. The thermal resistance between the beads was assumed to be negligible. The analytical 1D model utilized by Bellehumeur et al. [66] was developed as a lumped capacity model based on the small diameter of the extruded beads. A constant temperature was assumed for an elliptical cross section. As direction of interest, the printing direction was described by the model. Sun et al. [65] compared the two modeling approaches with experimental data. The authors found that the 1D lumped capacity model yielded a better prediction at greater material temperatures shortly after extrusion, while the 2D model provided a closer match at lower temperatures. Overall, the authors concluded that the two models were not adequate to capture the full heat transfer process in the bead. Especially for a full solidification analysis of a printed part, models are needed that consider the transient phenomena in material deposition. Costa et al. [77] developed a heat transfer methodology to predict 10

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 13. Modeled 2D temperature (left) and crystallinity distribution (right) during a print simulation of a simple 2 by 2 bead cross section. Beads 1–3 are active (deposited), while bead 4 is still inactive [80].

and stress simulation based on thermoelastic deformation of the utilized neat ABS material. The printing process was modeled with a stepwise element activation method in ANSYS©. In this way, the elements were subsequently activated at their extrusion temperature and subjected to the current surrounding temperature distribution to depict the material deposition. The overall printed geometry was built as solid mesh neglecting details like interbead voids or elliptical bead cross sections. Consequently, the beads were represented by a line of single elements of cubic shape. By employing this model, the authors were able to model stress accumulations and part distortions based on different printing patterns for a simple 3D plate geometry. In a subsequent study [88], the same model was used in a design of experiments to investigate the effects of different printing parameters like printing speed, bead width, layer thickness and the interactions of these parameters on the final residual stress state and the resulting deformations. In this study, the printing speed could be identified as the most significant process parameter. In a comparison to experiments, the same warped shape of the printed plate could be found qualitatively as for the simulations, however, a quantitative comparison had yet to be established. In the paper of Hebert et al. [89], the authors report a process simulation for the FFF process as well. Assuming a constant specific heat and thermal conductivity, as well as, temperature dependent thermal expansion and Young’s modulus, the final warpage was modeled for a

time step during the printing simulation with this model. As the material cools down due to convection and radiation losses and the contact with the colder printing bed, the modeled reinforced PEEK material crystallizes. When hot and molten material is deposited onto already crystallized material, re-melting takes place, which is crucial for the bond formation.

4.4. Deformation and residual stresses The resulting deformation and the residual stresses of a printed material are of interest since printed shape and strength are issues. With the objective of modeling the warpage of simple printed geometries, a couple of papers [85,86] derived elastic solutions utilizing rigorous assumptions for the printing process. Between the melt temperature and the glass transition temperature, no thermal stresses were assumed to build up. Furthermore, layers were modeled to be deposited instantaneously. Consequently, the resulting temperature gradient was considered in the stacking direction of the print only. While qualitatively correct deformation behavior was modeled with the resulting analytic formulations in these papers, numerical approaches are required in order to predict accurate deformations. One of the first full 3D FEA models was developed by Zhang and Chou [87]. The authors coupled a thermal analysis with a deformation

Fig. 14. Temperature distribution (left) and crystallinity distribution (right) for a 3D analysis. The re-melting of previously deposited material can be captured [84].

11

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 15. Predicted distortion due to thermally induced stresses (left) and residual stresses after the print developed during the transient printing process [90]. Reproduced with permission.

mechanical analysis to model residual stresses and the resulting deformation state. Fig. 16 illustrates the modeled temperature and crystallinity distribution for a modeled print of a NACA air inlet duct autoclave tool. As printing material, a 50 wt.% carbon fiber reinforced PPS (polyphenylene sulfide) was utilized in the simulations. For the mechanical analysis, both crystallization shrinkage and thermomechanical shrinkage of the material were considered. Fig. 17 depicts the final residual stress state for the transverse in-plane stress component σ22 and the final tool warpage after the spring back analysis in which the constraint was released that held the part to the printing bed during the analysis. Based on the partial infill at an angle of 45°, the part warped asymmetrically in the spring back analysis. To generate the results in Fig. 17, temperature dependent elastic orthotropic material properties were used. Barocio et al. [93] extended the process simulation work to include a performance simulation for the simulated autoclave tool. Autoclave temperatures and pressures were modeled and the resulting deformations of the printed tool were investigated. Here, the analysis was informed with the residual stresses from the process simulations. The authors found that the resulting deformation during the simulated autoclave cycle was mostly caused by the thermoelastic deformation induced due to the temperature change in the autoclave, while the influence of the residual stresses was minimal. The summarized work illustrates that most papers on the solidification behavior and the prediction of residual stresses and deformations are limited to simplified, reduced order models that do not consider the full complexity of the process. Herein, most work focuses on the description of the thermal analysis during the printing process. However, the development of the required simulation capabilities is evolving fast and several publications exist that present full 3D simulations of actual printed parts to predict residual stresses and deformations. Nevertheless, important physics are still left out in the simulations. The presented papers utilize constant elastic or temperature dependent elastic properties in the mechanical analyses. The viscoelastic nature of the reinforced materials is not considered yet. Furthermore, for semi-

simple 3D part considering two different build directions. The first work regarding the full three dimensional analysis of a 3D printed geometry made from composite materials was published by Talagani et al. [90]. Here, the Extrusion Deposition Additive Manufacturing process was modeled. The authors developed a strongly coupled thermomechanical FEA model in order to predict residual stresses and deformations of a printed car chassis made from carbonfiber reinforced ABS. Based on the machine code for printing, a voxel type mesh was built to represent the car geometry. During the analysis, the elements were activated to depict the real printing process. Depending on the local bead directions, local material orientations were defined to assign orthotropic material properties. The relevant data was determined based on reverse engineered constituent properties. The thermal and mechanical analysis was strongly coupled with the aid of experimental temperature data achieved from printing experiments. The resulting set of simulations was used to predict residual stress states and deformations of the whole car chassis. By implementation of linear fracture mechanics, potentially critical regions for interbead crack failure could be identified as well. Fig. 15 illustrates the final deformation state and the residual stress state after the printing simulation for the modeled car chassis. Recently, Purdue Universities’ efforts in simulating the Extrusion Deposition Additive Manufacturing process were published in a series of papers. For these simulations, a newly developed toolset in Abaqus 2017 was utilized. The printing process can be modeled through a progressive element activation which enables building the part in the simulation as it would be built in an actual print. The element activation is based on the machine code that would be used for the real print as well. Also, based on this machine code, local material orientations can be defined to assign orthotropic material properties. These functionalities were made available through a user subroutine suite in Abaqus. More detailed information on this can be found in the paper by Favaloro et al. [91]. The structure of the developed simulation toolset is described in the paper by Brenken et al. [92]. A 3D heat transfer and crystallization kinetics simulation was sequentially coupled with a

Fig. 16. Temperature profile (left) and crystallinity (right) during the printing simulation of a NACA air inlet duct tool.

12

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Fig. 17. Final residual stress state for transverse in-plane stresses σ22 in MPa (left) and the final deformation magnitude (right) after release of the constraint that held the part to the printing bed during the analysis. The deformation is exaggerated with a scale factor of 10.

should be controllable within tight margins. In a detailed discussion of the FFF process and its phenomena in section 3, three main research areas were identified: the material flow and resulting fiber orientation, the wetting of previous material as well as the bond formation between adjacent beads, and the thermomechanical solidification behavior of the deposited material including the internal stress generation and deformation during the printing process. Most studies in the field of flow and fiber orientation assume an uncoupled, isotropic Newtonian flow, while the material should be modeled as a viscous fiber suspension flow. In the research area of bond formation, no publications have considered the effect of the fibers on the resulting bond formed between adjacent beads yet. Here, it has to be investigated how the viscosities transverse to the printing direction change with the addition of fibers and how that affects bond formation. Finally, three dimensional solidification analyses of deposited material are published and the methods for simulating the process exist; however, important physics like the viscoelastic material behavior or the potential for sagging at high temperatures after extrusion are not accounted for yet. In summary, the review of published research work revealed that to date there is only a limited amount of relevant work published in the identified research areas. Often studies investigate significantly simplified versions of the FFF process where not all involved phenomena are taken into account. Therefore, there is a strong need for more fundamental understanding and subsequent research considering the inherent complexity of the FFF process. Current research efforts at Purdue University focus on the described research areas in order to fulfill this need. While significant technical accomplishments have been achieved in recent years, the FFF process itself is still empirically calibrated. In order to change that and ensure the success and competitiveness of this manufacturing method, the research community has to keep up with the technical progress and provide reliable tools to model and predict the process and its outcomes. By summarizing both the current mechanical performance of FFF and past research work in the discussed main research areas, this paper is intended to serve as a foundation for the future simulation development in the FFF process.

crystalline polymers, the mechanical properties are strongly dependent on the degree of crystallinity. In addition, due to the very limited stiffness at high temperatures, sagging of the material should be included. Finally, a lot of work still needs to be done to validate the predicted results and compare the simulations to actual printed parts. However, all of the required physics need to be included in the simulation to achieve both qualitatively and quantitatively correct results. In a model extension at Purdue University that is yet unpublished, full thermoviscoelastic, crystallization informed material properties were included in the process simulations for a 50 wt.% carbon fiber reinforced PPS composite. As outcomes, both the results for stresses and deformations change significantly, which underlines the need for physically complete simulations before validation cases can be investigated. 5. Conclusions The present paper summarizes past published work on Fused Filament Fabrication (FFF) with fiber-reinforced polymeric materials. Many reviewed papers focus on investigating and reporting mechanical properties of printed reinforced material systems. Therefore, summaries of tensile mechanical properties are provided in Section 2. These include data for both discontinuous and continuous fiber systems. For the discontinuous fiber materials, properties both parallel and transverse to the bead direction were shown. Several important conclusions can be drawn from this information. First, discontinuous fiber-reinforced printed polymers do not compare to aircraft aluminum yet. As reasons for that, fiber damage during the printing process and printing defects with related void formation can be named. Second, the reported properties transverse to the deposition direction for thermoplastic, fiber-reinforced materials are especially low; thereby, identifying the transverse directions as one key limitation of printed materials. Finally, for continuous fiber systems loaded parallel to the bead direction, strength values can reach the ones of aluminum. However, efforts have to be undertaken in order to ensure a proper wetting of the fibers during printing. Furthermore, the printable fiber volume fractions have to be increased. Despite of the identified mechanical limitations, the FFF method shows promising potential for the application to composite tooling. Significant time and cost savings are possible in comparison to traditional tool manufacturing, especially for prototype and low series tooling. In addition, the optional tailorability of the CTE of printed tools is an interesting feature. By modifying the CTE of the printed tool to approach the ones of the laid up part, the resulting part tolerances

Acknowledgements This work was supported by grant no 4000132567, Oak Ridge National Laboratory, UT-Battelle, LLC, acting under contract DE-AC0500OR22725 with the U.S. Department of Energy.

Appendix A See Tables A1–A3. 13

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

Table A1 Summary of extracted mechanical data for discontinuous fiber systems loaded parallel to the printing direction. Source

Material

Tensile Modulus (GPa)

Tensile Strength (MPa)

Extraction

Tekinalp et al. [11]

ABS/CF 10 wt.% ABS/CF 20 wt.% ABS/CF 30 wt.% ABS/CF 40 wt.% ABS/CF 13 vol.% ABS/CF 3 wt.% ABS/CF 5 wt.% ABS/CF 7.5 wt.% ABS/CF 10 wt.% ABS/CF 15 wt.% ABS/CF 20 wt.% ABS/CF 20 wt.% ABS/GF 20 wt.% ABS/GF 40 wt.% ABS/CF 13 vol.% Epoxy/SiC/CF 10 wt.% PPS/CF 50 wt.% PEI/CNT 4.7 wt.% ABS/VGCF 10 wt.% ABS/VGCF 5 wt.% ABS/5 wt.% SWNT ABS/Jute fiber 5 wt.% Epoxy/CF 15 wt.% ABS/chopped CF 20 wt.% ABS/CF 15 wt.% PEI/CF 20 wt.% PLA/CF 15 wt.%

7.7 11.5 13.8 13.7 8.91 2.1 2.45 2.5 2.15 2.25 8.4 11.9 5.7 10.8 8.15 24.5 26.4 3 0.8 1.27 1.74 1.54 4.05 10.87 11.88 8.36 7.54

52 60 62 67 70.69 40.8 42 41.5 33.8 35 66.8 65.7 54.3 51.2 53 66.2 92.2 125.3 37.4 27 32.5 25.9 66.3 47.7 61.9 61.1 53.4

Graphical Graphical Graphical Graphical Number Graphical Graphical Graphical Graphical Graphical Number Number Number Number Graphical Number Number Number Number Graphical Graphical Number Number Number Number Number Number

Love et al. [5] Ning et al. [21]

Hill et al. [22] Duty et al. [23]

Kunc [25] Compton & Lewis [24] DeNardo [6] Gardner et al. [36] Shofner et al. [35] Shofner et al. [37] Perez et al. [26] Mahajan et al. [28] Duty et al. [29]

Ferreira et al. [30]

Abbreviations: CF – carbon fiber, GF – glass fiber, SiC – silicon carbide, CNT – carbon nanotube, VGCF – vapor grown carbon fiber, SWNT – single-walled carbon nanotube.

Table A2 Summary of extracted mechanical data for discontinuous fiber systems loaded transverse to the printing direction. Source

Material

Tensile Modulus (GPa)

Tensile Strength (MPa)

Extraction

Love et al. [5] Hill et al. [22] Duty et al. [23]

ABS/CF 13 vol.% ABS/CF 20 wt.% ABS/CF 20 wt.% ABS/GF 20 wt.% ABS/CF 13 vol.% Epoxy/SiC/CF 10 wt.% PPS/CF 50 wt.% Epoxy/CF 15 wt.% ABS/chopped CF 20 wt.% ABS/CF 15 wt.% PEI/CF 20 wt.% PLA/CF 15 wt.%

1.52 2.6 2.1 2.5 2.2 8.06 2.6 2.84 1.98 1.83 1.1 3.92

7 12.8 10.3 15.3 13 43.9 9.72 46 6.8 5.8 4.3 35.4

Number Number Number Number Graphical Number Number Number Number Number Number Number

Kunc [25] Compton & Lewis [24] DeNardo [6] Mahajan et al. [28] Duty et al. [29]

Ferreira et al. [30]

Abbreviations: CF – carbon fiber, GF – glass fiber, SiC – silicon carbide. Table A3 Summary of extracted mechanical data for continuous fiber systems loaded parallel to the printing direction. Source

Material

Tensile Modulus (GPa)

Tensile Strength (MPa)

Extraction

Matsuzaki et al. [38]

PLA/CF 6.6 vol% PLA/Jute fiber 6.1 vol% Nylon/CF 6 vol.%a Nylon/CF 18 vol.%a PLA/CF 34 vol.% Nylon/AF 4 vol.% Nylon/AF 8 vol.% Nylon/AF 10 vol.% PLA/CF 10 wt.% PLA/AF 8.6 vol.% Nylon/CF 11 vol.% Nylon/AF 8 vol.% Nylon/GF 8 vol.% Nylon/AF 10 vol.% Nylon/GF 10 vol.% ABS/CF 10 wt.%

19.5 5.11 14 35.7 23.8 1.77 6.92 9 20.6 9.34 8.46 4.23 3.29 4.76 4.91 4.19

185.2 57.1 140 464.4 91 31 60 84 256 203 198 110 156 161 212 147

Number Number Graphical Number Numberb Graphical Graphical Graphical Number Number Number Number Number Number Number Number

van der Klift et al. [40] Li et al. [41] Melenka et al. [42]

Tian et al. [43] Bettini et al. [45] Dickson et al. [46]

Yang et al. [47]

Abbreviations: CF – carbon fiber, GF – glass fiber, AF – aramid fiber. a The volume fractions were determined in correspondence with the author. b The stiffness was estimated based on data provided.

14

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

References

2017.05.013. [31] F. Ning, W. Cong, Y. Hu, H. Wang, Additive manufacturing of carbon fiber-reinforced plastic composites using fused deposition modeling: effects of process parameters on tensile properties, J. Compos. Mater. 51 (2017) 451–462, http://dx. doi.org/10.1177/0021998316646169. [32] J.P. Lewicki, J.N. Rodriguez, C. Zhu, M.A. Worsley, A.S. Wu, Y. Kanarska, J.D. Horn, E.B. Duoss, J.M. Ortega, W. Elmer, R. Hensleigh, R.A. Fellini, M.J. King, 3D-Printing of meso-structurally ordered carbon fiber/polymer composites with unprecedented orthotropic physical properties, Sci. Rep. 7 (2017) 43401, http://dx. doi.org/10.1038/srep43401. [33] R.W. Gray, D.G. Baird, J.H. Bohn, Thermoplastic composites reinforced with long fiber thermotropic liquid crystalline polymers for fused deposition modeling, Polym. Compos. 19 (1998) 383–394, http://dx.doi.org/10.1002/Pc.10112. [34] R.W. Gray Iv, D.G. Baird, J.H. Bøhn, Effects of processing conditions on short TLCP fiber reinforced FDM parts, Rapid Prototyp. J. 4 (1998) 14–25, http://dx.doi.org/ 10.1108/13552549810197514. [35] M.L. Shofner, K. Lozano, F.J. Rodríguez-Macías, E.V. Barrera, Nanofiber-reinforced polymers prepared by fused deposition modeling, Appl. Polym. Sci. 89 (2003) 3081–3090. [36] J.M. Gardner, G. Sauti, J.-W. Kim, R.J. Cano, R.A. Wincheski, E.V. Barrera, Additive manufacturing of multifunctional components using high density carbon nanotube yarn filaments, SAMPE Conf. Proc. Long Beach, CA, 2016. [37] M.L. Shofner, F.J. Rodríguez-Macías, R. Vaidyanathan, E.V. Barrera, Single wall nanotube and vapor grown carbon fiber reinforced polymers processed by extrusion freeform fabrication, Compos. Part A Appl. Sci. Manuf. 34 (2003) 1207–1217, http://dx.doi.org/10.1016/j.compositesa.2003.07.002. [38] R. Matsuzaki, M. Ueda, M. Namiki, T.-K. Jeong, H. Asahara, K. Horiguchi, T. Nakamura, A. Todoroki, Y. Hirano, Three-dimensional printing of continuousfiber composites by in-nozzle impregnation, Sci. Rep. 6 (2016) 23058, http://dx. doi.org/10.1038/srep23058. [39] Y. Koga, F. van der Klift, A. Todoroki, M. Ueda, Y. Hirano, R. Matsuzaki, The printing process of 3D printer for continuous CFRTP, SAMPE Conf Proc. Long Beach, CA, 2016. [40] F. Van Der Klift, Y. Koga, A. Todoroki, M. Ueda, Y. Hirano, R. Matsuzaki, 3D printing of continuous carbon fibre reinforced thermo-plastic (CFRTP) tensile test specimens, Open J. Compos. Mater. 6 (2016) 18–27, http://dx.doi.org/10.4236/ ojcm.2016.61003. [41] N. Li, Y. Li, S. Liu, Rapid prototyping of continuous carbon fiber reinforced polylactic acid composites by 3D printing, J. Mater. Process. Technol. 238 (2016) 218–225, http://dx.doi.org/10.1016/j.jmatprotec.2016.07.025. [42] G.W. Melenka, B.K.O. Cheung, J.S. Schofield, M.R. Dawson, J.P. Carey, Evaluation and prediction of the tensile properties of continuous fiber-reinforced 3D printed structures, Compos. Struct. 153 (2016) 866–875, http://dx.doi.org/10.1016/j. compstruct.2016.07.018. [43] X. Tian, T. Liu, Q. Wang, A. Dilmurat, D. Li, G. Ziegmann, Recycling and remanufacturing of 3D printed continuous carbon fiber reinforced PLA composites, J. Clean. Prod. 142 (2017) 1609–1618, http://dx.doi.org/10.1016/j.jclepro.2016.11. 139. [44] X. Tian, T. Liu, C. Yang, Q. Wang, D. Li, Interface and performance of 3D printed continuous carbon fiber reinforced PLA composites, Compos. Part A Appl. Sci. Manuf. 88 (2016) 198–205, http://dx.doi.org/10.1016/j.compositesa.2016.05.032. [45] P. Bettini, G. Alitta, G. Sala, L. Di Landro, Fused deposition technique for continuous fiber reinforced thermoplastic, J. Mater. Eng. Perform. 26 (2017) 843–848, http:// dx.doi.org/10.1007/s11665-016-2459-8. [46] A.N. Dickson, J.N. Barry, K.A. McDonnell, D.P. Dowling, Fabrication of continuous carbon, glass and Kevlar fibre reinforced polymer composites using additive manufacturing, Addit. Manuf. 16 (2017) 146–152, http://dx.doi.org/10.1016/j.addma. 2017.06.004. [47] C. Yang, X. Tian, T. Liu, Y. Cao, D. Li, 3D printing for continuous fiber reinforced thermoplastic composites: mechanism and performance, Rapid Prototyp. J. 23 (2017) 209–215, http://dx.doi.org/10.1108/RPJ-08-2015-0098. [48] E.A. Starke, J.T. Staley, Application of modern aluminum alloys to aircraft, Prog. Aerosp. Sci. 32 (1996) 131–172. [49] Markforged, Composite 3D Printing, (2017) https://markforged.com/composites/ (Accessed September 14 2017). [50] R.B. Pipes, J.W.S. Hearle, A.J. Beaussart, A.M. Sastry, R.K. Okine, A constitutive relation for the viscous flow of an oriented fiber assembly, J. Compos. Mater. 25 (1991) 1204–1217, http://dx.doi.org/10.1177/002199839102501008. [51] A. Bellini, Fused Deposition of Ceramics: A Comprehensive Experimental, Analytical and Computational Study of Material Behavior, Fabrication Process and Equipment Design, Drexel University, 1991, http://dx.doi.org/10.16953/deusbed. 74839. [52] H.S. Ramanath, M. Chandrasekaran, C.K. Chua, K.F. Leong, K.D. Shah, Modelling of extrusion behaviour of biopolymer and composites in fused deposition modelling, Key Eng. Mater. 334–335 (2007) 1241–1244, http://dx.doi.org/10.4028/www. scientific.net/KEM.334-335.1241. [53] H.S. Ramanath, C.K. Chua, K.F. Leong, K.D. Shah, Melt flow behaviour of poly-ecaprolactone in fused deposition modelling, J. Mater. Sci. Mater. Med. 19 (2007) 2541–2550, http://dx.doi.org/10.1007/s10856-007-3203-6. [54] N. Mostafa, H.M. Syed, S. Igor, G. Andrew, A study of melt flow analysis of an ABSiron composite in fused deposition modelling process, Tsinghua Sci. Technol. 14 (Suppl) (2009) 29–37, http://dx.doi.org/10.1016/S1007-0214(09)70063-X. [55] J. Nixon, B. Dryer, D. Chiu, I. Lempert, D.I. Bigio, Three parameter analysis of fiber orientation in fused deposition modeling geometries, 72nd Annu. Tech. Conf. Soc. Plast. Eng. Plast. Conf. ANTEC 2014. 2 2 (2014) 985–995 http://www.scopus.com/ inward/record.url?eid=2-s2.0-84938266374&partnerID=40&md5=

[1] A.A. Hassen, R. Springfield, J. Lindahl, B. Post, L. Love, C. Duty, U. Vaidya, R.B. Pipes, V. Kunc, The durability of large-scale additive manufacturing composite molds, CAMX Conf. Proc. Anaheim, CA, 2016. [2] L.J. Love, Utility of big area additive manufacturing (BAAM) for the rapid manufacture of customized electric vehicles, (2014) 1–11. 10.2172/1209199. [3] Cincinnati Incorporated, Big Area Additive Manufacturing, (2016) www.e-ci.com/ baam/ (Accessed December 12 2016). [4] Thermwood, Large Scale Additive Manufacturing, (2016) http://thermwood.com/ lsam/lsam_main.htlm (Accessed December 15 2016). [5] L.J. Love, V. Kunc, O. Rios, C.E. Duty, A.M. Elliott, B.K. Post, R.J. Smith, C.A. Blue, The importance of carbon fiber to polymer additive manufacturing, J. Mater. Res. 29 (2014) 1893–1898, http://dx.doi.org/10.1557/jmr.2014.212. [6] N.M. DeNardo, Additive Manufacturing of Carbon Fiber-Reinforced Thermoplastic Composites, Purdue University, 2016. [7] T.Z. Sudbury, R. Springfield, V. Kunc, C. Duty, An assessment of additive manufactured molds for hand-laid fiber reinforced composites, Int. J. Adv. Manuf. Technol. 90 (2017) 1659–1664, http://dx.doi.org/10.1007/s00170-016-9464-9. [8] V. Kunc, J. Lindahl, R. Dinwiddie, B. Post, L. Love, M. Matlack, R.L.J. Fahey, A.A. Hassen, Investigation of In-autoclave additive manufacturing composite tooling, CAMX Conf Proc. Anaheim, CA, 2016. [9] A. Hassen, J. Lindahl, X. Chen, B. Post, L. Love, V. Kunc, Additive manufacturing of composite tooling using high temperature thermoplastic materials, SAMPE Conf Proc. Long Beach, CA, 2016. [10] N. DeNardo, E. Barocio, B. Brenken, A. Favaloro, R.B. Pipes, Economics of composite tooling made via additive manufacturing, SAMPE Conf Proc. Long Beach, CA, 2016. [11] H.L. Tekinalp, V. Kunc, G.M. Velez-Garcia, C.E. Duty, L.J. Love, A.K. Naskar, C.A. Blue, S. Ozcan, Highly oriented carbon fiber-polymer composites via additive manufacturing, Compos. Sci. Technol. 105 (2014) 144–150, http://dx.doi.org/10. 1016/j.compscitech.2014.10.009. [12] e-Xstream, Additive Manufacturing—Print It Right the First Time!, (2017) www.exstream.com/page/additive-manufacturing (Accessed June 30 2017). [13] J. Fort, S. Sett, How Simulation Can Help Advance Additive Manufacturing Technology, (2015) https://www.3ds.com/products-services/simulia/resources/ how-simulation-can-help-advance-additive-manufacturing-technology/ (Accessed June 30 2017). [14] N. Guo, M.C. Leu, Additive manufacturing: technology, applications and research needs, Front. Mech. Eng. 8 (2013) 215–243, http://dx.doi.org/10.1007/s11465013-0248-8. [15] H. Bikas, P. Stavropoulos, G. Chryssolouris, Additive manufacturing methods and modeling approaches: a critical review, Int. J. Adv. Manuf. Technol. 83 (2016) 389–405, http://dx.doi.org/10.1007/s00170-015-7576-2. [16] B.N. Turner, R. Strong, S.A. Gold, A review of melt extrusion additive manufacturing processes: I. Process design and modeling, Rapid Prototyp. J. 20 (2014) 192–204, http://dx.doi.org/10.1108/RPJ-01-2013-0012. [17] B.N. Turner, S.A. Gold, A review of melt extrusion additive manufacturing processes: II. Materials, dimensional accuracy, and surface roughness, Rapid Prototyp. J. 21 (2015) 250–261, http://dx.doi.org/10.1108/RPJ-02-2013-0017. [18] S. Kumar, J.P. Kruth, Composites by rapid prototyping technology, Mater. Des. 31 (2010) 850–856, http://dx.doi.org/10.1016/j.matdes.2009.07.045. [19] X. Wang, M. Jiang, Z. Zhou, J. Gou, D. Hui, 3D printing of polymer matrix composites: a review and prospective, Compos. Part B Eng. 110 (2017) 442–458, http:// dx.doi.org/10.1016/j.compositesb.2016.11.034. [20] W. Zhong, F. Li, Z. Zhang, L. Song, Z. Li, Short fiber reinforced composites for fused deposition modeling, Mater. Sci. Eng. A 301 (2001) 125–130, http://dx.doi.org/10. 1016/S0921-5093(00)01810-4. [21] F. Ning, W. Cong, J. Qiu, J. Wei, S. Wang, Additive manufacturing of carbon fiber reinforced thermoplastic composites using fused deposition modeling, Compos. Part B Eng. 80 (2015) 369–378, http://dx.doi.org/10.1016/j.compositesb.2015.06.013. [22] C. Hill, K. Rowe, R. Bedsole, J. Earle, V. Kunc, Materials and process development for direct digital manufacturing of vehicles, SAMPE Conf Proc. Long Beach, CA, 2016. [23] C. Duty, T. Drye, A. Franc, Material development for tooling applications using big area additive manufacturing (BAAM), ORNL Tech. Rep. (2015) ORNL/TM2015/78. [24] B.G. Compton, J.A. Lewis, 3D-printing of lightweight cellular composites, Adv. Mater. 26 (2014) 5930–5935, http://dx.doi.org/10.1002/adma.201401804. [25] V. Kunc, Advances and challenges in large scale polymer additive manufacturing, 15th SPE Automot. Compos. Conf. Novi, MI (2015). [26] A.R.T. Perez, D.A. Roberson, R.B. Wicker, Fracture surface analysis of 3D-printed tensile specimens of novel ABS-based materials, J. Fail. Anal. Prev. 14 (2014) 343–353, http://dx.doi.org/10.1007/s11668-014-9803-9. [27] O.S. Carneiro, A.F. Silva, R. Gomes, Fused deposition modeling with polypropylene, Mater. Des. 83 (2015) 768–776, http://dx.doi.org/10.1016/j.matdes.2015.06.053. [28] C. Mahajan, D. Cormier, 3D printing of carbon fiber composites with preferentially aligned fibers, Proc. 2015 Ind. Syst. Eng. Reserach Conf. (2015) 2953–2963. [29] C.E. Duty, V. Kunc, B. Compton, B. Post, D. Erdman, R. Smith, R. Lind, P. Lloyd, L. Love, Structure and mechanical behavior of big area additive manufacturing (BAAM) materials, Rapid Prototyp. J. 23 (2017) 181–189, http://dx.doi.org/10. 1108/RPJ-12-2015-0183. [30] R.T.L. Ferreira, I.C. Amatte, T.A. Dutra, D. Bürger, Experimental characterization and micrography of 3D printed PLA and PLA reinforced with short carbon fibers, Compos. Part B Eng. 124 (2017) 88–100, http://dx.doi.org/10.1016/j.compositesb.

15

Additive Manufacturing 21 (2018) 1–16

B. Brenken et al.

[76] M.A. Yardimci, T. Hattori, S.I. Guceri, S.C. Danforth, Thermal analysis of fused deposition, Solid Free. Fabr. Proc. Austin, TX, 1997, pp. 689–698. [77] S.F. Costa, F.M. Duarte, J.A. Covas, Towards modelling of Free Form Extrusion: analytical solution of transient heat transfer, Int. J. Mater. Form. 1 (2008) 703–706, http://dx.doi.org/10.1007/s12289-008-0312-9. [78] S. Costa, F. Duarte, J.A. Covas, Using MATLAB to compute heat transfer in free form extrusion, Matlab—A Ubiquitous Tool Pract. Eng. InTech, Rijeka, Croatia, 2011, pp. 453–474. [79] J.F. Rodriguez-Matas, Modeling the Mechanical Behavior of Fused Desposition Acrylonitrile-Butadiene Styrene Polymer Composites, University of Notre Dame, 1999. [80] B. Brenken, A. Favaloro, E. Barocio, N.M. DeNardo, R.B. Pipes, Development of a model to predict temperature history and crystallization behavior of 3D printed parts made from fiber-Reinforced thermoplastic polymers, SAMPE Conf Proc. Long Beach, CA, 2016. [81] Y. Zhou, T. Nyberg, G. Xiong, D. Liu, Temperature analysis in the fused deposition modeling process, 2016 3rd Int. Conf. Inf. Sci. Control Eng. (2016) 678–682, http:// dx.doi.org/10.1109/ICISCE.2016.150. [82] S.F. Costa, F.M. Duarte, J.A. Covas, Thermal conditions affecting heat transfer in FDM/FFE: a contribution towards the numerical modelling of the process, Virtual Phys. Prototyp. 10 (2015) 35–46, http://dx.doi.org/10.1080/17452759.2014. 984042. [83] K. Pooladvand, C. Furlong, Thermo-mechanical investigation of fused deposition modeling by computational and experimental methods, Mech. Compos. MultiFunctional Mater. vol. 7, Springer, US, 2016, http://dx.doi.org/10.1007/978-3319-41766-0. [84] B. Brenken, E. Barocio, A. Favaloro, R.B. Pipes, Simulation of Semi-Crystalline Composites in the Extrusion Deposition Additive Manufacturing Process, in: Proc. Sci. Age Exp. Conf. by Dassault Syst., Chicago, IL, n.d. [85] T.M. Wang, J.T. Xi, Y. Jin, A model research for prototype warp deformation in the FDM process, Int. J. Adv. Manuf. Technol. 33 (2007) 1087–1096, http://dx.doi.org/ 10.1007/s00170-006-0556-9. [86] L. Xinhua, L. Shengpeng, L. Zhou, Z. Xianhua, C. Xiaohu, W. Zhongbin, An investigation on distortion of PLA thin-plate part in the FDM process, Int. J. Adv. Manuf. Technol. 79 (2015) 1117–1126, http://dx.doi.org/10.1007/s00170-0156893-9. [87] Y. Zhang, Y.K. Chou, 3D FEA simulations of fused deposition modeling process, ASME Int. Conf. Manuf. Sci. Eng. (2006) 1–8. [88] Y. Zhang, K. Chou, A parametric study of part distortions in fused deposition modelling using three-dimensional finite element analysis, Proc. Inst. Mech. Eng. Part B J. Eng. Manuf. 222 (2008) 959–968, http://dx.doi.org/10.1243/ 09544054JEM990. [89] P. Hébert, S. Mathieu, L. Adam, D. Gianotta, C. Basire, Holistic multiscale simulation approach for additive layer manufacturing of plastics, SPE ACCE Conf. (2016) 1–15. [90] M.R. Talagani, S. Dormohammadi, R. Dutton, C. Godines, H. Baid, F. Abdi, Numerical simulation of big area additive manufacturing (3D printing) of a full size car, SAMPE J. 51 (2015) 27–36. [91] A. Favaloro, B. Brenken, E. Barocio, R.B. Pipes, Simulation of polymeric composites additive manufacturing in abaqus, Proc Sci. Age Exp. Conf. by Dassault Syst. Chicago, IL, 2017. [92] B. Brenken, A. Favaloro, E. Barocio, R.B. Pipes, Simulation of semi-crystalline composite tooling made by extrusion deposition additive manufacturing, SAMPE Conf Proc. Seattle, WA, 2017. [93] E. Barocio, B. Brenken, A. Favaloro, R.B. Pipes, Extrusion deposition additive manufacturing of composite molds for high-temperature applications, SAMPE Conf Proc. Seattle, WA, 2017.

c199db9b7d23d7bc87fd991510352456 .. [56] F. Folgar, C.L. Tucker, Orientation behavior of fibers in concentrated suspensions, J. Reinf. Plast. Compos. 3 (1984) 98–119, http://dx.doi.org/10.1177/ 073168448400300201. [57] A. Garcia, Nozzle geometry effects on fiber orientation, SAMPE Conf Proc. Seattle, WA, 2017. [58] B.P. Heller, D.E. Smith, D.A. Jack, The effects of extrudate swell, nozzle shape, and nozzle convergence zone on fiber orientation in fused deposition modeling, Am Soc. Compos. Proc. East Lansing, MI, 2015. [59] B.P. Heller, D.E. Smith, D.A. Jack, Effects of extrudate swell and nozzle geometry on fiber orientation in Fused Filament Fabrication nozzle flow, Addit. Manuf. 12 (2016) 252–264, http://dx.doi.org/10.1016/j.addma.2016.06.005. [60] B.P. Heller, D.E. Smith, D.A. Jack, Computing mechanical properties from orientation tensor for fiber filled polymers in axisymmetric flow and planar deposition flow, SPE ACCE Conf. (2016) 1–13. [61] S.G. Advani, C.L. Tucker, The use of tensors to describe and predict fiber orientation in short fiber composites, J. Rheol. (N. Y. N. Y.) 31 (1987) 751–784, http://dx.doi. org/10.1122/1.549945. [62] S. Montgomery-Smith, D. Jack, D.E. Smith, The fast exact closure for Jeffery’s equation with diffusion, J. Nonnewton. Fluid Mech. 166 (2011) 343–353, http:// dx.doi.org/10.1016/j.jnnfm.2010.12.010. [63] R.P. Wool, B.-L. Yuan, O.J. McGarel, Welding of polymer interfaces, Polym. Eng. Sci. 29 (1989) 1340–1367, http://dx.doi.org/10.1002/pen.760291906. [64] J.P. Thomas, J.F. Fodriguez, Modeling the fracture strength between fused-deposition extruded roads, Solid Free. Fabr. Proc. (2000) 17–23. [65] Q. Sun, G.M. Rizvi, C.T. Bellehumeur, P. Gu, Effect of processing conditions on the bonding quality of FDM polymer filaments, Rapid Prototyp. J. 14 (2008) 72–80, http://dx.doi.org/10.1108/13552540810862028. [66] C. Bellehumeur, L. Li, Modeling of bond formation between polymer filaments in the fused deposition modeling process, J. Manuf. Process. 6 (2004) 170–178, http://dx.doi.org/10.1016/S1526-6125(04)70071-7. [67] O. Pokluda, C.T. Bellehumeur, J. Machopoulos, Modification of Frenkel’s model for sintering, AIChE J. 43 (1997) 3253–3256, http://dx.doi.org/10.1002/aic. 690431213. [68] L. Li, Q. Sun, C.T. Bellehumeur, P. Gu, Investigation of bond formation in FDM process, Solid Free. Fabr. Symp. Proc. Austin, TX, 2001. [69] Q. Sun, Bond Formation Between Polymer Filaments in Fused Deposition Modeling Process, THE University of Calgary, 2004, http://dx.doi.org/10.16953/deusbed. 74839. [70] P.K. Gurrala, S.P. Regalla, Part strength evolution with bonding between filaments in fused deposition modelling, Virtual Phys. Prototyp. 9 (2014) 141–149, http://dx. doi.org/10.1080/17452759.2014.913400. [71] N. Aliheidari, R. Tripuraneni, A. Ameli, S. Nadimpalli, Fracture resistance measurement of fused deposition modeling 3D printed polymers, Polym. Test. 60 (2017) 94–101, http://dx.doi.org/10.1016/j.polymertesting.2017.03.016. [72] D. Young, J. Kessler, M. Czabaj, Interlayer fracture toughness of additively manufactured unreinforced and carbon-fiber-reinforced acrylonitrile butadiene styrene, Am. Soc. Compos. Thirty-First Tech. Conf. (2016). [73] V. Kishore, C. Ajinjeru, C.E. Duty, A. Nycz, B.K. Post, J.M. Lindahl, V. Kunc, Infrared preheating to enhance interlayer strength of components printed on the big area additive manufacturing (BAAM) system, SAMPE Conf Proc. Long Beach, CA, 2016. [74] A.K. Ravi, A. Deshpande, K.H. Hsu, An in-process laser localized pre-deposition heating approach to inter-layer bond strengthening in extrusion based polymer additive manufacturing, J. Manuf. Process. 24 (2016) 179–185, http://dx.doi.org/ 10.1016/j.jmapro.2016.08.007. [75] M. a Yardimci, S. Güçeri, Conceptual framework for the thermal process modelling of fused deposition, Rapid Prototyp. J. 2 (1996) 26–31, http://dx.doi.org/10.1108/ 13552549610128206.

16