Large-format additive manufacturing of polymer extrusion-based deposition systems: review and applications

Material jetting printers are characterized by their excellent accuracy and precision, allowing the printing of parts with various colors and materials, in a process that requires little concern with support structures [37]. Its characteristic good accuracy and resolution are usually dependent on a limited volume and relative speed, which inevitably imply that printers are typically expensive. From our market research, the model Mimaki 3DGD-1800 from Mimaki is the one which allows to print larger parts (up to 2.90 m³) [38], with a layer resolution of 800 µm. The only reported value for its building speed is 350 mm/hour (vertical direction), which is a high speed, when compared with the building rates of other printers of this AM process family.

Powder bed fusion printers are limited in size by the available volume of the powder bed [39], from all studied models, the TPM3D S800DL from TPM3D has a maximum build volume of 0.29 m³ [40]. Even so, the printing process is time consuming since the volume build rates (10–25 mm³/h) are low. For instance, considering a volume build rate of 25 mm³/h, the required time to print a volume of 0.29 m³ will be almost one week. The price of equipment and consumables is also high when compared with other AM technologies.

A key advantage of binder jetting over other AM processes is that bonding occurs at room temperature. This means that dimensional distortions associated with thermal effects are not a problem. As a result, the build volume of binder jetting machines is amongst the largest among AM technologies (up to 2200 × 1200 × 600 mm, roughly 1.6 m³) [41]. These large machines are generally used to produce sand-casting molds. Metal and polymer binder jetting systems typically have smaller build volumes (up to 1000 × 600 × 500, 0.3 m³) for the polymer printer Voxeljet AG VX1000 from Voxeljet AG [42]. Typical layer height depends on the material, which for polymers will be around 100 µm.

The build volume of vat polymerization systems is primarily defined by the vat that contains the photopolymerizable resin. In some systems the produced part can be larger than the photopolymer vat, since the part is produced using a bottom-up approach. In this configuration the optical source is projected on the bottom of the vat. The print starts by lowering the build platform to touch the bottom of the resin-filled vat, then moving upward the height of one layer and successively adding new layers to the object [43]. This bottom-up approach allows to obtain parts up to 2.1 m long on the Materialize machine [44]. Compared to traditional vat polymerization, these larger systems can provide unique parts (without assemblies), which can increase their mechanical performance, commonly identified as one of the downsides of vat polymerization technology.

ASTM defines material extrusion, as “AM process in which material is selectively dispensed through a nozzle or orifice” [1]. Concerning polymers extrusion, the process can be done with thermoplastic-based materials, where the raw materials are either in the form of filament (FFF) or pellets (FGF, fused granular fabrication), both processes involve the melting of raw materials on a vertical extrusion system. An extrusion system is a system constituted by a feeder, a hot end, and a nozzle. The function of the feeder is to place filament/pellets into the hot end. The hot end will melt the polymer and guide the melted polymer to the nozzle, that is responsible to define the diameter of melted material to be deposited. The polymer extrusion process can also be performed with thermosets (without melting), using an extrusion head that combines resin and hardener during deposition [45, 46]. In general, the relationship between nozzle diameter, head speed and material feed rate are coordinated such that material deposition is precise, creating the desire product layer by layer [47].

Within all AM process families, FFF has become one of the most widely used AM processes in the industry for functional prototypes and low volume production series, using polymers such as Acrylonitrile Butadiene Styrene (ABS), Polycarbonate (PC) or Polylactic Acid (PLA), with some restrictions regarding mechanical performance. To increase the mechanical performance (strength, stiffness, and creep) of parts obtained by FFF, engineering plastics like polyphenylsulfone (PPSU), polyetherimide (PEI) and polyether ether ketone (PEEK) are under continuous development. These materials offer improved specific bulk properties but are often limited by defects inherent to FFF technique, namely, a layered structure with difficult to predict porosities between adjacent rasters, and low inter-layer adhesion.

From our market research, the commercially available material extrusion printers with larger volumes are the Jupiter from ATMAT using filaments [48], and the LSAM 1540 from Thermwood using pellets [49], with building volumes of 2.0 and 84.6 m³, respectively. The Jupiter model from ATMAT is equipped with two printing heads specifically designed for large-format production and owns a granite heated bead to ensure a stable operation. The Jupiter printer also had a heated chamber and can deposit up to 0.5 kg of material per hour. The LSAM 1540 manufacturing system from Thermwood has a proprietary print head extruder called “MELT CORE”. This new design uses a servo-controlled plastic extruder with a plasticizing screw to heat and soften the polymer. After the bead deposition, the bead is compressed through a system based on a servo-controlled compression wheel, that flattens and fuses each new printed layer to existing layers. The LSAM 1540 printer from Thermwood has a maximum printing temperature of 450 ºC and can deposit between 91 and 227 kg of material per hour, using 40- and 60-mm print heads, respectively. The LSAM 1540 printer can also be used with the “Vertical Layer Print” set-up. In the “Vertical Layer Print” set-up parts are printed along the length of the print bed, with the layer stack direction along the length of the part. The LSAM 1540 printer does this by adding a second moving table, mounted perpendicular to the main fixed horizontal table. Thermwood state that the employed “controlled cooling” print technology minimizes sag, which is a common problem in thermoplastics 3D printing. With the “Vertical Layer Print” set-up, the LSAM 1540 printer can print parts with a length up to 12 m.

After the market research analysis, results for all AM families involving polymer materials reveal that, currently, the AM process family most suitable to produce LFAM parts is most likely based in material extrusion configurations, mostly due to the capability of producing larger parts with low costs, Table 1.

Besides all of its benefits, conventional material extrusion imposes fundamental limitations on speed, and scale of parts [50‐52]. As a way to overcome the limitations of scale, LFAM methods such as medium size AM (MSAM) [53] and high size AM (HSAM) [54], with build volumes ranging from 1 to 7 m³ in MSAM and higher than 7 m³ on HSAM, have received a lot of attention in the last years.

Also, a limitation of the conventional FFF process is the rate of deposition that is constraint by the head speed, acceleration, and extrusion diameter (0.1–0.5 mm), which limits the deposition rates to less than 0.5 kg/h, on printers usually with small build volumes (< 1 m³). LFAM methods can employ large extrusion head diameters (0.5– 4 mm in MSAM and 4–7.6 mm in BSAM) that are commonly feed by polymer pellets [55, 56] instead of filaments, which beyond increasing the deposition rates (to 0.5–4 kg/h in MSAM and 4–50 kg/h in HSAM) can also represent a reduction in costs of raw materials in more than one order of magnitude. LFAM surges as a solution to manufacture large parts while avoiding joining/welding of multiple parts, as extensively addressed in a research review by Tiwary et al. [33].

Tiwart et al. stated that their paper is the first literature review related with joining or welding of FFF 3D printed parts to obtain a bigger volume 3D printed component. This element of welding after fabrication of smaller components is a competition for systems capable of 3D printing of volumes directly. On this review, different joining techniques including adhesive bonding, mechanical interlocking, use of fasteners, LFAM and welding are presented with particular emphasis on the welding of 3D printed parts. The authors describe the procedures for applying the above referred joining/welding techniques, the correspondent advantages and limitations and present the challenges and future research direction in the context of 3D printed parts.

When evaluating processing costs, the energy requirements of LFAM methods (4 × 10⁶ J/kg), are one order of magnitude lower than the one reported for the small-scale AM (4.1 × 10⁷ J/kg) [52], due to several factors, including the removal of the heated chamber [32]. The elimination of the heating chamber in LFAM systems, brings associated problems of part warping during the cooling process. If unoptimized, the material’s anisotropic shrinkage and nonlinear compression characteristics cause severe delamination, cross-sectional tapering, and warpage [57, 58], and also increasing the risk of sagging structures [59]. One repeatedly reported solution to minimize warping, is the use of thermoplastics reinforced with small fibres [60‐63]. Reinforced thermoplastics have an average thermal expansion coefficient one order of magnitude smaller than the unreinforced thermoplastics. An alternative approach is the utilization of thermoset materials using an extrusion head that combines resin and hardener during deposition [45, 46]. In LFAM, time is commonly compromised or, using larger extrusion widths, the tendency is a decrease in resolution and a consequent decline in part surface finish [64, 65]. Therefore, the lack of print resolution and extruder flowrate control, characteristic of HSAM systems, can lead to potentially significant geometric deviations in the printed part [55, 58, 66].

On the last years, several new models of LFAM printers were introduced on the market. The website aniwwa.com reports 30 LFAM printer’s models with size higher than 1 m³. The prices from models with size higher than 1 m³, vary from 11 500 to up to $ 450,000 (for high-performance polymer). More than half of these models were released in 2021 and 2022. The complete list of LFAM printers can be found in Annex I.

For comparison, the most common process characteristics of small and LFAM methods are displayed in Table 2.

Felsch’s et al. [68] robotic arm system is an example that combines flexible and cost-effective industrial robots, with the authors making an evaluation of the manufacturing technology, the system architecture, and the motion planning. An articulating arm was also designed and implemented by Jasiel Minjares [69] into a BAAM machine.

Some authors approach to large extrusion systems starts by replacing the most common filament material by a pellet material feed [70, 71]. Most often, this modification will impose the need to incorporate a servo driven screw extruder, increasing machine complexity. Even so, when compared with filament extrusion, pellets solutions usually imply that material cost is decreased, and other problems such as filament runout is no longer an issue. In addition, pellet extrusion technology can increase the process rate, by more than 2 orders of magnitude, and decreases the electricity requirement per kg by about 2 orders of magnitude when compared to with filament extrusion technology [52]. A research by Zhiyuan Wang et al. shows that, for BAAM pellet extrusion, the process efficiency and deposition quality can be increased by using a variable pitch and progressive diameter screw [55].

Multiple and adaptable nozzles can be selectively used in different build areas of a part aiming at increasing surface quality of BAAM parts without significantly increasing print time. Chesser et al. explored the application of variable extrusion head diameter, through a poppet nozzle selector, achieving variable deposition rates in different regions of a part [64]. Using a proof-of-concept machine, Paritosh Mhatre developed an algorithm to control multiple nozzles mounted on the same print head for concurrent printing [65].

The manufacture of parts through multiple and collaborative extrusion modules rises as a potential solution to overcome the limitations of speed, cost, and scalability of common extrusion systems [80, 81]. The research team of Leite et al. termed the interdependencies between time, size and quality as the “3D printing of large parts trilemma” [73]. In collaborative 3D printing, individual mobile 3D printers or extrusion modules collaborate on building a part simultaneously [82]. Leite et al. [73] proposed the use of multiple collaborative deposition heads, and an experimental apparatus [74] was developed as a proof of concept. L. Poudel et al. devised an approach in which multiple print head-carrying mobile robots work cooperatively to print a desired part, which needs to be split into smaller sections that are then allocated to each robot [75]. This study also exposes the need and difficulties in generating efficient, collision free, printing paths, and suggests an innovative generative approach to tackle it. Wachsmuth [72] addresses various issues and concerns that arise while designing a multiple independent extrusion head systems and formulates suitable deposition strategies to reduce build time. Frutuoso [76] developed a slicing software to optimize part positioning and partitioning and proposing how to generate collision avoiding strategies. With the introduction of multiple extrusion heads, new bonding strategies are also needed to join the portions of the part that were produced by the different extrusion heads [72, 83], since the existence of independent extrusion modules imposes the existence of intersections areas in parts. Sardinha et al. [83] threated these intersections as seams, stating that the seams act in a part by creating a localized fragility, diminishing its tensile properties.

The work by Badarinath and Prabhu [77] proposes a modular solution based on retrofitting multiple robotic platforms to incorporate FFF extrusion heads. The authors approach includes a bed compensation algorithm based on bilinear interpolation and real-time synchronization of extrusion position, path, and velocity, with results having a less than 5% error regarding the extrusion widths and heights of manufactured samples. M. Layher et al. proposed a conceptual high-productive system composed of three extruders that use arbitrarily chosen plastics, also addressing model discretization into partial volumes using a specialized developed tool [67]. Similarly, S. Hongyao et al. proposed a large-format cooperative system composed of multiple robots, drafting considerations regarding the influence of the multi-robot layout on the maximum reachable area and geometry adaptability [78]. This work also proposes an optimized scheduling algorithm based on efficiency egalitarianism, and a robot interference avoidance strategy by dividing the printing layer into several safe areas and interference areas, with results showing that the print speed efficiency improvements with four robots exceeds 73% when compared with a general single extrusion method. Underlining the potential in multi extrusion, Zhang et al. developed a data-driven predictive model for the tensile strength of components fabricated by cooperative 3D printing [82]. In a complement to most common literature proposed solutions, the ORNL Manufacturing Demonstration Facility has created a system that works together with an existing BAAM system to ‘pick and place’ custom components into a part as it is printed [79]. Also mention worthy, Ni. Hopkins unconventional solution suggests that large-format material extrusion problems can be significantly reduced by using hollow tubes in place of solid extrusions, leading to a reduction in material usage and cooling related issues [59]. J. Shah et al. exposed some design consideration and challenges for large-format 3D printers, with a review regarding systems configurations and machine concepts [84]. Similarly, a review by A. Kampker et al. [23] describes machine design trends and highlights the strengths and weaknesses of existing designs.

Figure 1 displays some examples of systems configurations characteristics of LFAM printers.

With respect to thermal issues, most of the design constraints of small-scale polymer extrusion still apply to LFAM, and additionally, new limitations exist, because large-format systems significantly change the thermal properties associated with small-scale AM, and therefore, temperature profiles throughout large printed structures play an important role in determining its properties and manufacturing success rate [80, 87, 88]. The review of M. Pourali et al. on the thermal modeling of material extrusion AM provides a summary of approaches that have been applied mostly to small-scale polymer extrusion, which should be taken into consideration in BAAM [87].

Inter-layer welding is of special concern, as the large thermal gradients induce residual stresses that can overcome weak inter-layer welds, causing warping and cracking [60, 89]. Aiming at mitigating inter-layer inconsistency, R. Kurfess [89] developed a thermally driven design methodology focused on the interface temperature between layers, establishing a relationship between this temperature, flow rate and print velocity. To better understand the thermal expansion in large area extrusion parts, D. Hoskins et al. [90] compared the coefficient of thermal expansion of a finite element analysis that uses variations along the cross-section of a bead using thermomechanical analysis as input with strain maps using digital image correlation (DIC). K. Choo et al. [85] developed a model that aims at minimizing the risk of failure and slumping of overhanging features by predicting thermally driven fragilities and giving guidance in the decision process during the slicing of parts. Through this model, additional dwell times after each layer are introduced so that the next layer of printing initiates after the previous one is sufficiently cooled and, therefore, the structure is appropriately solidified. B. Friedrich [91] developed a transient thermal simulation model to aid in predicting failure behavior in BAAM, with results allowing to predict and adjust the model to avoid failure during the build process, with a 5% error achieved in analyzed used cases. Furthermore, researchers at ORNL investigated an in-situ control system comprised of a thermal camera and laser profilometer, which allows to control and adjust material flow and build speed to mitigate defects and uneven build surfaces [86]. F. Wang et al. [63] provided a framework to improve both quality and efficiency of LFAM by employing real-time data captured from an infrared thermal camera and, using a regression model to describe the surface temperature with high accuracy and then control the layer time, promoting high printing efficiency and quality.

To deal with common problems associated with large polymer extrusion such as temperature related issues described above, design methodologies that anticipate the manufacturing process workflow have proven to be highly efficient at improving manufacturing success rate [92]. A. Roschli et al. detailed significant physical and software-related design considerations for BAAM [80]. In a complementary research [93], the author suggests an overview of how slicing works, how to design for the slicing process, and how to configure slicing for the ideal output. M. Lee’s [94] case study of an excavator cabin is an example of how design for BAAM methodologies can improve the feasibility of large polymer parts. In another example, using a cantilever chair as case study, D. Albertini [88] extensively reviews and evaluates the influence of multiple variables on the mechanical behavior of large-format polymer parts, drafts challenges involved with this manufacturing field and derives an equation that aids to define the volumetric flow rate considering both tooling equipment and printing parameters. In line with these results, comparing two different methods, E. Fernández et al. [95] incorporated nozzle size and related constraints into topology optimization to generate suitable designs for large part resolution optimization. Dreifus et al. [96] discussed geometric restrictions of large-format polymer extrusion manufacturing, and by collecting data sets for qualitative analysis of BAAM, provides tools for designers and software developers to improve deposition strategies.

Increasingly complex systems impose a need of increasingly accurate control mechanisms. A current natural path in BAAM systems development is therefore to address manufacturing deviations by process control enhancements [55, 66, 70, 97]. Chesser et al. [66] proposed feedforward extruder control in BAAM to mitigate geometric inaccuracies and low-resolution problems. Silberglied [97] devised a bead characterization system that measures, and with feed forward control predicts and adapts the flow rate of a nozzle. Using finite element thermal modelling, D’Amico and A. Peterson [98] conclude that in BAAM, the temperature and thermal history of extruded beads has important implications for process monitoring, property prediction, and part performance. Using a closed loop control based in a framework for real-time data extraction from thermal images and a novel method for controlling layer time during the printing process, the research by Fathizadan et al. [99] was able to significantly reduce the overall printing time while preserving print quality. Borish et al. investigated the use of a laser profilometer and thermal camera to collect part data as it was being constructed, with results showing significant geometrical improvements [100]. Researchers at ORNL also investigated the use of a thermal camera mounted to a gantry to collect data on an object under construction, for feedback for an in-situ control system to adjust layer build times [101]. The research by Macdonald et al. [102] uses spatial frequency analysis during fabrication and uses this spectral information to infer on surface roughness, delamination, and deposition consistency, while the work by Roschli [70] controls the heat in a limited area in the proximities of the extrusion, allowing to produce consistent beads at variable speeds.

A research effort for improving interlaminar strength and process control of BAAM parts by Duty et al. [60] relates the structure of BAAM materials to the material composition, deposition parameters and resulting mechanical performance. Eyercioglu et al. [103] evaluated thin wall parts by investigating the cooling of single bead layers, and the interface and adjacent top layer temperatures. Kishore et al. studied the potential of using infrared heating for increasing the surface temperature of the printed layer just prior to deposition of new material, to improve the inter-layer strength of components [104, 105]. These studies found significant improvements in bond strength when the surface temperature of the substrate material was increased close to the glass transition temperature, using infrared heating.

In LSPED, due to the large size nature of the print beads, conventional test standards are usually inadequate. Schnittker [106] studied the feasibility of using DIC technology as a key enabler for robust data collection of strain measurements of large 3D printed parts. In complementary research, Schnittker states that DIC data correlated well with failure analysis performed on test coupons [107]. Using DIC, attempts have also been made to build a database of structural performance of large polymer AM parts [108].

The work by Kunc et al. [111] investigates the use of PEEK and its composites as potential feedstock for BAAM systems. Thermal and rheological properties were investigated and characterized to identify suitable processing conditions and material flow behavior. Using thermophysical and thermomechanical characterization techniques for fiber-filled ABS, literature reports suggest that Carbon Fiber (CF) reinforced ABS (ABS/CF) has the highest thermal stability to retain the shape at elevated temperature followed by Glass Fiber (GF) reinforced ABS (ABS/GF) and neat ABS [112]. Some studies have performed a rheological characterization of PEI, PPSU, PEKK, PPS, and ABS as well as their composites containing reinforcing fibers [113‐116].

A study by Ajinjeru et al. [117] characterized the dynamic rheological behavior of PEI and PEI/CF as BAAM feedstock material. The addition of CF to PEI enhances the shear thinning effect and significantly increasing viscosity, and the needed torque on the extruder. Fiber filled materials are used to counter-act the effects of thermal expansion, but the fibers stay in-plane meaning that no fibers span from layer to layer.

Roschli et al. [110] proposed a “z-pinning” approach, that is based in strategically positioning voids across multiple layers to be latter filled all at once. Duty et al. [118] suggested the same “z-pinning” technique for the reduction of anisotropy in AM polymer parts and improve their interlaminar strength, another possibility is to use a pos-tensioning structure [119].

Development of reinforced thermoplastic compounds that are specifically tailored to BAAM and possess the necessary properties for engineering structures is currently underway [54]. Sánchez et al. [120] developed and characterized CF filled Acrylonitrile Styrene Acrylate (ASA), with rheological, thermal, and mechanical properties of neat ASA and ASA containing 20 wt% CF to address LFAM. Regarding the parametrization of large-format AM techniques, using bio composites and multi-objective optimization, the work of Vijay et al. [57] demonstrated the ability to produce filaments of different shrinkage and tensile strength properties by solely changing process parameters and settings.

An alternative approach to fibre reinforced thermoplastics is the utilization of thermoset materials using an extrusion head that combines resin and the hardener during the deposition [45, 46, 121]. Large-format AM of reactive polymer systems offers significant improvements over thermoplastics such as potentially enhanced mechanical properties, faster deposition rates, and deposition at ambient temperatures. Kunc et al. suggested that the use of reactive polymers could reduce the thermally driven deformations and progressive buildup of residual stresses [122]. Unlike thermoplastics, reactive polymers cure during printing through a chemically or thermally initiated reaction and could potentially increase deposition toolpath flexibility and reduce the anisotropy of the printed components by reducing the dependence on temperature control of the process [45, 123], and yield printed parts with inter-layer covalent bonds that significantly improve the strength of the part along the build direction [46]. Lindahl et al. exemplified an AM system capable of depositing reactive polymers in a large format [45]. The work by Hershey et al. implements a design solution for bridging sparse infill patterns in additively manufactured parts while studying the thermal characterization of polymer composites and reactive polymers [124].

Integrating multiple materials into LFAM is a key for various industrial applications wishing to incorporate site-specific properties into geometrically complex designs, difficult to manufacture with traditional techniques. Brackett et al. [125] used a BAAM system to explore material transitions with a novel dual-hopper that enables in-situ material blending of a pelletized feedstock. The study characterizes a step-change transition between neat ABS and ABS/CF. Furthermore, the application of functionally graded materials and graded material properties transition could significantly advance the applicability of LFAM parts [126].

A list of commercial materials available for LFAM, including the most significant material properties reported by the manufacturers can be found on Annex II.

LFAM parts in aircraft interior and cabin parts are essentially applied in non-structural elements. China Eastern airlines explored the use of LFAM to produce spare newspaper holders, electronic flight bag supports [127], and door handle covers [12, 128]. These parts were produced with ULTEM™ 9085 resin, a high-performance thermoplastic material (PEI) with high strength-to-weight ratio compliant to relevant Federal Aviation Administration (FAA) and Civil Aviation Authority of China (CAAC) requirements. China Eastern airlines reports that lead times were reduced from months to days and costs reduced by more than 50%, using LFAM technologies. Lufthansa Technik redesigned and 3D printed a damaged air grill for a 747 Cockpit ventilation duct [129]. This new air grill is a certified part that benefits from better durability, reduced lead times, manufacturing, and maintenance costs. Spacer panels (which fills an end-gap in a row of overhead storage compartments) have been produced by a partnership between Airbus and Materialize for installation on board on the Finnair’s A320 aircraft [129]. The panels were produced with flame-retardant Airbus-approved materials and the panels shape has been optimized with a bionic design, to achieve a 15% weight reduction when compared to the original. In 2019, Diel Aviation announced the delivery of its largest fully 3D printed part, a Curtain Comfort Header, which was installed on a A350 XWB aircraft [130]. The Curtain Comfort Header is a complex enclosure for the curtain rail, that can have a length of up to 1150 mm. The curtains separate the classes from one another within the cabin. With the 3D printed Curtain Comfort Header, Diel Aviation states that achieved a reduction of tooling parts (complex aluminum tooling for composites were employed) and the advantages of the integration of cable channels, emergency escape route signage and specialized retaining clips on the 3D printed Curtain Comfort Header. Etihad Airways employed LFAM methods to implement a first proof of concept of a full-scale print of an Airbus A320 sidewall with print conductive tracks, antennas, and ornamental features [131]. In a partnership with BigRep company, the sidewalls were printed using 6-axis industrial robots with dual extrusion heads allowing to deposit conductive or capacitive materials.

Aerodynamic models are extremely useful to define mathematical models that can describe the aerodynamic forces and moments acting on the airframe of an aircraft during the flight. These models are tested in wind tunnels and are traditionally produced with computer numeric control (CNC) machined metal parts. 3D printed aerodynamic models are typically produced by stereolithography due to its excellent surface finish but are limited in size. The use of aerodynamic models produced by LFAM methods can be found for the development of an M-346 aircraft model [132] and for the F-35 prototyping of fuselage and wing skins by Lockheed Martin [133]. For the M-346 aircraft the test model was printed with PLA on a MakerBot Replicator Z18 3D printer. To obtain a good surface finish, after the printing process the model was polished with sandpaper to smooth the surface and painted applying an acrylic primer and topcoat [132]. Lockheed Martin reports costs savings of $65 K and time savings from 4 weeks to 5 days, by using LFAM methods when compared with conventional CNC machining. Lockheed Martin also saves time by using the LFAM printers to rapidly produce parts such as mockups for the cockpit floor of the F-22 aircraft, with cost savings of $86 K [133].

Surrogates are functional tools which are placeholders for assemblies, used on production floors and in training rooms. Bell helicopters uses LFAM to construct surrogates to assess a hybrid aircraft’s tail-wiring configurations, reporting a significant reduction in time (6-week reduction) and costs, when compared with the original cast aluminum parts [134]. These surrogates were produced to build polycarbonate wiring conduits that are placed on the vertical stabilizers for on-the-ground confirmation of the wiring path.

Several prototypes have been produced with LFAM to test a concept or part functionality in the aerospace sector. The use of LFAM for propulsion elements was first presented on an exploratory work by the National Aeronautics and Space Administration (NASA) in 2015, to produce a nonmetallic gas turbine engine. Polymer based components of the project included acoustic liners (Fig. 2a) and inlet guide vanes (Fig. 2b), produced with ABS, ULTEM and ULTEM/CF [135]. Taylor-Deal Automation is another company using LFAM for prototyping, through the production and modification of specialty fluid and air handling parts. Taylor’s 3D printing material of choice is ULTEM^® 9085, which meets FAA flame regulations. The 3D printed parts (toroid housings) contain less material, so their weight is approximately one-third of that of the metal parts they replace [134]. Marshall Aerospace and Defense presents a ducting adapter prototype for cooling avionics systems when the aircraft is on the ground. The ducting adapter prototype was made in Nylon 12 and achieved a significant cost reduction compared to machining the part out of aluminum, as well as a 63% reduction in overall weight [136].

Advanced Composite Structures produced layup tools for composite materials using AM, reducing costs and production time [134]. Piper Aircraft manufacturer uses LFAM to produce polycarbonate hydroforming tools as an alternative to the traditional CNC aluminum machining, taking advantage of the speed and waste reduction of the LFAM process [134]. Airbus in a collaboration with Autodesk developed a plastic mold for a partition wall with an optimized design. The partition wall was first projected for metal AM but now is defined by a plastic mold which is then cast in a flight-qualified metal alloy [129]. BAE Systems is using LFAM to produce high-temperature mold tooling utilizing a CF reinforced, PEI-based 3D printable resin, in collaboration with Airtech Advanced Materials Group and Ingersoll Machine Tools, Inc. who printed the mold tool on their MasterPrint large-format 3D printing platform. The mold tooling has supported over 250 autoclave cycles, without degradation [137]. CARACOL, an Italian company, has been producing large-format aerospace tools for the positioning and vacuum gripped drilling of airplane fuselage panels for aerostructure with a robotic extrusion system using polypropylene (PP), Polyamide 12 (PA12), polyphenylene sulfide (PPS) with CF and PPS/GF up to 40%. This innovative solution leads to a few benefits for clients such as: halving production times, eliminating the need for manual assembly, cutting material waste, and a significant cost advantage of around 30–50% [138]. Additive Engineering Solutions completed and shipped out a large-format 3D printed 12-foot-long NC vacuum mill holding fixture that will be used by Dassault Falcon Jet in the trimming and drilling of composite panels [139].

The use of AM in UAVs is currently a common practice, due to the less restrictive regulatory procedures that materials and processes are subjected to [134]. Aurora Flight Sciences and Stratasys embarked on an ambitious project: to build a jet-powered remotely piloted aircraft with a wingspan of 2.9 m. The UAV consists of 34 total components, 26 of which were 3D printed and make up about 80% of the aircraft airframe by weight. Using Stratasys Fortus® 3D Printers, the wings and fuselage were produced in ASA thermoplastic, to give the necessary strength and stiffness, but with low density [140].

The first application of LFAM technologies in rockets seams to date back to 2014 when the Lockheed & RedEye Team presented two rocket fuel tank prototypes, the bigger one being 2 m long. The tanks were made of a polycarbonate material using a Stratasys Fortus 900 mc FFF machine. Lockheed reports about half the cost, and a lot of save time by 3D printing the tanks in pieces and bonding them together, instead of machining them [141]. In 2016, the United Launch Alliance developed a duct system for the Environmental Control System (ECS) of the Atlas V and Delta IV rockets using LFAM technologies introducing ULTEM 9085 and the Stratasys Fortus 900 mc printer, slashing the system’s production cost by 57%, and reducing the ECS assembly from over 140 to 16 production parts [142]. In 2018, the startup Rocket Crafters made the first deployment trial with 3D printed rocket thrusters made of ABS, with a larger custom-made printer and signed a DARPA contract with a $600 K investment. After the hybrid fuel grains are printed, they are further processed by wrapping them in carbon for additional strengthening and then are test fired [143].

In 2013, Urbee was announced as the first “3D printed car”. Urbee is three wheels and two Seats small hybrid car. The 3D printing components of the car include the car body, bumper, and dashboard. These parts were printed with ABS on a LFAM system. The whole car (about 3 m long) takes about 2500 h to be produced [144]. Apart from the fuel drive cars, one segment in which LFAM technologies can be particularly efficient are in Electric Vehicles, due to his capability to provide polymer lightweight components.

Local Motors Strati’ Car and LM3D Swim are two examples that integrate LFAM technologies to produce parts for electric vehicles. The Strati car was presented in 2014 by a partnership between Local Motors and ORNL, producing a vehicle in days rather than months. This reduction is due to the use of pellets extrusion systems from ORNL and the produced parts include the chassis and skins made with carbon infused ABS [145]. The LM3D Swim was a production version of the Strati Car where roughly 75% of the car is 3D printed, including the body panels and chassis, using the same process and materials employed on the Strati Car [146]. In 2018, the YOYO electric vehicle from XEV received some media attention and is already in production today. The polymer fiber reinforced large parts are produced on a Big Rep PRO printer and for the final surface quality, the 3D printed parts also go through an automated robotic milling process to guarantee a smooth surface. XEV is betting on the customization of the vehicles design surface and body of the car [147]. The Project Chameleon from Scaled Ltd, a firm specialized in large-format 3D printing via a robotic extrusion process, is exploring the LFAM with the use of thermoplastics made from recycled plastics for the chassis of electric vehicles [148] Electric Vehicles truck manufacturers like Nikola Corporation are also exploring LFAM to fabricate fixtures, test models and final parts. Nikola Corporation installed a Big Rep PRO printer for continuous printing with engineering-grade filaments like polyamide 66 (PA66), ABS, ASA or fiber-filled thermoplastics [149]. BCN3D and Elisava Racing Team are introducing LFAM technologies in the electric DAYNA motorbike for producing fenders and fork covers. The Elisava team used BCN3D’s largest 3D printer (Epsilon W50) to fabricate the parts with high-temperature CF reinforced polyamide and PP/GF composite filaments [150].

ORNL developed parts prototypes for a Shelby Cobra car [151] and an electric utility vehicle [152] using the ORNL facilities and its polymer extrusion systems with ABS/CF. The prototyped parts include the chassis (Fig. 2c), supports and skins. For both vehicles the frame took approximately 12 h, the skins 8 h, and the supports 4 h to print. The chassis design incorporated threaded rods that could pass through the layers serving two purposes: providing attachment points for drivetrain components and putting the printed material under compression, thus improving the integrity of the chassis. Figure. 2d displays the completed electric vehicle prototype. Ford company is exploring the use of the LFAM to produce room-sized prototypes at Ford's research and innovation center with a Stratasys Infinite Build 3D printer. The printer uses an innovative printing set-up to print parts laterally instead of layer by layer, allowing to print large parts, theoretically with an infinite length. The printer uses a robotic arm to feed the cannisters with thermoplastic pellets and Ford is considering its use to print small-volume parts like spoilers, wings for race cars, molds, jigs and other production aids used in factories [153]. Also, in the prototype phase are non-Pneumatic (airless) tires from various manufacturers. In the recent work of Jafferson et Sharma [154], a review of the innovations in airless tires and the unique designs that are used are presented. The authors present a study considering a set of lattice structures potentially produced on the Big Rep—Thermoplastic Polyurethane (TPU) material fabrication platform.

Molds and tooling for the automotive industry are traditionally made with non-AM methods, but in the last decade, this paradigm has changed and today we can already find some examples of the use of LFAM technologies for molds and tools in this industry. Polimotor was the first company to develop an automotive engine with several polymer parts and today is working with Ford on a prototype composite engine. Polimotor in a partnership with ORNL studied the application of LFAM technologies in the production of an injection mold for an oil pan for use in a composite engine [155]. The mold was produced using a hybrid approach: the large bulk of the mold was printed with the ORNL BAAM system with carbon fiber-filled ABS and the detailed molding surfaces were printed on a Stratasys Fortus 900 out of ULTEM 9085. Another company that explores the LFAM technique to produce tools and molds for the automotive industry is DTG, which in cooperation with ORNL studied the production of a check fixture and a sand-casting mold [156]. The check fixture was printed using ABS/CF polymer. Sand-casting molds were produced using the same material, the larger section of the mold had a maximum size of 1.4 m and took 6 h to print. After printing, the casting pattern went through a machining process. On the work of Henderson et al. [157], cope and drag casting aluminium sand-casting molds for the automotive industry were also produced by LFAM technologies using a Stratasys Fortus 900mc system and ULTEM, with a nozzle diameter of 0.25 mm and layer height of 0.125 mm. The authors believe to have demonstrated the compatibility between printed patterns and traditional casting foundry practices and that both lead time and cost are reduced when compared with subtractive technologies.

Suntem 3D created a 3D printed architectural model of the Turning Torso building by architect Santiago Calatrava, using BCN3D Sigma machines. After 137 h of 3D printing, a physical PLA model of the building was ready. The mock-up was made in a scale of 1/135 and measured 1.40 m [167]. Big Rep is applying LFAM to implement architectural models using PLA on the Big Rep STUDIO printer for a Villa design. The total printing time for the 12 sections of the Villa design model is 120 h. To obtain a smooth painted finish, manually post-processing is done by sanding and painting the external surfaces. The full architectural model took 11 days to produce, a 50% reduction in both production time and labor costs [168].

In 2014, the project “3D Print Canal House” developed by the designing company Dus Architects used a biodegradable thermoplastic to demonstrate the construction of a building. Printed with a large 3D printed called Kamer Maker, the house components were directly produced on site, minimizing building waste and transport costs [169]. The Additive Manufacturing Integrated Energy (AMIE) demonstration project includes the fabrication of a house with LFAM, Fig. 3a. Approximately 80% of the house, including the segments, were 3D printed with ABS/CF using ORNL system with a deposition rate of 45 kg/h. The house has an area of 19.5 m² and was built to serve as an example of the capabilities of AM in the construction industry, producing an energy efficient building with less material waste [9, 170]. The Chinese company Qingdao Unique Products Development Co Ltd developed a large size 3D printer (12 × 12 × 12 m) to print an entire house with GF reinforced graphene [171].

One disruptive application of LFAM in the construction sector may be to produce window frames. A techno-economic viability study performed by Love et al. [172], showed that LFAM technologies can enhance the customization level of window frames, with cost and energy savings, when compared with the traditional fabrication methods. The structural testing proved to be viable in both material strength and energy efficacy.

Furniture and decorative items employing LFAM include chairs, tables, or decorative claddings. Chairs have been produced in PLA in cartesian 3D printers [88] or a robotic arm [173], Fig. 3b. A sample table with dimensions of almost 1 m was printed on the BAAM machine developed by the ORNL in 1 h and 43 min [69]. A construction company that has explored the benefits of LFAM is Skanska, through the production of unique cladding for the Bevis Marks Building in London, Fig. 3c. Polymer parts produced for the Bevis Marks Building had decorative purposes only, even so, they test the safety and reliability of AM components, and their use in large-format structural applications [9].

ORNL worked with Gate Precast to demonstrate the viability of using ABS/CF and BAAM technology by manufacturing molds for the precast concrete industry. The authors state that using BAAM, they are able to demonstrate a significant reduction in costs and that future research will attempt to employ materials, such as ABS/GF or PLA reinforced with wood flour, to bring the cost down even further [174].

Figure 3 displays some applications of LFAM in the architecture and construction sector.

One of the main applications of LFAM in the energy sector is wind turbines, that can be whether for the development of prototypes, wind turbine blades, diverters or nacelle covers.

In 2014, Abdelrahman and Johnson [176] tested the application of LFAM to produce a prototype for a wind turbine test rig with a modular blade (based on the S833 airfoil design) using 3D printed PC-ABS material. The blade measure 1.7 m with a section width of approximately 0.2 m, Fig. 4a. The authors state that the implemented experimental set-up with 3D printed blades sections was effective in measuring small changes in strain within the wind turbine blade. Beyond the prototyping phase, LFAM methods have been applied in applications of horizontal and vertical wind turbine blades. In 2018, the ORNL in collaboration with XZERES Wind fabricated a small (in comparison to traditional wind energy systems) horizontal wind turbine blade using LFAM methods [177]. The prototype blades with 0.9 m were produced with a 2.5 mm nozzle using ABS/CF on the ORNL BAAM system and mechanical tests showed that the failure of blades occurred high above (3.4 times) the maximum ultimate load determined in the standards. Vertical wind turbine diverters were also produced using LFAM technologies. ORNL worked with Hover Energy LLC on the design of LFAM components [178]. A design modification to fabricate vertical wind turbine diverters using LFAM was performed, and the diverters were analyzed based on anticipated wind loading. The 1/3 scale diverters models with 1.2 m in length were printed with a 2.5 mm nozzle on the ORNL BAAM system. The authors expect reductions in costs and lead times, when compared with the previously employed hand-layup method to produce the vertical wind turbine diverters. The use of AM to produce wind turbines allow to fabricate these parts with internal structures or variable infill patterns which facilitates the design for functionality of these components, as well as reducing their weight. In 2018, a mid-span blade of a wind turbine blade from XZERES Corp (442SR) with a length of 2.8 m was printed with a variable infill pattern, adapted to the estimated deflection loads by finite element analysis (FEA) that the blade faces during service [179]. The applied infill pattern has a functionally graded honeycomb with higher density of hexagons where high stresses are estimated. The mid-span blades were printed using ABS/CF and TPUs with a foaming agent at the ORNL BAAM system. The authors also implemented a new slicing method for the graded honeycomb infill pattern. The obtained results represent a potential reduction in fabrication time and weight of wind turbine blades produced by LFAM methods. Another example of wind turbine blades with internal structures is the work of Sanandiya et al. [180]. In this work, a wind turbine (1.2 m) with an internal structure (Fig. 4b) was printed using a robotic arm, with cellulosic materials. The cellulosic material has a density less than one half that of the lightest commercially available 3D printed filament (Polyamide, 0.95 g/cm3), resulting in a lightweight finished blade of 5.28 kg. Nacelles covers are components of wind turbines that can provide aerodynamic shell around the wind turbine components and a safety barrier. A nacelle cover was produced by ORNL on the AMIE structure [181], and details from this construction were already described on this paper (Sect. 2.3.2).

LFAM technologies are being studied as a potential process to produce parts for low-head hydropower systems. ORNL worked with AMJET Turbine Systems to demonstrate LFAM of components for low-head hydro turbine/generator parts, namely: stator vane body, stator outer, bell plug and draft tube [182]. ORNL tested several LFAM systems that include the Stratasys Fortus 900mc printer, the Cosine AM1 printer and the BAAM ORNL system. The authors find some limitations of the BAAM to produce parts, since components did not adequately perform due to persistent leaks caused by porosity and suggest the vacuum infusion of resin of parts to solve this problem. Another example of application of LFAM technologies in low-head hydropower results from a partnership between ORNL and Cadens to produce low-cost parts [183]. The fabricated parts on the BAAM ORNL system include a draft tube, a thimble, and a runner housing layup mold. The project provided an opportunity for Cadens to scale up their modular AM hydropower parts using the capabilities of BAAM.

LFAM printed magnets are an emerging technology in the energy sector in direct drive wind turbines, since these systems require large permanent magnets. Direct drive wind turbines have some advantages when compared to gearbox wind turbines since the gearbox is the heaviest and the most maintenance costly part of a turbine. LFAM printed magnets were produced by BAAM in the ORNL using NdFeB [184] or NdFeB + SmFeN [185] powders on a Nylon 12 matrix. The prototyped magnets were a cylinder (15 cm length), and horse-shoe shaped magnet (5 cm square), Fig. 4c.

LFAM technologies are already a reality regarding its application to molds production for the energy industry, namely in wind turbines blades. The main use of LFAM is for the implementation of vacuum assisted resin transfer mold, for wind turbines blades [186, 187]. ORNL demonstrated this by building 13 m long windmill blades. The mold was produced with ABS/CF in multiple 1.8 m sections, that were aligned and joined together. The molds were coated with 8 mm of fiberglass that was machined to the final surface geometry and finish. To achieve a uniform temperature on the surface of the mold, optimized air channels were printed, to allow heating of the tools via forced heated air. These channels create time and cost savings by eliminating the step of making the traditional resistive heating system.

Figure 4 displays some applications of LFAM in the energy sector.

Table 4 summarizes the applications of LFAM technologies in the different activity sectors that were mentioned above.