* BUILDING CONSTRUCTIONS: AROUND MATERIALS, LABOR, AND THE ENVIRONMENT
* CURIOUS CONSTRUCTIONS
*° SCALABLE TACTICS
*° ROBOTIC PLASTIC PRINTING
*° SEMI-MONOCOQUE PLYWOOD STRUCTURES
*° SYNCLASTIC TENSILE STONE
*° FIBERGLASS WRAPPED STRUCTURAL NODES
*° TENSIONED TEXTILE COMPOSITES
*° THERMOFORMED BIOCOMPOSITES
DCS °Lenticular Column, Closed Loop, & Teeter Dock
DATE2021 - Present
LOCATIONIthaca, NY
TYPEMaterial & Fabrication Research
FUNDINGMui Ho Center for Cities Grant and Cornell Council for the Arts
PRIMARY INVESTIGATORDavid Costanza
RESEARCH TEAMZekai Lin
Riley Wimes
Jonahan Wilmers
OVERVIEWRecycled Robotic Plastic 3D printing explores the structural and architectural potential of recycled plastics and renewable bioplastics, through robotic fabrication in Architecture. It reframes plastics from disposable to essential building materials—durable, waterproof, affordable, and potentially sustainable. Through robotic 3D printing and full-scale prototypes, the work demonstrates a new fabrication workflow and promotes environmentally conscious construction practices. The research has been used for the production of three projects to date, the Lenticular column, Closed Loop, and Teeter Dock.
KEYWORDSRecycling, Plastic, Robotics, 3D Printing, Robotic Additive Manufacturing (RAM)
INTRODUCTIONDigital fabrication technologies enable the manufacturing of plastic components in unconventional ways. This research aims to redefine plastic, positioning it not as a short-lived disposable material but as a critical material for long-term use. The study builds upon these observations and the recent advancements in 3D printing, seeking to scale up the work by incorporating industrial robotics and offering greater control over geometrical precision.
In this project, additive manufacturing significantly reduces the conventional blow molding method and avoids the waste of disposable molds [1]. FDM (Fused Deposition Modeling) is a widely used additive manufacturing technique that constructs three-dimensional objects by stacking melted and extruded plastics. However, when a polymer is heated and extruded, its mechanical properties degrade [2]. One option to reduce the number of melt/extrude cycles for recycled plastics used in 3D printing is to print directly using recycled plastic in the form of pellets, flakes, reclaimed materials, or fragments, known as fused particle fabrication (FPF). This study explores the feasibility of FPF printing construction-grade builds using recycled polymer particles of non-standard shapes and sizes. The bench serves as a platform to explore and develop the production workflow while testing the structural capabilities of individual components.
Throughout the project, each iteration and prototype undergoes analysis and evaluation, with the findings informing the design and revision of future prototypes. Evaluation methods rely on empirical evidence, including material consumption, labor involvement, process replicability, and the ability to produce alternative forms using the same process. Additionally, the prototypes are 3D scanned to compare the intended geometry with the resulting prints, serving as a digital catalog of various iterations. Finally, the prototypes undergo structural performance testing, evaluating their compressive strength through axial loading and assessing the sections for modulus of elasticity in bending and flexural stress using a three-point flexural test.
GENERAL PRINCIPLESTo explore recycled plastic printing, the research team chose recycled PLA pellet plastic, a biobased polymer, which significantly reduces its carbon footprint compared to petroleum-based plastics. This material can withstand a tensile modulus of 3500 MPa and a tensile strength of 45 MPa. For large-scale additive manufacturing, this research uses a medium-sized 6-axis robot arm from ABB and a screw pump extruder (developed by Massive Dimension). Compared to general Cartesian printers, the 6-axis movement range of ABB 4600 provides greater printing angles and volumes while continuously keeping the extruder TCP perpendicular to the printing plane, achieving greater accuracy [3].
Considering the final volume of the object to be printed, employing non-planar paths for additive manufacturing is crucial. Non-planar paths maintain an almost constant deposition of material unaffected by overhang angles, resulting in angled parts having higher strength than vertical ones [4]. Non-planar paths allow printing without auxiliary support structures, which can reduce surface accuracy, material wastage, and post-processing challenges [5]. This significantly increases printing efficiency and reduces process complexity.
The combination of non-planar trajectories and multi-axis movements genuinely expands the capabilities of large-scale FDM printing. When printing large objects using a large-sized nozzle, a single-wall line is often sufficient to achieve the desired structural performance. Given the limited retraction capability of the screw extruder, vase-mode printing becomes an apt choice. These techniques can achieve consistent wall thickness, and overhangs can reach 75° [6].
ASYMMETRIC OBJECT PRINTING TEST PARAMETERSMultiple tests were conducted at different scales to ensure the success rate of additive printing, and the acquired information was used to refine the design of the printing path. The research team printed a series of items composed of multi-planar paths. Starting with an average TCP speed of 25 mm/s, an extrusion rate of 0.45 cm^3/s, and an extrusion head diameter of 2mm, the tests began with a tool path interval of 1 cm and then decreased to 4 mm. The average printing thickness was 2.3 mm, and the extrusion line width was approximately 4.5 mm. The extruded material from adjacent paths entirely overlapped, and there were no issues with excessive extrusion. In the tests, we analyzed the physical behavior of the material extruded on non-horizontal surfaces. At positions exceeding a thickness of 2.6 mm, even when reducing the TCP speed below the average, the material would exhibit self-diffusion, resulting in the extrusion thickness not reaching the theoretical value. This may be due to insufficient material viscosity [7]. Within a range of 2.5 mm, the extruded material can maintain its geometric shape.
PRINT LAYER THICKNESS CONTROLThe volume of objects with non-planar trajectories varies unevenly across different positions on the same plane. To achieve the variable print layer thickness required for non-planar paths, we developed a method using Grasshopper to calculate the extruder head's movement speed based on the toolpath's relative height to the working plane and output it through G-code. By controlling the TCP speed of the extruder head, the extruded geometry can maintain the extrusion width while reducing the extrusion thickness on the same plane, avoiding excess extrusion or insufficient foundation due to changes in printing angle.
BASE LAYER STRUCTURE TESTSThe research team printed a series of cavities to determine the degree of deformation after plastic shrinkage. The results indicated that, without a base layer structure, the object significantly bent at the corners once printed to a certain height. Moreover, the object required a particular base layer area to generate adequate friction with the heated bed to prevent it from collapsing under its weight. Printing tests were conducted to limit deformation and enhance rigidity to determine the optimal infill structure. Considering the weak retraction capabilities of the screw pump extruder head, a concentric circle pattern was chosen as the base structure, complemented by an additional sub-layer to reinforce stiffness.
PLASTIC RECYCLINGFused Deposition Modeling (FDM), commonly known as Fused Filament Fabrication (FFF), is a widely accessible additive manufacturing technique due to its cost-effectiveness; consumer-grade FDM 3D printers are available for as low as $100. However, as this technology gains popularity, it also leads to a surge in plastic waste. In particular, intricate shapes and failed printing processes often require support structures made of plastic, which contributes to waste. Polylactic Acid (PLA) is the most commonly used filament in FDM 3D printing. Without effective recycling procedures, most PLA is discarded instead of being reused. Several factors hinder the integration of 3D waste into the recycling sector. The material and quality control of 3D waste lacks standardization, societal acceptance of recycled filaments needs to be higher, and regulations like REACH (UK registration, evaluation, authorization, and restriction of chemicals) pose barriers to their integration [8].
Consequently, 3D waste faces challenges in the recycling industry. Organizations and individuals without adequate recycling resources often discard 3D waste directly into landfills. Due to contamination, this waste remains non-biodegradable and contributes to the ongoing waste generation. Only a few organizations with significant waste production or companies specializing in recycled plastic filaments invest in recycling machines to convert waste into new granules.
To advance the work, the research lab acquired an industrial grinder to grind plastic parts into usable shards for the printing workflow. The shards were sifted to remove particles too large or too small that might otherwise clog the extruder feeder. Introducing the industrial grinder to the workflow allows past prints, failed prints, and prototypes to be ground and reused.
STRUCTURAL OPTIMIZATION AND WORKFLOW STRATEGY
PRINTING PATHS, G-CODE, AND RAPID CODETo generate non-planar 3D tool paths and G-code, we developed a specialized Grasshopper script that calibrates the tool paths according to reciprocal parameters: the specific volume, speed, and range of layer thickness, with the drilled 2 mm nozzle diameter acting as a fixed value. The basic structure of the toolpath is akin to the vase mode, using a single-line wall as its foundational structure and creating reinforcement and mechanical structures at specific locations based on the input parameters. ABB's Robot Studio program and the 3DP PowerPac plugin are utilized to simulate motion trajectories and generate the RAPID code required to operate the 6-axis robot, as well as the simulation signals necessary for operating the screw pump.
STRUCTURAL DESIGNA closed-loop toolpath was designed and optimized specifically for the characteristics of the screw pump extruder, ensuring uninterrupted path changes between each surface and achieving structural variations according to function. All components use a rotating triangular section to provide structural strength, and a stepped structure is employed to implement section changes. Critical structural points of the components utilize multiple overlapping paths to achieve higher material density, ensuring that the couplings can withstand the shear forces during use while guaranteeing that the paths can self-close. The tension direction of each component is based on a central point; the compressive force brought by the post-tension strap maintains the circular balance and adds strength to the component joints.
RESULTS
30% SCALE COMPONENT TESTA succession of small-volume tests was conducted to optimize the tool paths. Challenges arose during testing at 30% volume due to excessive material extrusion in areas with high density adjacent to a multi-tool path. These issues occurred when the printing height exceeded a certain threshold. The excessive extrusion resulted in thermal accumulation, leading to volumetric expansion of the printed object.
50% SCALE COMPONENT TESTAfter increasing the test scale, the team printed several arcs at 50% volume. In this round of testing, the print speed was optimized. By incrementally increasing the speed of the extruder head, we calibrated the object's volume with the tool path dimensions, which had previously led to printing failures.
100% SCALE COMPONENT TESTThe final prototype was informed through iterative and heuristic tests. The nozzle diameter, print speed, variable layer thickness, and path thickness were all optimized for rapid printing or large-scale structural parts. Shifting to full scale also brought about new concerns to address. The full-scale prints took around 18 hours to complete. During that time, the heated build plate would cycle on and off, causing the aluminum ½” plate to expand and contract, impacting the accuracy and resolution of the parts, adjusting the center of gravity of the print bed, and turning off the heating element after the first few layers guaranteed proper adhesion to the oblique build-plate while limiting the expansion and contraction.
FINAL COMPOSED COMPONENTSThe final structure comprises six self-similar components assembled using spring pins for registration and to resist rotation, and a post-tension ring that compresses the parts perpendicular to the axis of printing. The design ensures that the continuously changing cross-sections within the structure concentrate the compressive force of the tension rings at a single central point, thereby maintaining their stability. Consequently, three stable contact points with the ground are formed. The parts lift off the ground between the three support points and act as a beam. The support points are custom aluminum brackets formed to accept the geometry of the bench but without any mechanical connection to the plastic.
FUTURE INVESTIGATIONSTraditional fabrication processes for liquid materials like plastics require the construction of molds, which can be limiting. However, the developed workflow that integrates 3D printing and robotics provides an alternative method that reduces labor and eliminates the need for additional materials in mold production. Furthermore, the digital fabrication workflow offers the potential for mass customization, deviating from the economies of scale typically associated with plastic manufacturing.
Future investigations aim to build on the developed workflow to construct three full-scale architectural components: two axially loaded columns designed to resist buckling and compression and one beam subjected to perpendicular loading, intended to withstand shear, bending, and moment of inertia.
Each architectural element serves as a platform for exploration. The two columns examine different approaches to reimagining plastics in architecture: recycled synthetic plastics, which are non-renewable but can be recycled, and renewable bioplastics derived from natural biomass sources like corn. As a spanning member under tension, the beam investigates the use of natural fiber-reinforced plastics. Renewable and low-density fibers from plants such as flax, hemp, and jute enhance the plastic's tensile resistance when embedded in a plastic matrix [9].
Compared to other plastic manufacturing techniques used for large-scale components, such as injection molding or rotational molding, one of the drawbacks of plastic 3D printing is that it can have insufficient mechanical properties caused by layer-to-layer adhesion [10]. Objects printed using this method can fail if the layer adhesion lacks proper mechanical bonding. For architectural constructions, it's imperative to have uniform and consistent strength in all directions. Future investigations will also seek to exploit the flexibility afforded by robotic 3D printing to introduce multi-directional prints, allowing for quasi-isotropic properties within a part rather than the directionality inherent to FDM, FFF, and FPF techniques.
The work will continue to build on colleagues' research exploring robotic additive manufacturing (RAM). One such method uses various composite materials, including Carbon Fiber Reinforced Plastic (CFRP) [11]. CFRP's superior material properties can assist FDM structures in withstanding shear forces along the Z-axis while maintaining a lightweight and high degree of freedom for the entire structure.
In addition to the future material and construction process investigations, the research team plans to further the research through the computational logic of geometry derivation. Topology optimization algorithms present a promising supplementary approach when combined and informed by the material and robotic workflow limits. This method uses mathematical techniques to find the best structural form for the constraints and loads [12]. Altering the internal structure of an object can achieve the required mechanical properties while minimizing the amount of material used. Combined with the high degree of freedom that robotic 3D printing offers regarding tool path direction, the material can be extruded in multiple directions to counteract forces from all sides.
CONCLUSIONThe 3D-printed bench offered a valuable scale to develop the workflow and demonstrate the potential of recycled plastic as a viable building material and robotic 3D printing as a construction technique. Closed Loop also helps to challenge pre-existing notions about the value of plastics in construction as a valuable building material. Integrating robotics and computational tools has provided insights into the potential advantages and challenges of scaling FDM, FFF, and FPF printing to address an architectural scale.
Future work is focused on enhancing the fabrication process's broader scalability and accuracy and expanding the use of recycled and fiber-reinforced plastic in multi-axis additive robot manufacturing. Certain limitations became evident, particularly regarding the consistent mechanical performance of 3D-printed structures for broader architectural applications. The envisioned directions for future research, involving using materials like CFRP and leveraging topology optimization algorithms, are just a few potential pathways to refine the process further.
Closed Loop, as a research investigation, offers a promising workflow that expands on existing and ongoing work around 3D printing in architecture. The project underscores the need for continued exploration, collaboration, and refinement to realize the full potential of recycled plastic robotic 3D-printed parts.
REFERENCES
[1] Bak, D. 2003. Rapid Prototyping or Rapid Production? 3D Printing Processes Move Industry Towards the Latter. Assembly Automation 23(4): 340–345.
[2] Oblak, P. & Gonzalez-Gutierrez, J. & Zupančič, B. & Aulova, A. & Emri, I. 2015. Processability and Mechanical Properties of Extensively Recycled High Density Polyethylene. Polymer Degradation and Stability 114: 133–145.
[3] Yang, Y. & Fuh, J.Y.H. & Loh, H.T. & Wong, Y.S. 2003. Multi-Orientational Deposition to Minimize Support in the Layered Manufacturing Process. Journal of Manufacturing Systems 22: 116–129.
[4] Dizon, J. R. C. & A. H. Espera & Q. Chen & R. C. Advincula. 2018. "Mechanical Characterization of 3D-Printed Polymers." Additive Manufacturing 20: 44–67.
[5] Zhao, H. & He, Y. & Fu, J. & Qiu, J. 2018. Inclined Layer Printing for Fused Deposition Modeling With out Assisted Supporting Structure. Robotics and Computer-Integrated Manufacturing 51: 1–13.
[6] Krčma, M. & Paloušek, D. 2022. Comparison of the Effects of Multiaxis Printing Strategies on Large-Scale 3D Printed Surface Quality, Accuracy, and Strength. The International Journal of Advanced Manufacturing Technology 119: 7109–7120.
[7] Darweesh, B. & Gutierrez, M.P. & Schleicher, S. 2023. Non-Planar Granular 3D Printing. Construction Robotics 7: 291-306.
[8] Peeters, B. & Kiratli, N. & Semeijn, J. 2019. A Barrier Analysis for Distributed Recycling of 3D Printing Waste: Taking the Maker Movement Perspective. Journal of Cleaner Production 241: 118313.
[9] Fahim, I. & El-Haggar, S.M. 2015. Innovation in the Plastic Industry 2: Upcycle of Plastic Waste: Natural Fiber–Reinforced Plastics. Sustainability and Innovation: The Next Global Industrial Revolution: 143–170.
[10] Peters, B. 2016. Solar Bytes Pavilion. Robotic Fabrication in Architecture, Art and Design 2016: 327–337.
[11] Kwon, H. & Eichenhofer, M. & Kyttas, T. & Dillenburger, B. 2019. Digital Composites: Robotic 3D Printing of Continuous Carbon Fiber-Reinforced Plastics for Functionally-Graded Building Components. Robotic Fabrication in Architecture, Art and Design 2018: 363-376
[12] Al-Ketan, O. & Rowshan, R. & Al-Rub, R.K.A. 2018. Topology-Mechanical Property Relationship of 3D Printed Strut, Skeletal, and Sheet Based Periodic Metallic Cellular Materials. Additive Manufacturing 19: 167–183.