Developing composite wood for 3d printing

P. Janssen, P. Loh, A. Raonic, M. A. Schnabel (eds.), Protocols, Flows and Glitches, Proceedings of the 22nd International Conference of the Association for Computer-Aided Architectural Design Research in Asia (CAADRIA) 2017, Paper 41 (Preprint). Š 2017, The Association for Computer-Aided Architectural Design Research in Asia (CAADRIA), Hong Kong.

DEVELOPING COMPOSITE WOOD FOR 3D-PRINTING

RACHEL TAN1 , CHIN KIAT SIA2 , YONG KIAT TEE3 , KENDALL KOH4 and STYLIANOS DRITSAS5 1,2,3,4,5 Singapore University of Technology and Design 1,2,3,4 {rachel_tan|chinkiat_sia|yongkiat_tee|kendall_koh}@mymail.sutd.edu.sg 5 dritsas@sutd.edu.sg

Abstract. We present the initial findings of our research project aiming at development of a 3D printing process for wood composites. The 3D printing method employed is based on material extrusion principle and utilizes industrial robotics for position and motion control. The unique characteristic of our approach is in the development of the material where we employ exclusively organic components for both the matrix and reinforcement; a decision informed by prioritizing environmental considerations.

Digital Fabrication; Additive Manufacturing; 3D Printing; Wood Composites; Robotics.

Keywords.

1. Introduction The current focus on 3D printing in design research centres about its two major challenges: scale and the material composition. Our project investigates the potential for a novel type of additive manufacturing process based on renewable materials. The majority of available engineering materials for 3D printing are based on petroleum products such as thermoset, thermoplastic polymers and composites. There are also commercially available technologies for printing ceramic materials as well as metal alloys including steel, aluminum and titanium (Materialise, 2016) . In recent years there have been research developments in investigating materials such as concrete (Khoshnevis, 2004 ), sand (Gramazio and Kohler, 2011), clay (Sun et al, 2014), wax (Gardinger and Janssen, 2014) for the architecture, engineering and construction industry.

3D printing is also being extensively researched for biomaterials suitable for medical applications (Gillemot et al, 2010). Our investigation is inspired by developments in organic biocompatible materials.

2

R. TAN ET AL.

We explore the spatial deposition of a particulate composite material comprised of wood fibre and an organic matrix originating from animal byproducts. Both components are of organic source, renewable and biodegradable. There exist methods for 3D printing wood composites using preimpregnated thermoplastic polymer matrices such as Polylactic Acid (PLA) in particulate or filament form using either extrusion or sintering methods (3DSystems, 2011; Krassenstein, 2015). Compositionally those materials are not very different from engineered timber laminates, wood plastics in general, such as plywood, cross laminated timber and particulate composites such as fibreboards. Those generally suffer in recyclability because the separation of the constituent components is extremely difficult (Frank and Roffael, 1998; Pizzi, 2006) . A driving factor behind this problem is the challenge of binding (Lu et al, 1998; Clemons, 2002) cellulose, the primary organic component of wood and the most abundant on earth, with other materials thus requiring matrices such as phenolic and formaldehyde resins. We approached the development of our material process by prioritizing its environmental characteristics, its potential for being used for additive manufacturing processes and then study its mechanical properties and potential applications.

2. Project Overview The project is comprised of multiple research tasks: Material composition design and analysis, where we investigated various matrices from organic origin, selected a bonding agent which has been traditionally used in the fabrication of musical instruments and art restoration, performed testing for material mixing ratios, shear strength, temperature control, flow rate and curing time; Design and development of the electromechanical tools for 3D printing based on extrusion, integrated the end-effector with an industrial robotic system for motion control; and Investigated the special characteristics of fabricating extruded wood composite materials in terms of geometric limitations, machine motion control and instructions via computation.

Figure 1. Overview of Extrusion Process

3. Material Properties Timber has been used as a naturally sourced, renewable, low embodied energy (Falk, 2010) material with rich history in architecture, engineering and construction (Deplazes, 2005; Ross et al, 2009). In its natural solid state, it has anisotropic properties derived from the composition of the grain, which allows

DEVELOPING COMPOSITE WOOD FOR 3D-PRINTING

3

it to be both structurally sound yet unpredictable at the same time (Wiedenhoeft, 2010). The compressive and tensile strength of wood lies in the organization and build-up of fibres, composed of lignin and long chains of cellulose (Kretchmann, 2010). These properties allow timber structures to be light and strong. We are able to harness similar qualities while manipulating the material density and mechanical properties by breaking down wood into fibres or particles. Plastic wood composites is the broad family of engineered materials comprised of short timber fibres or wood particles bonded with a thermoplastic or thermoset matrix (Frihart and Hunt, 2010). The main objective of those materials is to overcome the deficiencies of natural wood towards mechanical predictability and weathering (Stark et al, 2010). The environmental impact of those materials is dominated by the characteristics of the matrix in which thermosets perform worst, followed by thermoplastics, followed by bioplastics (Schwarzkopf and Burnard, 2016).

The intent for our material design was guided by the desire to achieve 3D printed wood, a material that is traditionally formed through subtractive fabrication processes, and to explore the potential of an environmentally more benign product than polymer wood plastics. The material we are developing is a particulate composite comprised of saw dust filler, an industrial by-product of timber processing also used in particle boards. Using wood particles, though mechanically inferior to short fibres, was guided by the need for consistent flow in extrusion process. We experimented with various bonding agents and focused on the organic adhesives which are also by-products of industrial processes. The objective of the first phase of the material study was to determine the appropriate composition of material components for controlling fabrication such as mixing volume consistency, phase separation, viscosity, temperature, flow rate, curing time and mold growth before we even begin understanding mechanical characteristics which is planned for the second phase of the project.

We conducted tests with various natural binders exhibiting gelatinous properties to determine the most suitable sample. Starch is a polysaccharide derived from plants with adhesive properties which can be controlled by water solution and has been used for 3D printing (Lam et al, 2002). Rosin gum resin is a natural low molecular weight polymer extracted from pine trees with thermoplastic properties. It is used as adhesive, food additive and generic tackifying agent for inks, rubbers and fluxes (Sivestre and Gandini, 2011). Fish glue, derived from the skin and bones of fish is an effective binder for paper but is also applied to wood musical instruments. Hide glue, obtained from animal hide, is a thermoplastic binder that is used for woodworking and in musical instruments.

Polyvinyl acetate, (PVA) also known as white or carpenter’s glue, is biodegradable despite being derived from petrochemical processes.(Chiellini et al, 2003) It is also effective in binding porous natural materials such as paper and wood.

Preliminary testing with those adhesive products using an Instron universal testing machine showed that even though PVA glue exhibited the highest shear strength out of all others products, both hide and fish glue were close.

Starch was substantially weaker and rosin gum test samples failed completely perhaps due to wrong application. Nevertheless, the ability to achieve encour-

4

R. TAN ET AL.

aging results with animal processing adhesive by-products motivated further exploration towards properties related to the intended use of extrusions for 3D printing.

Figure 2. Testing Bond Strength of Glues using Instron Universal Testing Machine

Viscosity is a rheological property examined for conveying and depositing materials but it is very difficult and expensive to measure accurately (Brummer, 2006). As the viscosity of our materials are higher than fluids like paints and oils, we devised a method similar to the Zahn Cup viscosity measurement which is also indicative of the flowrate of the material passing through a hole at the bottom of the cup (Patton, 1979; ASTM D4212, 2005). Hence we designed a setup by which a finite amount of material is extruded via a syringe with a constant weight applied. While this approach is rather bespoke it allowed us to replicate tests and pin point range of acceptable compositions.

Figure 3. Properties of Different Glues and Glue Mixtures

For each matrix we qualitatively evaluated its ability combine with wood dust to determine the maximum amount added while maintaining similar viscosities. Not all the binders combined well with wood sawdust. Due to high percentage of water in PVA and starch glue the mixture did not evenly mix and separated easily from the wood, leaving a stiff and un-extrudable compound.

DEVELOPING COMPOSITE WOOD FOR 3D-PRINTING

5

Rosin gum was difficult to emulsify with sawdust even with the use of solvents.

Fish glue had increasing difficulty of mixing with sawdust and its working time was limited as upon contact it became extremely tacky. Hide glue mixed with sawdust had reasonable working time before it completely cooled and hardened. Overall, hide glue had the most favourable characteristics particularly for its ability to harden by temperature control. This is an advantage for 3D printing as it is desirable for beads to solidify as fast as possible in order to allow stacking of layers.

The consistency and flow of the wood composite could be adjusted with addition of water, as hide-glue is water soluble; as such we continued to fine-tune the constituent mix ratios. Mixtures became smoother, uniform, less viscous and easier to extrude with higher proportions of binder and water but slower to cure and more difficult to stack. Mixtures with higher proportions of sawdust had rougher texture, were more difficult to extrude but also deformed more by curling as the material hardened.

Water loss via evaporation from the composite caused shrinkage in the samples. To determine the amount of shrinkage, we set up tests, measured and computed the differences in size of the samples, as defined by the average of two lengths of the square samples. We prepared samples of 2mm, 4mm and 6mm thicknesses to verify if the effects of exposed surface area affected the rate of shrinkage. In general, the samples shrunk and stabilised at 13.7% after a period of 10 days. We observed that 2mm samples decreased in size at a much faster rate than the thicker samples, due to warping. However, upon applying heat via a heat gun and some pressure to un-warp the samples (on day 10), we noticed that the samples had only shrunk to an average of 11.2%, with a much smaller deviation range.

Figure 4. Shrinkage Results

Visually, high matrix compositions failed to yield a wood-like qualities and took up the appearance of the binder instead. At higher proportions of wood sawdust, the mixture had more familiar aesthetic due to exposed grain. Unlike natural wood the samples exhibited waxy sheen and emitted a characteristic for hide-glue but not unpleasant odor that becomes less noticeable as the material cures. In order to experiment with the visual properties of a large scale product, as well as to test the machinability and finishing quality of the cured

6

R. TAN ET AL.

mixture, we cast a stool prototype using the sawdust-hide glue compound. The prototype also allowed testing of the extent to which a large volume-mixture would warp during curing.

Figure 5. Cast Prototype of Stool Product to Evaluate Visual Properties and Machinability

4. Tool Building Given the high viscosity of the material and large pressure required to drive it through a small diameter nozzle, we adopted the auger screw design. An auger or Archimedean screw mechanism moves material forward using a captive rotating helix within a cylinder transporting materials from liquids to solids such as grains. As the wood composite material is much more viscous than water, there was a need to create an airtight system to ensure continuous positive displacement and prevent backflow against gravity and internal pressure built up.

The additional benefit of the auger screw design in comparison to reciprocating pump with a solenoid flow control system is that it ensures pulsation free bead extrusion with capability of flow reversal.

Figure 6. Tool Iterations from Hand Tools to Electromechanical System

We approached tool building starting from tools operated by hand and studying their effectiveness before moving towards an electromechanical system. The early tool prototypes were used for studying the process of extru-

DEVELOPING COMPOSITE WOOD FOR 3D-PRINTING

7

sion and finding a suitable material consistency. We used manual extrusion methods, employing squeeze bottles, syringes and simple augers. Next we developed iterations of the auger system that mounted onto the 6-axis robotic arm. The drive system is based on a lightweight, high gear ratio servo motor controlled via serial communications using an Arduino microcontroller. The mechanical parts were prototyped by 3D-printing. The total weight of the end effector was kept at minimum to allow more material to be loaded without interfering with the payload limitations of the robot. In addition, the effector was designed such that it is easily disassembled for cleaning. An injection port was also incorporated into the final design for externally refilling the system with material. The plastic material of the 3D-printed parts was also beneficial as a thermal insulator for retaining material temperature about 50°C where the material exhibits good extrudability.

Figure 7. Tool Assembly and Robotic Setup

5. Robotic Fabrication We integrated the extruder in a Universal 6-axis robot system with 10Kg payload. This machine was chosen as a prototyping tool, because it is essentially a flexible CNC machine that can provide the positional control required for three-dimensional material extrusion. For our experiments, we restricted of the machine to 3-axis movement in order to simplify the extrusion process.

The nozzle TCP follows a sequence of coordinate systems generated from machine paths created in Rhinoceros/Grasshopper, converted to machine code and transmitted to the robot controller using the Jeneratiff digital design and fabrication library. Additional components for motion control were created using the C-sharp programming language. The designs tested follow the conventional CAD/CAM for CNC and 3D printing approach converting geometry

8

R. TAN ET AL.

to machine paths and the machine code in principle. However, as many parameters are under investigation, we approached design generation backwards from motion to form.

Figure 8. Motion planning and Translation to Robot Code using Jeneratiff on Grasshopper

As such we tested rather simple planar shapes for calibrating robot motion and material flow speeds and spiral geometries for evaluating the vertical material behaviour. For accessing the overhang behaviour and ensure stability of wall thicknesses, we employed a protruding multi offset layer strategy which is typical in filament deposition 3D printing; also used for ensuring best exterior finish as opposed to interior. A more robust strategy left as future exercise is the development of internal webbing using material lattices which is also a common strategy in most 3D printing processes. We tested two methods for vertical motion transition between layers, namely u-turn and cross over, as well as various vertical step over ratios such that the extruder compresses material to increase adhesion between layers. Due to slow curing time which restricted the speed of extrusion and resulted to collapsed walls we used a fan to convectively cool the prints such that we could build faster and higher.

Figure 9. Testing Robotic Motion with Material Extrusion

6. Conclusions We presented the initial steps in the development of a wood composite 3D printing process highlighting the multitude of complex parameters required for successful production of sustainable products using traditional material and contemporary digital fabrication processes. There is a lot more work required to fully understand the material in both its mechanical as well as its aesthetic properties before we can even begin the design of products. Moving directly

DEVELOPING COMPOSITE WOOD FOR 3D-PRINTING

9

into building construction is even more challenging as it will require to address issues such as behaviour against fire hazards, thermal insulation performance and long term resistance against weathering.

Nevertheless, inspired by the current progress made we can envision some potential applications. Our matrix possesses exceptional properties such as its thermoplastic behaviour and self-adhesion. Those are key characteristics for its use in artwork and musical instrument repair. It is thus potentially interesting to investigate both additive repair and even full reconstruction of architectural ornamentation for heritage buildings. In past studies (Dritsas and Yeo, 2013) we witnessed the challenges in periodic restoration of timber temples which is dependent in an art and craft progressively vanishing. With careful consideration a 3D printing process may be a valuable tool for built heritage conservation.

Using 3D printing process may also allow us to exactly revisit those traditional forms of wood working (Kaijima et al, 2016) deploying purely material and geometry for construction without use of nails, screws and extraneous fasteners (Seike, 1977). Thus a contemporary application relevant to the resurrection of timber as building material supported by the need for sustainable design and construction processes is in the domain of timber joinery. The process of 3D printing is exactly relevant because connection between elements such as posts and beams are the most complex to design and fabricate. We look forward for further design, experimentation and prototyping that will help bring some of those ideas into practice.

Acknowledgements The authors would like to express gratitude to the SUTD-MIT International Design Centre, the SUTD Digital Manufacturing and Design Centre and Singapore Ministry of Education for supporting this research. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the aforementioned organizations.

References “3DSystems, Freedom of Creation develops Tree-D Printing in Wood” : 2011. Available from <http://www.3dsystems.com/blog/foc/freedom-of-creation-develops-tree-d-printing> (accessed November 2016).

“Materialise, 3D Printing Material Properties Data Sheet” : 2016. Available from <http://manu facturing.materialise.com/sites/default/files/public/AMS/Datasheets/datasheets_oct2016_1. pdf> (accessed November 2016).

Black, J.T. and Kohser, A.R.: 2012, Degarmo’s Materials and Processes in Manufacturing, Eleventh Edition, John Wiley & Sons Inc. Brummer, R.: 2006, Measuring Instruments, Rheology Essentials of Cosmetic and Food Emulsions, Springer Laboratory.

Chalcraft, E.: 2013, “How 3D printing will change architecture and construction” . Available from <https://www.dezeen.com/2013/05/21/3d-printing-architecture-print-shift> (accessed November 2016).

Chiellini, E., Corti, A., D’Antone, S. and Solaro, R.: 2003, Biodegradation of Poly(Vinyl Alcohol) based Materials, Progress in Polymer Science, 28, 963-1014.

10

R. TAN ET AL.

Clemons, C.: 2002, Wood-Plastic Composites in the United States, The Interfacing of Two Industries, Forest Products Journal, 52(6), 10-18.

Deplazes, A.: 2005, Constructing Architecture: Materials Processes Structures, A Handbook, Birkauser.

Dritsas, S. and Yeo, K.S.: 2013, Undrawable Architecture: Digital Workflows for the Conservation of Heritage Buildings and the Discovery of Digital Tectonic, Proceedings of CAADRIA 2013.

Falk, R.H.: 2010, Wood as a Sustainable Building Material, Wood Handbook: Wood as an Engineering Material, Forest Product Laboratory, General Technical Report, 6-13.

Franke, R. and Roffael, E.: 1998, Recycling of particle and fiberboards (MDF). Pt. 1: On the hydrolysis resistance of cured UF-resins in particle-and fiberboards, Holz als Roh- und Werkstoff, 56(1), 79-82.

Frihart, C.R. and Hunt, C.G.: 2010, Adhesives with Wood Materials Bond Formation and Performance, Wood Handbook: Wood as an Engineering Material, Forest Product Laboratory, General Technical Report, 228-251.

Gillemot, F., Mironov, V. and Nakamura, M.: 2010, Bioprinting is coming of age: Report from the International Conference on Bioprinting and Biofabrication in Bordeaux, Biofabrication, 2, 1-6.

Kaijima, S., Conti, Z.X., Sakhaei, A.M. and Dunn, M.L.: 2016, Parametric Finite Element Contact Analysis and Optimization for Topologically Interlocking Joinery Using Design of Experiments, Proceedings of the IASS Annual Symposium 2016.

Khoshnevis, B.: 2004, Automated construction by contour crafting—related robotics and information technologies, Automation in Construction, 13, 5-19.

Krassenstein, E.: 2015, “Selective Laser Sintered 3D Printed Wood is Now a Thing, Thanks to Materialise” . Available from <https://3dprint.com/70787/selective-laser-sintered-wood/ > (accessed April 2016).

Kretschmann, E.D.: 2010, Mechanical Properties of Wood, Wood Handbook: Wood as an Engineering Material, Forest Product Laboratory, General Technical Report, 146-171.

Lam, C.X.F., Mo, X.M., Teoh, S.H. and Hutmacher, D.W.: 2002, Scaffold Development using 3D Printing with a Starch-based Polymer, Materials Science and Engineering, 20, 49-56.

Lu, Z.J., Wu, Q. and McNabb Jr., S.H.: 2000, Chemical Coupling in Wood Fiber and Polymer Composites: A Review of Coupling Agents and Treatments, Society of Wood Science and Technology, State of the Art Review, 32(1), 88-104.

Patton, T.C.: 1979, Paint Flow and Pigment Dispersion: A Rheological Approach to Coating and Ink Technology, Ed. Wiley.

Pizzi, A.: 2006, Recent developments in eco-efficient bio-based adhesives for wood bonding: opportunities and issues, Journal of Adhesion Science and Technology, 20(8), 829-846.

Ross, P., Downes, G. and Lawrence, A.: 2009, Timber in Contemporary Architecture: A Designers Guide, TRADA/RIBA Publishing.

Schwarzkopf, J.M. and Burnard, M.D.2016, Wood-Plastic Composites, Performance and Environmental Impacts, in A. Kutnar and S.S. Muthu (eds.), Environmental Impacts of Traditional and Innovative Forest-based Bioproducts, Springer Science+Business Media, 19-43.

Seike, K.: 1977, The Art of Japanese Joinery, Weatherhill/Tankosha.

Silvestre, A.J.D. and Gandini, A.2011, Rosin: Major Sources, Properties and Application, in M.N. Belgacem and A. Gandini (eds.), Monomers, Polymers and Composites from Renewable Resources, Elsevier, 67-88.

Stark, N.M., Cai, Z. and Carll, C.: 2010, Wood-Based Composite Materials Panel Products, Glued-Laminated Timber, Structural Composite Lumber, and Wood-Nonwood Composite Materials, Wood Handbook: Wood as an Engineering Material, Forest Product Laboratory, General Technical Report, 252-279.

Starr, M.: 2015, “World” . Available from <https://www.cnet.com/news/worlds-first-3d-printe d-apartment-building-constructed-in-china/> (accessed November 2016).

Sun, J., H.o., K. a.n.d. W.a.n.g., S.: 2014, Clay Robotics, Master’s Thesis, University College of London, The Bartlett School of Architecture.

Wiedenhoeft, A.: 2010, Structure and Function of Wood, Wood Handbook: Wood as an Engineering Material, Forest Product Laboratory, General Technical Report, 61-79.