1. Successive fabrication of complex innovative prototypes

Seriality as methodology is a main characteristic of the author’s research practice and represents his conscious attempt to gain expertise by repetition and iteration, ‘in the domain of scientific observation that is absolutely objective, the “first time” doesn’t count. Observation, then, belongs in the domain of “several times”’ (Bachelard 1994). In this project, seriality meant a succession – parallel or sequential in time – of exercises, and with a large number of similar or contrasting techniques and project partners. Ongoing confrontation with several problems provided a contextually broad record of research significance with weight and impact for wider application in the industry (28–9).

28

28 Clay Column, 2018. Extrusion-thickness tests to evaluate and develop a clay extruder.

29 3D-printed concrete. Several repetitive tests were necessary to develop a printable mortar mix.

2. Scaling-up of existing and new additive manufacturing techniques using bespoke industrial workflows and craft-related processes

Increasing the size of the manufacturing objects necessitated a switch from desktop-sized 3D printers to bespoke techniques involving robotic arms and special tools. Such scaling-up implies an exponential increase of difficulties, as detailed below.

1. Transport logistics: the design of the large prints such as Pahoehoe Beauty were planned so that all individual elements were printed directly onto

EUR-pallets (1800 × 1200 × 145 mm) to simplify handling and transportation

(33–4).

2. Assembly and disassembly strategies: the elements comprising Coralloid

Cocoons and Venice Taevatiib were designed to be light and small enough to be assembled by hand (36–8).

3. Material behaviour: 3D extrusions and depositions deploy a large amount of material, with a higher response to forces such as gravity and shrinking.

For the concrete and clay prototypes, all elements had to be designed in order to withstand their own weight once fully printed and solidified, as well as during the process of printing when the material was still wet and had little structural performance. Overall shape and local contouring were also considered.

4. Predictability: There is an inherent deformation in the clay prints due to material shrinkage (35). Experience and data, such as 3D scanning to measure the deviation of the digital with the actual piece, helped the team to create a predictable data set for future production.

5. Equipment capabilities: Large-scale prints made with desktop extruders take a long time to make and deploy a lot of material, which has a repercussion on the reliability of the tools used. Failure is inevitable due to overheating and breakdown of parts.

Industry-standard extruders and end effectors thus needed to be improved, with a rigorous and methodological testing of alternatives (36).

30 Bespoke analogue equipment developed for Porous Cast, 2015. Digital automated processes included metalworking furnaces, crucibles and moulds to cast bronze pieces. 31 Bespoke analogue equipment developed for Pahoehoe Beauty, 2018, and Terrestrial Reef, 2019. Garden spray bottles and pressure pumps were used to spray stabilising agaragar onto 3D-printed soil. 32 Terrestrial Reef, 2019. Bespoke equipment included a rotary compost sifter to separate coarse and fine soil.

30 31

32

33 33 Pahoehoe Beauty, 2018. Several tonnes of soil were printed on EURpallets and were shipped to Ars Electronica in crates. The solid parts were concealed within unprocessed soil before being excavated onsite.

34 Coralloid Cocoons, 2016, was subdivided into small and light segments to be easily transported and assembled at the venue without a need for cranes or lifting aids.

35

36 35 Clay Column, 2018. Detail depicting unequal shrinking of the top and bottom of each segment.

36 Bespoke end effectors were developed, designed and built several times to achieve high-quality extrusions.

37

37 350/360 Cohesion/ Cocoon, 2019. A segment, manufactured at incremental3D, with inclined print paths and variable extrusion thickness.

3. Development of bespoke software

Large-scale robotic 3D printing with complex workflows and a wide range of materials, forms and tasks requires specific software, including programs, parametric tools, scripts and simulation parameters, to function properly. Tailored software adjustments and scripts were therefore required to run the robots in new situations and to control adapted end effectors with bespoke materials and processes.

1. The plugin Taco for Rhinoceros (38) was developed by Shih-Yuan Wang, Yu-Ting

Sheng and Florian Frank to program, simulate and control ABB industrial robots directly within Grasshopper and Rhinoceros. Through an active connection with the controller, the user can control the robots with a fast and low-risk workflow. Even without a robot controller, it is possible to use Taco together with ABB RobotStudio as an offline programming solution. The aim is to reduce the expert knowledge required for complex robot movements and to feedback the necessary information to the user in an easy and comprehensive way. It makes it possible to develop movements parametrically and to create a familiar interface for users. A variety of different ABB robot models are supported and others can be added per user request. Taco also provides non-movement related commands for the robot to create more advanced tasks. Through the use of IO ports, it is also possible to connect other devices and custom tools to the head, which communicates with Taco. The head reads and uses the calibration data from a connected controller to ensure that the programming and simulation matches the robot’s movements. This provides a precise digital representation of the physical environment and guarantees the digital-to-physical workflow. The ability to create custom tools as an end effector for the robot allows for a wide range of applications.

2. For Liquid Rock, a variable growth algorithm was developed for Grasshopper, with the use of Anemone, Boid Library and Kangaroo Physics 2.42 to encode material and process behaviour, capacities, affordances and constraints (39). This algorithm controls growth variation for print optimisation and stability, i.e. amount and degree of overhang to avoid local drips at each deposition layer, and ensure overall stability for the object during 3D printing and in its final morphology.

The agent-based approach proved to be computationally lighter, easier to write and optimise, more flexible to support variability and locality, and more intuitive than the common physics-based pipeline to simulate growth. Parallel comparative tests for differential growth were conducted in Houdini. In addition, the script allowed the team to move from continuous layer height to a discrete variation in feed rate due to the forgiving nature of the mortar.

38 Taco for Rhinoceros, a plugin for Grasshopper to run the multimove ABB system at REX|LAB. 39 A software script was developed for Liquid Rock by Eftihis Efthimiou, which minimises fabrication time and evaluates each element for stability during and after printing.

38 39

3. Layercake is a script for Grasshopper developed for Guardians of Time by Damjan

Minovksi. It renders concrete-additive 3D-printing extrusions while considering material viscosity (40).

4. A multi-step, iterative script was developed to generate the design of Pahoehoe

Beauty and Terrestrial Reef (41–2). This process followed three actions: first, the creation of a multi-agent system expressing a behaviour based on natural systems, i.e. rooting of plants. Specifically, the agents represent a 3D description of a fractal tree index, a peculiar data structure that generates branching entropy with a tendency to merge in one specific direction. Second, isosurfacing operations commonly used in computer modelling were used to generate rough volumetric data from 2D information (lines or points).

Third, details were added to accentuate specific sleeping features, such as venations and openings in the mesh, originating from the first volumetric data.

5. A music/scanning script was developed for

Pahoehoe Beauty. Two robot arms were equipped with webcams to record sound and create a unique composition (42). The robots traced the 3D prints along a predefined path, scanning the surface and gathering information about its structure.

These live images were processed individually and as a pair, and were ingested into an algorithm. This converted the grayscale value of each pixel, and their position in relation to each other, to control various parameters of digital synthesizers, including pitch, length and volume to generate tones. The two digital images were then compared and fed back into each other.

41 42

40 Guardians of Time, 2017. Render showing the impact of 3D-printing texturisation. Scaling-up of 3D printing is not a linear direct process but instead requires an equal amount of rigour and invention. 41 Scripts developed to design the 3D-printed soil mounds for Terrestrial Reef, 2019.

42 Pahoehoe Beauty, 2018. A bespoke end effector with camera that scans the landscape and provides input to the music/scanning script by Jonathan Hanny.

4. Development of bespoke hardware, processes and equipment

In parallel to software, end effectors and other tools had to be designed, built and tested, from the most basic, such as the metal scoop used for carving clay for Pahoehoe Beauty to more sophisticated elements such as material-specific extruders.

1. For Porous Cast, a simple bent metal tool was used to scoop out the forms from a clay bed for the Stewalin cast, which could be used again (43).

2. Plastic printing heads for 1 mm extrusions were developed over three generations for Quaquaversal

Centrepiece and Crystal Net 3.

Based on desktop printing hardware, they work in mid-air and have a springy nozzle that can print on irregular surfaces (44).

3. The industrial hand-held Dohle-Herz

Robot DX125 extruder was adapted, tested and evaluated for Plastic

Column’s 3D-printing processes, to reach an extrusion velocity of 0.8 kg/h dependent on the performance of the extender. In parallel, a bespoke heated glass printing base (1050 × 1600 mm) with 12 self-adhesive silicon heating mats (200 × 400 mm, 230 V, 533 W) was developed, to reduce bending and distortion in the plastic as it cools down (46). 4. Industrial handheld extruders and concrete mixing pumps and mixers were adapted, tested and evaluated for 3D-printing processes for Coralloid

Cocoons, Liquid Rock and 350/360

Cohesion/Cocoon, resulting in the

Baumit BauMinator® (47).

5. Industrial hand-held extruders were adapted, tested and evaluated for clay 3D-printing processes (48).

6. End effectors that deposit agar-agar and/or Sporosarcina pasteurii bacteria and tools for compacting soil were developed for Pahoehoe Beauty (45) and Terrestrial Reef. In parallel, bespoke end effectors were built for the scanning of the soil (as input for the sound installation) and covering ‘socks’ were tailored to protect the robotic arms from dust, sand and liquid.

43 Porous Cast, 2015. Robotic carving of reusable casting moulds.

44 Quaquaversal Centrepiece, 2015. Bespoke 3D extruders ‘stitching’ plastic onto laser-cut fabric.

45 Pahoehoe Beauty, 2018. Bespoke ‘compacter’ and ‘squirter’ end effectors.

43 44

46

46 Plastic Column, 2018. The 3D-printing set-up required specific tool and material solutions, such as a heated bed so that the model does not detach from the base and distort.

47 The Baumit BauMinator® system.

5. Development of bespoke 3D-printable materials

Various materials were researched and tested in the scaling-up of 3D printing for additive extrusion processes. In order to experiment with alternative rapidprototyping methods, several materials that undergo a phase-change process – melting from hard to soft to hard (plastics) or from powder to liquid to hard (mortar, clay) – were tested, including hot-melt adhesive, plaster, plastic, concrete, mortar, Stewalin, gel, clay compound, sand, soil, agar-agar, mycelium and Sporosarcina pasteurii bacteria. The larger and heavier the prototypes were, the more apparent the ecological aspect of the chosen material became. Clay was introduced and, in the later stages of the research, the biodegradable 3D-printed soil process was developed.

48

48 Several generations of 3D extruders were tested for Clay Column, 2018.

49

49 Gel inlays were used to create a porous bronze cast for Porous Cast, 2015.

50 Porous Cast exhibited at the Tallinn Architecture Biennale, 2015.

51

51 Porous Cast, Tallinn Architecture Biennale, 2015. Prosity was achieved in both the bronze and Stewalin casts. 52 (overleaf) Crystal Net, 2015–8. Filigrane and elastic extruded-plastic structures were fabricated and covered with crystals. The lines were extruded in such a way as to resemble hand-drawn sketches. 53 (overleaf) Crystal Net, 2015–8. Patterns, printing techniques and materials were developed to embed crystals into a flexible and customisable fabric.

53