The research addresses the investigation of Viscous Thread Instability (VTI) for the additive manufacturing of Functionally Graded Structures (FGS) at the architectural scale. The deposition of a thread of viscous material in space is an inherently unpredictable process, characterized by a high degree of instability non-linearity.
Material instances of a similar fabrication process results in an intricate mass of matter exhibiting a uniform or variable resolution surface texture, which is the product of a layer-by-layer deposition process. Such a process of additive manufacturing can be conceptually interpreted as an extension to the third dimension of the process of production of non-woven textiles.
Gaining control over the spatial deposition of a liquid-phase material and acquiring data at high-frequency time intervals requires a reliable on-line evaluation strategy to feed back a computational fabrication model. The envisioned fabrication setup includes an industrial robotic arm equipped with a pneumatic extruder, able to adjust position, orientation and flow rate by reacting to the information produced by a vision system, ideallly in a real-time loop.
First tests are currently being conducted with a small desktop 3D printer (Renkforce 100) extruding thermoplastic filament (Poly Lactic Acid) coupled with one affordable Time-of-Flight sensor (pdm CamBoard pico flexx) which frames the fabrication scene off-board. The process will be later scaled-up by using a robotic arm and, after experimental considerations and planning, by eventually refining the setup in terms of vision (by embedding extra cameras to expand the field of view) and/or in terms of acquired data (thermal cameras or other kind of sensing devices depending on desired outputs and selected material systems).
The output of the sensing process in our current experimental setup is a point-cloud representation of the XYZ position of the scanned geometry, coupled at each point with depth cinfidence, grey and noise values, which are accessed at a pre-defined time interval. This data is processed inside a 3D modelling environment (McNeel Rhinoceros/Grasshopper) which allows for design and fabrication data management as well as for customized computational workflows via scripting interfaces in different programming languages (Python is the preferred option). Therefore, other types of input data could potentially be managed and easily integrated in the workflow.
Regarding the translation of these data into useful information to feed the fabrication process back, two possible scenarios come to mind:
- A large scale evaluation of statistical properties – e.g. density, porosity, mass distribution and fiber orientation – of the last-deposited layer is generally required for assessing quality and functionality of fibrous structures. As a further refinement step, since viscous materials are involved in the process, other material- or environment-related properties – temperature, moisture-content – could possibly be integrated. The output of these sensing processes my be rendered as either color, grayscale or binary bitmap or, eventually, as simple parsable text file.
- A small scale geometric evaulation of the current position (plus, eventually, orientation) of a thread center-line, to be output as a polyline (or, eventually, as an array of reference planes).
Marco Palma, Univ.Ass.Dott.Mag.