Nonlinear Fabrication: A Data-Driven Framework for Evaluating and Calibrating the Toolpath Design of 3D Printing Cementitious Materials
View/ Open
NonlinearFabrication_Alothman_Dissertation.pdf (13.05Mb)
Access Status
Full text of the requested work is not available in DASH at this time ("restricted access"). For more information on restricted deposits, see our FAQ.Author
Alothman, Sulaiman F S A A
Metadata
Show full item recordCitation
Alothman, Sulaiman F S A A. 2023. Nonlinear Fabrication: A Data-Driven Framework for Evaluating and Calibrating the Toolpath Design of 3D Printing Cementitious Materials. Doctoral dissertation, Harvard Graduate School of Design.Abstract
Clay, just like other natural, paste-like materials, offers a potential reduction in the embodied CO2 that the production of buildings using conventional materials emits, yet its large tolerances during printing remain an obstacle. In paste-based 3D printing, the material dries and shrinks at unpredictable rates while new layers continue to be deposited, causing increased self-weight onto the lower layers that are subjected to variable displacement. The ability to anticipate and correct the complex material behavior during the extrusion process is important in the effort to achieve accurate building components and assembly. While a possible approach is highly specialized models or workflows guiding designers to understand and model material and process variances, comprehensive models or workflows dealing with the nonlinearity of paste-based 3D printing processes are still lacking. Nonetheless, these processes promise efficient, waste-free, and sustainable production workflows at the architectural and building scale. This dissertation investigates how computational techniques based on machine learning models can enable rapid assessment and calibration of design solutions before fabrication, allowing for the prediction and simulation of final geometrical outcomes for accurate printing.The research contributes to digital fabrication by connecting digital design processes with material outcomes through a data-centered framework that leverages machine learning in novel ways. The framework offers: 1) a scanning method for a real-time calibration workflow that corrects the printing trajectories of the design object and serves as a rapid data-collection technique for machine learning applications; 2) a method for building an optimized dataset to evaluate the printability of design solutions; and 3) a method for training neural network models to calibrate the printing trajectories before fabrication. Tested in the context of clay lattice printing, an unorthodox extrusion scenario characterized by a large feature space and high material uncertainty, the framework demonstrated the ability to evaluate and calibrate the toolpath geometry of clay lattices with sufficient accuracy for manufacturing while using minimal resources, presenting an important step toward next-generation solutions for sustainable 3D printing.
Terms of Use
This article is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAACitable link to this page
https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37377305
Collections
Contact administrator regarding this item (to report mistakes or request changes)