Extrudate formation modelling and variable geometry nozzle system design for extrudate geometry control in 3D concrete printing

• Main Author: Lao, Wenxin
• Other Authors: Li Hua
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/143015
• Institution: Nanyang Technological University

3D printing is a fast-developing technology which allows for solid part fabrication in layer-by-layer manner. This technology has been introduced to construction industries and it draws a lot of attention due to its potential for higher customization, reduced human intervention and shorter construction time for complex structure. Recently, large-scale house and structures can be fabricated using this technology with fresh concrete material. However, the lack of control on extrudate formation has become a major drawback for this technology. Without the control of extrudate formation, the design freedom is restricted, and printing quality is limited. This research focuses on the computational and experimental studies to control the extrudate formation process to improve the output quality of 3DCP. In order to allow for a controllable process, the relationship between the printing parameters to the resulting extrudate shape needs to be established. Preliminary studies show that an appropriate strategy to adjust nozzle outlet shape can be utilized to control the extrudate formation to improve the printing quality. Moving forward, the nozzle-adjustment strategy suitable for on-site printing applications needs to be developed. A machine learning method is used to analyze the relationship. A data-driven predictive modelling methodology based on Artificial Neural Network (ANN) is adopted to develop the relationship between the nozzle shape and the extrudate cross-sectional geometry. This approach allows for establishing a nonlinear relationship between nozzle shape, which is represented by the finite geometrical points, to the extrudate shape. The results show that the ANN method successfully maps a nonlinear relationship between the nozzle shape and the extrudate formation with encouraging results. The accuracy of the predictive ANN model is experimentally validated, and surface finish quality is improved in several case studies. Based on the developed model, an active mechatronic nozzle, which can adapt its shape during printing, is designed to control the extrudate formation during the printing. This nozzle is featured for its capability to control extrudate cross-sectional by adjusting the nozzle outlet shape. With this nozzle, the extrudate geometry can be controlled to comply to the designed outer surface shape of the structure, and hence improve the surface finish quality. A slicer program is also developed to analyze the CAD file of the designated printed structure and to extract the desired geometries for every extrudate layer for printing. The controller of the variable-geometry nozzle is developed to regulate the nozzle shape to achieve the desired extrudate geometries. Finally, an arch structure was printed with the active nozzle and exhibited obvious improvement in surface finish quality. The current study could be further enhanced in three directions. The monitoring system could be integrated on printing system to allow for close-loop control of printing quality. The predictive ANN model could be improved to predict the extrudate geometry for different materials. The variable-geometry nozzle could be applied to solve other quality problem in 3DCP, such as material distribution problem in curved-path printing.