Please use this identifier to cite or link to this item:
Title: Optimization algorithms for additive manufacturing
Authors: Fok, Kai-yin
Degree: Ph.D.
Issue Date: 2019
Abstract: This thesis is dedicated to the study of 3D printing optimization algorithms. By minimizing the time needed by a printing nozzle to traverse a tool-path, the fabrication time of 3D objects can be shortened. Apart from fabrication time, other physical properties of fabricated parts such as dimensional accuracy and visual quality are considered as important evaluation criteria in this thesis. The tool-path planning problem of 3D printing applications is formulated into an undirected rural postman problem (URPP), which can be solved using existing URPP solvers. However, the computational times required by them increase rapidly with the number of print segments in a model. Some of these solvers are too time-consuming to be applied in real-life 3D printing processes. This thesis aims to develop algorithms to enhance the efficiency of 3D printing processes for both industrial and household applications. This thesis begins with a detailed problem formulation, followed by an introduction to some essential tweaking and modification techniques, which include a tool-path optimizer developed based on Christofides' algorithm and a 2-opt local search algorithm. Furthermore, a segment-consolidation scheme is proposed to shorten the optimization process. The tool-path optimizer is operating at the inter-partitions level and the intra-partition level sequentially, with an aim to find a sequence with a low time-cost to visit all dissected parts on the same print layer of a model. Moreover, to represent curves on a layer, massive volumes of chained print segments were normally utilized. These print segments are consolidated into replacement segments in the optimization process to reduce its computation complexity. The replacement segments are reverted to their corresponding print segments at the end of the optimization process. Simulation result using actual 3D models showed that the proposed tool-path optimizer can shorten the fabrication time of printed parts and required less post-processing time than its counterpart's.
Then, an ant colony optimization (ACO) based tool-path optimizer is proposed to further accelerate the printing processes. By utilizing the stochastic mechanism of the nature-inspired meta-heuristic and some unique properties in 3D printing processes, the proposed optimizer performs a more thorough search in the shrunk search space which helps to find better solutions when compared to generic ACO. Experiments were conducted using a domestic 3D printer and models to evaluate different tool-path optimizers. The proposed tool-path optimizer outperformed its counterparts in terms of both time-saving on fabrication times and post-processing times. Moreover, experiment results suggested that the proposed optimizer does not downgrade the dimensional accuracy of print parts but indeed improves the visual quality. Finally, a computationally efficient tool-path optimizer is proposed based on a new detour search algorithm and an efficient local search algorithm. The new detour search algorithm focuses on replacing unnecessary overheads on a path with detours which have relatively lower time costs. While conventional k-opt local search algorithms are widely used to solve routing or path planning problems, their computational complexity increases rapidly with the parameter k. A new implementation is proposed based on some unique properties in 3D printing applications. The new implementation shrinks the searching space significantly. It is mathematically proven that the new implementation does not degrade the quality of the solutions generated. Both experiment and simulation results showed that the proposed tool-path optimizer can significantly accelerate the 3D printing process by reducing at most one-third of the fabrication time when printing a model, which took insignificant post-processing time. The proposed tool-path optimizer is a local search algorithm, therefore, it is capable of cooperating with other algorithms, including the former mentioned two proposed optimizers, to further improve the quality of solutions.
Subjects: Hong Kong Polytechnic University -- Dissertations
Three-dimensional printing
Mathematical optimization
Pages: xxii, 92 pages : color illustrations
Appears in Collections:Thesis

Show full item record

Page views

Last Week
Last month
Citations as of Jun 4, 2023

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.