Geometry-aware support design for additive manufacturing

Abstract

In recent years, additive manufacturing (AM) has drawn significant attention and interest from both academia and industry due to its remarkable advantages. However, one critical challenge in AM is tunability of mechanical properties for AM parts. Therefore, this dissertation focuses on the development of systematic approaches for enhancing mechanical property estimation and improving geometric accuracy through support structure design. Specifically, the study introduces data-driven models to predict mechanical outcomes based on process parameters such as feed rate, support distance, and nozzle temperature. In parallel, novel support generation techniques—such as conformal and Pythagorean-Hodograph (PH) curve-based structures—are proposed to reduce geometric errors and optimize support material usage. The presented methods aim to improve both predictability and efficiency in AM processes, paving the way for smarter design strategies and more reliable part production. Through experimental validation and comparative analysis, this work highlights the critical role of support design in achieving functional and structurally sound AM components. This dissertation first addresses the challenge of poor bridging and mechanical performance in additive manufacturing (AM) parts by examining the influence of key process parameters and support distances. Poor bridging, which occurs when material extruded between unsupported points sags or deforms, not only affects part quality but also compromises mechanical integrity. To mitigate these issues, two estimated models (EMs) are developed to predict the ultimate stress and strain of AM parts based on feed rate, support distance, and nozzle temperature. The EMs are derived using a central composite design (CCD) method and refined through systematic sampling for improved accuracy. Extensive tensile tests are conducted to validate the models. Finally, the EMs are integrated into a process optimization framework that automatically suggests optimal process parameters to meet predefined mechanical performance criteria. This approach enhances both the reliability and efficiency of AM production by enabling data-driven tuning of manufacturing parameters. An efficient support structure design methodology is then introduced for three-axis additive manufacturing, with a particular focus on conformal print-path strategies. Recognizing the trade-off between support density and geometric accuracy, the proposed approach integrates print-path geometry into support placement, ensuring controlled support distances and minimizing the risk of unsupported features. The method has been validated across eight diverse test cases, demonstrating a notable reduction in support material usage by an average of 16\% and a significant enhancement in geometric accuracy by 73\% compared to conventional support generation techniques. These results highlight the potential of geometry-aware support planning to enhance both material efficiency and dimensional fidelity in AM processes. Finally, a novel framework is proposed for generating supports in additive manufacturing (AM) using Pythagorean Hodograph (PH) curves. PH curves offer significant advantages in AM by enabling smooth geometry transitions and demonstrating favorable offset properties, which promote uniform support spacing, minimize poor bridging effects, and reduce machine vibrations. We propose two distinct PH-based support (PHs and O-PHs) design schemes integrated with a type of Traveling Salesman Problem (TSP) based print sequencing algorithm. The results show that both O-PHs and PHs significantly improve geometric accuracy in printed parts and reduce their print times.