Authors: Zhi Li
Time:Â 2024/12/03
Abstract
Topology optimisation techniques are increasingly used for creating innovative and efficient structures by redistributing given materials to their most needed locations. However, these techniques typically focus purely on structural performance, often neglecting subjective design requirements, such as aesthetic quality. This thesis is aimed at advancing topology optimisation techniques to produce structures that effectively balance creative forms with structural engineering principles. The thesis makes the following contributions.
First, a novel smoothing method is proposed to smooth structural designs obtained by element-based topology optimisation, specifically for the widely used bi-directional evolutionary structural optimisation (BESO) method. The proposed method can efficiently smooth the zig-zag boundaries of the optimised structures by using pre-built lookup tables modified from marching geometry algorithms. Additionally, the smoothing method employs a bi-section approach to preserve the structural volume, ensuring that both unsmoothed and smoothed structures are equivalent. The effectiveness of this smoothing method is demonstrated through a series of 2D and 3D examples; the results show that the structural stiffness can be slightly improved after smoothing.
Second, the BESO method has been advanced to SP-BESO by integrating designers’ subjective preferences (SP) into optimisation for 2D structural designs. This novel method introduces interactive systems that allow designers to input preferences by explicitly scoring and drawing. Subsequently, the inputs are converted into weights to guide material redistribution to create topologically different and structurally efficient solutions. The proposed smoothing method is used to refine the solutions so that designers can decide their subjective preferences in subsequent design exploration. Additionally, a user-friendly digital design tool, iBESO, has been developed. This tool can simultaneously execute four SP-BESO optimisers to assist designers in 2D structural design. Experiments with iBESO reveal that the combination of parameters used in the scoring and drawing systems can effectively control whether the final solutions are performance-driven or preference-driven designs.
Third, a novel design exploration strategy is proposed, which integrates virtual reality (VR) with SP-BESO to create desirable 3D structures through effective human--computer collaboration. This strategy uses VR sculpting to realise immersive visualisation, intuitive design exploration, and real-time feedback, with the sculpted models influencing material redistribution in topology optimisation. The sculpting--optimisation workflow can be repeated in multiple cycles. This iterative process allows for the continuous refinement of subjective preferences until a final design meets all design requirements. A digital design tool, VR-BESO, has been developed to implement and demonstrate the proposed strategy. The results show that this strategy can effectively harness the strengths of human creativity and computational power to enhance the efficiency of design exploration and the quality of optimised structures.
Overall, this thesis has significantly advanced structural design and optimisation by integrating subjective preferences into topology optimisation. It has introduced innovative methodologies, developed advanced digital tools, and enhanced human--computer interaction, laying a strong foundation for future research and practical applications. These contributions mark a substantial progression in creating engineering solutions that are more adaptable and responsive to user needs.