Piecewise length scale control for topology optimization with an irregular design domain

• Published in: Computer methods in applied mechanics and engineering Vol. 351; pp. 744 - 765
• Main Author: Liu, Jikai
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.07.2019; Elsevier BV
• Subjects: Algorithms; Constraints; Control methods; Decomposition; Design optimization; Irregular design domain; Length scale control; Level set; Segmentation; Structural skeleton; Topology optimization; Topology optimization; Length scale control; Level set; Structural skeleton; Irregular design domain
• ISSN: 0045-7825; 1879-2138
• DOI: 10.1016/j.cma.2019.04.014

This paper presents a piecewise length scale control method for level set topology optimization. Different from the existing methods, where a unique lower limit or upper limit was applied to the entire design domain, this new method decomposes the topological design into pieces of strip-like components based on the connectivity condition, and then, the lower or upper limit for length scale control could be piecewise and dynamically defined based on each component’s real-time status (such as position, orientation, or dimension). Specifically, a sub-algorithm of structural skeleton identification and segmentation is developed to decompose the structure and its skeleton. Then, a skeleton segment-based length scale control method is developed to achieve the piecewise length scale control effect. In addition, a special type of length scale constrained topology optimization problem that involves an irregular design domain will be addressed, wherein the complex design domain plus the length scale constraint may make the conventional length scale control methods fail to work. Effectiveness of the proposed method will be proved through a few numerical examples. •Realize the piecewise length scale control for level set topology optimization.•Dynamically and separately define the length scale target based on each component’s real-time status.•Use image processing techniques for structural skeleton identification and segmentation.•Address the length scale constrained topology optimization problem that involves an irregular design domain.