Node-based body-fitted topology optimization with a boundary-preserving sensitivity filter

Topology optimization techniques are widely employed to design efficient and innovative structures. However, achieving optimized results with smooth boundaries remains a significant challenge, primarily due to ambiguous boundary definitions with zigzag features induced by intermediate-density elements and fixed meshes. In this study, a node-based body-fitted optimization approach is presented. A probabilistic body-fitted mesh algorithm is integrated into element-based topology optimization frameworks to achieve optimized solutions with well-defined, smooth boundaries. To enable high-quality body-fitted mesh generation, a boundary-preserving sensitivity filter is developed, which effectively removes intermediate-density elements by suppressing filter effects around material boundaries, thereby ensures precise boundary representation without the need for projection techniques. Numerical examples demonstrate that the proposed filter successfully removes intermediate-density elements. By leveraging the node-based body-fitted optimization approach, optimized designs with smooth boundaries are inherently obtained, eliminating the need for post-processing while simultaneously enhancing structural performance. Quantitative analyses demonstrate approximate 15% improvement in objective functions for the SIMP method and 10% for the BESO method. The proposed approach enhances both computational accuracy and practical manufacturability, providing a solid foundation for applications of topology optimization in real-world engineering structures. • A node-based body-fitted optimization approach is presented. • A probabilistic body-fitted mesh algorithm is integrated into topology optimization. • A boundary-preserving sensitivity filter is developed to eliminate intermediate-density elements.