Topology optimization is a design method used for conceptual design of structures. It has great impact on the final performance of structures compared with optimization methods in lower hierarchy such as size or shape optimization. The homogenization method [1], SIMP (solid isotropic material with penalization) [2] and ESO (evolutionary structural optimization) [3] are in the mainstream of topology optimization. A major drawback of these methods is their high computational cost, which is particularly critical when solving practical design problems with large numbers of design variables. Parallel processing [9-10] would help, but this approach does not resolve the inherent problem of computing inefficiency in the method. In this paper, we propose a novel topology optimization method that has high computing efficiency for large-scale problems. This method is developed based on Design Space Optimization by Kim and Kwak [4]. First, the fundamental concept of design space adjustment is introduced briefly, and then our simple but effective implementations are explained. Second, design space refinement scheme is proposed based on fixed grid. The design domain is adaptively refined in order to utilize computing resource most efficiently. A numerical example of automotive suspension part, knuckle, is optimized using the proposed method.
展开▼