A parallel parameterized level set topology optimization framework for large-scale structures with unstructured meshes

In addition to the requirements of full-scale optimization, the adaptability to structures with arbitrary geometries and complex boundary conditions is also important to topology optimization in practical engineering applications. A parallel parameterized level set topology optimization framework for large-scale structures with unstructured meshes is proposed in this work, in which the full-scale optimization is realized with distributed memory parallel computing technology while the arbitrary geometries and complex boundary conditions are conveniently handled with the usage of unstructured meshes. To realize the combination of distributed memory parallel computing technology and parameterized level set topology optimization using unstructured meshes, several means are taken: (1) the shape functions in finite element analysis are employed to parameterize the level set function; (2) the data structure called directed acyclic graph is adopted to represent the unstructured mesh; (3) the passive domain and boundary conditions are imposed directly on the geometry entities of the structures; (4) a multiple averaging filter is introduced to reduce the tiny structural members in the optimized results for the requirement of manufacturability. Several computing tests are presented in this paper, which verify the stability, efficiency, scalability, and the potential to discover new structure styles of the framework.