Abstract: The Cartesian mesh method has the advantages of easy adaptation, high degree of automation and good quality. However, its non-body fitted characteristics leads to unnegligible massive amount of grids when addressing complex configuration or complex flow problems. The efficiency of Cartesian mesh generation directly affects the whole computation cycle, entailing the development of efficient generation technology. The key to Cartesian grid generation efficiency is the grid data structure, which directly influences the amount of computation and storage. In this paper, Cartesian grid generation is conducted for three-dimensional configurations, and the Fully Threaded Tree data structure with faster neighbor query and higher memory utilization is developed, which is applied and further improved in this framework. Meanwhile, to effectively determine the type of grid cells, a fast retrieval method of facets is constructed, and the painting algorithm introduced to further improve the grid generation efficiency. In addition, a robust singularity detection algorithm is proposed to ensure the robustness of the grid cell type determination. With respect to mesh adaption based on the flow field solution, a three-dimensional criterion combining velocity divergence and curl is adopted to ensure the ability to capture various flow characteristics. Some test examples such as spheres, missiles, wing-body and wing-aileron configurations are used to verify the method. Comparison shows that the adaptive position of the mesh is in good agreement with the theoretical solution, the generation time cost of the mesh is low, and the average time is less affected by the distribution of the facets, thereby proving the reliability and efficiency of the proposed method.

Key words: Cartesian mesh, fully threaded tree, singularity check algorithm, flow feature based adaptation, painting algorithm