A fast adaptive LBM multi-layer grid generation method
The lattice Boltzmann method,which is based on a Cartesian grid,offers advantages such as reduced computational complexity,easy adaptability,and the ability to handle complex boundary conditions.However,its non-conforming characteristics require an excessive number of grids when complex geometries or flows are simulated.To alleviate this issue,local grid refinement can be used for complex geometries.The achievement of local grid refinement depends on rapid grid generation and the mapping relationship between grids of different scales.This study proposes a new rapid grid generation algorithm by designing a novel data structure and a ray intersection algorithm.The algorithm is applied to the multi-layer grid lattice Boltzmann method under two basic elements,thus further developing efficient grid generation technology.The ray method algorithm's characteristics allow for a significant reduction in the number of surface elements that need to be retrieved through the new data structure.For geometrically adaptive multi-layer grids with two basic elements,the cell-centered scheme constructs the grid mapping relationship between different layers through a quadtree structure.The vertex-centered scheme determines the mapping relationship through the overlap of the background grid.In the grid results of the two formats,the multi-layer grid of the cell-centered scheme is less affected by the number of encryptions.It reaches a generation efficiency of about 5.89 million grids per second at the level of tens of millions of grid points with 9 encryptions.The efficiency of the vertex-centered scheme is significantly impacted by the number of encryptions.However,it still achieves a generation efficiency of approximately 660000 grids per second,even at the level of tens of millions,which is about 40 times more efficient than existing methods.The accuracy of the grid algorithm is finally verified through the solutions of two-dimensional cylinder flow,three-dimensional sphere flow,and three-dimensional submarine flow.
万方数据
