Parallel Algorithm Research Of Compressible Fluid Based On Two-dimensional Structure Grid

Shock tube problem is a classic problem of compressible flow,rich analog phenomenon,widely used numerical methods and analysis of the results of verification,in order to improve accuracy while reducing computing resources,computing adaptive shock tube problem is often encrypted technology,with the development of computational fluid dynamics to calculate the size of shock tube problem is also increasing,parallel computing is an inevitable trend.In this paper,shock tube numerical simulation,flow field simulation based on structured grids,equations are NS equations,discrete way to fourth-order Runge-Kutta method,using explicit calculations and adaptive encryption technology,to carry out parallel compressible fluid optimization of the main work and achievements are as follows:(1)Based on a two-dimensional structured grid based grid,we designed three parallel algorithms,parallel algorithms are blocking,as well as on the basis of non-blocking communication optimized parallel optimization algorithm and MPI + OpenMP hybrid programming algorithm.(2)Based on two-dimensional structure of a multi-layer adaptive refinement of the grid mesh,an adaptive parallel optimization algorithm is designed based on workload migration load balancing algorithm to solve the cause of local mesh refinement after the load is too large,resulting in load imbalance.(3)This design is based on a variety of parallel optimization algorithm,the Milky Way on the 2nd supercomputer were carried out to test the accuracy and performance testing,performance testing has been weak and strong scalability and scalability testing,test results show no the use of encryption technology,the basis of the number of meshes 40 million,extended to 8192 to calculate the scale of nuclear,speedup can reach 6896.03.Adaptive refinement layers to three layers,set the number of 16 million base mesh,calculation of the expansion to the 1024 nuclear speedup can reach 294.71.