Research On SPH-GPU Parallel Computing Method Of Sand Group Movement

At present,land desertification and its wind-sand environment are not a problem of a region or a country,but a major global problem.Desertification directly leads to the loss of human agricultural land and the increasingly bad living environment,which restricts the social and economic development of the country.Therefore,in order to find a reasonable technology to prevent desertification,we should start with the study of the law of wind and sand flow.In the numerical simulation of hydrodynamic problems,according to the different discrete methods of the problem domain,the numerical methods can be divided into grid method and grid-free method.The Lagrangian grid method can easily track the movement and interface of any point on the material through the grid nodes.However,for situations involving great deformation,numerical accuracy and solving difficulties are often caused by severe grid distortion,and it is difficult to effectively simulate wind and sand movement.The Euler grid method does not have the mesh distortion caused by the large deformation of the object.It is suitable for analyzing the large deformation and fluid flow problems,but it can not track the motion of the material with a fixed mesh or the interface of the material.Smooth Particle Hydrodynamics(SPH)meshless method discretizes the solved problem domain into a series of arbitrarily distributed particles.these particles carry their own material information,such as physical quantities such as mass,density and velocity,avoid problems such as mesh distortion or deformation boundary,and can easily track the trajectory of a single particle,so this method has a unique advantage in studying wind and sand flow.However,with the increase of the number SPH particles in the wind and sand flow,the calculation efficiency of this method is low,and the defects of large scale calculation are especially obvious in the process of wind and sand simulation.Hence,in order to improve its computational efficiency,this paper uses SPH-GPU large-scale parallel computing technology to realize the acceleration operation in numerical simulation of wind and sand flow.According to the process of wind and sand flow,the corresponding mathematical model is constructed,and the complex governing equation used in the simulation of wind and sand flow is introduced and SPH discretized.Furthermore,the boundary particle integration is truncated in the SPH method,so the virtual particle method is used to solve the problem of error.Meanwhile,in order to improve its computing efficiency,a gas-sand two-phase coupling model SPH-GPU parallel acceleration is established on the CUDA hardware and software platform supporting parallel computing.Firstly,the hot program of SPH serial algorithm is analyzed in detail.Find out the most time-consuming and suitable for parallel hot programs.Secondly,the parallel neighborhood search of sand-sand-flow particles based on-pair search method is used to simulate the wind and sand flow.After that,the parallel computing model SPH-GPU wind-sand flow is verified,the spatial and temporal variation of sand mass motion is obtained macroscopically,and the track of single sand particle motion is obtained microscopically.Lastly,in order to illustrate the efficiency of SPH-GPU large-scale parallel computing method,the computational efficiency of CPU and GPU under different particle numbers and the influence of the number of lines on the computational efficiency are compared.Simulation results show that SPH-GPU parallel computing method can be further applied to the numerical simulation of wind and sand flow.