Design And Implementation Of A Large-scale Flow Field Visualization System Based On Parallel Particle Tracking

In recent years,with the continuous development of computer hardware technology and numerical simulation technology,researchers have chosen parallel technology to develop flow visualization software.However,with the gradual expansion of the scale of the flow field and the continuous improvement of the accuracy of the flow field simulation,the traditional flow field visualization system faces many challenges such as uneven data distribution and closed areas in the flow field.Therefore,this paper studies and designs a large-scale flow field visualization system based on particle tracking.The system is divided into three parts:(1)Parallel particle algorithm:with the continuous increase of flow field data and simulation calculation,particle tracking has changed from traditional serialization to parallelization,and multiple processes are parallel computing multiple data blocks.The conventional division method uniformly divides the entire flow field into several data blocks,but with the progress of the calculation tasks,the distribution of data particles in each data block tends to become uneven,which will cause a few calculation tasks to be concentrated in some processes,resulting in a large number of calculation processes being empty,which is not conducive to improving the calculation efficiency.In the case of a small amount of data,the imbalance between computing tasks is still in an acceptable range.Therefore,this paper designs and implements the partition algorithm based on RCB and the partition algorithm based on KD on GPU.During the calculation process,the flow field area is re divided to balance the task load between each area.(2)Limit cycle detection algorithm:there may be some closed areas in the flow field,which are called limit cycles.After entering these areas,the particles will move repeatedly in this area and cannot leave the area.There is no doubt that the existence of such regions in the flow field poses a challenge to the performance of the particle tracking algorithm.Therefore,the system designs and implements the detection and scheduling strategy of the limit cycle region,detects the limit cycle region in the flow field and schedules the particles entering it separately.(3)Visualization system:by expanding paraview plug-in function and integrating the above two modules,this paper realizes the architecture design of visualization software for large-scale flow field.