Parallelization of the Fluid Behavior Modeling Algorithm in Real Time

Abstract

A parallel algorithm based on Compute Unified Device Architecture (CUDA) technology is proposed to accelerate fluid behavior simulation and real-time decision making capability. Three main steps were highlighted: implementation of the fluid flow simulation method, distri-bution of work between CUDA threads, and collection of results. A software product was de-veloped to analyze the obtained results. As a result, it was found that the minimum acceptable refresh rate of the simulation environment is achieved on an environment with a size of 512  512 and is an average of 51.54 FPS (number of frames per second) for both states (quiet and ac-tive simulation). An analysis of literary sources was carried out, where the current state of this scientific problem is outlined and the advantages of the proposed approach are indicated. Among the simulation methods, the method using the Navier―Strokes equation for the flow of incompressible matter was chosen because it is simple and has good possibilities for parallelization.