Modelagem Computacional de Fluxo em Aplicador de Herbicida Amparada por Simulação com Ênfase em Exploração de Paralelismo

Abstract

The present research analyzes the functioning of the selective herbicide applicator Campo Limpo, which promotes the control of weeds in pastures by direct contact with the applicators soaked with a herbicide syrup. In order to better understand its operation, with a focus on flow behavior, to improve the accuracy of its application, computational experiments were performed using the simulation with the construction of the background mesh, simulation mesh, mesh refinement, initial condition settings and simulation parameters. It was also necessary to develop series of laboratory experiments with an apparatus close to the reality of the operation of the applicator implement. The simulation experiment in different computational environments was conducted in order to execute behaviors in the laboratory. Some field experiments were carried out to enable cross-validation studies of the proposed models, contrasting in a laboratory with a real Campo Limpo scale. The construction of a geometry model for use in the construction of mesh and, later, resolution of calculations requires high performance computing solutions. The application of parallelism techniques dividing of computational work in different processing units provided an increase in speed of comuputacional environments processing by up to 4 times and increased simulation complexity of simulation mesh from 7-sided to 37-sided scenario with more than 400 thousand cells. A performance evaluation made it possible to compare computational environments and indicated an architecture based on use of distibuted computing with a reduction of processing time in approximatly 40%, with special attention to parallelism techniques, seeking a resolution of case studies with time restrictions. The different case studies point out possible ways to improve the construction metrics and operational conditions of Campo Limpo based on the data obtained in experiments and simulations.