Adaptação do modelo de combustão turbulenta do reator parcialmente agitado em simulação multidimensional de um motor diesel

Abstract

The RANS numerical models, due to its low computational cost, are the most commonly used in Computational Fluid Dynamics simulations. However, in the analysis of reactive turbulent flows, the use of this approach must be linked to a method capable of determining the concentration variation rates of the chemical species present in the combustion process. Turbulent combustion models arise as a way to determine the values of such rates, however, their implementation must be validated with experimental data. This study proposes the analysis of the effects of the application of an adapted version of the turbulent combustion model PaSR in multidimensional simulations of a diesel engine. The adaptations seek to obtain greater control over the influence of turbulence in the progress of the chemical reactions. The results should be compared to those obtained in a simulation using a model that neglects the effects of turbulence on chemical kinetics. In addition, a micro-genetic algorithm, developed in PYTHON, is used to generate simulations with variable input parameters, in order to adjust the values of two terms present in the adapted model. The same algorithm is also used as an optimization tool and should have its ability to perform such an evaluated task. The results show that the implemented model was able to describe the premixed combustion phase in a way close to the experimental data, slowing down the energy release rates in that period and obtaining a peak only 13% greater than the experimental one. However, in the diffusive phase, which shows less influence of the turbulence on the progress of the reactions, the results obtained by the two models were similar, both sub-represent the experimental energy release, the implemented micro-algorithm was able to obtain satisfactory results and showed to function as expected. However, the operation of the algorithm did not characterize an optimization process, since the best result obtained appeared at the beginning of the process and it was not possible to perform a greater number of simulations.