Prediction of Temperature Field and Pollutants in a Turbulent, Non-Premixed CO/H2/N2 Flame

The applicability of the laminar Flamelet concept for the prediction of temperature and mass fractions of important pollutant species such as NO and CO in a turbulent CO/H2/N2 jet flame has been studied, using Artificial Neural Networks (ANN). In the first step, by means of the solution of counter-flow diffusion flames, OPPDIF, temperature and species concentrations have been calculated in different flame strain rates. The results of this step are related to the mixture fraction and scalar dissipation rate. Then, turbulent fluctuations were applied to the calculated profiles through numerical integration with presumed shape probability density functions. Ultimately, a Flamelet library was created. In order to interpolate in this library, two artificial neural networks were built for the mean species mass fractions, and temperature, respectively. The simulations done in this research revealed a drastic decrease in computational time of ANN approach in comparison with the traditional Flamelet library computations. Aclose relationship between the accuracy of NO and temperature predictions was observed, as the thermal NOx plays an important role in the total NOx pollution in this flame.

کلیدواژه ها:

Turbulent Combustion ، Flamelet Model ، NOx ، Artificial Neural Network ، Non-premixed Flame