Comparison of the Numerical Models for the Temperature Distributions of Non-premixed Swirling Methane Flame


TUNÇ G., YILMAZ İ.

JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI, cilt.22, sa.4, ss.819-826, 2019 (ESCI İndekslerine Giren Dergi) identifier

  • Cilt numarası: 22 Konu: 4
  • Basım Tarihi: 2019
  • Doi Numarası: 10.2339/politeknik.448529
  • Dergi Adı: JOURNAL OF POLYTECHNIC-POLITEKNIK DERGISI
  • Sayfa Sayıları: ss.819-826

Özet

This study aimed to investigate the numerical modelling parameters that are in good agreement with experimental data for the temperature distribution of a non-premixed swirling methane flame. All numerical calculations have been performed with FLUENT, a computational fluid dynamics code. P-1 radiation model has been chosen for all numerical calculations. In addition, the swirl number has been taken 0.4 value to validate the results with respect to reference experimental data. All comparisons have been performed in axial and radial temperature distributions according to experimental data. Firstly, the number of swirls has been defined as a user-defined function. Thus, the effect of defining user-defined functions has been examined in the swirl number. Secondly, the model constant (A) of the eddy dissipation combustion model has been investigated to determine suitable value. After that, the eddy dissipation and PDF mixture fraction combustion models have been compared with each other. Finally, the k-epsilon standard, realizable and RNG models have been analyzed to determine the proper turbulence model. The results showed that to define the swirl number as a user-defined function in the comparison tests has not been an important effect to obtain good agreement with the experimental temperature distribution data. It has been found that the value of one for the Eddy dissipation model constant is suitable for this model. It has also been found that the experimental results in both combustion models give approximate results, but the PDF model is particularly better at axial temperature distribution. Moreover, it has been seen that the k-epsilon realizable turbulence model is more suitable for this model.