Comparative performance study of two simple soot models for the prediction of soot level in atmospheric turbulent non-premixed flames

The increase of current fossil fuel consumption has led to an increase of soot emission into atmosphere.  Accurate prediction of soot production and destruction in a combustion system is not only important for the purpose of the design of the system, but also vital for the operation of the combustor. Numerous soot models have been proposed to predict the soot production and destruction in a flame, categorized as empirical, semi-empirical and detailed soot models.  Although the detailed model represents the highest level of soot modelling, its use has been impaired by substantial requirement of resources of computer and time. Therefore, empirical and semi-empirical approaches still have their position in soot modelling of practical combustors. In this study, two soot models, single-step and two-step models are examined in the simulation of atmospheric turbulent non-premixed sooting flames. The soot models are compared and evaluated for their performance in predicting soot level in methane and ethylene non-premixed flames.  The commercial software Fluent 6.3 was used to perform the calculations of flow and mixing fields, combustion and soot. Standard k-ε and eddy dissipation models were selected as solvers for the representation of the turbulence and combustion, respectively.  The two soot models used in the study are available directly from the code for evaluation. The results show that the two-step model clearly performed far better than the single-step model in predicting the soot level in both methane and ethylene non-premixed flames. With a slight modification in the constant a of the soot formation equation, the two-step model was capable of producing prediction of soot level closer to experimental data.  In contrast, the single-soot model produced very poor results, leading to a significant under-prediction of soot levels in both flames.