Application of transport PDF approach for modelling MILD combustion

A transport PDF (TPDF) approach was used for modelling turbulent non-premixed Methane/Hydrogen (1/1 by volume) flames issuing from a jet in hot, oxygen-diluted coflow. A comparison between the TPDF results and detailed experimental data of ξ, temperature, major species and OH is presented for flames with different oxygen level in the hot coflow (oxygen mass-fraction of 6%, and 3%). Results from a previous study using the eddy-dissipation concept (EDC) solver, is also presented. A comparison in performance between the TPDF and EDC models is presented. Introduction In moderate and intense low-oxygen dilution (MILD) Combustion [2,17], fuel is mixed with highly diluted and heated air to create a distributed reaction zone with a reduced peak temperature. Attractive features of these flames include a semiuniform temperature field, higher radiation flux and low emission of pollutants. Whilst MILD combustion can be classified as nonpremixed jet flames, studies of diffusion flames cannot be directly extended to characterise MILD combustion. This is because most studies of non-premixed flames are conducted in cold air surroundings. Although the concept of MILD combustion has been extensively studied experimentally [5,6,10,11], mathematical modelling of this regime has received relatively little attention [4,7,9]. At first glance this regime seems relatively straightforward to model as it does not feature highdensity gradients and complex turbulence-chemistry interactions processes, which are prominent in conventional turbulent jet flames. However, the conditions of elevated and uniform temperature distribution and low oxygen concentration in MILD regime, lead to slower reaction rates and enhances the influence of molecular diffusion on flame characteristics. These two effects in particular challenge the applicability of simple combustion models that assume fast chemistry and neglect the effects of differential diffusion. Motivation In a previous study [3] the authors used Reynolds-Averaging Navier-Stokes (RANS) approach to model the flow, compositions and temperature fields of a fuel jet issuing into a hot, oxygen-diluted coflow. That study focused on modelling Jet in Hot Coflow (JHC) flames measured by Dally et al. [5], and examined the effects of various combustion and turbulence models, chemical kinetics mechanisms, thermal radiation and differential diffusion, on the accuracy of the predictions. It was shown [3] that the standard k-ε turbulence model with a modified dissipation equation constant (Cε1) provided the best agreement with the experiment. Differential diffusion effects were found to have a strong influence on the accuracy of the predictions and therefore should always be accounted for. It was also found that conserved scalar based models, i.e. the ξ/PDF and flamelet models, are inadequate for modelling jet in hot coflow (JHC) flames. The representation of the chemistry in the model was also found to play an important role in accurately predicting flame characteristics. Using detailed chemical kinetics , rather than global or skeletal mechanisms, with the eddy-dissipation concept (EDC) model was found to improve the accuracy significantly. In general, the EDC model performed reasonably well for flames with higher O2 concentration in the hot coflow, such as flames with 9% O2 and 6% O2. The agreement with the measurements however was poor for the 3% O2 case. However, the largest discrepancy was noted at the 120mm axial location where the model did not perform well, particularly for 3% and 6% cases. This is due to the intermittent localised flame extinction that the EDC model could not capture. This paper is an extension of our previous study [3] and focuses on examining the performance of transport PDF (probability density function) approach in modelling JHC flames.