Flame Structure

In this section, the flame structure of all the three flames is discussed. The influence of several parameters on the results will be examined in the next section. Figure 5.4 shows instantaneous snapshots of temperature for the three flames. They are taken from the 16 field, 15-step chemistry mechanism in the fine grid simulation.

Figure 5.4: Temperature snapshots of F1 (left), F2 (middle) and F3 (right) flames. The F1 flame exhibits thin and interrupted high temperature contours while the F3 flame exhibits an uninterrupted high temperature structure.

The differences in the flame structure among the three flames are apparent. Contrary to the low Reynolds flame (F3), the F1 flame exhibits thin and interrupted high temperature contours on the axial slice. The F3 flame on the other hand exhibits thick and uninterrupted high temperature structure in the axial slice. The above observation suggests the proximity of the F1 flame to the distributed reaction zones regime (see Figure 5.2). In Figure 5.5, two temperature iso-surfaces for the F1 flame are shown. These surfaces are coloured with instantaneous OH values (left image) and mean CO₂ values (right image). The “holes” in the flame surface of the two images are apparent (blue regime on the CO₂ plot and green regime in the OH plot). The interruption of the reaction is probably due to large scale mixing, suggesting that the F1 flame is closer to the broken reaction regimes than the conventional distributed reaction regime (as observed for example in [1, 4]) where intense burning is still present.

Figure 5.6 shows the curve-fitted mean flame-front position for the three flames. Based on experimental measurements [15], tabulated data of the temperature limits for the flame front

Figure 5.5: Temperature iso-surface (T = 1500 K) with instantaneous OH (right) and CO₂ values (left) coloured on the surface of the F1 flame.

at different axial positions are obtained. In all three flames, in the radial station (close to the nozzle exit), the flames extend radially to the edge of the outer nozzle (6 mm in the radial direction). Further downstream, the flames burn inward and the nozzle exit velocity affects the radial position of the flame. At a given axial distance from the nozzle, the radial distance of the flame is larger in the F1 flame, which has the highest nozzle exit velocity. The figure shows a very good agreement between the LES-PDF simulated flame front and the experimental data, especially in the first radial stations of flames F1 and F2.

The stochastic fields method allows to easily extract the instantaneous sub-grid PDF. In figure 5.7, the CO₂ mass fraction marginal PDFs are shown for the three flames at the flame front region at two axial locations. The histograms were obtained as follows: A specific location at the shear layer where the maximum temperature gradients were observed was selected. A total of 10 cells in this region was selected and for each one of the cells, the data from each stochastic field was extracted. All the instantaneous field values that were extracted were

“binned” and the histgrams were created. Despite the fact that this is only an instantaneous plot of a modelled PDF, it can give a qualitative indication of the flame regime. At the first axial

Figure 5.6: Curve fitted mean flame front position (c = 0.5) of the three premixed flames. The dots denote experimental values [15]. The flame front is defined as the iso-line of the mean temperature of the temperature limits at each axial position [15].

position, in the F1 flame, most of the values are clustered around 0.04, indicating that the flame has large probability of burning away from equilibrium (equilibrium CO₂ mass fraction is 0.15 at stoichiometric conditions). The exact regime of the flame cannot be directly extracted from the PDF, as all sub-grid scales are modelled and the extracted regime will only be a “modelled”

one. However such distribution suggests that locally and instantaneously, the flame belongs to the distributed or broken reaction zones regime. The F2 flame shows approximately the same image in the first axial position, suggesting that even though the flame globally is not close to the distributed or broken reaction zone regime it can be so locally. Finally, the F3 flame in the first axial position shows a more uniform distribution. Regarding the second axial position, it is observed that as we move from the F1 flame towards the F3 flame, the variance diminished and PDF becomes narrower, with the F2 and F3 flames in the burn side of the flame.

Figure 5.8 shows the scatter plot of the OH mass fractions as a function of the temperature, using an analysis similar to Duwig and Fuchs [26]. The results show that the data follow an exponential distribution with very low values of OH mass fraction at low temperatures and a rapid increase around 1650 K. As we move further downstream from the nozzle exit, the data points are highly scattered, indicating the effect of turbulence and entrainment of cold

Figure 5.7: Instantaneous sub-grid PDF CO₂ mass fraction expressed as a probability distri-bution histogram in the mean flame front location (location of steep temperature gradients) at two axial positions for the F1–F3 flames.

gases upon the reacting layer. Similar conclusions were presented in [53], where LES simulation results of a piloted lean premixed jet flame were compared to experimental measurements. At the same axial station, the scattering is broader in the F1 flame (with higher OH mass fraction values) with large deviations from the 1D laminar flame structure, suggesting the F1 flame is strongly affected by large scale turbulence and close to a broken reactions zone.

Figure 5.8: Scatter plot of OH mass fraction as a function of temperature at three axial positions. The solid line is a 1D Stoichiometric premix flame, the dotted line is the rich branch of a diffusion flame at a strain rate of 40s⁻¹.