The confined TECFLAM swirling natural gas burner
Application Challenge 2-06 © copyright ERCOFTAC 2004
Comparison of Test data and CFD
In order to yield a general quantitative characterization of the flames, radial profiles of the mean values and rms fluctuations of the temperature, concentrations, and mixture fractions have been extracted from the PDF's at 8 different heights. These profiles reflect, for example, the position and downstream development of the mixing zone, the turbulence intensity, and the overall temperature level. Calculated radial profiles are displayed together with experimental results for the velocity in figure 14. Figure 15 shows radial profiles for the turbulent kinetic energy and the mixture fraction. The mass fraction of CO and CO2 are displayed in figure 16.
The two recirculation zones of the flames, i.e. the inner one near the flame axis and the burner mouth and the outer one which reaches from the flame region to the burner walls, can be clearly distinguished in the profiles. The inner one exhibits temperatures close to 2000 K and near stoichiometric burnt mixtures, whereas the outer one contains exhaust gas with temperatures around 1200 K and with a mixture fraction value that corresponds to the air/fuel ratio of 1.2.
Figure 14: Calculated velocity profiles (U- and W-Velocity)
Figure 15: Calculated profiles for the kinetic energy and the mixture fraction
Figure 16: Calculated profiles for the CO2 and CO mass fraction
A deeper insight into the turbulence-chemistry interaction and the thermochemical state of the flame was gained from the correlations between various quantities. The scatter plots of temperature vs. mixture fraction revealed, for instance, the coexistence of unreacted fuel and oxidizer, even for stoichiometric mixtures in regions of intensive mixing. Furthermore, the temperature reduction due radiation, flame stretch, and wall contact could be quantified from these correlations. For a swirl number of 1.8 the average temperature is 100 - 200 K lower than in flames with lower swirl numbers. This effect is explained by a faster mixing and the small influence of higher strain rates.
The comparison for the scatter plots of temperature and mixture fraction was carried out for 4 particular regions: I - the central recirculation zone; II - the mixing zone of the fuel; III - the mixing zone of the air; IV - the outer recirculation zone. The results are displayed in figure 17 for a distance of 10 mm from the burner exit.
Figure 17: Scatter plots for measured and calculated temperatures at h = 10 mm
From these plots the interaction of the turbulence and the chemistry in the regions I and II gets obvious. Especially in region II ignitable mixtures are apparent due to fast mixing. In region I the simulation show comparable high temperatures which means that heat losses (due to radiation or convection) can be neglected and adiabatic boundary conditions are adequate to simulate the central part of the flame. Further results at a distance of 40 mm from the burner exit are shown in figure 18.
Figure 18: Scatter plots for measured and calculated temperatures at h = 40 mm
These plots show that at a distance of 40 mm from the nozzle, the mixture fraction is concentrated in a more narrow band compared to a distance of 10 mm. In particular in region II the temperature at h = 40 mm is higher than for h = 10 mm. In region IV the predicted temperatures are reasonable higher than the experimental data, which is due to significant heat losses due to convection and radiation in this region.
For the calculations of EKT Darmstadt it can be concluded that:
• The flow field is well predicted up to axial positions of 30 mm.
• For axial positions further downstream (h > 70 mm) the spreading of the central recirculation zone is over estimated.
• Computed distributions of major species as well as their variances are in reasonable agreement to the experimental values.
• CO representing a minor species sensitive to finite chemistry effects is predicted fairly well using an ILDM which is spanned by two progress variables.
• The retroactive effect of chemistry model on flow field is marginal.
• There are differences observed between the presumed b-PDF and the Monte Carlo PDF method but these are not crucial.
© copyright ERCOFTAC 2004
Contributors: Stefan Hohmann - MTU Aero Engines