I am simulating natural gas combustion in ANSYS Fluent. If i use the mixture fraction combustion model, the outlet temperature of the boiler is overpredicted according to the experimental datas. However, if i use the eddy dissipation model under the same boundary conditions, the outlet temperature of the boiler is well predicted, despite the much higher flame core temperature obtained, which overpredicts the NOx emissions. Anybody who can give some tips are appreciated. Thank you
There could be several possible reason. The discrepancy may be due to the boundary conditions for temperature, or due to the chemical mechanism.
Additionally, I think the the combustion mode could be partially premixed mode so that the simple flamelet model may not be sufficient to predict the flame properties.
We made a similar experience a few years back when comparing simulations of a gas-fired boiler and its heat exchanger. We used PDF-equilibrium as a mixture-fraction-based combustion model, and a finite-rate/eddy dissipation model approach in combination with a 2-step reaction mechanism as a species transport model. At the time, we were told that the coupling of the combustion model and the energy equation is stronger in the case of species transport models than with mixture fraction-based models but I had thought that ANSYS has since fixed this. We regularly do comparisons of measurements and CFD simulations and usually, flue gas temperatures agree quite well with measured data.
How much do you trust the measurements? Was radiation taken into account?
Thanks for your responses. I know that our measurements are accurete enaugh for the experiments. The radiation heat transfer were taken into account for both combustion models. The eddy dissipation model, for my case, calculates more radiation heat transfer rate to boiler walls than PDF flamelet model, since the flame temperature values are calculated much higher than PDF model does. I think, this is as to why there is a discrepancy for flue gas temperatures.
I was actually thinking about whether a radiation correction had been carried out for the temperature measurements. Depending on the location of the measurement probe and the overall temperature levels, the differences between corrected and uncorrected temperature measurements can be substantial.
Concerning the differences between the combustion models: in theory, the maximum temperatures predicted by mixture-fraction-based models should be lower and closer to reality than those of eddy dissipation models as the reduced reaction mechanisms are usually unable to take dissociation into account. So, explaining a better performance of eddy dissipation models in terms of heat loss by their higher (and somewhat unrealistic) max temperatures doesn't really work in my opinion.
- Badges
- Science topic