[User12]

Hello everybody,


I am simulating a device to destroy a stream formed by NH3/CH4/H2O generating as low NOx as possible. The strategy is to burn an auxiliary CH4 stream in a quarl upstream the point where the NH3/CH4/H2O stream is injected in order to increase temperature and let NH3/CH4 react in a hot non-oxygen containing environment, producing mainly N2, CO and H2. This would be a sort of thermal decomposition/partial oxidation of NH3. I need to simulate the full device, i.e. the first part in which CH4 is burned with air and the rest in which the hot mixture of gases continues reacting. The aim is to directionally study the effect of different design parameters (for instance, location of the NH3/CH4/H2O injection) in the NH3 conversion and NOx generation. My simulation is a 3D geometry with rotational periodic boundary condition.


I chose EDC to model the turbulence/chemistry interaction, since it makes it possible to account for the finite rate chemistry needed for the hot gas mixture with the use of a detailed kinetic mechanism. However, I am getting rather unphysical results concerning the formation of the CH4/Air flame, which does not appear in the quarl, but downstream in the chamber designed for the reaction of the hot gas mixture. Mixture fraction-based models (which I could use by just considering the auxiliary CH4 and air streams) predict the flame to appear in the quarl, as I expected. I carried out the same comparison (EDC vs mixture fraction models) with a periodic 3D geometry of the BERL combustor, which is similar to the device I am trying to model, and I get the flame far from the mixing point too when I use EDC (pictures attached). I checked that this is not an issue with the kinetic mechanism I am using by running the cases with GRI 3.0, obtaining the same results.


Therefore, I strongly doubt about the convenience of using EDC for my problem. Does anybody have experience with the use of EDC for this kind of problems? Can anybody suggest a more convenient approach to solve it? Any comment is really appreciated.


Many thanks!


Diego

[User12]

Any suggestion? If we forget about my particular case, anybody has experience in simulating the BERL combustor with EDC? Which is the reason for the difference between EDC and, for instance, steady flamelet model?


Many thanks