Hi. I have used the non-premixed combustion model to simulate coal combustion in PF burner. The assumption of the chemical equillibrium was acceptable for my case, but I had other problems due to the 2 mixture fractions restriction (I can give more information on this). Therefore I would like to upgrade the combustion model I use, to EDC or PDF transport, but I am not quite sure, which are their main differences and which model is more suitable for my case (coal combustion).

Could anybody with experience on these models give me some guidelines?

PS. The computational cost of the two mixture fraction approach (non-premixed model) is already the most cost I can afford. So please relate it, if possible, to the cost of the other models (for a certain number of species for coal combustion)

EDC and PDF Transport are models to include *finite-rate* kinetics into CFD. In contrast, the Non-premixed model assumes chemical equilibrium (the Steady Laminar Flamelet model is very close to equilibrium).

So, first, you need a good kinetic mechanism. Second, the cost of EDC is about an order of magnitude greater than the 2 mixture fraction model, and TPDF is a least an order of magnitude greater than EDC!

Good luck!

If EDC costs an order of a magnitude more than the 2 mixture fraction approach does, the whole world's luck won't be enough for me! As it is getting more clear to me, these models can be practically used for parallel processing (for 3d domains), because the computation times will be otherwise prohibitive.

Could you please clarify something? The most expensive part of my (furnace) simulations is when I inject a huge number of coal particles to be combusted into the 3d domain. Is also the time of the DPM-iteration going to be increased by an order of a magnitude, or where you referring to the continous phase iterations?

Thank you for your time.

Regards, Ilias

No: DPM is independent of the gas-phase kinetics. You should only use EDC if you are interested in finite-rate kinetic effects. What are you trying to model?

I have already modelled a pulverized coal boiler (only the furnace) using the non-premixed combustion model (chemical equillibrium everywhere) and now I want to upgrade my model by incorporating non equillibrium effects in the combustion as well.

In the boiler domain I insert fuel, oxidiser and recirculating gas. The fuel (coal) is the discrete phase matter, therefore it is injected as particles. The interaction of the discrete phase with the continuous phase, takes place every 10 iterations. So the 9 iterations do not take into consideration the interaction of the 2 phases (these iterations do not cost much) and the 10th iteration calculates this interaction (very very expensive when the number of injected particles is huge!!!).

So I would like to know, if the transition of the non premixed (2 mixture fraction) model to the EDC model will also increase by an order of a magnitude the computational cost of this, 10th iteration (interaction between the discrete and continuous phase) as well as you said it will to the first 9 iterations (where the coal fuel is considered as 'gas').

Imagine that in a case, the cost of 1 of the 9 iterations was 20 minutes and the cost of the 10th iteration (with the interaction between the 2 phases) was about 8 hours!!! So if the EDC model won't increase the 8 hour iteration to 80h (!!!), I could try to solve the problem (maybe with a 2 processor machine).

Thank you for your time.

Regards

Ilias

No, the EDC will not affect the DPM time (i.e. the 10th iteration time).

In that case the total computational cost will be very high but still feasible. I think that I could give it a shot after all.

Thank you very much cg, for your help.

Ilias