[DomBadger]

Hello all,

I am studying a large scale swirl burner. As the first step, simulating the non-reacting swirling jet with high swirl level (Re ~ 1*10^6, swirl angle ~ 40 degree, swirl number ~ 0.8). A recirculation region is expected, as shown by many experimental studies.

I have tried standard ke, RNG (either with swirling dominated flow option selected or not), realizable ke, but none of them produced a satisfying result.

(1) standard ke converges well but no recirculation region is predicted.
(2) RNG ke shows the recirculation region but stays with high residuals. Seems no hope to get converged results.
(3) Realizable ke converges well. It also indicates some sort of birth of the recirculation region, but fails to successfully predict it.

I wonder how to deal with such a swirling jet. SST k-w ? Reynolds Stress turbulence model? Or use a transient solver? I am going to try these on relatively low density meshes but am afraid my trials will generate biased results due to mesh density. Therefore I really wish to learn from your experiences and suggestions.

Thanks a lot!

[DomBadger]

My swirling jet impinges into the atmosphere.

There are some researchers managed to predict high swirl flows for combustion burners, such as https://iieta.org/sites/default/file...HT/34.2_11.pdf

[LoGaL]

Can you show me the k-epsilon realizable results? What is unsatisfactory? Can we see velocity profiles?

Keep in mind that for high recirculation zones you eventually need LES or LES/RANS hybrids. Show me first what you get, I should be able to tell if there’s something wrong or it is reasonable for a RANS

[DomBadger]

Thank you Lorenzo for your help!

Here are velocity profiles and vector fields using realizable k-e, RNG swirl, standard k-e models and SST k-w model.

Sorry I need to cut out part of the velocity profile around the area of the air inlets, fuel inlet, and swirl blades due to IP reasons.

Basically I have a burner that includes air inlets (Fan, pressure jump B.C. in Fluent, blowing air into the burner), fuel inlets (currently set as Wall B.C. because I want to test non-reacting flows). Air and fuel are supposed to mix and go through swirl blades, then go out of the burner from a nozzle. In the velocity profiles you can only see the nozzle part. The B.C. for the plenum are set as: pressure inlets for four sides, pressure outlet for top, all at 0 gauge pressure.

Realizable k-e velocity profile.jpg

Realizable k-e velocity vector.jpg

RNG swirl k-e velocity profile.jpg

RNG swirl k-e velocity vector.jpg

[DomBadger]

Standard k-e velocity profile.jpg

Standard k-e velocity vector.jpg

SST k-w velocity profile.jpg

SST k-w velocity vector.jpg

[LoGaL]

Ok, RNG simulation is defnitely not converging.

Concerning the realizable one, It seems reasonable, but I don't see any comparison with the data you are trying to match, I guess due to IP reasons.

So I will post you a crappy picture of what I expect from the realizable model, compared to some experimental velocity profile I invented. If you get something like this, in which recirculation is somehow underpredicted, you should be fine.

Remember, high reciruclation in combustion chambers (which is alwais there to stabilize the flame position) = LES or LES/RANS hybrid. No way to escape