Hello,

i´m simulating a pulverized fuel cyclone burner and I am experimenting a lot of problems, mainly related to convergence.

I have just begun with air flow and using these models and options:

-Hexaedra mesh with about 300000 elements. I have put cyclone outlet and vent outlet into the same discharge box, which represents a limited zone of the boiler.

-Boundary conditions: velocity inlet (10 m/s) in the duct entering the cyclone and atmospheric pressure outlet in the surface exiting the boiler.

-RSM model. Quadratic pressure strain (I´m not sure about the differences between this, linear and low-re), Wall BC conditions from k equation

and non-equilibrium wall functions

-No energy equation

-Presto for pressure and quick for the others.

-Steady solver

The problem is that the solution diverge after approximately 1000 iterations. I´m using very small URF, even until 0,001 for momentum, k, eps and stress

Can you give any advice? and references where I can find strategies for cyclone burners modelling?

I´m also dealing with a cyclone problem. I suggest you try first using k-e until it reach convergence. Then turn on RSM using small URF

Get hold of "Best practice advice for AC3-03 cyclone separator" (2003) from qnet-cfd or FLUENT

I think you can try to solve this question with steady solver and unsteady solver. According to my experience, you can obtain good results by this means. Hope this can help you. Good luck.

With regards to the RSM model try it without the Quadratic pressure strain model. It will mean less accurate results but ive found that even with residuals at less than 10-4 on 2nd order RSM before switching on the Quadratic pressure strain model the residuals will explode due to it. Seeing as your inlet velocity is only 10m/s you should not really need it anyway as the Quadratic pressure strain option is there to better resolve large amounts of shear in the flow caused by intense swirl resultant of large inlet velocities.

Hope this helps

Neil

Thank you very much Neil. I will try your suggest.

Actually I good like to make some more cases varying inlet velocity, even until 30 m/s. When should I use quadratic pressure strain?

Do you mean that I should switch to unsteady when I get convergence in steady results? Actually I don´t get convergence in steady solver.

Ideally you should use the quadratic pressure strain model for all the inlet velocities you use to gain maximum accuracy in the results. But I have found that it is a trade off between trying to get a more accurate solution that won't converge or a solution that converges without the added accuracy. The RSM model alone though will provide accurate results in relation to capturing the W tangential velocity profile as opposed to the V shaped tangential profile obtained with the k-e RNG model. By all means try to get convergence with the quadratic pressure strain model activated its just that I can never get convergence when I run it. I would suggest though to increase accuracy use the QUICK discretisation scheme for turbulent quantities and for the Reynolds stresses, also for species aswell. This will minimise the effects of 'false numerical diffusion' during iterations increasing accuracy. With regards to the unsteady solver I would only switch to this once you have a steady state converged solution. I'm assuming your only dealing with a single precessing vortex core? There are a few papers on the 'tornado' combustor which uses a co-axial bi-directional vortex structure which provide good modelling hints for cyclonic flows, worth checking out.

Neil

In my case, I´m not dealing with a cyclone combustor. It´s just a coal cyclone burner working in an arc-fired anthracite boiler, which has 48 cyclones.

At the moment, I´m not interested in combustion. I just want to evaluate the impact of a retrofitting, mainly in the primary air to coal relation leaving the cyclone at the underflow

Do you know where could I find information related to cyclone working as a burner, not as a particle separating device? I have information about the last one, but I´m not sure that the air flow in my burner should look like the ones of the cyclone working as a particle separating device

Sorry for my english!!

Probably one of the best places to find in depth information would be to use scopus.com to find papers on the subject, which there will be alot of. Apart from that with regards to the fundamental flow physics I would suggest a couple of books being:

-Swirl Flows by A.K.Gupta, D.G.Lilley and N.Syred (Abacus Press)

-Combustion Aerodynamics by J.M.Beer and N.A.Chigier

Alot of the material in these books brings together material from papers published in the international combustion symposiums. Although these are not available on the internet papers from the combustion institute are and you should be directed to these when using Scopus.com.

Hope this is of use

Neil