CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   a hydrocyclone modeling problem (

tom_smith_2012 April 24, 2012 21:52

a hydrocyclone modeling problem
2 Attachment(s)
Hi guys:
I'm recently doing a hydrocyclone simulation with CFX12 and having some problems, I hope I can get some sugguestions or idea here.
This is a solid-liquid seperator with tow tangential inlets, I meshed it with structure grids with ICEM, the quality seems to be OK.
In the CFX-Pre, two phase massflow rate and volume fraction was assinged dependtenly, 0 static pressure was assigned on the outlets(overlet and underlet) , the fluid turbulence model was selected as SSG RSM, but covengence problems occured.
The massage window showed that:A wall has been placed at portions of an outlet..... I did as it recommended to swith the outet as openning, but it didn't work, even to single phase simulation. I also tried with K-E, RNG K-E, and BSL RSM model, all could converge well.
So, I want to know, if someone has the similar problems and how can I sovle the problems. By the way, In the simulation, with other converging turbulence models(k-e, RNG K-E and BSL-rsm), I didn't capture the aircore, so how can i get in, is any special setup needed in the CFX-Pre?
Any suggestions or idea is appreciated!:)

ghorrocks April 25, 2012 07:53


A wall has been placed at portions of an outlet
You are getting back flow at the outlet. If this is not real then you have an error in your setup. Have a look at it in the post processor to see if you can understand what is causing the problem. And probably the easiest fix is to move the outlet further down stream.


k-e, RNG K-E and BSL-rsm
You won't get an air core with k-e, RNG k-e or similar turb models. They do not work well for rotating flows. RSM or LES will be required to capture it, but as you have found RSM is harder to converge. Your mesh which is adequate quality for k-e based turbulence models is not good enough for RSM, you will need to improve mesh quality further.

tom_smith_2012 April 25, 2012 20:19

thank you very much for your idea, I have lenthened the outlet as someone suggested, but it didn't work, even for singlephase flow.After sveral steps of calculation, the residual go down sharply below1e^-6 and the computation stopped "normally". So I don't understand why did this happened.
My cyclone has two tangential inlets, so the mesh of the tangential area seems to be OK,but the angle check is poor. So I want to know if this influenced the RSM model calculation.

lffabiani April 25, 2012 22:52

I have seen a number of hydrociclones simulated with SST + curvature correction with good results

Ou local Ansys representative once advised to extend not only the outlet but the inlets as well.

Hope it helps

Best regards

ghorrocks April 26, 2012 06:08

Good point Luis, the SST model with curvature correction may help here. You may be able to get that to run without all the hassle of RSM.

Yes, any poor quality elements will give RSM a hard time. You need an excellent quality mesh for reliable RSM convergence.

Also have a look at the flow with the reverse flow are the outlet in the post processor. You might be able to see where the reverse flow is coming from, and whether it is real.

tom_smith_2012 April 26, 2012 20:11

Hi,ghorrocks and lffabiani
Thank you for your attention for my problem.
Maybe ghorrocks are right. After I do some check and modification in CFX-Pre, some model seems to be OK and could be computed,but some models still couldn't. From OK models, the flow field and preesure is not so good as many orther simulation of cyclone, no air core was obtained.
I got the massflow rate of the solid phase from the two outlets, it didn't equal to that from the inlets. For small particles, it seems to be OK, but the outlet massflow rate deviates seriously for largger particles, the result is incredible completely.
I am having a headache about it these days. But thank you agian for your ideas and I will have a try!

All times are GMT -4. The time now is 23:28.