Cyclone modelling
 

Hello everybody,

i have to modell and simulate a separation cyclone due to my studies. I am using fluent and gambit for my work.

Since i am familiar with the basics of gambit and fluent i am now searching for any kind of tutorial or help for this project.

I already searched this forum as well as google, but i find no specific help, just general descriptions.

Maybe someone here has good tutorials especially about the turbulence modelling and the boundary conditions and can send them per mail to me? (zweeper[at]gmx.net)


Kind regards,

zweeper

I have done a few industrial cyclone projects which can't be shared, but we found that the RSM model for turbulence gave good agreement with measured pressure drops for cyclones. You can find experimental data in many basic textbooks or on the internet. If you are trying to match volumetric flow data, why not use velocity (or a mass flow) inlet and outlets (with a small amount of flow for particle outlet). If you have pressure data, use pressure boundary conditions but it may need a nudge to get started.
I have used a range of single-diameter particles, in combination with a mass flow rate versus particle diameter distrubtion, to calculate an efficiency. Good luck with your project.

thanks for the tips, that was already a big help for me!

Hello Zweeper

Designing cyclone in GAMBit is easy. yes good meshing requires some skill. I have done some work on cyclone during my internships etc. At the beginning this following tutorial helped me to get started.

http://www.itc.polsl.pl/gwecel/Mns/Fluent_tutorial.pdf

I hope this will help u also. Good luck.

Mohsin

Hi Allan Walsh,

I have some issues in cyclone separator simulation with particle tracking. Could you help me?

With out particle loading, I got good results like pressure drop and flow pattern. Then I tried with particle loading. After some time solution was diverged due to more number of particles (more than 60000) inside the cyclone separator. I got turbulent viscosity limited to ... number of cells and simulation was diverged. How to avoid this.

Note: I am running transient case with RSM turbulence model.

Thanks Mohsin!
I already had this tutorial and it helped me a lot with the geometry creation in gambit. This part is done.

I'm just searching now for the right mesh and then i try some different turbulence models (with the expectation that the RNG is the most suitable for the cyclone).

What about a boundary layer in 3D, you think thats necessary? Boundary layers over the whole computational domain (internal cyclone walls) or just check the y+ value within fluent to get a rough range between 30 and 500?

You are welcome Zweeper.

I have never used boundary conditions in cyclone. In my case i got acceptable results without using boundary conditions (flow had good convergence, results were similar to what was expected). But yes i used FLUENT "adapt" feature to refine geomtery. Generally for cyclone my recommendations are as follows:

RNG model (swirl Dominiated flow)
standard wall functions
Second order discretization scheme for other parameters and for pressure: PRESTO scheme
For turbulence intensity: there is a formula: 0.16*(RE)^(-1/8).

Calculate it and get the results. Good luck.
Mohsin

good help thanks. if i dont get good results with my boundary layer mesh, i will try it without it.

any idea how to calculate an reynolds number for the inner part of the cyclone? the inlet part is no problem, i use the hydraulic diameter with the inlet velocity, but i dont know which i can take for the inner velocity... some problems there

Why do u need Reynolds number calculation in the inner part? I think u dont need.

After calculation u can do post processing and u can get Reynolds number at any inner area as velocity field would be there. and yes at the inlet hydraulic dia is taken for the calculation.

okay thanks, i will start now with my analysis and i think i will get good results now :)

okay i am having big trouble to get a structured mesh with gambit and an boundary layer to control the y+ value. i will post some screenshots tomorrow, hope someone can help me...

I am doing a project on hydrocyclone, want help in meshing.
Thanx

Hi,

I'm doing simulation of cyclone separator with different inlet tube angle. I've used steady-state solver with RSM model and inlet as velocity inlet, overflow at outlet and no-slip BC on wall. The problem is that after some iteration it got diverged and shows some error called Divergence detected in AMG solver: epsilon. I tried to fix the URF, but it does not work. I think its the meshing problem. I have used the hex mesh with 124314 nodes. Can anyone help me with the meshing??How could I fix the problem??

Hi
What is your cyclone dimension and velocity inlet value? I think your mesh cell number is low or maybe too low.
Regards

Thanks CFD-fellow for your reply.
Cyclone Body diameter 8.31 mm, Gas outlet diameter 2.24 mm, Gas inlet diameter 1.99 mm, Height of cylindrical part 26.13 mm and Height of conical part 18.42 mm. I'm using 40°, 30°, 45° and 60° inlet tube angle. Velocity inlet value is 20 m/s. I'm also using DPM for particle tracking.
How could I fix the mesh problem??

hi dear Allan Walsh
i m trying to model cyclone but got confuse for somthing.
i have some question if you can help me:
1.have we need to define 1 outlet or 2 outlet flow? i saw some model use define bottom B.C as wall and used dpm trap wall.
2.i found that RSM is better than other method with using simple for run as steady flow with out define any DMP at first. after converging we can define DMP with small time run with transient, is it true?
3. how structure meshing is important for meshing?
thank for help