[Cloudseeker]

Hello everybody

I am trying to simulate the separation rate of an axial cyclone.
The particles are sand grains with a diameter of 0.1 mm, liquid is water.

The cyclone is a double walled pipe with an inner diameter of 0.08 m and an outer diameter of 0.12 m. The liquid enters with a spin. 5% of the overall massflow (80l/min) are separated in the outer 3.25 mm of the pipe. Particle mass flow is 100 g/s.

There was an actual experiment. The separation rate is about 15%. It was done with two different lengths (around 1200mm and 1800mm) of the cyclone and the difference in the separation rate was minimal. In fact the shorter cyclone had a slightly higher separation rate. I think this comes from the reduced circumferential speed due to friction and the resulting lower centrifugal force. Which means that after a minimum length only the turbulence near the wall is important.

I simulated the cyclone and got a separation rate of 96%. That's far away from the experimental results.
https://dl.dropboxusercontent.com/u/16679073/Rohr.cfx
https://dl.dropboxusercontent.com/u/...3/Rohr_001.res

I used Non-Drag Forces: Turbulent Dispersion Force and the Pressure Gradient Force.
I tried different turbulence models: k-e, SST, BSL, SSG, each with different meshes for the right y+.
I tried the Sommerfeld Collision Model, but because of the low particle load it has nearly no effect.
I also tried Sommerfeld Rough Wall Model, but also not much of an effect.



I exported a few particle paths for better visualization. You can see, that most of the particles are distributed close to the outer wall.

Did i miss some important settings?

Thanks


PS: Sorry for the bad English.

[ghorrocks]

So let's make a big assumption that your flow field is correct.

Then I can think of two issues: You are over estimating the amount of particles which leave the domain, or in the real experiment once particles leave the domain they come back in again.

I would:
* Check the location where the particles go after the cyclone is correct. You may need to model this region of the flow more accurately.
* Check that your particle model is accurate. Is there some form force which keeps particles away from the outlet?

[Cloudseeker]

I modeled the region where the particles leave the cyclone and also the spin generator. But I am not allowed to show this parts, but the problem there is the same as in this simple case.
The experiment should be fine. The separated flow leaves the cyclone in direction of the velocity and enters a separate tank, so there should be no problem with reentering particles. The rest goes back to the central tank. The period is short enough to ensure no particles can go through the cyclone a second time.
We did velocity measurements in the straight part of the cyclone. The axial velocity distribution matches pretty well. The circumferential velocity is in the simulation a bit smaller but the distribution looks similar.

The pipe is made of plastic. Could there be a force due to electrostatic charge? But wouldn't that pull the particles to the wall?

[ghorrocks]

Yes, electrostatic might play a part. Also some particle flows have repulsion forces near walls. And there are Van Der Waals forces, Brownian motion and many others. I would do a literature search and find out what is relevant to your flow.