[DPM problem] infinite particles loop in domain (twin cyclone - double cyclone))
 

Good morning all,
I'm simulating a double cyclone system (twin cyclone), to separate a stream of particles (ash) from a fluegas.
The model consists of about 2.300.000 cells. Model is steady k-epsilon RNG with swirl dominated flow, no energy equation, incompressible gas. After about 4.000 iterations I stopped the simulations, residuals were all constant and under 10^-3 except continuity which was around 1*10^-2; however inlet and outlet mass flows were identical, and velocity and pressure profiles seem to be ok.

After that I enabled the DPM model to track my particles (steady) and to see the global efficiency.
So I set up multiple surface injections (at inlet) (I didn't use the Rosin-Rammler distribution) for each size of particles (I tried 5 - 15 - 82 and 2500 microns) and set the number of steps to an initial value of 200.000.
Outlet was setup as an escape boundary, dust collector (below cyclones) was setup as a trap boundary, all the other walls were setup as reflect (reflection condition --> normal -->constant=0,8, tangent-->constant=0,8).

However I saw a 100% incomplete path for all the injected sizes (except the 5 microns size), as you can see in the attached pictures (in picture I changed the injection type from surface to point).
I increased the number of steps till 5.000.000 but the results are the same: particles recirculate in the domain, without exiting or being trapped.
Pictures with tracked particles are all calculated with 1.000.000 steps.

What does this mean?It cannot be real an infinite loop of the particles.

Thank you for any replies

Daniele

Found my answer in article "NUMERICAL STUDY OF GAS-SOLID FLOW IN A CYCLONE SEPARATOR"

Hi Daniele
I am simulating a cement cyclone and faced with the same problem as you mentioned, that particles with diameter more than 50 micron keep spinning at a level of cyclone.
how can I reach a realistic result as the real cyclone in factory has 83 percent efficiency factor, but mine is less than 10% because most of the particles remain in cyclone.
Thanks for any reply and consideration.

Hi,
consider incomplete particles as trapped particles: as I wrote that particles will interact with other particles and they will probably spin down and collected.

thanks for reply
I got your note,but there is 200 pascal difference in droplet pressure of real cyclone and stimulated one that I guess is due to the thing I mentioned, How can I valid my stimulation?

I think this is more related to turbulence model; if you didn't do it, simulate with reynolds stress model (better), or k-epsilon RNG with swirl activated.

Hi Elham,
I am facing the same problem (particles remaining at a certain level), could you overcome this? How?

Thanks