[arjun3020]

Hi all,

i am doing analysis of Cyclone separators using DPM, i have used following schemes,
1. k-w with swirl dominating flow.
2. velocity inlet and pressure outlet BC.

i run the analysis with air and after some iterations (after convergence of primary phase) i set injections using DPM, i have used R-R distribution, when i run the setup with injections then residuals goes suddenly up, but they dint come to converge smoothly, see the attachment.

also i set up monitors of pressure and mass flow rate but they dint remain constant.

also i am getting reversed flow when i set the injections.

please help me.

Thanks.

[doronzo]

hi

how many iterations are you using for the fluid phase per one iteration for the discrete phase? how concentrated is the initial mixture? how dense is the discrete phase? all these could be critical in analysis like that.

may be you also need to decrease the under-relaxation factors. how big is the time step?

regards

[arjun3020]

Hi,

Thank you for reply,

i have 10 iterations per DPM iteration.
i am doing steady state analysis.
and volume fraction is less than 10% as per DPM requirements.

[byronsmith]

Hai I think u should continue your simulations for more time and see whether the residuals are coming down. if not try changing the underrelaxation factors for convergence. the reversed flow that u have mentioned is it for particles??

[doronzo]

you have a strong increase of the residuals in few tens of iterations, starting from when you set injections, which probably means that the number of iterations used for the fluid and particles could be critical. are you using a segregated or coupled solver?

sometimes just one iteration on the particles may defeat the whole decreasing trend of the residuals due to iterations on the fluid. by the way, those residuals are for the fluid only.

also, have you the two-way coupling on? the coupling between the phases can be both physically and numerically very important

regards

[arjun3020]

hi,
i am using segregated solver,
10 iterations per DPM iteration.

when i run the case now i am getting pressure dorp of 200 Pa,
but if i find by experiments it should be 900 Pa.
where is the mistake?

also i have coupled two phases.
but minimum quality is 0.4
and maximum aspect ratio is 64.
what to do?

[doronzo]

you could try, as an extreme case to see what happens, with 1 interation per DPM.

also, why don't you set injections from the beginning? how long roughly is the expected flow (flow time, not run time)?

[Katja]

Hello Arjun,

I'm facing a exactly similar problem to yours in my particle separator! After I iterate the continuous phase and residuals decline by about 3 orders of magnitude I enable interaction with discrete phase, then i get the normal spikes, but after a while the residuals start to increase and I get the warnings about "temperature limited to" and "turbulent viscosity ratio limited to...in xxx cells"
Can you please tell me how did you resolve this issue in your simulations?*
Also what method did you use to calculate the efficiency of the cyclone?
I use the mass flow rate of the trapped particles relative to the injected mass flow rate, but that seems to change when under discrete phase relaxation factors change! any idea about this matter?