missing particles with variable inlet velocity

mbaldini's picture
Submitted by mbaldini on Fri, 01/13/2017 - 15:29

Hi all, I'm running an unresolved simulation in which particles are injected in a Helle-Shaw like cell.
The cell has a variable (linear) inlet velocity, as showed below, and the particles are inserted using insert/rate/region.
inlet

{
type uniformFixedValue;
uniformValue table
(
(0.0 (8.0 0.0 0.0))
(9.0 (8.0 0.0 0.0))
(10.0 (0.0 0.0 0.0))
(11.0 (0.0 0.0 0.0))

);
}

When I check the number of particles I see that some particles disappear around t=9s, right when the inlet velocity is "turned off". It is a bit strange because the simulation runs smoothly, the Co number is very low the DEM ts seems to be appropriate. Does anybody have an idea of what might the problem be? Thanks

Cheers,
Mauro

AttachmentSize
PDF icon particles.pdf27.45 KB
PDF icon courant.pdf72.73 KB

Forums:

j-kerbl's picture

j-kerbl | Mon, 01/16/2017 - 11:52

Hi Mauro,

hard to say, is it possible the particles are too slow to leave the insertion region and "jam" it? Are there any warnings from LIGGGHTS concerning the insertion?
Do you see particles leaving the domain when you look at the results with paraview?

Cheers,
Josef

mbaldini's picture

mbaldini | Mon, 01/16/2017 - 20:52

Hi Josef, thanks for your answer.
I've checked and there are no warnings when the particles start to disappear. There are a few but at the beginning of the simulations because the specified particle rate cannot be reached because of the "jamming" that you mention.
It is hard to see particles leaving the domain because there are a lot in the domain, but the DEM box has fixed BC and also physical walls at every limit of the simulation domain . Now I wonder if the simulation box should be a bit larger than the "physical box".

Cheers,
Mauro

mbaldini's picture

mbaldini | Thu, 01/19/2017 - 20:54

Hi all, I have some updates regarding the missing particles problem. I've tried several things until I solved the problem.
- First I checked that the DEMts was appropriate using fix check/timestep/gran and the timestep was OK. But just in case, I reduced the timestep and the problem still remained.
- I reduced the CI and CFDts and the problem remained.
- I increase the simulation box in one particle diameter to see if particles were disappearing because the physical walls where just on the limits of the simulation box. Problem not solved.
- I use a time constant velocity and the particles stopped disappearing.
- I use a codedMixed BC to give the time variation to the velocity instead of a table. Again, problem not solved.
- Finally I changed the BC on the DEM side to boundary m m m and now it works! No particle is disappearing when I reduce the inlet velocity.

Cheers,
Mauro

j-kerbl's picture

j-kerbl | Fri, 01/20/2017 - 09:55

Hi Mauro,

good to hear!
A short hint: with m m m always keep an eye on the overall boundary. If there is a problem and just one particle is trying to leave the domain, the simulation box will be increased to keep it. This can lead to sudden box-expansion which leads to sudden increase of memory usage, which can cause a crash.

Happy simulating,
Josef

mbaldini's picture

mbaldini | Fri, 01/20/2017 - 13:54

Hi Josef, thanks yes I know that. But at the moment is the only thing that worked. It's strange because the walls are well defined, the DEMts is appropriate and the particles star to disappear when the inlet velocity starts to decrease. If I use a constant inlet velocity everything is OK.

Cheers,
Mauro