Particles in the computational domain: two abnormal phenomena

Hi everyone,

(1) In my case, particles seem to be partially beyond the wall. Also, the bed height is obviously lower than that in a reference (http://dx.doi.org/10.1016/j.powtec.2015.09.037) despite the same parameters (domain width and depth, particle diamater and number) as those in a reference. How to solve this problem?

(2) The domain depth is three times larger than the particle diameter, but particles do not full fill the computational domain in thickness. How to solve this problem?

image

Futhermore, when I set the inlet velocity to zero, trying to let those particles fall freely, the time step become too small to continue running.

image

The above questions share a same case.
Cohesive_newVDH.mfx (11.6 KB)

Thank you.

  1. You are using cyclic BC in the z-direction, so there is currently no wall along z=zmin and z=zmax planes. Remove the cyclic BC if we want walls.
  2. This is not really a problem, you are getting exactly the 9522 particles you are asking for in the IC region. Using a cubic lattice gives two columns instead of one, but this won’t change the results.

My recommendation if you want to settle the particles is to run a first simulation without the gas phase (granular simulation), this will be much faster.

Thank you. I will use pure granular flow to settle the particles at first. For the first question, however, the z-directional cyclic BC is also imposed in the reference (http://dx.doi.org/10.1016/j.powtec.2015.09.037). They use OpenFoam-LIGGGHTS as a solver. The bed height in my results is much lower when compared with that in the reference with same amount of particles and computational domain (in width and depth). I am puzzled about this.

You mean after you turn the gas phase back on and let it fluidize for a while?

I am sorry I did not formulate my words well. Yes, I mean the result after the case runs for a while and check it at the same moment as shown in the reference.

Hi @jeff.dietiker , these two simulations are run in the a domain with same parameters, such as same domain width and depth as well as same particle number and gas velocity.

Result with no-slip wall boundary condition
image

Result with cyclic boundary condition in z direction (front and back faces)
image

It obvious that the bed height under cyclic boundary condition is much lower and it seems that many particles is lost. I think the bed height in these two simulations should not exhibit significantly difference. I am not sure the reason.

You have much fewer particles with the cyclic BC. Do you start with the same number of particle? If so, can you see where the particles are leaving the domain (are the elutriated from the top BC)?

Yes, I started with the same number of particle in both simulations. For the cyclic boundary condition in z direction one, the following figures show the bed height against time. It can be seen that none particle seem to leave the domain from the top outlet and the particle number in the last figure is much less than that in the first figure. I am not sure whether particles would leave the domian from cyclic boundaries. P.S. The domain thickness is assumed to be equal to the solid diameter in this case.

Cohesive_noVDH.mfx (12.7 KB)





Something is wrong when using 1 cell in the z-direction and cyclic BCs. Please use 2 cells in the z-direction and this should work.

Thank you very much. I will try out.