Error in particle inspection and void fraction calculation

Submitted by Mahsa Bagherzadeh on Wed, 05/21/2014 - 16:34

I have faced a problem in particle inspection and void fraction calculation by CFD side, which I think might be a bug.
When, I insert a particle in the middle of my domain, (100*100*160 mm cube) only a quarter of the particle has been identified by the CFD. On the other hand when I move the center of the particle from p1=(50,50,120) to a bit to the corner to p2=(30,30,120), the whole particle becomes visible.
As it can be seen also in the attached figures the values of the void fraction makes it clear as well that when particle is in the center the whole volume of the particle is only assigned to a quarter of cells.
For the void fraction I used BigParticle with scale up volume of five, the diameter of the particle is 15 mm and the cell size is 2*2*2 mm.

The code by which I insert the particle in Liggghts is as follows:

# particle insertion
#======================================

create_atoms 1 single 0.050 0.050 0.12 units box
set atom 1 diameter 0.015 density 1120 vx 0 vy 0 vz 0

#======================================

Thanks for your time

Mahsa Bagherzadeh | Tue, 05/27/2014 - 18:48

Hello,

From a week ago I checked many other parameters that I thought might be influential, like changing the time step, coupling interval, mesh size, changing the viscosity and density. But apparently, none of them is really playing a role. On the other hand, when it comes to geometry some changes seem to be influential. For instance I scaled down the model to 1/10 of my real domain it seems that the results are more acceptable. I appreciate any help.

Regards

cgoniva's picture

cgoniva | Wed, 06/04/2014 - 09:00

Hi,

I guess it is a matter of parallelization.
You use the cfdemSolverIB, right? Why don't you use the IB voidfraction model with it? That should solve the problem.

Cheers,
Chris

Mahsa Bagherzadeh | Wed, 06/04/2014 - 20:35

Thanks Chris for your response.

But the point is that we have already solved this problem using IB solver and IB void fraction and we got acceptable results.

Then we were wondering whether it is possible to solve this problem without IB method or not. Therefore, we used PISO solver and BigParticle void fraction model, in which, we faced this problem with the void fraction.

Also I have this question that, is the combination of PISO solver and BigParticle void fraction model able to solve a resolved problem?

Regards,
Mahsa

cgoniva's picture

cgoniva | Thu, 06/05/2014 - 17:27

Hi,

if the particles are really bigger than the flow structures you want to solve for and as a consequence bigger than the mesh, there is no way around using the IB method.

Kind Regards,
Chris

e.derakhshani | Fri, 06/06/2014 - 10:49

Hello,
Thanks for kind replies.

Right now the accuracy of the simulation is not important for us, because at the first step the void fraction of a single particle cannot be created perfectly in CFDEM.
Indeed, the main question is why the big particle voidfraction model is not able to model the whole volume of a particle. This is clearly shown in Mahsa's figures.

How can we figure out that this problem is because of an unknown bug or it is because of a mistake?

Best regards,
Ebrahim