OpenLB – Open Source Lattice Boltzmann Code Forums on OpenLB General Topics Boundary material numbers setting question(Poiseuille3D)

Viewing 3 posts - 1 through 3 (of 3 total)
• Author
Posts
• #9180
araeli
Participant

Hello everyone,
I’m studying the Poiseuille3D test case, trying to build a model very similar to the micron scale, and I’ve found myself asking some questions about the assignment of material numbers.

1. At the very beginning the material number 1 (internal domain) is set on an enlarged cylinder using an origin point of center0(-converter.getPhysDeltaX() * 0.2, radius, radius);
Why this 20% of displacement is added?

2. In the forced case this quantity is expanded by 3 cells on each side. Why exactly three?
a. The voxel are intended by cells and assigned by point values on the direction vertices?

3. In the non-forced case to describe the inlet (and similarly the outlet), two cylinders that are 4 cells thick are used (with origin[0] = -converter.getPhysDeltaX() * 2; and extend[0] = converter.getPhysDeltaX() * 2;).
Then, if I understood correctly, using the function superGeometry.rename(2, 3, 1, inflow), the material number is assigned by exploring only the two neighboring voxels in the chosen direction. The goal is to obtain a surface, so why is a cylinder that is 4 cells thick used instead of directly defining a surface?

4. Going back to the previous question (2a), does the assignment on neighboring voxels made by superGeometry.rename(2, 3, 1, inflow) explore points or cells?
If the assignment is made on the cells, when is it converted to point values?

I suspect that when using Paraview, cell values are somehow shifted to the points in the geometry visualization. I can’t find an exact match between what I see and what I expect in the code. If this is the case, I apologize in advance.

Thank you in advance for any clarification.
Alice

#9256
FBukreev
Keymaster

Hello Alice,

1. We use 20% randomly, it could be also 30 or 40%. This delta makes sure that the point at the x coordinate 0 is converted to material number 1.

2. 3 cells in both cases we need to garantie that all cells are converted to MN 1 in the x direction, because there is a cell with MN2 in the negative x-direction.

3. We can make this inlet cylinder arbitrary thick, because with the command rename(2,3,1,inflow) only one layer of cells that borders to MN1 will be renamed to MN3.

4. In OpenLB we have actually only points and we call cells also the points. We do not have here distinction between cells centers and cell surfaces like in FVM meshes.

#9257
araeli
Participant

Dear FB,

Thank you so much. The fact that points are called cells wasn’t clear to me, but it’s crucial in clarifying my doubts.

Best regards,
Alice

Viewing 3 posts - 1 through 3 (of 3 total)
• You must be logged in to reply to this topic.