Owner list and its relation with the total number of cells

AliE · September 12, 2018, 14:23

Hello Foamers,

I am actually stack on a wired problem and I appreciate your help

. I have generated a tetra mesh inside a cube in the classical openFOAM format. I was looking into the file "owner" and surprising I have noted this: the maximum owner index is not equal to the total number of cells... For example the number of cells is 2750 while the maximum value in the owner list is 2746.

How can this be possible? Am I missing something?

Thank you very much!

Tobi · September 27, 2018, 02:21

Hi,
interesting point. I never checked it out. Can you provide your case? I would like to know what is going on there. For that purpose, I need the OpenFOAM case.

AliE · September 27, 2018, 03:55

Hi Tobi, thank you very much for your interest!

Working a bit on this I have find out that "usually" for hexa meshes the maxium index in owner list is also the total number of cells. However, in general, this might be not the case (e.g. doing a mesh in a cube using only tetra)! Why? Well, because if you think a bit about this, you can build one element using faces belonging only to neighbour cells and this depends on how your mesher export the mesh in OF format.

Thus the answer is (and I am 99% sure) that the total number of cells is: max(a,b), where a=max(owner) and b=max(neighbour).

Tobi · September 27, 2018, 05:36

I already made a sketch too, and it should be clear that

for pure hex meshes, the owner size should be similar to the number of cells
for other meshes, the owner size should be identical to the number of cells too

However, I might be wrong, but it is a conclusion of my drawings. Thus, I was thinking of free-standing cells. Therefore, I would like to get your mesh, to analyze the mesh and the corresponding files, to visualize the cells that do not have a neighbor; it is wired because each cell should have a neighbor.

Because, based on fundamental analysis, each numerical cell should have a neighbor, either way, it is hex, tet, poly, or any other volume type. Therefore, - for me - the only possibility is to have separated cells or a configuration I am not aware right now. In your example, there are two cells which do not have any neighbor (or maybe I am misinterpreting the files), and I would like to visualize and analyze the mesh in order to know what is going on and why there are two entries missing. Moreover, I know that, e.g., snappy makes some crazy cells sometimes.

Even though you have your analysis and statement, there is 1 % of uncertainty

Maybe the mismatch is related to the declaration of owner/neighbor (should be related to the normal surface vector)

September 12, 2018, 14:23	Owner list and its relation with the total number of cells	#1
AliE Senior Member Join Date: Dec 2017 Posts: 153 Rep Power: 8	Hello Foamers, I am actually stack on a wired problem and I appreciate your help . I have generated a tetra mesh inside a cube in the classical openFOAM format. I was looking into the file "owner" and surprising I have noted this: the maximum owner index is not equal to the total number of cells... For example the number of cells is 2750 while the maximum value in the owner list is 2746. How can this be possible? Am I missing something? Thank you very much!

September 27, 2018, 02:21		#2
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Tussenhausen Posts: 2,708 Blog Entries: 6 Rep Power: 51	Hi, interesting point. I never checked it out. Can you provide your case? I would like to know what is going on there. For that purpose, I need the OpenFOAM case. __________________ Keep foaming, Tobias Holzmann

September 27, 2018, 05:36		#4
Tobi Super Moderator Tobias Holzmann Join Date: Oct 2010 Location: Tussenhausen Posts: 2,708 Blog Entries: 6 Rep Power: 51	I already made a sketch too, and it should be clear that for pure hex meshes, the owner size should be similar to the number of cells for other meshes, the owner size should be identical to the number of cells too However, I might be wrong, but it is a conclusion of my drawings. Thus, I was thinking of free-standing cells. Therefore, I would like to get your mesh, to analyze the mesh and the corresponding files, to visualize the cells that do not have a neighbor; it is wired because each cell should have a neighbor. Because, based on fundamental analysis, each numerical cell should have a neighbor, either way, it is hex, tet, poly, or any other volume type. Therefore, - for me - the only possibility is to have separated cells or a configuration I am not aware right now. In your example, there are two cells which do not have any neighbor (or maybe I am misinterpreting the files), and I would like to visualize and analyze the mesh in order to know what is going on and why there are two entries missing. Moreover, I know that, e.g., snappy makes some crazy cells sometimes. Even though you have your analysis and statement, there is 1 % of uncertainty Maybe the mismatch is related to the declaration of owner/neighbor (should be related to the normal surface vector) __________________ Keep foaming, Tobias Holzmann

September 27, 2018, 03:55		#3
AliE Senior Member Join Date: Dec 2017 Posts: 153 Rep Power: 8	Hi Tobi, thank you very much for your interest! Working a bit on this I have find out that "usually" for hexa meshes the maxium index in owner list is also the total number of cells. However, in general, this might be not the case (e.g. doing a mesh in a cube using only tetra)! Why? Well, because if you think a bit about this, you can build one element using faces belonging only to neighbour cells and this depends on how your mesher export the mesh in OF format. Thus the answer is (and I am 99% sure) that the total number of cells is: max(a,b), where a=max(owner) and b=max(neighbour).