CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   STAR-CCM+ (http://www.cfd-online.com/Forums/star-ccm/)
-   -   Cells disappearing when a z plane is taken (http://www.cfd-online.com/Forums/star-ccm/108178-cells-disappearing-when-z-plane-taken.html)

otq October 16, 2012 14:13

Cells disappearing when a z plane is taken
 
I am modeling flow around a cylinder that is flowing in the axial direction of the cylinder. The mesh scene displays cells in all the regions; however, when I take a z=0 cross-section (the base of the model), the following image appears. This image was taken after an initialization, just to add clarity to the problem. Has anyone encountered a similar problem? I have double checked the meshes validity and the missing cells reappear when I move just above the z=0 plane (z=1 picometer).
http://i.imgur.com/Scwiy.png

siara817 October 17, 2012 01:41

Is it a symmetry plane?

otq October 17, 2012 07:43

In the model that area is defined as an inlet boundary. The z=0 plane is a derived part. The model was generated in STAR-CCM+ based on sketches and extrusions, etc from the x-y plane.

siara817 October 17, 2012 11:59

I think there is no problem. The reason is that you add a plane at the edge of the domain. I you move the plane section further ( for example z=0.001) then you will see there is no problem.

otq October 17, 2012 13:33

Why should cells that exist on the boundary, not show up when a z plane of that boundary is taken? How is that not cause for concern? As it is, this shows that there is no value at those locations (in this case it is temperature). How does no temperature at z=0 not be a problem?

abdul099 October 17, 2012 14:44

The reason why you don't see values there is the plane section, just like siara817 already said.

But let's start from the beginning:

"Why should cells that exist on the boundary, not show up when a z plane of that boundary is taken?"
This plane section should exactly go through your boundary, at least in theory. The issue is, you're not modelling something on a sheet of paper, where you can say 0.000000000000000000000000001 = 0.000000000000000000000000001. You're modelling something with a computer. And what you're modelling requires floating point numbers. And there are no exact floating point numbers for a computer, you will always have some deviation. See
http://en.wikipedia.org/wiki/Floatin...dern_computers
for more details.
Let's say the deviation is some 1e-7m. Then the plane section you put ON the boundary is NOT everywhere ON the boundary. There might be positions where it is just outside the domain. Even when it's only some 1e-7m, it IS outside.
Or the cell faces are slightly moved towards the domain - also just some 1e-7m, but it's enough for the plane section not to cut through the cell face.

"How is that not cause for concern?"
Pretty simple - you just need to know why you don't see any values there and you know that it is nothing to be concerned.

"As it is, this shows that there is no value at those locations (in this case it is temperature)."
Right. But who cares? It's outside the domain, so there is no need for values.

"How does no temperature at z=0 not be a problem?"
The same reason why a temperature at z=-989898435798745 is no problem. The solver doesn't care if there is a value at z=0 - just because it is OUTSIDE THE DOMAIN.
It is just a visibility problem. That you can't see a value there doesn't mean, the solver will have problems, as long as there is a value provided on every boundary face. Since this face might lie behind you plane section, there can a value be provided, although you don't see it.

So what I suggest: Don't try to visualize a plane section which is located exactly on a boundary surface. Visualize the boundary surface itself. Just put it in the displayer instead of the plane section, and you will see, it's all okay.

otq October 17, 2012 16:18

I get it now, thank you for both of your responses.


All times are GMT -4. The time now is 03:22.