Learning Forum

[Jimmyhan]

Hi

I am Jimmy,

I found in some papers that mesh density has been mentioned.

In order to compare with them, I would like to calculate mesh density

But I do not know how to realize this in CFX generator.

Can CFX do that, and how?

Thank you!

 

Jimmy

 

[Rob]

The paper probably uses mesh density to show how refined the mesh is: ie how many cells are in/over a region. The reports in ANSYS Meshing will give quality etc, and you set the sizes when building the mesh. 

Does the paper refer to a cell count or cell size anywhere? 

[Jimmyhan]

Thank you for your reply.

Yes, they mentioned the cell number in some regions. As you know, we have to refine the mesh or enhance the mesh quality in some regions, especially for boundary layer, But I don't know how to use CFX checking the cell number or other information about meshing details. CFX always gives out the total value. Can you give me some suggestion? 

[Rob]

I always check mesh characteristics in ANSYS Meshing, rather than the solver.  You're looking for statistics or quality (depending on version) under the Mesh part of the ANSYS Meshing tree. 

[Jimmyhan]

The version I used is 19.0, By by the way it does not have ICEM, right? I still did not understand the way, If you were me,  how to calculate the cells in the connecting region of two solids parts.?

[Rob]

In a mesh you use the settings to set what mesh you want. You can then check visually, or look at the quality metrics. I'd suggest going through some of the materials on ANSYS Meshing. ICEM CFD is an excellent tool, but potentially not necessary for what you're doing: you're trading power/functionality for usability. 

[Jimmyhan]

If I understand correctly, I still can't get a value about the mesh density. All I can do just control the mesh quality visually and set maximum value like 10^-6 unless I count one by one by myself, right? So people should show mesh figures in their papers directly.

[Rob]

Pretty much. We set the cell sizes (along with growth rates and size function settings) which combine to give us a mesh. We can then look at the mesh resolution, cell quality, cell count etc.  I suppose the density could be measured as number of cells in a given volume (you built the CAD so you know the volumes, or take it from the software). 

Many papers are trying to show an application or technology. Some aspects of the work may not be well explained, and this could be by deliberate omission if they're trying to hide something or there's an assumption that the audience will know. Note, if you're taking data from a single paper look to see what else the authors have published. I've had papers published in Pharma journals where we had to explain the technology fully using very simple terms: publishing essentially the same work in a CFD magazine meant we could summarise the CFD part (the audience understand it) but had to really explain the results. 

[Jimmyhan]

Thank you for your reply. It is a great suggestion. And from your explanation, I can learn some tricks in CFD paper writing. 

[Romain Falla]

In this article, I proposed a formula based on the internal angular sum of the mesh cells: (2) Mesh adaption for two-dimensional bounded and free-surface flows with the particle finite element method | Request PDF (researchgate.net).
The idea is to interprete internal angles as nodal fraction. The sum of the internal angles divided by 4 pi * V (2 pi * A in 2D), V (A) being the cell volume (area), then gives the dodal density or mesh density, i.e. the number of nodes per unit of area / volume