CFD Online Discussion Forums

CFD Online Discussion Forums (
-   NUMECA (
-   -   General mesh quality criteria (

daniel.struwig February 7, 2011 02:41

General mesh quality criteria
I'm new to NUMECA and was wondering if someone could give me a general guideline as to what values should be attained or not to be exceeded regarding the orthogonality, expansion and aspect ratio. I'm currently simulating a KKK K27 compressor, but it seems that the b2b mesh quality at the hub is rather poor because of the large hub fillet (1.85mm). If the best mesh quality that I can attain still does not satisfy the minimum requirements is there any other way I could improve the mesh quality even if it means altering the geometry slightly.

Any help will be greatly appreciated.

oldshoes81 June 10, 2011 02:52

Hi Daniel,

The quality of mesh is the most important issue that you have to consider, in fact, it leads you to a better accuracy. Before looking at the orthogonality, first you would need to make sure there will not be a single negative cell. Then the range of expansion ratio and aspect ratio should be followed accordingly and lastly you should look into the orthogonality.

If you have difficult to obtain a better quality of mesh that you expected, you should alter the geometry of hub and/or shroud by expanding in Z direction.

I hope this is useful.

Lazarus September 2, 2017 20:33

Dear friends,

I am trying to carry out a grid independence study with 3 different mesh size using HEXPRESS. I want to set the grid refinement ratio(r) to 1.4.

My question is; should we apply this value(r) only for Initial Mesh step at the beginning and keep the other parameters same? However, when i applied this refinement ratio for only in Initial Mesh step, the difference between medium and fine grid counts is very small, i.e medium count is 2M and fine is 2.7M. Is it also necessary to change the refinement values ​​on the surfaces and curves? How can i ensure that the grid refinement ratio will strongly enforced in the x, y, z direction as we move from the coarse to the medium, or from medium to fine? Can you help me if you have any knowledge or experience about this issue?

Thanks in advance.

CFDUser12 October 3, 2017 03:53

Grid independence study
Hey Lazarus

As you might have read in the paper from Tokyo Workshop 2015 NUMECA recommends to perform grid dependence studies based on changing the initial cell sizes. This allows to refine the whole fluid volume and not only the areas right next to the solid. Hexpress defines the initial mesh as an integer number of divisions in each direction. That is why the refinement ratio 1.25 is ideal, because it allows to keep integer number of cells (6, 8, 10 for example).

In practice:
- Medium mesh: the initial mesh needs to have a number of cells in each direction that is multiple of 4. For example, 20 cells in X direction, 8 cells in Y direction, 8 cells in Z direction. Try to keep the initial mesh as isotropic as possible, the default proposed by Hexpress is isotropic.
- N coarse = 3/4*N medium, N fine = 5/4*N medium, where N is the initial number of cells in X, Y and Z directions. If you need more meshes just continue with the same ratio.
- The diffusion (Adapt to geometry > Global > Advanced > Refinement diffusion) needs to be 1 for coarse, 2 for medium and 3 for fine. This diffusion is a measure of how far from the solid surface the refinement propagates in number of initial cells. The reason to change the diffusion is to ensure the distance covered by refinement is the same in all meshes.
- All surface refinements need to be defined with a defined "Max nb of refinements" and "target cell size" = (0,0,0) --> this way the different initial meshes will lead to different mesh sizes.


All times are GMT -4. The time now is 11:44.