Learning Forum

[hasan.mustafa.moayad]

Hi, I'm confused regarding the choice of mesh metric that can be used to check the mesh quality?? which metric can give a good representation of mesh quality, is it the skewness or orthogonal? I read some comments provided by the ANSYS community and they mentioned that the min. orthogonal greater than 0.1 will be acceptable. For me, I deal with CFD Fluent and I need to know:

1) which mesh metric I can use skewness or orthogonal?

3) To which value should I look for checking the quality, (the Min., Max., or Avg.) values of these metrics (i.e. skewness and orthogonal)?

2) what is the range of skewness and orthogonal to be defined as (accepted/good/ very good) quality?

[SRP]

Hi,

The choice of mesh quality metric depends on the specific requirements of your simulation and the physics involved.

For tetrahedral elements, skewness is a more relevant metric. For structured meshes (hexahedral or quadrilateral elements), orthogonality is crucial.

Generally try to keep the orthogonal quality greater than 0.1 and maximum skewness below 0.9.

Hope this helps.

[hasan.mustafa.moayad]

Thank you so much for your reply.

I made the mesh for the geometry, the mesh is structured for both sides of YZ-plane, but it is unstructured for both sides of the YZ-plane (as you can see in the picture below). Generally,  I got an orthogonal quality of (Min.=0.297). 

1) is it acceptable to proceed to the set up and solver or I must refine the mesh more and more before that?

2) If I have to refine the mesh, what should I do to accomplish this? I tried to improve the mesh but could not reach a good mesh and sometimes mesh is failed. Could you please help me?

Thank you again.  

[Rob]

Cell quality is one metric (which you're looking at) but also consider the mesh quality. Ie have you resolved the expected flow/heat transfer and are the changes in cell size low/smooth enough to capture flow gradients? 

You may want to review Fluent's wall boundary options as some of the cell zones may be better modelled using a thin wall. 

[hasan.mustafa.moayad]

dear Rob, sorry I didn't get you.

could you please explain clearly? 

I attached a picture showing the details of "mesh" if this can help.

Thanks.

[Rob]

Look at the change in cell size between the various zones in your image. If flow passes from one block to the next how well do you think the gradients will be handled? Ie how well will cell/volume weighting be dealt with? The solver has several tools to deal with this, but it's best to have a good mesh rather than rely on numerics. 

Note, one degree resolution may be a little excessive, 8-15 degrees is more normal. But 0.1mm is the lower limit so that may be the local cell size.