As i am new CFD ACE User, I like to know how to check the grid quality such as grid independance. And what all the basic things and parameter that is to be considered to generate good quality grids can any one help in this.

Thanks in Advance.

lalgudi

No Reply!!!!!!

Hi,

Correct, no reply. Why? Because I think I have answered this question about 50 times before. I wish there was a FAQ we could refer common questions to.

Grid quality parameters can be measured in CFX in ICEM or CFX-Post. They give a number of measures of grid quality.

Grid independance checks are done by generating a number of meshes with different grid resolutions and finding the mesh resolution where the parameters of interest stop changing. The methodology is the same regardless of which CFD code you use.

Regards, Glenn Horrocks

Thank you Mr.Horrocks,

As i am new user of CFD, i feel i asked an basic question.Thanks for answering. ok! should i change the number of grids in all directions (I,J,K) together or in one direction alone, Sugest me the pattern to change the number of grids which gives good result in less number of grid generation.

Thank you

change it in the direction(s) that has or have high gradients. godd luck.

[waseeqsiddiqui]

Hello everyone,
I would like to ask in regards to the procedure for performing grid independence using ANSYS ICEM. I have created a very fine unstructured mesh with each surface mesh accurately capturing the surface profile. I then simply increased the scale factor for creating coarser, even coarser etc. meshes. Is this the right way? Or do I manually assign coarser parameters on each individual surface?
Cheers.

[ghorrocks]

Performing mesh studies can be difficult. The ideal situation is you take a given mesh, then halve the element edge length of all elements (so a single hex element becomes 8 hexes in 3D, a single tet becomes 5 tets) and compare the results of the two simulations. But in practise this is tricky as it often leads to meshes too large to be practical. You then need to use some judgement in determining what the mesh regions which require the accurate mesh and refine there only - leaving areas which do not affect the accuracy as much unchanged, or only refined a little bit.

So your approach using the ICEM scaling factor can work. But note it is usual to be required to refine the mesh compared to the starting mesh, not coarsen it. So be open to the possibility of a finer mesh being required.

[waseeqsiddiqui]

Originally Posted by ghorrocks

Performing mesh studies can be difficult. The ideal situation is you take a given mesh, then halve the element edge length of all elements (so a single hex element becomes 8 hexes in 3D, a single tet becomes 5 tets) and compare the results of the two simulations. But in practise this is tricky as it often leads to meshes too large to be practical. You then need to use some judgement in determining what the mesh regions which require the accurate mesh and refine there only - leaving areas which do not affect the accuracy as much unchanged, or only refined a little bit.

So your approach using the ICEM scaling factor can work. But note it is usual to be required to refine the mesh compared to the starting mesh, not coarsen it. So be open to the possibility of a finer mesh being required.

Thanks ghorrocks.