[Hybrid]

Hi,

How can I do Grid Convergence Study using constant refinement ratio. I have gridRefine.glf script from Pointwise Git Hub page? How to use the script for constant refinement i.e. ratio2 (fine Mesh/Medium Mesh) = ratio1 (Medium Mesh/ Coarse mesh).

[RcktMan77]

Hi Ali,

I would recommend building your initial coarse mesh, and then edit the script with the project file name for your coarse mesh, use a refinement factor ~1.5 for a medium refinement level, and then specify a factor of 2.0 for your fine grid level.

A couple of other suggestions I would proffer would be to also turn off volume meshing in the script. Check the surface mesh quality and double-check the T-Rex wall boundary initial spacing (as this will get changed) for each resulting refined surface mesh. Then, you can go ahead and generate the volume mesh. The script works best if you have manually set all of the end spacings of connectors in your original coarse mesh. Hope that helps.

[Hybrid]

Thanks alot RcktMan77.

Kindly clear more...as the near wall spacing or Y+ value for each level of grid will change, so is it correct for grid convergence test or Y+ should be same for every level of grid i.e. Coarse, Medium, Fine e,g Y+ = 1.0?

By doing refinement factor 1.5, 2.0, etc.. can I get the constant refinement ratio i.e. Medium/Coarse = Fine/Medium = constant?

[RcktMan77]

Yes, you're correct. The initial wall boundary y+ spacing should remain the same for each grid refinement level. However, the script will change this value in the T-Rex Boundary Conditions Panel, so you will want to be sure to re-set this value to it's original prior to re-initializing the volume grid of your unstructured block(s).

The level of refinement should be uniform across the surface grid when you specify your refinement factor (e.g. 1.5, 2.0, etc.). I believe in this case your refinement ratio here for each level will be 2^3, or about 8 times as many surface grid elements for each successive refinement when using these parameters. Since the grid is unstructured, and it is difficult to get the same interior mesh from a given surface mesh, this refinement won't be exact, but pretty close to the constant refinement ratio once you have initialized the volume.

Give it a try, and see if it does what you're looking for...

[Hybrid]

Dear Zach Davis,

One thing, I forget to tell that I am making Structured grids for 2d-airfoil study but will plan to pally on unstructured and hybrid grids as well.

As closing remarks to this thread (from my side):

1. Wall Y+ or initial cell spacing from wall for boundary layer resolution should be same for all levels of grid (i.e. Coarse, Medium, Fine, etc.).
2. Constant refinement ratio is easy to apply Grid convergence formula and Richardson extrapolation.

Kindly correct me, if I write anything wrong.

[RcktMan77]

Hi Ali,

I think you have the gist of it. Note that the grid-refine script currently only works with unstructured grids, so it isn't advised that you utilize it in your structured examples.

Best Regards,



Zach