How to set up sizing function where there is a large transition in grid sizes
 

I am trying to mesh a slightly complicated 2d geometry in gambit. I am facing problems while using the pave method to generate unstructured quads.

I am not able to post the geometry here, so I will provide the description about a part of the geometry that can be easily understood. |I am facing problems only in this region.

Consider a rectangular domain height = 10000 width =20000. node distribution on bottom face = first length 1, number of nodes = 500, node distribution on left face = first length = 40 number of nodes =50.

I need to mesh this domain using the pave method using the least number of cells as possible as not much flow physics is expected here. I have tried using various values in sizing functions, I have not been to able to successfully mesh it.

Anyone who can provide any clues to how to set up an acceptable sizing function is very much appreciated.

create a size function with meshed option
select both mesh edges as source and your face as attachment.
grow factor 1.2 for example and max size 1000
adapt those 2 parameters

Dear Sir,

I have previously tried using size functions with the exact parameters mentioned with a start size of 40. Gambit fails to pave after 4700 cells.
Gambit suggests to adjust mesh sizes and try again.

I have tried several values of size functions. Not been able to pave successfully. I think due to the large edge growth rate in both directions, the algorithm has failed.

I have attempted several combinations including growth ratios from 1.05 till 1.95, none of them successful. Please let me know if there is a more methodological approach to determining the correct growth ratio, start size and max size.

Thank you for your help!

just try with several parameters
there is no start size in the size function if you select the "meshed" type
Attachment 44032

Thank you very much sir. The mistake was that I was trying to use a fixed size function, due to this meshing failed.

For future cases and others who may view this thread, in case of meshing failing while paving or any other method, where the cells are generated in order of advancing front, is it possible to find out which region caused the mesher to stop? Is it possbile to view any file where the mesh is temporarily written and try to understand the problem.

Thank you,

the only info you can have will be displayed in the transcript windo (also *.tr file if I am right)