CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Siemens (https://www.cfd-online.com/Forums/siemens/)
-   -   cell size (https://www.cfd-online.com/Forums/siemens/55025-cell-size.html)

Baskar November 24, 2005 07:45

cell size
 
hi when i am making trimmed mesh, what is maximum and minimum cell size i have to give. for the same problem i am getting converged solution with some value of max and min cell size, but if it changes i am not getting convergence. what is the optimal cell size recommended in star-cd. Thanking you with reg Baskar

Ben November 24, 2005 08:41

Re: cell size
 
how long is a piece of string?

Tom November 25, 2005 23:58

Re: cell size
 
Baskar, you have just hit upon one of the problems of CFD. Just because something may work, doesn't mean it is correct. There are a couple of best practices you might want to check out.

Basically, when building a mesh for a problem a mesh sensitivity study is done. The mesh size is increased until the answer doesn't change, then go back to the smaller mesh size that gives the same answer. Different kinds of problems have different mesh size requirements.

Tom

Siddharth January 17, 2006 12:04

Re: cell size
 
Tom, you are quite right. Can you please suggest me the steps to carry out mesh sensitivity study for a problem? What parameters i need to focus for one-to-one comparison.

Actually i want to carry out a comparative study of cell-size effect as well as cells-type ( Hexa, Pyramid, tetra etc. ) effect for a fixed problem.

Thanks in advance

Sidd

Tom Keheley January 18, 2006 13:34

Re: cell size
 
Siddharth, the ERCOFTAC best practices recommends that for each class of problem a grid dependency study be performed. They recommend using three grids that are signficantly different in grid size, where ideally you double the grid twice in each direction and then apply the Richardson extrapolation (P.J. Roache(1997), "Quantification of Uncertainty in Computational Fluid Dynamics," Annual Review of Fluid Mech., Vol.29, pp 123-160).

What is interesting is that sometimes it is not the entire grid that is important. I just finished a study where with one grid size it took 5000 iterations for the residulas to converge to E-04. I did a grid refinement on just one section of the mesh and the problem converged to residuals of E-06 in 1500 iterations and agrred with the test data.

I hope this helped. Tom


All times are GMT -4. The time now is 20:08.