Grid independence study
 

Dear friends,

To study the grid independency/dependency, how should the mesh size be varied in order to check the solution at different grid sizes and get a range at which there is no variation in the solution.

Thanking you for a kind response!

prasanth

Re: Grid independence study
 

Hi,

You can adapt your grid and continue you simulation with the refined grid to see whether there is a significant change in the parameters.

regards

Re: Grid independence study
 

Do you mean that we don't have to redo the meshing in gambit, instead, we only need to adapt grid in fluent? Thanks.

Re: Grid independence study
 

Hello Ozgur

Thanks for your reply. But, I know what you are saying, adapting the grid to refine mesh at locations of high gradients!!

but I am talking about grid independence studies!! you will take 3-4 different grid sizes and like that, repeat the simulation until we get no change in solution!

Thanks prasanth

Re: Grid independence study
 

I believe you can modify the grid based on different methods (not just high gradients), and when you are convinced that adapting the grid anywhere will not change your outcome, then you are grid independent. Your solution is 3D, the continuous adaption may not be practical. You can also refine the grid in Gambit as it seems you are about to do, but do decide where you want to make your changes based on what you are studying in the flow. May I ask what you are simulating, what model you are using, and how many cells you are going to have in your various meshes.

Re: Grid independence study
 

A good rule of thumb I use is 15 to 25% difference between subsequent grids. Sometimes it is necessary to refine grids in order to achieve values of pressure, velocity, turbulent reynolds number, y+ (if turbulent flow), etc that are appropriate. Then once those values are "believable" or close to experiment, refine grid with something like 3 or 4 or more densities. If you are in hexahedrals, you can refine in one direction at a time to ensure there is no effect from grid. Or if in tetrahedrals, a gross refinement of total cell sizes would be O.K. Finally, if tetrahedrals are use, it is always a good idea to try to create hex grid that is very close to the density of the tet grid so you can see whether there is effect of grid type. Hope it helps.

hii
after doing some solutions with different grid sizes..am gettting matching result in some parameter and different result in some other parameter.....
is this a grid independent solution?

You can make the number of cell in order 1.5 time incresing, and do calculation with same condition, if the solution is unchange to much, so you can get grid independent and chose the smallest grid fir your calculation

thanks sir....
is this 1.5 times increasing is a rule of thumb or can we take any other ratio for increment?

No, It is just my work, you can do whatever you want

thanks sir...for your valuable suggestions....

sir...
after completing some run with different number of cells....am getting a matching result within a range of number of cells....and a different result after that range.....so can we take the range at which we r getting same solution for our actual testing....is that a valid method......
please reply....
regards
gafoor

So it is not conv yet, I think need more grid point

sir.....can you please explain this....i didnt get ur point in your reply

FYI http://www.grc.nasa.gov/WWW/wind/val.../spatconv.html

In my opinion, grid independence is almost impractical, especially for engineering

thanks sir.......for your valuable suggestion.....and can you suggest me that if am getting same solution in a range of 60000 cells(maximum number of cells are only 1.2 lakh)...can I go further with my solution with this grid setup...pls reply...