CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   grid dependent results (

newbie June 1, 2001 15:54

grid dependent results
Dear all

I always obtain a difference (small or not so small) in my results when I run a problem in grids of different cell sizes.

My question is; is there an accepted criteria for rejecting a result in CFD based on such difference?.

I mean what's an acceptable size of such difference?

I'm starting to work in turbulence (k-e) and the difference in my results in grids with 10000 and 70000 cells can be as big as 30%, is there any remedy?

I would thank some reference(s).



newbie June 1, 2001 16:03

Re: grid dependent results
I've searched in the forum and found answers to this "stupid" question Sorry

John C. Chien June 1, 2001 19:02

Re: grid dependent results
(1). There is no such thing as a "stupid question". (2). Whether a solution is acceptable or not depends on the situation. (3). If your boss is looking for a solution ASAP, the perhaps a solution is better than no solution. (4). If you are fighting for the survival of your company, and your analysis of your design is the key to the new product, then you probably would like to have the solution as reliable as possible. (5). In any case, if you tell your friends that your solution is function of the time you run the code, then it is going to change their habit as to when is the best time to get an answer from you. (6). If your calculator is as accurate as your cfd solution, then it must be hand-made in ? (7). The interesting thing about CFD is, the solution is changing in time, even though you are solving the steady-state equations.

clifford bradford June 6, 2001 20:05

Re: grid dependent results
If you do grid study and you get x% difference in your result of interest then you need to decide whether x% is too big. If not stop, else refine further. That's the simple answer.

At the same time there are ways to choose grid sizes so that you can be more confident that your initial grid is sufficient (note I didn't say "that you can be assured") as there are times when you don't have time or resources to do a simulation more than once. the grid requirements depend on the type of scheme you are using, what is being simulated and the flow conditions.

All times are GMT -4. The time now is 09:51.