CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Grid Verification (

julie corish January 21, 2003 11:17

Grid Verification
Hi there, could anybody please explain the difference between grid verification and grid independence?

Is there any good literature available on grid verification?

Thank you, J.

Aaron January 29, 2003 22:36

Re: Grid Verification
This could be an issue of semantics; it may be possible that they are intended to be the same thing.

Verification, in general, can be thought of as answering the question, "Did we solve the problem correctly?" Usually, verification is associated with software and its ability to do the intended job. For example, when a solver is verified, it means the discretizations were coded correctly and the software is expected to correctly solve the equations, provided the boundary counditions are appropriate.

Perhaps those using the terminology, "Grid verification" mean that the software used to create the grids has been verified to actually create grids.

Independence, however, is a slightly different ballgame. Grid independence means that no matter how much more we refine the grid, our solution will not get better. When this happens, we choose the largest grid size that will give us our final solution and that becomes our solution grid. The converse is true as well: we make the grid coarser and coarser until the solution is no longer consistent from grid to grid.

There is a wealth of information in the CFD literature about these topics. Check the CFD-Online References section regarding books for some places to start. Another good starting point is AIAA G-077-1998, Guide for Verification and Validation of Computational Fluid Dynamics Simulations.

Rami January 30, 2003 09:50

Re: Grid Verification
As far as I understand, grid verification means verifying that the grid is admissible. By this I mean that the grid passes certain tests to verify, as some examples:

o There are no cells with negative area/volume (as may occur if care is not taken).

o Cells are not overly skewed.

o Aspect ratios are within an acceptable range.

o Cell faces are not wrapped (e.g. in quarilateral face, no face is distorted into two triangles).

There may be other requirements, specific to the the specific software you use.

The grid independence is indeed, as mentioned by Aaron, a convergence test to prove the solution achieved by a certain grid remains essentially the same on finer grids.

I hope this helps.

julie January 30, 2003 10:47

Re: Grid Verification
Thanks very much to both of you, I appreciate your responses. Julie

All times are GMT -4. The time now is 10:15.