Hi guys,

usually to demonstrate the grid indipendency it is used the Richardson extrapolation. Anyway, it supposes that the refinement is homogenous...but if I have adapted my grid adjusting gradients can I still use Richardson extrapolation?

Thanks

Cynthia

Theoretically you cannot. Richarson assumes that you refine in all directions homogeneously and methodically. That is by doubling the cell count for each new grid (you can actually multiply by any number as long as it is the same for each new grid). you can find more information on www.grc.nasa.gov

You can also check the book by Ferziger and Peric for details.

Esteban

Check this paper, C. J. Freitas, U. Ghia, I. Celik, P. Roache, and P. Raad, "ASME'S QUEST TO QUANTIFY NUMERICAL UNCERTAINTY", 41st Aerospace Sciences Meeting and Exhibit, 6-9 January 2003, Reno, Nevada.

The authors provide a nice procedure for CFD error estimation. There is a huge amount of publication in this regard .. but this paper outlines what I think is the most important part!

Very interesting paper!!! As I understand the refinement cannot be local but has done for the whole grid... Are you aware of how richarson and GCI's can be applied with local refinement only?

Thank you

Unfortunately I was not able to find such a procedure in open literature. However, I did some work on that and I calculated r locally "mesh refinement" and I used it in the same procedure .. I am not sure though if it is the right thing to do!