[RodriguezFatz]

Dear all,

I'm reading the "Computational Methods for Fluid Dynamics" book by Ferziger and Peric and have a question concerning verification of results.

When I change the grid-size to smaller lenghts, the results in some regions change, others don't.
1) Now, would you restore the coarser grid, where the results don't change and only make it smaller where you saw the differences?
2) If I do so, can I use Richardson-extrapolation to estimate the error in the refined regions, without regarding the coarse regions at all? / Does the Richardson-extrapolation need a completely refined grid to work reasonably, or is such a partial refinement ok, too?

Thanks for any help,
Philipp.