I've performed few grid refinrments, but difference between them is still high (the further the higher). What is the problem here? Maybe there is some 'critical' cell size that all the cells should be kept bigger than it?

there probably isn't a problem, you're just capturing more detail. When you say a few refinements what do you mean? Generally one should refine in areas of steep gradients, you can do this in the grid refinement panel. Also you should be solving uvw with mars.

allan

I do the refinements in order to check results on grid independence. Every time I make a refinement I compare results of the new run with the results of the previous one. The idea is to obtain a small difference between results. Then I can tell that the model is grid-independent.

I am solving uvw with mars scheme and double precision.

are you just solving to the default convergence criteria, if so ar the residuals still falling or have they levelled off on each of you models?

allan