There are two issues confusing me over cfd now: 1.AS to wall function, y plus constrains the thickness of near wall layer, thus finer interanl cell will result in size jump between two layer of cells, whick cause numerical error. Since finer grid go to ideal solution. 2. Are Some coefficients of k-epsilon model gride-size related, or other model? If so, finer size does not lead to better solution.