Influence of the boundary layer refinement on convergence
 

Hello everyone!

This is my first thread in a forum, so I apologize for possible mistakes like posting this in the wrong place. Also I´m not a native english speaker.

I´m working on a 2D simulation of a flow through cascade with sheet metal blades, but i am quite new to CFD (about half a year) and have a few questions.

I made a structured grid using ICEM CFD and realized that minor differences in the refinement of the near wall region have an influence on wether my solution converges using Spalart-Almaras or k-omega SST as turbulence model in ANSYS FLUENT.
With 40 layers in the near wall region plus linear refinement to the wall I achieve convergence only with k-omega SST, while using 40 layers in the near wall region without any kind of refinement in wall direction leads to convergence only with Spalart-Almaras.
Also using 30 layers without refinement will cause a faster drop in residual decrease. (I used for every approach second order upwind interpolation scheme).

So my question is just how this is possible? I never would´ve thought that this small differences could cause non-convergence for one turbulence model, but lead to convergence for an other one.

Greetings