[Hqjen]

Dear Community

I have been trying to figure this problem for quite a while, reading this forum thoroughly, but noone seems to have encountered the same problem as I. So here goes - my first post.

I am currently trying to validate the Gamma Reynolds Theta model on a 2d circular cylinder, however, when performing a mesh-analysis, the results are somwhat inconclusive. Reynolds number is 10^6

The problem lies in the shear stress distribution. It seems there is some continuity problems (??). This happens only when i decrease the target surface size to a certain point. I will let the attached picture speak for itself Target_Konfig6_Shear Surf.png
As can be seen, there exists a laminar separationbubble, followed by turbulent reattachment, and finally there is turbulent separation. The Gamma Reynolds Theta model is doing a great job at this, if slightly on the aggressive side when it comes to turbulent separation (acccording to http://download.springer.com/static/...ed13549b34db9d p460), which states that the angle from the front should be about 130 degrees. Separation occurs at about 115 degress here.

What i do not understand is the large gradients of shear stress between the cells.

The mesh is quite fine, as can be seen. Target_Konfig6_Mesh Scene 1.png Max y+ is below 1, 60 prism layers, and a cell length parallel to the cylinder of about 2.5e-4 (diameter of the cylinder is 10cm)
convective courant number i likewise under 1, and the number of allowed inner iteration has been cranked up to 20 and the residuals are stable, and dropping 3-4 orders of magnitide in each timestep. Oh, and it's a transient simulation of course
The tesselation of the geometry is set to "very fine", and the mesh matches this geometry without detectable gaps - and as said, this happens when i increase the number of cells parallel to the surface.

So, can enyone guide me towards a solution of eliminating these large gradients?

If i have forgotten some crucual information, please just ask, and i will provide.

Kind regards