I use SST k-omega (transitional version) to model flat plate transition.

Skin friction at 8000 iterations looks realisticaly, but then diverges. BL after transition point behaves like fully laminar (unstable waves).

I have tried to lower y+ (from 1.5 to 0.3), lower courant number (3 -> 0.3), lower k-w underrelaxations, but with no success. It always diverges (even after 100 000 iterations).

Do you have any ideas? I have put example on: http://www.iol.sk/~mgasparovicova/cfd_test/diverg.htm

man, are you crazy? for a velocity of 10 m/s you use coupled implicit solver? it will always diverge at such a small velocity! use segregated solver. it was designed just for that. coupled is only for high velocity flows (M>0.6).

Thank you Razvan. As uninitiated self-taught person I always used coupled implicit for 2D problems (where the memory demand is not issue). Sometime I had to use low Cour.n. during whole computation, but I accounted complex turbulence models for it. Sometime I had strange divergence problems even with very simple geometries. Now I see the reason. Thank you much again.

Hello Peter. First of all, I want to say that I am sorry for the language used in the first message I posted for you, but also to tell you that I have very big expectations for myself and the others I work with. So you will understand why I was quite rough. Anyway, I also want to encourage you in your work, do not give up no matter what(I am also a self-taught CFD user), but first get your hands on a Fluent doc and study it from start to end, page by page, and you will find out almost everything you need to know as a user. Further you will need to study a little bit of theory, can't do anything serious and most important CORRECT, without it. Best wishes, Razvan

Don't apologise - you helped me a lot. Actually it was funny to call me crazy I have very good orientation in Fluent doc, but there is nothing about poor convergence of Coupl-implicit in low Re flows. I can read there: "Both approaches are now applicable ...".

Now back to SST k-omega. I hoped it will allow me to simulate BL transition on some barely measurable objects. But the result of the airfoil case is similar to the fully turbulent k-e simulation - i.e. very short laminar zone and the same drag. I woul like to delay transition distance more than 6 times boh in favourable and adverse pressure gradient zones.

How to tune this model to better correspond to experimental results? Which k-omega constants to change? I am just beginning with turbulence math, so am not well oriented yet.