I've evaluated several turbulence models on flow past circular cylinder at Re=3900.The cubic CLS(Craft,Lauder and Suga) model predicts well,but all the linear models say SST predict rather poorly.(centreline velocity profile, Cp distribution, separation angel etc.) I guess it is due to the wrong prediction of separation angel. Numerous experiments and DNSs give a separation angel of 84<sita<88 degree,however these linear models give an angel of 98 degree. I have tried fully turbulent numerical simulation, transition imposed at 90 degree, laminar simulation all field(separation at 102 degree), and they all give the separation angel over 90 degree. It is said that someone have employed SKE on this flow and the results agreed with the experimental data quite well. Could anyone experienced give me some advice? Thank you very much!

but each linear model say SST give an angel of 98 degree. I've tried fully turbulent numerical simulation, transition imposed at 90 degree and laminar simulation through the whole flow field which give an angel of 102 degree, and the separation angels are all over 90 degree. It is said that someone have employed SKE on this flow successfully. Could someone give me some advice?Thanks a lot!