[HELP]mixing plane convergence strategy??

Pak

Hi all users. I have some serious and urgent problem.

The model, i solve now, is below (1) supersonic turbin system consist of 8 converging-diverging nozzle, 80 blade. Over expanded flow in nozzle hits turbine blade.

(2) So, I modeled 1 nozzle and 1 blade with periodic BC's and mixing plane.

(3) BC's are 6.8 Mpa pressure inlet at nozzle inlet, 0.5 Mpa pressure outlet. Rotor speed is 50,000 rpm

(4) I started with 1,000 rpm, mixing plane relaxation factor 0.5, and update interval 1.

(5) But, Reversed flow were not disappeared at mixing plane and Residual showed oscilation after 4000 iteration (about). Before 4000 iter, the Residual shows good convergence.

Is there anyone who had solved turbine problem with fluent like this?, or who has any related articles? reply or mail me plz.

Any comment about solution strategy will help me..