diverges with RSM, converges with k-e
 

Hi,

I am simulating the flow around a cube in a turbulent channel at Re = 40000. I use enhanced wall treatment and my y+ values are around 2. First I solved with realizable k-epsilon, and I had good results, in accordance with articles. Now, I switched to RSM and it explodes. I have x velocity of 1e6 in the channel. I decreased relaxation parameters yet it always diverges. It says turbulent viscosity is limited to 1e5 and after a few iterations backflow at the outlet.

May anyone have an idea about it?

Thanks

Now, I also changed the mesh and the domain. I increased the length of the channel behind the cube. But there is still no difference. Always, turbulent viscosity limited to 1e5.