Strange things in SRF (urgent)

Cem

Hello, I have a problem in the simulation process of a turbine meter since I recognized it a month ago. I am working on a turbine meter model with a volume around one blade. The best way to simulate flow is using single reference frame in a steady-state condition. In theory if a denser fluid enters the volume at a certain speed it should produce more torque on the walls than a less dense fluid if the blades were stationary. Additionally a denser fuid causes the turbine meter blades to rotate faster than a less dense fluid. I use Fluent to simulate this and the most important part is to find the rotational speed for different conditions.

Since the turbine meters rotate freely around their rotation axis, the rotational speed should be updated according to the moment coefficient on the walls of the blade. Now the strange things happen in this simulation. Wh ile testing two different densities for the same inlet velocity speed, the following happens. First I give 0 rpm for the fuid and monıtor CM on the blade walls. In reality the denser one should cause more torque. The simulation results match with the reaity in this case, the denser one causes more torque. However I begin to increase the rotational speed to a value where CM would be 0( actually osciate around zero) My idea is if I rotate the fuid in the same direction that moment is produced, the torque should decrease. For a certain value of rotational speed CM should be zero. Actually this process happens as expected. As I increase rpm CM decreases then I find a rpm value for CM=0. But as I increase rpm the denser fluid converges to a smaller rotational speed for CM=0. This is really strange. The fluid registering for a more torque (more CM) for w=0 happens to have a less rotational speed for CM=0 when compared with a less dense fluid. In this comparison everything is the same the only thing I do is going to the materials panel and changing the density. I have done this for gases with incompressible ideal gas model in different operating conditions. The result is the same.

I use k-e realizable turbulence model with enhanced wall treatment, 3d segregated solver. For 15 days I am trying on finer models. I check yplus values of the walls to be under 1. I have used meshes over 1 milion elements. The results are aways the same. After some value of rotational speed, the denser fluid begins to register less torque when compared with a less dense one. If anyone has such an experience or has an idea to clearify this problem, you are very much welcome. I appreciate any comment or help Regards,

Cem