Feedback - Evaluate Residuals
 

Hey guys,

I need your help to evaluate the attached Residuals. My set-up is described in the following:

-Computational Domain: Centrifugal fan with volute
-1500 million cells - polyeder unstructured
- k-e Modell : implicit transient - sliding mesh
- Time Step: 2,35e-05 (1 degree rotation each timestep
- 2nd order High Resolution
- Segregated Flow
- 7000 U/min

Following questions do I have:
- Where do the peaks in the residuals come from?
- Why are they in average on a very high level? The mesh, geometrie and set-up is OK.
-Next step I am trying coupled solver improving the residuals
- Between which Re number can I switch from segregated to coupled?
- General ideas and suggestions from your side

Thanks for help

From the figure, it seems that the solution did not converge. So, check it to improve the mesh quality.
The peaks could point to the start of new time step, I do not know how many iteration you had for every time step.
Good luck

First of all, the residuals in Star-ccm+ are normalized. Don't bother about the level of the residuals.
Second, I don't know where the peaks are coming from. But like siara817 said, check the mesh quality. Also the time step etc. Maybe you need more inner iterations (don't know if you changed that). Set the trigger type for residual monitors to iteration, you should be able to see what happens within a time step.
Using the coupled solver will not help very much since something else seems to be wrong. That's like trying a different saddle on a dead horse.

By the way, are you sure, your mesh contains 1500 MILLION cells?!?