local timescale weird convergence
 

Hi everybody,

Simulating a radial gas turbine a came to this case which is very hard to make converge.
Just to get a feeling of what the different timesteps will do i simulated with different ones. In the picture there are the convergence criterions plottet with different timesteps. The last one, the local timestep, looks weird. I saw that before in another case. Why does it not converge donewards. It looks like it would converge to a "convergence criteria which doesn't make sense.

Any Ideas?

Your simulation is clearly very numerically unstable. Local time stepping is not helping you much - so do not use it. Physical time stepping will probably work better for you.

That is why all these options are available - different simulations need different approaches.

Thanks for your rapid answer Glenn,
Could it also be that it is phisically unstable instead of numerical? Because with different mass Flows and Speeds of revolution i don't have that sort of problem.

I mean, i am doing a steady state analysis, but what if the flow is unsteady. Thats what i mean with phisicaly unstable.

Yes, physical instability will result in numerical instability as well.

So if the flow is truly transient then a steady state run will be unstable. This is all explained in the FAQ this thread has linked to.