CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   setting Timestep (http://www.cfd-online.com/Forums/cfx/118408-setting-timestep.html)

Cecco May 27, 2013 08:31

setting Timestep
 
Hi everybody,
I'm running a steady simulation of a centrifugal blower.
Usually I do the following:
1. I set a really small timestep order to E-06 with Upwind scheme;
2. after 1000 iterations I switch timestep to E-05 with upwind scheme;
3. when physical properties which I monitor are stable, I switch to Blend Factor =0.75 (so not completely High resolution) and wait for a new stable state.

This process usually works, but not now.
Until I use upwind solution seems reach the convergence, but when I switch to Blend factor 0.75 ( i tried 0.6 as well) it doesn't happen.
I changed a lot of timestep between E-5 and E-7, but seems no works.
Someone has an idea about this behaviour and about what I should do?

Thanks

ghorrocks May 27, 2013 18:50

This is typical behaviour when you switch to a more accurate discretisation scheme. It means your simulation is not very numerically stable. In your case this probably means you need better mesh quality.

Cecco May 28, 2013 03:08

I'll do a new mesh.
Thanks man. U precious!


All times are GMT -4. The time now is 09:19.