Adjustable Time step 'Minimal timestep'


#1

Hi,

I am currently trying to do a slosh analysis and I keep diverging, however instead of crashing my adjustable time step keeps reducing the step:

This continues for many hours, and I’ve had instances where it does continue increasing the tilmestep and converging again however this took an entire day! I would like to specify the minimal time step and if it can’t solve at that time step then it should crash.

There is a possibility that the simulation hasn’t actually diverged but crossed the maximum courant number and got a little over excited, in this case the minimal time step would still be handy.

Thoughts? also any suggestions on my simulation would be appreciated :smiley:

Kind regards,
Darren Lynch


#2

Hi Darren,

Once this divergent behavior happens as shown on the graph, there’s no going back. The simulation will always diverge but the point I see you are getting at is that it will keep running and running with smaller time steps so you have to be watching the simulation to hit STOP otherwise you waste a bunch of core hours.

I will add this to our list of feedback and see if there’s anyway to remedy this.

Thanks again,
Anna