Posted Mon, 23 May 2011 09:58:21 GMT by dvb
Time to time it happens that during a TS run the time step stays to its initial value (e.g; 0.001 day). A different time integration scheme doesn't solve that. I would be currios to know what is the reason for that behaviour and how to "break" that.

Thanks !

Posted Wed, 25 May 2011 08:07:21 GMT by Alexander Renz
There are a number of potential reasons for this behaviour, which problem class is used for the model?
At first it should be made sure that the model is not unstable. Afterwards, try to change the numerical parameters (different initial time step size and the error tolerance).
Posted Wed, 25 May 2011 13:38:18 GMT by dvb
Effectively I noticed that a pumping was drying easilly. Once this was solved, the time steps curver was more normal.
Thanks Alex !

Posted Fri, 27 May 2011 14:42:08 GMT by dvb
I had this problem with a quite stable model (unconfined, run in transient ok, more difficulties in SS). With the automatic time step control and the default options the problem occurs. It disappears if I restrict the maximum change ratio. I don't understand this as with or without a constraint, the time step should freeze, isn't it ?
Posted Fri, 03 Jun 2011 11:39:40 GMT by Denim Umeshkumar Anajwala
If the freezing acutally occurs in the second step (after successfully finishing the first one), the limit on the change ratio may improve it. Otherwise FEFLOW might try a too-long time step, lowering it iteratively, which might take very long. If you experience such difficulties, still the model seems to be quite sensitive on the time stepping as FEFLOW is relatively careful in increasing the time step already internally withoug an external constraint on the ratio.

You must be signed in to post in this forum.