search

idaholab / moose

Multiphysics Object Oriented Simulation Environment
https://www.mooseframework.org
GNU Lesser General Public License v2.1
1.77k stars 1.05k forks source link

Picard sensitive adaptive time stepper #7436

Open snschune opened 8 years ago

snschune commented 8 years ago

Description of the enhancement or error report

Currently we cannot cut the time step in case Picard residuals do not reduce. We see this in Bison/Relap problems or RSN/Bison/Relap problems supposedly when we go into contact. Sometimes Picard its perform fine, at other times residual does not reduce at all.

Rationale for the enhancement or information for reproducing the error

I would suggest to implement a TimeStepper that gets _picard_timestep_begin_norm, _picard_timestep_end_norm and keeps its own old versions of it. Then by looking at the residual history, it makes a decision whether to cut the timestep.

Identified impact

Adding two getter functions to Transient for getting the _picard_timestep_begin_norm, _picard_timestep_end_norm

permcody commented 8 years ago

@snschune - Is this ticket still valid? I'm hoping we've dealt with this problem.

snschune commented 8 years ago

I still think we need a time stepper that is sensitive to Picard convergence. I am not sure when I will be able to implement something that works properly, though.

friedmud commented 8 years ago

I agree - I would still like it too. I might even have to have it for some of the work I'm doing. I would say leave this open.

It's something that might be able to be helped / dealt with in #5229