34 ramberg osgood plasticity #45
Conversation
examples/ramberg_osgood.py
Outdated
| """ | ||
|
|
||
|
|
||
| class RambergOsgood3D(object): |
There was a problem hiding this comment.
Have you tried using @jitclass here? In my memory jitclass was not as fast as jit functions, but maybe they have progressed in the last years.
There was a problem hiding this comment.
I did not try jitclass yet, but I also think, in the old version I did not use a class, because it was user to jit a function. I have no strong opinion on this. I can try @jitclass, but would not want to start a comparison of performance here.
examples/ramberg_osgood.py
Outdated
| self.QT_views.append((QT_parent, QT_child, QT_sub)) | ||
| self._tangent.append(df.fem.Function(QT_sub)) | ||
|
|
||
| self.stress = df.fem.Function(QV) |
There was a problem hiding this comment.
If we want this problem definition to work with incremental laws, we need the stress two times. One time for the updated state and one for the initial state. Same for history variables
| self.stress.x.scatter_forward() | ||
| self.tangent.x.scatter_forward() | ||
|
|
||
| def eps(self, u): |
There was a problem hiding this comment.
On main, there is already a pre-defined mandel strain function
There was a problem hiding this comment.
Something like
from .stress_strain import ufl_mandel_straininside interfaces.py will lead to a circular import.
My current solution is to let IncrSmallStrainProblem.eps be an abstractmethod that the user has to implement. I would consider this the safer option. If there was some default eps and the user does not have to think about it, then this could potentially lead to the IncrSmallStrainProblem computing mandel strain, although user has implemented some material routine expecting standard voigt basis.
Hi @srosenbu ,
this is a draft for the ramberg osgood example. It's now working with dolfinx version 0.6.0, but the formulation is not incremental and the compilation of the forms has to be explicitly called by the user at the moment. Also, I would like to reduce the complexity and remove certain parts of the code that are not necessary here.
We could also have a talk in the next days and add a general
fenics_constitutive.interfaces.IncrSmallStrainProblem, such that the class is simply imported here (and potentially in the other examples).Also, I did not generate the documentation yet.