[dune-pdelab] Enquiry: testelasticity.cc

[Nagesh Eranki]

Hello

Yes, it seems most appropriate to implement a traction, or a stress vector,
rather than a stress tensor. This will more closely represent reality. I
hope we can access the global coordinates of the entity to control where
the traction is being applied. But what about the concentrated force at a
particular node of the mesh? Can this be implemented by 'g' as well?

Thanks


On Sun, Jun 3, 2018 at 12:34 PM, Christian Engwer <
christian.engwer at uni-muenster.de> wrote:

> Hi Nagesh,
>
> > I have looked at the function 'g' that you have mentioned; it seems that
> > this function assigns to the local variable 'y', no matter the
> coordinates.
> > However, I am interested in imposing Neumann conditions only at the end
> of
> > the plate. (Apart from the "Tip loading")
>
>
> No, the funktion g takes an entity and a local coordinate, so it can
> return different values for each point. The main question is which
> types of loads to consider in the Neumann boundary condition. If you
> support arbitrary stresses this is a full tensor, but perhaps you are
> fine with a prescribed traction, so that g only returns a vector?!
>
> Depending on the formulation the return value of g changes and the
> weak form changes.
>
> So this is what needs to be answered first and then you can update the
> implementation.
>
> That said, we are happily accepting pathces to improve the current
> implementation ;-)
>
> Best
> Christian
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.dune-project.org/pipermail/dune-pdelab/attachments/20180603/6af6c790/attachment.htm>