boundary faces

[edljk]

I wanted to share an observation that you may have in mind but which could get some troubles for a naive user.
Defining faces using a characteristic function of the boundary like

addfaceset!(grid, "∂Ω", x -> onboundary(x))

may select faces that are actually not on the boundary. See the left bottom corner on the illustration

Perhaps it could be interesting to define an addfaceset! function which only selects boundary faces which is perhaps the most common expected behavior defining a faceset.

[edljk]

Just for the record: to avoid to select interior faces, I always intersect my facesets with the boundary faceset (which is identified as the set of faces which are associated to only one cell)

[termi-official]

We should be able to resolve this with the help of the topology. A boundary face is by definition a local face with no corresponding neighbor.

[termi-official]

How have you generated and loaded the grid?

[edljk]

I am using the Triangle mesh library to get nodes and cells.
The grid is loaded by a direct call

grid = Ferrite.Grid(cells, nodes)

[graphitical]

I'm interested in this idea and the open issue.

How can I check/plot the facesets individually? None of the FerriteViz plots seem to take a facesets kwarg. Maybe this is the first step, implementing a simple way to show facesets on a wireframe?

[termi-official]

Hi Daniel, thanks for reaching out. I had no time to implement any kind of infrastructure visualization for boundary conditions yet but I opened an issue here for the facesets. Should be straight forward to implement if you are interested in these. If you run into issues feel free to reach me again.
However, while it is nice to have visualizations at hand, #595 should be straightforward to solve without any kind of visualization. So, if you are not interested in visualization and only in FEM, then feel free to to skip Ferrite-FEM/FerriteViz.jl#14 . :)

[termi-official]

This should be fixed in #606 .