*tie constraint for multiple mesh

[S-Y.Chen]

If there are separate boxes for master and slave faces when selecting the faces for tie constraint, and if you can find the nodes/elements on master and slave faces, separately, there should be no confusion for master/slave nodes ?

You are right, but my comments above were about the current situation (bodies with space between).
Currently if you analyse a multi-solides model with contacts on coïncident slave and master faces (lets say Part_Compound with 2 plates) you will not get expected behavior. Slave and master nodes seem to be fused. I assume that's due to single mesh as said in early posts of this thread. If you want I will send an exemple later, no time right now sorry.
BTW I fully agree with your request in order to allow contacts on multi-bodies meshing with coïncident faces with no space between. Would be a huge enhencement.

I believe a lot of persons think these features are important. But how can we make it happen ?

[-alex-]

But how can we make it happen ?

Ah, good question. By coding I assume... but I have no skills about that so...

If I understand correctly there is 2 strategies:
- improve the current algorithm as @ur said
- implement multi-mesh analysis

I would prefer the second strategy, I assume this would be a convenient way to allocate each slave or master face to the suitable meshed body, efficiently in non ambiguous manner.
IIRC @bernd has an experimental multi-mesh FEM in his branch right?

ping

BTW how can we access your multi-mesh branch? Just by compiling it?

The reason is FEM's algorithm which decides which node belongs to tied face A and which node belongs to tied face B.
Decision is made by incidence only.
We need a much more sophisticated algo.

Ok, so do you think multi-mesh analysis would ease to know which node belong to the suitable contact face?
Any opinion much appreciated.

[davidosterberg]

I believe a lot of persons think these features are important. But how can we make it happen ?

As I tried to say above. I am planning to implement this. Unless somebody else is already working on it of course.

[S-Y.Chen]

I believe a lot of persons think these features are important. But how can we make it happen ?

As I tried to say above. I am planning to implement this. Unless somebody else is already working on it of course.

Good to hear this. I have been working on some FEM algorithms now and then. So please let me know if I can help.

[Jee-Bee]

issue #4560

[S-Y.Chen]

I have attached an example in the issue report.

Also in this post
https://forum.freecadweb.org/viewtopic.php?f=18&t=71154