I’m trying to connect 4 slabs so that the Tie holds the 4 vertical edges. I’ve tried various methods.
RigidBody or converting the slabs to solid works best.
I’ve tried setting the constraint range and making several TIE constraints, but I can’t seem to get it to work. In a simple model with two plates in one plane, there’s no problem.
Is it physically possible to connect 4 slabs so that one slab edge is simultaneously connected to 3 other edges in a soft manner?
Christmas tree ? Did you try the Model → Node → Merge Coincident Nodes tool (works if nodes are coincident or nearly coincident - you can set the tolerance).
If merging coincident nodes doesn’t work (part compounding may fail in such cases with shells too), rigid body constraint can make sense too. Or try merging parts in CAD software. Tie constraints and tied contact can be a bit tricky to use here.
“Merge Coincident Nodes” is great, but in this case it doesn’t help. I moved the slabs apart to make selecting edges easier. The first version was a precise fit and a tight mesh. Merging the slabs into a single object in FreeCAD also causes a problem. One object and two surfaces have intersecting normal directions.
I had to create an intermediate 1mm x 1mm solid for the entire height and create a slab + cuboid surface constraint. This is a last resort, as it will make the model larger
Regarding your tie constraint setup, you don’t have to pick the entire faces - their edges in the middle should suffice. You may just have to increase the position tolerance. And keep in mind that two tie constraints with the same surface used as a slave lead to overconstraint.
You may need to use large position tolerance for tie constraints here because the parts are really thick and extruded elements overlap. CalculiX takes the shell thicknesses into account for tie constraints. Abaqus has an option to ignore them.
Still, it would be best to merge the geometry in CAD software so that the faces share the same edge. It might be possible with more manual workflows in FreeCAD. I can’t test it now, but you could ask on the FreeCAD forum if it’s doable for this geometry without messing up the normals.
Sometimes mixed approach also makes sense - you merge some geometries/meshes and connect the merged regions together with tie constraints.
I think the model almost works. They’re merged into two plates. A similar mesh must be present on the shared edge. The model was created to practice TIE binding. I don’t have time to check it more thoroughly because I’d have to slice the plates more precisely. Nice result of the frequency analysis
