I created 4 models out of shell elements primarily and I want to merge them together. I have moved the node coordinates of each model separately so that they are in the correct geometrical position to be merged.
1. Should I renumber the nodes, shells and members to avoid overwriting and losing these elements?
2. For plates (for example: bent plate flanges) that mate or partially mate with the plates of the adjacent model, should the node numbers for the corresponding nodes match if they occupy the same coordinates? I assume that this will become a fixed connection in the merged model. True?
3. For plates that mate in the same plane but the nodes are not in the same position, Will Ram mesh the portion of the plates that overlap in a similar pattern such that the FEM nodes are shared or connected?
4. Most of the 4 models are constructed with 1/4" plate and bent plate. There are a few plate stiffeners made from small HSS square tubes.The stiffeners have nodes within the plane of the shell elements but not necessarily at the outer edge of the shell it is attached to or welded. Also, I set the cardinal points of the stiffeners to correspond to the plane of the shell element. Will these stiffeners behave the way I am thinking they should?
Your answers help a lot and bring up more questions. From my sketch below, the casing is in the middle model above and the bent plate double channel with the stiffeners are in the base model on the right above. Based on your answers:
1. I think the top flanges of the double channels will mesh with the casing.
2. But the stiffeners will only attach to the edges of the flanges and not the casing. I would have to define separate casing shell elements that match the ends of the stiffeners to make them work.
Am I on the right track?
There are two basic approaches.
The main topic is still "Merging Models with Shell Elements". I understand from your post above that the plates will overlap after merging. My question today is about the resulting connection between the two models.
See my sketch below for a simple example. The Green model has Plate A that will be in the same plane with the Red model Plate B when they are merged using the OpenRE merge tool. How should I model the shell elements in each model so that the physical plates are "welded" along the edges of the overlapping plates?
Should I subdivide the shells before merging to provide boundaries along the "future" overlapped areas? The models are fairly complex and it would be difficult to do after a merge.
Shells that intersect with no common edge do not mesh together:
If the weld edges are lines with members (any section), then that would influence the mesh:
In that image, the shells in the intersection region are doubles. You could leave it that way or mesh the model manually, using Process - Segment Selection, then delete the doubles and make the remaining shells in the intersection thicker.
From your example, nodes with the same coordinates as Nodes 5,7,9,10 should be present in both models before merging. Beam elements should be present in one of the models and they should be connected between these nodes. As shown above, 3 beams would mesh and connect the plates on 3 sides. A 4th beam would connect the plates on the 4th side (N9-N10).
For a beam element to be meshed with a shell element, the nodes at each end of a beam must be listed in the definition of that shell. Or the beam must be connected to another beam that is being meshed with the shell as the beams in your example connect to N5 and N7.
Did I get it right? I won't need to have a thicker plate. A good connection (weld) around the perimeter intersection is all I need. However, it looks like I need to do some more work before I do any merging!
It looks like the 4th beam N9-N10 is not required to mesh and connect or "weld" the two plates together. I think that it meshes because the edge of the left plate cuts completely across the right plate. But the edges of the right plate do not cut completely across the left plate, so beam elements are needed to make the meshing and connections work. I'm not sure why exactly this is true.
Seth, could jump in, look at these 2 last posts and straighten me out?