Joining more mesh

Andrea

Hi have a big plated structures (support for a foundry ladle) and I have a big problems to mesh all the structures at the same time. Do I decided to mesh separately. Importing on Mecway the meshes I have two plates that have to been joined with bonded contact but I receive the attached message. I tried to move the parts (mesh) and to adjust the tolerance on bodend contact window with no succes. Any idea?

Thanks in advance

On the pictures only two parts of the structures...other parts must be joined with bonded contact

Victor

Have you tried increasing the tolerance to bigger than the 2e-2 (m) distance shown in the error message? The tolerance it identifies is kind of weird, I suppose that's the automatically generated one?

Or, if you use the non-TIE option, it allows unlimited distance as long as the slave nodes are in the direction of the master face normals. So for flat plates like this, it should connect with no problems.

Andrea

With *TIE I have always problem on MPC! But the worst problem is the error on Jacobian which false all results

I'll switch to *EQUATION but I will work on a simple model to understand if the problem is due to the type of elements.

Regards

Andrea

I think that's better to avoid *TIE with shell.
I thought that it has no problems with 2D elements because CCX expands them to 3D elements but with shell not every nodes partecipates to *TIE, so the results is not acceptable.CCX manual about this section must be interpreted literally

Victor

I'd like to isolate what's going wrong here because it sounds like a general problem. If you can post or privately email me the .liml file, that would be helpful. Maybe I need to put in some warning about this.

Andrea

I share here because is a simple model of a plate to plate joint.
Look to the solution when the master/slave nodes are placed on the same plane.
I tried to change tolerance but I never obtain the right solution.
Last chance is to try applying shell offset.

Other problem is how CCX expands shell elements. If the master/slave surface are not placed at the right distance you will obtain two different thickness

Regards

Andrea

I'm trying to glue the plates with RBE. Seems to be efficient and doesn't build thousands of *Equation card.

In the meantime I tried to apply offset to the shell and also these parameters has effect on the results

Victor

Any luck with rigid body?

I see what you mean about reading the manual. It explicitly says shells aren't allowed so I've added an error message to Mecway if you try to use *TIE with shells or beams.

I did find that it seems to work OK as long as the shell T-joints aren't on the Master surface. But I wouldn't be confident that it's completely right.

Sergio

That´s the things that keep me away from using shells on CCX, have know that we cannot use TIE, but also the artifacts that it creates at the angles (knots) or find weird how the boundary conditions are traslasted to the expanded nodes. Another drawback is that is not possible to compute stress linearization with Mecway.

I always preffer to make a mesh extrusion and work with solid elements, at least with small models.

Andrea

Creating double T-Joint there is no right solution.
I think that's no possible to assign a general rule for shell/tie combination

Andrea

I'm doing test about RBE to joint the mesh.
Agreement with classic theory (Saint Venant beams) is very good. I'm testing a frame modelled with two mesh block bonded with RBE (Mesh onto Salome and imported). The same frame is modelled with beam element with a FEA for steel structures (that I utilize to design steel frames) and in modal analysis the results are very close.
Obviously stress field is different because in shell model there are singularity not present on beam model. Some differences I found on reactions but the model weight is slightly different because of the presence of the plates on shell model