Global-local w/ solid model - coupling boundary nodes to new mesh?

[McLeod]

I'm trying to do a global-local analysis of a solid model in Nastran for Windows 2003. The global model has been solved, and the original nodes on the sub-structure boundary surfaces have 6dof constraints and displacement loads derived from the output vectors. The tricky part is trying to tie the original boundary nodes to the new local mesh. I've tried keeping the original discretization on the boundary surfaces, but the mesher consistently fails to include all of the hard points, or generates distorted elements that fail Nastran's checks during the solution. Remesh>Refine only works with 2D elements. The automatic commands for mesh transitions, closest link, etc. don't connect the correct nodes.

Is there a straightforward method for doing 3D global-local in N4W?

 

[SWComposites]

I don't think it is very easy to do in N4W. However, it is extremely easy to do in Abaqus even with completely different meshes in the two models.

 

[fkmeyers]

If you are talking about NEiNastran for Windows you do this using the build in interpolation commands. It would be a DISPINTERPOLATE command using the displacements from the global model which are automatically interpolated onto the local one.

Otherwise for MSC you are out of luck. Some suggestions:

1. Keep the nodes on the boundary meshed the same. Then convert these output displacements in FEMAP to enforced displacements as loads for the local model.

2. Use MPC equations to enforce the displacement field of the global model.

3. Use MPC equations to interpolate the global displacments to local nodes.

 

[ThomasH]

If the problem is to include the hardpoints in the mesh it's not Nastran (MSC or NE) that fail, its FEMAP.

Why not make sure the nodes have identical id's in both models, then read the data from local to excel, and then from excel to global.

If you have the full Nastran (from your post I unfortunately doubt that) you kan use superelements. Otherwise you can study the comcept of superelement, it might give some ideas.

Good Luck

Thomas

 

[McLeod]

This is MSC.Nastran for Windows, not NENastran or full Nastran. I've read about Nastran's superelements and Patran's displacement fields, but have access to neither at the moment.

Although I understand how to apply MPC equations, I have no idea how to formulate such equations for an arbitrary surface with an irregular distribution of dependent and independent nodes.

What does transferring the data to and from excel accomplish?

 

[ThomasH]

Once you have your local mesh you can create the load using excel. You take the displacements from the global analysis and apply them as loads in the local analysis.

Or you can write your own Refine commands in the API.

Regards

Thomas