FEM: mesh of hole does not match mesh of round bar

[melomania]

Hello,

I have an assembly that has somewhere a hole with a round bar that exactly fits in this hole.

But when meshing the assembly, the mesh of the hole and the mesh of the bar are different.

The result is that the different parts do only seem to touch each other at the edges of the mesh,

resulting in high stresses. See picture.

I tried to made a dirived part from the assembly, but that one does not finish meshing. After 48 hours

the progress bar is still half way. While the one above did finish in about 15 minutes.

What can I do to make the meshes match?

[henderh]

Hi Melomania,

  <<What can I do to make the meshes match?>>

  Inventor Stress Analysis doesn't need conforming mesh nodes to transfer loads.  We use contact-based technology.

<<The result is that the different parts do only seem to touch each other at the edges of the mesh, resulting in high stresses>>

  As you've found the meshed model's geometry is not exactly the 'true' model geometry in some cases.  The way to not have polygonal cylindrical faces is to:

1)  Check the Mesh Settings option "Create Curved Mesh Elements"

2)  You've already found the "Maximum Turn Angle" setting in the Mesh Settings, which will increase the number of elements for curved faces

3)  Local mesh control to have a tighter mesh near areas of interest.  It will yield better results than a global setting to increase the mesh density.

  I've attached two images showing before and after option 1)

<<I tried to made a dirived part from the assembly, but that one does not finish meshing. After 48 hours

the progress bar is still half way. While the one above did finish in about 15 minutes.>>

  When we had only part based SA (R2009 and previous) we would caution users against deriving the assembly into a single part. This still applies for Assembly SA.  You no longer have the ability to create contacts like separation, have unique materials for parts, and cannot turn off the visiblitly of components to get a better view.

Now that we have parallelized meshing and solving in R2013, there is more reason not to derive the assembly. Most likely, it will much faster and more robust to mesh each component individually in it's own process simultaneously (there can be as many as cores available) than the derived assembly as one body in a single threaded mesh process.

A modeling analogy is like trying to model an entire complex machine as one part​ instead of an assembly of components. The feature tree will get very long and complicated...the same sort of issues will have to be faced by the mesher processing a derived assembly.

Hope this helps...please let us know if you have additional questions or comments.