Component mode synthesis reduced-order matrices do not match analytical solution. Bug?

Aaron McCrary

I created a component mode synthesis model of a cylinder to export the reduced-order mass and stiffness matrices; however, when I compare the matrices to those determined analytically, I see completely different results. This results in two very different solutions to a flexible body dynamics model. The cylinder has two boundary nodes: at the face center of either end. I am using the CBN method of arriving at the matrices. Is this a bug with the software? I confirmed that the matrices determined analytically match what we would expect, so the issue comes from Hypermesh.

Output from hypermesh (mass matrix top, stiffness matrix bottom):

 

Analytical solution (mass matrix top, stiffness matrix bottom):

 

Thank you

yuhaohe

There is some assumptions behind the analytical solution, and we need to match those assumptions in simulation to get the results aligned.

In this case, the two rigid RBE2s where the boundary is applied need to include all the nodes at each ends as slave nodes. Currently, the RBE2 only connects to a few nodes in the center of each end section.