Contacting structure in frequency response analysis

[netjack]

Hi,
I used RBE2s to connect two bolted structures together and run frequency response analysis (shell-solid). Ofcourse, the regions where there is no RBE2, structures may penetrate each other in modal analysis as well as in frequency response. My collegue suggest that I should use glued contact (so that shell is rigidly connected to solid in all contact area) will give more accurate results in freq response. What is the best method to model bolted structures for frequency response analysis.

 

[Stringmaker]

Have you by chance tried connecting them with some beam elements having similar stiffness to the bolts? I've used this for certain situations in the past with what I thought were good results. It eliminates the need for contact elements as well. The only limitation I can see is if you're concerned about the stresses close to the interface itself.

 

[netjack]

I am not interested in stresses in and around bolts. I am only looking for a good method to model contactig surfaces in modal analysis. Some softwares have linear contact (glued) which you can run modal analysis. Actually, I believe that this contact is nothing but a lot of RBE2s generated automatically between contacting surface nodes. So, if I use RBE2s to connect all nodes at contacting surfaces, would it be correct for modal analysis.

 

[GregLocock]

I suggest you build a separate model of the bolted joint, establish a linear stiffness model for that, and then substitute that value back into your dynamic model.

Contact conditions have no place in a linear modal analysis. In the real world we call them rattles and eliminate them before testing.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.