How to model Interference fit and Welded joints

[thirumuruga]

Hi,
I performing a frequency response analysis of an assembly containing two concentric cylinders - A cylindrical component and its casing. To simulate the interference fit,I simply defined both of the cylinders as deformable bodies and gave a GLUE contact with ERROR=0.0001 . Is this the right way to model this?

Another task I need to performance is that I need to check the frequency response when the components are not assembled by press fit but welded with each other. How do I proceed with modelling the welded joint between my components?

Using MSC NASTRAN and PATRAN 2012

 

[aerostress82]

Deformable Bodies + GLUE contact looks about right for the frequency response analysis. However I'm not an expert with this, but that's how I would model it, too.


For welds, if you are expected to check the fatigue life at those weld locations, you can model them as vertical shell elements connecting 1st part to the 2nd part at the exact weld locations - thickness of these shell elements should be assigned depending on your weld thickness (you can probably see it from the drawing).
If your weld region has the 2 parts touching each other at 90degrees, then you may define the weld as 1 diagonal + 1 vertical row of shell elements between the 2 parts.


IF you don't need to analyze weld regions in detail at all and only need them for the frequency response analysis, then instead of the shell elements I mentioned above, you can just put RBE2 elements. Though, your nodes should be at "nice" locations on both parts to have good (vertical) RBE2 connections. You want equal number of nodes on both sides at the best projection locations on both sides. You don't even need to apply "vertical + diagonal" for RBE2s that I mentioned above. They were only for shells.

Hope this is enough..

Spaceship!!


Aerospace Engineer, M.Sc. / Aircraft Stress Engineer with 7 years of experience
(United States)