Extracting reaction forces from a shell model

[imagitec]

I am analyzing a thin-walled bellows that is welded to a thick ring; they are of different materials. The loading is due to the differential thermal expansion between the two. Because of its thin section, the bellows must (and should) be modeled as shell elements. I'm using CosmosWorks 2004, which does not support mixing of solid and shell meshes in the same model, so I'm using component contribution analysis: I'm analyzing the bellows by itself, with the ring replaced by boundary conditions. Then I'd like to use the reaction forces at that boundary as the input for a model of the ring. Although I'm using CW, I hope people using other packages might have some insight, too.

I'd also appreciate criticism of my approach to solving this problem. I have done a hand calculation, treating the bellows as a thin walled cylinder, as a first estimate.

I modeled the bellows using shell elements, defined a uniform input temperature of 500 C in the study properties dialog, and applied a forced displacement to the face that is welded to the ring. The magnitude is equal to the differential thermal expansion between the two parts; this is conservative, since it assumes the ring is infinitely stiff. I ran the analysis.

I measured the reaction force on the face with the forced displacement, and it was nearly zero. That made sense when I thought about it: it's a statics problem, so the normal forces on the cylindrical face cancel.

I added split lines so I could extract the forces from just a quarter of the cylinder; x- and y-components would not cancel. I ran the analysis again and got nearly the same result.

As I'm writing this, I'm thinking it through. In CW, you pick a solid model face and CW selects the underlying shell mesh that the face was used to define. So the force summation also includes the resultant of the hoop stress over the cross-section, which again cancels out the forces normal to the face.

How can I extract the normal force that would be transferred to the mating part? If you use another package but could describe the process in it, that might me find the right path in CW. As of now, my next step will probably be to export the node results to Excel.

Thanks,
Rob


Rob Campbell

http://www.livejournal.com/users/robcampbell

 

[vonlueke]

Rob: When you say "normal force" in your last paragraph, are you referring to the radial force the ring applies to the bellows? Or the radial force the bellows applies to the ring? I.e., are you referring to the bellows shell element model and analysis, or the ring solid element model? Is the bellows inside the ring, or is the ring inside the bellows? Off-hand, your enforced displacement method sounds like a valid approach. So far, it sounds like your split lines might change the problem at hand entirely.

Without using the split lines, to obtain the hoop force in the shell elements, output the shell element hoop-direction element forces. To obtain the radial force the shell elements apply to the ring, perhaps try displaying the shell element z-direction element forces.