Nonlinear buckling of thin walled cylinder: loading and unloading

[futer8]

I have been trying to calculate accumulated plastic strains in cylindrical storage tank (20 m height, 20 m diameter) that lost stability due to internal underpressure. The shape of the container after being subject to buckling is given. So basically the analysis should consist of 4 steps:
- recreating loads that caused the given post-buckling shape and calculating it
- removing the loads (underpressure)
- if needed, pressurizing the container so that it comes back to its original shape
- checking the plastic strains.

I cannot get past the point 1. I recreate the post-buckling shape but in order to achieve convergence I'm forced to use numerical damping. When I try to unload I have to ramp the load to zero with damping working from the 1st substep (no convergence at all if I remove the stabilization). The problem is that the deformation is getting larger when I'm removing the load. I understand that I simply passed the bifurcation point so in general case it may happen, but the force-deflection curve I get during the loading phase is monotonic due to stabilization forces. How do I "move back" the same way?

I use Ansys, regular 4-node bending-membrane elements (SHELL181).

 

[rb1957]

"the deformation is getting larger when I'm removing the load" ... isn't that the point ?

i think that the deformation due to the underpressure is less than the final state (assuming your FEA is correct); and it sounds reasonable that the deformed structure can't support it's weight (like it could in the undeformed state).

so you have to apply some loading to initiate collapse, then unload (return to atmospheric conditions) to get the final deformed shape.

Quando Omni Flunkus Moritati