Hi,

I'm performing a (linear) buckling analysis on a tank filled with liquid.

I've fixed its legs and put variable pressure to simulate liquid load:

I'm using shell elements.

In this kind of tanks, the main concern is buckling of the legs, therefore it's important to perform FEA to determine optimal geoemtry and sheet metal thickness.

When I run the analysis I get all negative buckling load factors caused by pressure. Like this:

This tank is going to be emptied very slowly, therefore negative pressure is not expected, and so it won't buckle that way.

I need to find positive buckling load factors, but I can't because the max number of modes you can set is 200. The 200th mode is still negative.

My questions are:

- is there a way to "filter out" negative buckling modes ?

In frequency for example there's an option to find frequencies above a certain value in order to exclude low frequencies (like rigid body motions). I was wondering if buckling had a similar option.

- Given the limit of 200 modes, how can you calculate positive BLF's

I was thinking about replacing pressure with a remote load with rigid connection: this way you can find some positive BLF's.

My fear is that this would make my buckling load factor results anyway invalid.

Any experiences with this situation or suggestions ?

Alex

Reducing the model to just the legs if the tank is much less stiff than the legs seems like a sensible idea. I would do a linear buckling estimate of the complete leg assembly with a vertical load to see what sort of modes are likely and then do a NL arc length or displacement control analysis - I doubt there is any snap back that would be of interest and the arc length control has been a bit busted since like 2012 or so - seems to take inordinately long time to solve. They may have addressed it in later releases. Either way, you typically need to add some out of plane displacements/loads in an NL analysis of this type if they will not develop on their own which may not be the case here as the legs look tapered. For other modes to emerge (like an axial rotation) it might be required. It doesn't hurt in any case as they are insignificant in the big scheme of things. I would add some trivial load to any surface involved in a low order linear buckling mode. This will allow the out of plane forces to develop. I would strongly discourage using any sort of symmetry model as it will suppress any non symmetrical modes, though in this case, if it is leg crippling that initiates the failure it wouldn't matter much. Then again I can't really tell from the image exactly what you are dealing with here in terms of leg construction like whether its an enclosed tube or 3 plates with a side open.

In other codes there is often a capability to add very small displacements from a frequency analysis or buckling analysis to impose non perfection into a model so that out of plane forces develop naturally. Though, I don't think the jury if fully in on how well it mimics actual imperfection.