Thin Shells vs Thick Shells

[rh142]

I have sqare tube that is 9 inch wide x 1 inch long x 1 inch thick. It is going to get loaded in compression along 1 inch length. Is it ok to model it as a shell element?

Thanks in advance
rh142

 

[corus]

No, use solid elements. It's clear that your minimum dimension is relatively large to the overall size.

corus

 

[rb1957]

tube ? what wall thk ?

 

[rh142]

rb1957,

The tube is 1 inch thick and 1 inch long. So its a short, stubby tube.

 

[rb1957]

sort of 9" square, 1" thk, 1" long ...

 

[GBor]

Is that like a 1"x1" square BEAM that is 9" long? I guess my coordinate system is confused by something that is wider than it is long.

Typically, you are looking for a thickness that is greater than 10% of the shorter aspect dimension before using solid elements, but this one is probably close enough considering the overall dimensions.

The real thought process is primarily how the structure will flex and probably how it is supported. If it is simply supported so that the end rotation is significant, you may want to try it with shells because brick elements are rotational rigid. If it is fixed on the ends, brick elements are likely more reasonable.

Since this is a fairly simple structure, I would get out a hand calculation and run a test. Build a model with bricks and one with shells. Simply support an end, or some other support scenario that Roark's has. Run both models, refine the mesh, run them again and see which one comes closer to the hand calculation.

What software are you using and how is your beam supported?

 

[rh142]

GBor,

I am using LS DYNA, I was thinking about using the TSHELL (Thick shell) option for this. At one end the tube is supporting a much larger plate (100x100x2)and it will probably by fillet welded on the outside. At the other end it will be supported on a plate again fillet welded, so the conditions would be close to pinned at both ends.

Thanks
rh142

 

[GBor]

Unfortunately, I have no experience with LS-DYNA, so I do not know the theoretical basis for their TSHELL option and its application. The only thick shell to which I am accustomed is for cored composites.

I suggest the model to hand calculation comparison approach. Just run a simply support beam with some load on the other end with each element type option and see which one gives better results. LS-DYNA, if I'm not mistaken, is a p-element convergence code with an explicit solver, so you may be able to get reasonable results either way.

 

[kellnerp]

If you really mean a 9" square tube with 1" thick walls and a 1" length this is more like a piece of drop.

In the first place if you weld something like this the weld is going to play a big part in the load path. Normally a shop would bevel the ends of the tube and with such a short length and the bevel you are going to end up with the weld playing a big part in the behaviour of the connection. Definitely not a place for shell elements although the plate would be. You might consider modeling the connection with solids and the plate with shells and join the two.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."