boundary conditions axisymmetric tube

[moppe]

Hey guys,

I have a question which is pretty basic, just started on a company and I am right now doing analysis on a aluminium tube which has pressure on a section inside. I of course want to approximate this as follows:

------------------------------------
^^^^^^^^
pppppppppp

My question is the boundary conditions, since I cannot lock the ends (we are interested in the movement there as well). What is your opinion.

Also, what element do you think I should use?

Regards

Mårten

 

[GBor]

moppe,

I want to make certain I understand your situation. You have a tube and you have only modeled 1/4 of its cross-section, but you have modeled its length? You want to place proper boundary conditions on it, but the end needs to be able to freely extend?

If this is the case, then place the symmetric boundary conditions on each "cut" edge, and place an axial boundary condition in the center of the pipe. Basically, if you have a cylinder, model 1/8 of the cylinder by slicing it with each of the three coordinate axes. At eac of the slice planes, place the appropriate boundary conditions.

You may also want to search these threads for a paper by JohnHors that address minimal constraints.

 

[rb1957]

you have to take out the 6 rigid body motions.

one end (either one) needs to be constrained longitudinally. i'd assume no end moment restraint.

the longitudinal sides end to react tangental load. these could have moment restraint if you think it's relevant (probably isn't).

a problem i think you'll have with modelling a axi-symmetric piece of the pipe is that the displacements probably won't work out right. i think you'll have better luck with a complete pipe ... i'd try both to see !

think too about the endload due to pressure. maybe your pipe is free to expand/contract so the poisson effects won't create load.

 

[Ingseven]

I am not an expert in FEM, but shell elements would be reasonable. The boundary question is always there. The best way is to model the whole tube and you should avoid rotation by restraining the model. You don’t need to restrain in axial direction because the pressure is radial (if I understood right). But a restrain in the center is correct (as GBor said).

 

[corus]

For a tube you need axisymmetric elements. If you're wanting to model details, say a change in shape whilst remaining axisymmetric, then use 2D axisymmetric elements, otherwise you can get away with axisymmetric shell elements. For the restraints, extend the model 6 wavelenegths away from any detail, where a wavelength is given by the fourth root of (R^2t^2/(3(1-v^2))). At this distance any sinosoidal change in displacment will be damped out sufficiently. At one end apply an axial restraint to prevent rigid body motion, and also apply a rotational restraint for quasi symmetry conditions. At 6 wavelengths way though any rotation of the 'free' end would be minimal anyway. You don't need any radial restraints to the model unless, say, the end is connected to another momeber, or has a bolted on flange, say.

corus

 

[yugabalan]

Hello all,

Just a silly concern. If this is axisymmetric problem, why not just use an axisymmetric element with 2 DOFs. Is it necassary to to model a fourth or an eighth model using shell elements? Or is there something I am not catching here?

This should be a pure 2D problem. Cheers!

Yugabalan K