Shell buckling

[MUMAR12]

Hey!!!

I designed a simple long thin open ended cylinder in ANSYS and the type of loading is hydrostatic pressure.
I calculated the buckling pressures for different modes shapes (n=1,2,3...) using Von-mises and Donnel's equation.Both have used classical simply supported boundary conditions.I constrained all translations(x,y,z) on one end and two translation(x,y) on other end as classical simply supported conditions.There is a huge difference in theoretical and ansys results.

I am sure that that is happening because of boundary conditions.Can anyone explain or tell me what are classical simply supported boundary conditions for cylindrical shell?

Regards

 

[rb1957]

if you're constraining the perimeter in X- and Y- i suspect there'd be a problem.

i'd play with your model with different constraint scenarios untill you get correct hoop behaviour (under tension).

i've also seen odd behaviour depending on how i created the geometry.

 

[MUMAR12]

u r right about constraining the perimeter! what would u suggest as boundary conditions?

 

[rb1957]

i'd contrain the perimeter axially (X-) at each node, laterally (Y-) at two nodes (say a diameter), and up/down (Z-) at one node (pick one of the Y- nodes).

then see if you get good "hoop" behaviour.

 

[MUMAR12]

Thanks..I am going to try the ones suggested by you..I am attaching my file in case you want to have a look at it...

 

[TGS4]

If you want good "hoop" behavior, how about starting with a cylindrical coordinate system and inputting your BCs in that csys?

 

[MUMAR12]

I actually haven't used cylindrical coordinates?

 

[TGS4]

You're modelling a cylinder under internal pressure and you haven't ever used cylindrical coordinate systems before and you're wondering why you have problems?

 

[MUMAR12]

I am not designing a cylinder under internal pressure.I am designing a cylinder under hydrostatic pressure.Going through literature,I came to know that we can model it using Cartesian Coordinates.It shouldn't effect the solution.

Regards

 

[TGS4]

Um - OK. If you want to believe that, sure.

It's a cylinder. That's the reason that cylindrical coordinate systems were invented... Next you'll be asking about spheres, and then admitting that you've never used spherical coordinate systems either.

 

[missil3]

@TGS4 : I am confused, how will choice of coordinate systems affect solution if the BC's can be accurately described with a Cartesian co-ordinate system.

Isn't there a way to use a local co-ordinate system instead of the global co-ordinate system ?

 

[TGS4]

Please tell me how you can define u-theta=0 or u-r=0 from a cylindrical coordinate in a Cartesian coordinate system? Or even better - how would you define rot-theta=0?

Sure, you could use a local cylindrical coordinate system. That's the point.

 

[missil3]

You can define those with a local Cartesian coordinate system

 

[TGS4]

You can define those with a local Cartesian coordinate system

Sure - with a lot of trig functions: sine, cosine, etc. My point is that if you are:
a) modelling something that is a cylinder, and
b) applying boundary conditions that would be best described in a cylindrical coordinate system, and
c) yet haven't ever used cylindrical coordinate systems before, and finally
d) have unexplained problems

then you may have a problem with your boundary conditions (and the coordinate system that you have defined them in).