Pressure rating of thick cylinder

[vtmike]

Hi,

I have been trying to figure out the pressure rating of a thick cylinder under pressure and ran into a few problems,

1) I have come across different criterias for distinction between thick and thin cylinders (i.e. Thin cylinder if D/t>10 or if t<r/10 or if r>5*t). So I went with the one that occured in most books including Roark's handbook which is t<r/10.

2) I calculated the Von Mises stress using Lame's equations (attached file). But the Von Mises using these equations does not match with the FEA simulated results with a difference of 11% (attached file showing boundary conditions).

So, the question is am I using the right approach (i.e. correct formulaes) and why do you think my analytical Von Mises stress value is so far from the FEA Von Mises stress value?

 

[vtmike]

Here is the file showing boundary conditions

 

[metengr]

vtmike;
If you have access to ASME B&PV Code books, I would review Section VIII, Div 1 for thickness formulas. Despite this being structural with pressure, the same concept and formulas would apply for internal or external pressurized cylindrical components (pressure vessels).

 

[gwolf2]

That's a CAD picture, what does the mesh look like and what elements did you use?

gwolf

 

[vtmike]

I used solid brick elements with 4 elements across the thickness. Attached is an image of the mesh.

 

[LSThill]

Reference check review example on the PRESSURE DESIGN DRIECTORY EU design code. ANSYS SOFTWARE.

 

[corus]

It's difficult to say without looking at the FEA results, but your formula for Von Mises is only applicable at the inner radius. You really need to look at the mean (membrane) stress, say at the mid-thickness. At the external radius the principal stress, sig 3 = Po (the pressure) and hence the von mises stress will be larger than the value you calculated. At the mid-thickness the radial stress is some value not equal to zero, and hence the von mises stress will be larger also. For external pressure you should also look at buckling of course.

corus

 

[corus]

Incidentally, there's not much need for an FEA model in this case, but if anything you should have used a 2D axisymmetric model.

corus