FEA help and linearisation ..........

[sampurna]

Hello ppl ,

I have a query regarding a FEA analysis of a pressure vessel:

Suppose i have a shell and inside to it a bracket is welded and along with a pad ,

I have got the stress results form Ansys , i see that the maximum stress in the shell is the corner of the bracket pad .these are due to the stress singularities in the model or due to the contact stresses.

Now i want to do the linearisation for the model. what i have been told is to check the results at one node away from the maximum node ,

Now after doing linearisation at the one node away from maximum , i am into confused state , on comparing the membrane stresses and membrane+bending stresses.

I have heard of a term called the linearisation at local area and linearisation at transition area where we compare the local area linearisation membrane stresses to allowable and membrane +bending to 1.5 Allowable.

and linearisation stress at transition area-- membrane stress is compared to 1.5 allowable and membrane +bending to 3 times of allowable .

Now my query is which location will be the local area and which one will be the transition area in a analysis .

Please find the attachment.

 

[MikeG7]

Sampurna

You don't say what code you design to but if you are doing this to ASME VIII Div.2 there is guidance in that code where to take the linearization at.
Mikeg7

 

[sampurna]

Hello Mike ,

Yes i am using ASME code for the design.

 

[TGS4]

Please ensure that are are using the most recent edition of Division 2. There is plenty of guidance in there regarding how to perform the linearization. If, after going through that you still have questions, please come back as ask - with specific reference to the specific paragraphs that are giving you problems. Note that it is my opinion (and the Code also does not contain any specifics in this regard) that you should not perform te linearization away from the peak stress, unless in doing so you use an inappropriate SCL.

 

[sampurna]

TGS4 so u mean to say m, that linearisation should be done at the peak node.

 

[TGS4]

Yup. But it is a little more nuanced than that. Read Annex 5-A. And WRC 429. And a whole host of ASME PVP papers on the topic.

 

[sampurna]

Ok , what if my peak stress are at the weld location, ie suppose if i have a bracket and a pad , then the peak stress will be at the corners of the bracket where it is touching the pad , where in reality weld will be der, So i was takng one node away from the peak stress and doing linearisation ,

is it correct approach.

 

[TGS4]

If you explicitly model the weld, then your SCL will start at the town of the weld. If you don't model the weld, then your SCL will start at the edge of the pad.

This whole business of starting one node away is meaningless. I have no idea what your mesh density is - nor should it matter. Follow the guidelines in Annex 5-A, we out it in for a reason...

If you think that some peak surface stresses will mess up your linearization (they won't), then do an elastic-plastic analysis.