FEM Analysis, AISC, API, ASME codes

[zkormos]

I'm trying to analyse a welded structure in Ansys Workbench (linear analysis). Mainly the principal and shear stresses are under the allowable stresses (according AISC principal<0.6*Fy, shear<0.4*Fy) but in some places (for ex. sharp edges, connections between plates) I get stresses over the allowable. With finer mesh these areas will be smaller but even higher stressed.
Is there any standard (AISC, API, ASME) which describes how to handle these areas/results? Or generally an american standard which describes how to handle FEA?

Your comments would be really appreciated.

 

[rickfischer51]

I dont think you can get accurate weld stresses directly from FEA. For a variety of reasons, it just doesn't work very well. I think the best thing to do is to take force and moment reactions from FEA and use them with the closed-form calculations specified by the code you are using. I write the reactions to a text file and then paste them into a spreadsheet. I then calculate section properties for each weld group using the methods given in Design of Weldments by Blodgett or Shigley's mechanical design text, which I use with the reactions to get weld stresses. I then compare the weld stresses to my code specified allowable.

 

[JoshPlumSE]

Where to start:

1) It sounds like you may be referring to AISC 9th edition (1989 or earlier) as your reference. Referring to a code that is 25 years old seems to be unwise to me. We're now on the 14th edition of the AISC code, making your code reference out of date by 5 code releases.

2) Even back in the 9th edition days, the allowable stresses were not so simplistic. They will instead depend on b/t ratios, buckling calculations and such. So, there is much more variation than that.

3) The newer codes are force based, not stress based. Therefore, they would calculate a total allowable force that is allowed to be used for that connection, not a stress. If you have to extend that concept back to the 9th edition days, then I suppose it would be an average stress in the area of loading... The 0.4*Fy value, in particular, was the average shear stress in the web, not the maximum shear stress.