Compute Weld Stresses using Shell Mesh, ASME VIII Div 2, Annex 5-A.9

[Doodler3D]

Hi,

New FE user here. I ran a few test simulations for a nozzle pipe junction to be evaluated using shell elements. For the analysis, the internal pressure, constraints, mesh size, etc. have been kept identical. Thickness = 7 gauge carbon steel. I am particularly interested in the nodal forces of element on the toe of the weld on the longitudinal plane.

3D tet mesh:

2D mesh based on ASME VIII div. 2 Figure 5-A.9, weld throat length = 0.15" (oversized)

2D mesh based on ASME VIII div. 2 Figure 5-A.9, weld throat length = 0.135" (correct size)

What is the correct approach to size the weld throat length for the analysis? I'm a bit lost as I'm unable to validate the 2d and 3d analysis. I would believe that the top surface Von Mises stress at the node under consideration would be in the 600-650 psi range, but the method i've followed based on the guidelines in the code yields values that are about 100 psi higher.

Thank you

Reference:

 

[goutam_freelance]

To model the fillet consider the geometry as variable depth beam and calculate stiffness. Then consider uniform depth beam of same length. The stiffness should be same. Find the equivalent thickness by equating the above two stiffnesses.

 

[DriveMeNuts]

I've never used this method. It just seems too....... arbitrary and approximate.

 

[goutam_freelance]

I think where stress concentration exists it is better to use 3D elements.There are software like Nozzle Pro etc which are more convenient for these.

Engineers, think what we have done to the environment !

 

[BJI]

You are modelling mid-plane thickness so the second approach is not correct, you must consider shell thickness and external fillet weld leg length to determine the node junctions. However, if you did use the second model, then the stress should still be evaluated at the location of the physical weld toe, not the junction of the intersecting shells (likely resulting in a similar stress magnitude). The shell model should be slightly more conservative than the solid model, so don't expect an exact match, especially at discontinuities where you can't approximate the same localised restraint.

The OP is actually using 3D elements in all examples, one being solid elements and the latter being shell. Why not linearize the stress through the thickness in your solid model then directly compare the bending and membrane stress results between all approaches?

 

[Doodler3D]

Gentlemen, I appreciate the inputs, if it's in the ASME code, then why so much ambiguity and hate for Shell mesh at nozzle-shell interface?