FEA of welded constructions 2

netjack,
your question is actually very wide ranging and there is no one answer - or rather the answer is "it depends on ..."

1) What you are designing and the types of welds. This will influence if there is a design code (eg British Standards, API ..) available.

2) The factor of safety that is required, do you need a minimum weight design. If you do then design codes may not be the best way to go.

3) What are the modes of failure that are being designed for. Static strength, collapse load, fatigue, creep ... All of these will require different approaches.

4) Are weld defects being assessed?

As an example if you are designing against fatigue failure using BS7608, only nominal (direct and bending) stresses are required and so the global shell model will be sufficient.

Be more specific and I will answer in more detail

Best wishes
TERRY

netjack: (1) In general, I'd say your method is correct. Joint analysis is often nontrivial, to say the least, and any suggestions from anyone to streamline the process economically for a very large, shell-element global model are appreciated. The local model is necessary where global model joint geometry does not reasonably match actual joint geometry general configuration. How do you get loads applied accurately to boundaries of your local models? Mapped 3-D, nodal, enforced displacements?

(2) No comment on this product, except it sounds like it might give completely wrong answer if global model joint geometry does not reasonably match actual joint geometry general configuration.

Yes ..it is correct,provided you analyse the welds with appropriate boundary conditions.The actula loads acting in the welded region can be obtained from the stic(beam)model and applied on the 3D model.

U need to model the actual welds(including any deterioration in them in service,if already put to service)and do 3D analysis.Get the maximum stresses and compare with the allowable stresses to know the adequacy or otherwise of the welds.This method gives best results,though it requires some effort.No pains no gains!

Best Wishes

There is another way to do this, provide say 0.1mm distance between surf.1 and surf.2 and model 3d weld with say 0.5mm in diameter at your actual location and length and perform 3d analysis.

I would suggest the FEMFAT software. This can anylize several different weld seams from the fatigue side. I think this is the best fatigue software!