Welding in structural member FEA 2

[kaffy]

Hi guys,

I am trying to run a linear static FEA on a structure made up of thick plates welded together. I would like to include weld in my simulation. I am using solid works simulation standard.

1: first idea was to use edge weld but for this I have to use shell mesh. As my the plate is 3/4" thick, I want to use solid mesh elements and I couldn't do it
2: Second idea was: I created a triangular weld as separate part and then in contact sets I use bonded wherever I was planning to weld it but since it was triangular and has sharp corners, I had stress singularities which I couldn't get rid of
3: Third idea was to use contact sets. I used face and edge bonded at weld locations and all other joint as non penetration. That works but doesn't reflect reality as actually a 1/4" surface is bonded instead of just an edge.

Any ideas about how to simulate that?

Thank You
Newbie

 

[FEA way]

Check the article "Modeling Welds for Finite Element Analysis (FEA)" on the Apollo Engineering website, it discusses several methods of weld modeling (including those that you’ve mentioned). All the approaches are used in an example and the conclusion is that the most accurate results can be obtained with welds modeled as separate parts attached to plates using bonded contacts.

 

[kaffy]

Thank you very much but what about the material data for welding electrode?
I could only find yield and tensile strength. Is there any reference showing the minimum young's modulus, poison ratio of welds?

 

[NRP99]

kaffy

It may be late to answer but still let me put some points. I do not use solidworks FEA but there are generic comments.

1-The normal practice we follow is to not explicitly model welds in the analysis. If welds are full pen, no need for weld evaluation/modelling. For welds other than full pen, take nodal resultant forces and moments at weld location and evaluate the weld with analytical calculation (based on structural code AISC/EN or relevant code applicable). For more details on calculation part, refer Shigley's Mechanical Engineering Design. The modeling of weld metal in the simulation is time consuming and may not change results much. If the welds are really needs to assessed, analytical calculations are sufficient.

2-The point 2 you wrote and the suggestion by "FEA way" is the correct approach to model weld, if you are still interested to explicitly model weld. But make sure that the load transfer happens only through weld metal in that case. Components to be welded should not be in direct contact with each other and the stress singularities in welds can be minimized by applying "small fillet radius" (~1/2mm) at the corners. But still it may remain there due to inherent corner stress. The evaluation of weld metal stress can be done based on engineering judgement which you may find difficult to convince the approver.

3-The weld metal properties generally are better/superior than the components to be joined due to the controlled manufacturing of welding rods. But for conservatism and as a normal practice in analysis you can consider that weld metal has same properties as that of component material.

 

[kaffy]

Thank You very much,