how to model border conditions :part welded to a non rigid body

[SofieneFEA]

Hello Everybody,

I’m now working a welded structure. The purpose of the study is to decrease the weight and keep a safety factor equal to 1.5

I tried to do a local study of one part of this structure, so I decided to isolate this part (to refine mesh and gain calculation time). The problem is that this part (formed by plates and sections) is assembled to the whole structure by welding.

I m asking how to model the border conditions in case of an isolated study of a part welded to a non rigid body. (Zero displacements border condition seams to give wrong results).

Thank you and Best regards.

 

[corus]

Tie the nodes across at the weld, or make the structure continuous and later calculate the forces that the welds must transmit, if you can't do it by hand.

corus

 

[GBor]

Depending on your situation, there are a couple of generally used methods:

1) calculate the equivalent spring stiffness of the rest of your part, or at least the parts to which your local model is attached, and attach springs to your local model, then fix your springs with stiff boundary conditions. This is particularly easy if your local model is welded to beams.

2) use your global model to determine displacements at locations specific to your local model, then apply prescribed displacements to contort your local model. You may have to be creative about how you apply the displacements so that you don't end up with some localized stresses overpowering your "real" results, but it can be done.

Garland E. Borowski, PE
Borowski Engineering & Analytical Services, Inc.
Lower Alabama SolidWorks Users Group

 

[rb1957]

i'd add elastic supports (springs, axial elements) between your model and the hard suppression. i'd use three at each grid (one in each direction).

 

[prost]

Doesn't tying the nodes together cause unnatural (that is, not really there) singularities at the tied together nodes? Sounds as if you are creating more problems then you are solving.

If it were me, I'd take rb1957's suggestion to model the supports as elastic. The question is how elastic? That's something you'll have to experiment with--forgetting about Poisson's entirely, just changing the stiffness of the elastic support could or could not change your results significantly. If you use springs, you could estimate the spring constant by dividing the Young's modulus of the support structure you are trying to model by the thickness of said support structure where the welds are used. It's not perfect, but it gives you something to start with.