how to set a 2D analysis for 1D results

[flyforever85]

I have a complex model with 1D beams and for some specific joints we would like to create the 2D model and have a better insight of the stress distribution.

for each beam that belongs to the joint I have the nodal forces (3 forces and 3 moments) in the local coordinate system. I also have different loading conditions and each force has the worst value in one of them
For example for Beam 1 the most critical axial load is in flight condition 1 while the most critical bending is in flight condition 3

Should I apply all the worst loads simultaneously to the 2D model? I'm afraid I'm going to oversize the structure at the point. Or shall I apply all the flight conditions ? Very time consuming but closer to reality

 

[FEA way]

This would be nice problem for submodelling analysis but unfortunately in Abaqus beams are not supported in submodelling. Solving the worst case scenario is usually a good idea since you can increase safety (which is of course especially important in aerospace enginnering). But in this case I think that it would be exaggeration. Thus I would check all flight conditions (you can use load cases to speed up the process).

 

[flyforever85]

Let's say I have a joint with 4 beams converging. I have the loads and the moments for each beams. Shall I apply these to only three beams and fix the fourth? Can I chose randomly?

 

[FEA way]

Yes, you will have to fix at least one beam (maybe it will be better to distribute partial constraints on more than one beam to avoid overstiffening it). The selection depends on the loads in that joint - try to model the worst case scenario.

 

[flyforever85]

Apparently I haven't been able to find a way out so I prefer to post here some pic to see if there is something I do completely wrong
The first pic id a 1D model of beams converging in a node (53). In blu the labels of the elements and in yellow the labels of the nodes.

https://files.engineering.com/getfi...583-4912-bc3f-c650a3ac54b0&file=beam-1303.tif

The second pic shows the same components in 3D. At each free surface I applied the displacements measured with the 1D model (so in nodes 65, 336, 333 and 64) and I obtain something pretty crazy since the stress goes to 100k psi instead of 4k psi of the 1D model. to apply the diplacements i created a reference point in the middle of each free face of the pipes and connected rigidly to that surface. There are no boundary conditions

https://files.engineering.com/getfile.aspx?folder=9ace86f2-3242-4699-a169-a557a35a5d11&file=2D.tif

 

[FEA way]

Are both these simulations static or dynamic ? Without boundary conditions how do you prevent rigid body motions ? Using inertia relief ? Also does the first picture show results from global analysis with many beams (and you just suppresed the visibility of remaining ones to show only one joint) ?

It may be really difficult to correctly represent only node from the whole model in 3D. Unless the structure is extremely complicate, it would be best to model whole thing with solid elements. Or try modeling selected joint/joints as solid while leaving remaining parts represented by beams (use coupling to connect solid part with beams) in the same analysis.

 

[flyforever85]

This is a static analysis and I found out that by applying only the displacement you don't need boundary conditions.
The first pic shows a small portion of a bigger structure. From a 1D analysis I have to move to the 2D analysis to better understand how stress is distributed in the joints since the beams will be welded. I was hoping that I could transfer the results from the 1D analysis to the 2D (that's what I was told by people with 20 years of more experience in this field). I applied the loads and displacements but nothing seems to work. Unless the results I obtain are real and the 1D model is way under sized.

 

[FEA way]

Abaqus will never stop surprising me in terms of what can be solved without BCs (contrary to classic FEA assumptions).

This would be perfect case for beam-to-solid submodeling available for a long time in ANSYS but unfortunately still not available in Abaqus.

Did you try applying only forces measured in beam elements instead of displacements (you will have to fix the model somehow but still it may work better) ?

 

[flyforever85]

I did and I still got very high stress. I asked the same in the structural engineering of this forum and I got good answers if you want to know some possible solutions