Connection between beam and shell element problem

[Nikolai_Damgaard]

I am trying to connect a beam and a shell element using the method described here:
How to connect beam and shell elements in Nastran (section "Beam in plane of Shell").

When I use beam elements for line BC, the connection to the shell element lacks rigidity.

When I use bar elements, the connection seems to deflect correctly.

Why does this happen?

Considerations:

I am certain the shell and the beam share the same node B.

I have tried with 2D sketches and 3D sketches, but I get the same problem.

The model is a surface body model with split faces and a single vertex at the connection.

I am using Inventor Professional 2024.

I know I could just add a connector between the beam elements or just use bar elements, but I want to know if i am missing something or if there is a bug I am unaware of. 

[John_Holtz]

Hi @Nikolai_Damgaard. Welcome to the Inventor Nastran forum.

Can you attach your model?

• If it is just a part file, attach the .ipt.
• If it is an assembly, please compress (such as zip) the assembly file (.iam) and all the part files (.ipt).

Having the model will make it easier to understand the setup and what the results are showing. For example, if the beams are on the surface of the shells, you do not want to use any sketches. You need to split the faces of the shell and define the beams by selecting the edge of the shells. (You only need a sketch if some of the beams extend off of the face of the shell.)

Thanks.

John

[Nikolai_Damgaard]

Attaching the model is not an option, unfortunately. It is too confidential for that.

Regarding: " You need to split the faces of the shell and define the beams by selecting the edge of the shells. (You only need a sketch if some of the beams extend off of the face of the shell.)"

I have done exactly this.The only sketches are of the beams protruding from the shells. One end of each sketch line is attached to a shell vertex, specifically the same vertex of the in-plane beams.

[John_Holtz]

Hi @Nikolai_Damgaard 

I could not reproduce the problem using a test model. Perhaps from looking at my model attached, you can see something that is different or reproduce the issue with my model.

My only thought is you are using continuous meshing. Occasionally that will miss items associated with nodes or edges that are merged together.

Let us know if you find something.

John

[Nikolai_Damgaard]

Hi John,

Disabling continous meshing did not seem to fix it, unfortunately.

It is worth noting that the difference likely lies within the difference between bar elements and beam elements, since bars behave as expected while beams do not.

Another detail: the beams are C-channels with placement relative to their centroid axis.

[John_Holtz]

The difference between bar and beam elements happens in the analysis. The mesh should not be any different. Did you try to reproduce the problem with my test model?

I am not sure what is shown in the original image "beams.png". The beams are either:

• connected and the strange displacement is due to the cross-section or material properties, or
• not connected which would cause an error in the analysis, unless you changed something in the setup to ignore the error. (If the beams are not connected, they should have a much larger displacement. Since their displacement is about the same as the shell displacement, perhaps the problem is with how Inventor shows the deformed shape. In other words, if the displacement value is correct, the result it correct. How Inventor would show a different deformation than the displacement value would be the question.)

Let us know what you find out about the setup or the result.

John

[Nikolai_Damgaard]

The only difference between beams.png and bars.png is the application of beams vs. bars elements on the same sketch lines.

The sketch lines are applied the same way as in your model.

I have not changed anything in my setup.

I tried recreating my entire model and FE-setup at one point, thinking something was messed up in the original, but the same problems occured.

I have also tried to decrease the gravitational accelation by a factor 10 in my model. I get the same results.

I tried to recreate it in your model, by implementing the differences from my model:

• Implementing UNP beam elements
• Only including a downwards gravitional force, scaled to 100 m/s^2, as a test.
• Changing supports to the protruding beams' ends in the -y direction. Similar to supports in my model.
• Stitching all surfaces together, similar to my model.

I was unable to replicate the problem in your model.

[Nikolai_Damgaard]

If I get the time, I will investigate further, possibly create a quick model that I have permisson to share.

I will use bar elements for my model until then.

[John_Holtz]

Hi @Nikolai_Damgaard 

It looks like both of your beams deflect "up".  When I run your version of my model, one beam deflects up and one down. Perhaps this is round off error since the load (gravity) is in the Y direction and parallel to the beams. I will investigate it more to see if I see something else.

John