Static linear analysis - using bolt connectors

[arun.sankar]

Hi

We modeled two plates using solid elements connected via 16 M20 bolts. Attached to each of these solid  plates are two beams modeled as line elements, The midsurface of the solid plates were considered and the midsurface was attached to the line element using Offset Bonded contact with automatic activation distance. "Separation" contact was ascribed for the midsurfaces. An axial load is applied at the top most node of the line element whilst constraint was applied to the lower line element as a pin. Please refer to the attached pics for further clarity. The results show a huge displacement of 500 mm for an applied tensile load of 5000 N. The beam particulars are : square tube 400 x 250 x 10 mm. Please advise if the model settup followed is correct as the displacement seems to be superficial. 

[John_Holtz]

Hi @arun.sankar . Welcome to the Inventor Nastran forum.

The results look wrong, but it is difficult to know what is wrong by looking at the images. If you can attach your model, someone will take a look at it and determine what is wrong. (Zip the part files and assembly file, and attach the zip file to the forum. You can also follow the  video by Roelof Feijen.)

[arun.sankar]

Thank you sir for your prompt reply. We are using Autodesk Inventor Professional 2021. We have attached the requested files.

[arun.sankar]

Sir,

 Can we have solution to the previously addressed query. 

[John_Holtz · Accepted solution]

Hi @arun.sankar 

Those results are strange. I am not sure what is causing the problem, but removing the separation contact between the plates will give correct results.

If you must have the separation contact between the plates, the problem might be caused by one of these items:

• Trying to use the same face on the plates to define contact with each other AND with the beams.
• The normal directions of the plates are in the wrong direction.

The solution to both of these potential issues is as follows:

1. Draw the plates using surfaces in the modeling environment instead of using solids.
2. Created a circular region on the plate around where the beam connects. Use the circle to split the plate.
3. In Nastran, define the normal direction of the plates so that each circular region is pointed at the corresponding beam, and the rest of the plate is facing the other plate. (The normal direction is changed by right-click on Elements > Reverse Normals.)
4. Define the offset bonded contact between the beam and plate using the small circular region on the plate. Define the separation contact between the plates using the remaining face. (Be sure to specify the Penetration Surface Offset of 15 mm so that the shell elements behave as if they are in contact. In other words, the 15 mm gap between the shells is not a real gap of 15 mm.)

Let us know what you decide to do.

[frantisek.thorz]

Hi@John_Holtz,

would it be possible to do the same analysis in a bit different way?

1) Adjust both sketches representing beams that they actually touch the plates (shells).

2) Connect them using the mesh control by creating nodes at the ends of the beams.

3) Fix the existing separation contact -penetration surface offset -  15mm, Max activation Distance - 18 for the 10 element size mesh?

BR

F. 

[John_Holtz]

Yes, that would work. Physically connecting the beam to the plate would eliminate the need for the offset bonded contact. In fact, using the mesh control is one of the suggested methods in the article How to connect beam and shell elements in Nastran.

The only thing that I do not like about this method is that you need to "predict" where Nastran will create the midplane mesh and draw your beam element to the predicted location. It is no problem with flat plates (and should not be a problem with any geometry in theory), but I can imagine how something could go wrong. For simple geometry, I think it is just as easy to create the surfaces in Inventor (instead of creating solids and converting to a midsurface) and connect the beam sketch lines to the surface. In the end, both methods are the same.

[arun.sankar]

Hi @John_Holtz 

 thank you for the solution. We have sorted it out and it is working fine.