Simple Bracket Post

[darrenlovesmusic]

Hi, 

I have a few doubts about setting up a surface mesh simulation. 

1. The contact between the vertical post and horizontal bracket should be separation, offset bonded or sliding/no-separation? The gap between the surface mesh is 0.125", while the sheet metal is 0.125" thick.  

2. How can we set bolt connectors to be perfectly rigid as they are not expected to fail in operation?  

3. How can it tell whether the problem is going to be large deflection? 

Thank you! 

 

[darrenlovesmusic]

@Roelof.Feijen

Any advise? 

[John_Holtz · Accepted solution]

Hi @darrenlovesmusic 

1. Offset bonded would defeat the purpose of the bolts, so I think offset bonded is not what you want to use. Then whether you want to use separation (allows the faces to separate and slide) or sliding/no separation is a matter of what effects you want to include. (Separation is more realistic.)
2. You can make the material used for the bolts to be a higher modulus of elasticity.
3. Large displacement is important when "P-delta" effects become significant. That is, the displacement either becomes harder as the load is increased or becomes easier as the load increases. Without doing some calculation, is may be hard to know, especially if it is a borderline case between linear and nonlinear. In these cases, you can do the linear analysis first (because it is generally faster). Then follow-up with a nonlinear analysis.

[darrenlovesmusic]

@John_Holtz 

Thank you for the response. 

I'm modelling a large gusset to hold 30000 lbs. I was trying to formulate the problem correctly, to move beyond a 'bonded analysis', which is quite close to hand calcs. As the cross section of the cantilever arms are tapered, the design needs to be verified by FEA. However, I get wildly different responses when I try and use bolted connectors in Nastran inCAD 2018. Additionally, the T2217/G3051 error shows up on a bunch of elements at random locations. 

When I try bolts+separation, the deflections are off by 100%. 

[darrenlovesmusic]

@John_Holtz  upper pic bolted (SAE A5), lower pic offset bonded, A36 sheet metal. All constraints are the same, the only difference being contact.  Hand Calcs deflection at tip ~ 0.075".  

[John_Holtz]

The shape is correct in the upper figure but wrong in the lower figure (unless the undeformed shape is off in the lower figure). Otherwise, not too much can be learned from the images.

You mentioned that these are shell elements (from sheet metal if I remember correctly). The shells are separated by the thickness of the shell in the model, but of course are touching in real life. The offset bonded eliminates the gap (simulated they are touching), but the separation contact does not eliminate the gap unless you change the "Penetration Surface Offset" for the contact. If the larger result occurring because of the gap between the shells?

If that is not the situation, then you need to attach the model files so that someone can look at it. (See the video by Roelof Feijen for creating a pack-and-go file.)

[darrenlovesmusic]

@John_Holtz , I was not aware of the  "Penetration Surface Offset" for the contact. I will re run the analysis and see the response.

Thank you! 

[darrenlovesmusic]

@John_Holtz I'm attaching a part file, incase you wish to take a look. Load on each arm = 2500 lb.

[John_Holtz]

Hi @darrenlovesmusic 

The file that you attached is not ready to analyze. It either was not saved after setting up the analysis, or the model was changed after setting up the analysis. When I open it in the Nastran environment,

• There are dozens (hundred?) of items lost due to "Part Modification".
• All of the contacts are offset bonded.
• There are no bolt connectors.

You may get reasonable results by using a rigid connector instead of bolts. This would assume that the bolts are tight enough that the friction prevents the plates from rotating around the bolts. The rigid connector would also solve the problem that will occur where you have three plates stacked up which I assume are all bolted together. See this article for the challenges of trying to bolt three plates: Bolt through multiple plates does not solve as expected in Nastran.

My suggestions are as follows:

1. Exclude all of the racks except for the top rack, at least until you get the setup working as expected. This will reduce the model size and amount of contact, so the analysis will solver faster.
2. Decide how to handle the three-plate stack that is bolted together.
3. If the results are still unexpected, save the model, copy it to a different folder (and maybe rename it) and open the copy to make sure the setup comes into Nastran.

[darrenlovesmusic]

Thank you @John_Holtz . I think I can neglect the three plate joint and focus on the cantilever. I will post a solution based on your suggestions. 

[darrenlovesmusic]

@John_Holtz 

I was able to model this correctly, I think. Thanks for the advice!