How to correctly use midsurfaces in Nastran

[enzanelli]

Good morning,

I have a large trailer assembly that isn't appropriate for analysis via solid elements, so I'd like to use surfaces. 

My problem is that when I use the "Find Thin Bodies" feature within the in-CAD environment, the midsurfaces of the different plates are separated by half of the plate thickness. 

My question is: what is the correct procedure for bonding all of these disconnected midsurfaces? Am I going to have to go in and manually bond all of these plates? Should I instead prepare a derived surface model outside of the in-CAD environment specifically for analysis? 

Thank you.

[Roelof.Feijen]

Hello @enzanelli ,

I've seen people struggle with midsurfaces quite often. I don't even start with it anymore if it is a large complex model. My advice: make a surfaces model. I always say, you have a model that you use for production drawings and a model for FEA analysis. These are not the same. Some engineering managers aren't happy when I tell their employees that, but they've probably never done FEA analysis either. Another saying: one analysis is not an analysis, it all starts after you get the first analysis results.

[enzanelli]

Thank you for the input and advice. I think it's very true that the complexity of these studies is underrated.

As for the surface model, I derived a shell model from my solid, but when I open this new shell in Nastran, I cannot create any idealizations using the shells.

What should be the workflow here? 

[darrenlovesmusic]

enzanelli, Create a surface model from the get-go. Best workflow for me, since I work extensively with thin walled structures. However, if you need to modify the structure frequently, you must design the model parametrically to save time. 

[Roelof.Feijen]

Hello @enzanelli ,

Please share the surface model and I will have a look.

[enzanelli]

Hi @Roelof.Feijen , attached you'll find the surface model I extracted from the solid. I used the thicken/offset tool to build the midplanes and then extended the surfaces such that they'd all touch. I'm unable to do anything with this in Nastran. 

Thank you in advance! 

[Roelof.Feijen]

Hello @enzanelli ,

When you use the Midsurfaces or Surface Offset command, Shell idealizations are automatically generated.

In this case you need to setup the Shell Idealization yourself. The picture below should give you enough guidance, I hope. Otherwise have a look at this tutorial.

[enzanelli]

Thank you @Roelof.Feijen this allowed me to make my idealizations. My follow up question is do I now need to use cotoniuous meshing to bond the surfaces? I'm trying to run the simualtion and I'm getting 2 errors: 1) the auto mesher cannot create a mesh using continuous meshing and 2) stiffness error when trying to run the simulation. 

Thank you in advance.

[Roelof.Feijen]

Hello @enzanelli ,

If everything is bonded, I would suggest to use the Solver contact with the Offset Bonded contact type. Use a appropriate max activation distance. A simple rule of thumb in this case for the max activation distance is 1.2 x the distance (openings) between the surfaces.

The Offset Bonded contact will connect 6 dof (translational and rotational), while a Bonded Contact will connect 3 dof (only translational). A Bonded Contact with shell elements will result in a hinge effect. I don't think this is what you want. 

[enzanelli]

Yes, the entire weldment should be bonded together. Are you suggesting that I have to manually add the (offset) bonds between all of the surfaces? Or do I only need to manually bond the surfaces that are not touching?

[Roelof.Feijen]

No, with the Solver Contact you can specify regions on which the Solver Contact should be applied, but if you don’t specify any region, it will be applied to the whole model. There’s no need to select a anything. The Solver Contact can also be used in combination with Manual Contacts. The Solver Contact will not create extra contact elements if Manual Contacts are specified between two surfaces/edges.

[Roelof.Feijen]

You might also be interested in this article.