Large Language Model (LLM) for pathfinding feature identification using nodes, elements, normals

[Cody Godines]

It seems pyNastran lends itself to be an LLM for all the input and output, the trick I need to learn is to use it efficiently in a pathfinding manner.

I would like to setup a LLM for the FE entities, nodes, so far and easily thanks to pyNastran, I have the nodal information and, for example, connected elements.

I would like as a beginning to get the normals for the element faces/edges connected to these nodes, and the average normal.

I've read about normals on the elements, is there a general method for all element types to get the normals once I have element id's connected to a node?  The type of element would be uncertain until querried.  Also, for example, for shell elements, there are face normals and edge normals.  If i wanted to find a fastener hole I'd need to know the edge normals.  But the same fastener hole would be determined using face normals for solid elements.

Thank you,

Cody

[Steven Doyle]

I’m not familiar with the term edge normal, but you can get a shell normal with the .Normal() or do it yourself without cross referencing. I’ve implemented average normals in the past for other codes, but haven’t found it to be useful for Nastran.

Solids also support normals, but the interface is geared around the way Nastran defines pressure faces, which is probably not ideal.

--
You received this message because you are subscribed to the Google Groups "pyNastran Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pynastran-disc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/pynastran-discuss/01ae66d6-2730-4eed-a838-76d528522542n%40googlegroups.com.

[Cody Godines]

I believe pressure faces would be the same face id I see in FEMAP. 

I will try the .Normal() to see if I can get a general approach, quads, ...

And I will explore normals for 3d faces using pressure faces but if you mention.  I'm really looking for the outer surfaces to identify the 'terrain' so to speak.

The edge normal for a shell would be perpendicular to the face normal and also perpendicular to the edge line from node A to B.

Thank you Steven for the quick reply,

Cody

[Steven Doyle]

There’s a skin_solid_faces or something like that that will make a shell model from solids. I don’t remember what it does with existing shells, but I think it removes bars.

A positive pressure acts in the direction of the normal to create a force, so you may need to flip things. Solids faces are backwards and a positive pressure induces compression.

To view this discussion on the web visit https://groups.google.com/d/msgid/pynastran-discuss/32ff67d7-25ff-4e2a-9496-f39dabeb3464n%40googlegroups.com.

[Cody Godines]

Thank you Steven for the detailed replies.  I keep learning more good stuff about PyNastran.