Weld Sizing

[imajar]

I am looking to size the weld where two steel members come together in a simple T configuration with bonded contact (solid elements).  For the weld calculation, I am hoping to extract simple axial and shear forces at the interface - which I intend to use with traditional weld calculations.

I found this article to sum the contact force, is this the best way?  In my scenario, the contact interface does not align with global coordinates - All I can output are x,y,z forces.  How do I get shear/axial forces?  (and verify the orientation of the shear forces?)

Thanks in advance.

[Roelof.Feijen · Accepted solution]

Have a look at this video "Predicting and Validating Welds with FEA in Autodesk Nastran In-CAD"

I think it might be helpful in your case.

[imajar]

Hi @Roelof.Feijen, thank you for your quick reply.  That video is actually one I have seen before, and I agree it is very helpful!

In the video, they demonstrate how to extract the element forces and moments from a shell, average them, and then use those loads to do traditional hand calculations.  That is exactly what I want to do, except with solid elements instead of shell elements, but I do not know the correct way to extract the average forces from the solid mesh like they did so simply from the shell mesh.

My model is statically indeterminant and not well suited to converting to shell elements, so I am hoping there is a way I can do it from either the solid elements or from the contact elements.  If I can just find out the actual load being transferred at the joint, I can do the rest by hand.

[darrenlovesmusic · Accepted solution]

You may use something like this - https://www.apolloedge.com/modeling-welds-for-finite-element-analysis-fea/, followed by the hot-spot method. There are some guidelines in the DNV standards. 

[John_Holtz · Accepted solution]

Hi @imajar 

I do not remember if the contact results include "shear" and "normal" directions. If they do, they will be available in the FNO file.

Most likely, the contact results are only available in X, Y, and Z coordinate system. Using FNO Reader, you can get the sum of the X, Y, Z forces, and then use the dot product with a unit vector in each shear and normal direction to calculate the components in the desired local directions. I believe the same vector/coordinate transformation can be used for the moments also, but you should research that a little better than trusting my memory. (Re-read the first 4 words of my reply if you have any doubts about my memory 🙄.)

[amirandaL4VAC]

Hi all, adding on to this question here. I want to merge the concepts of the Hot Spot Method with solid elements, and treating welds as lines.  I use solid elements with no weld representation, bonding the edges and leaving the faces with a separation contact (see the photos below).

Most solid element approaches I've seen assume you also want to model the weld, but I want to avoid this so that the program can tell me what size it should be. Have you seen this method used before in Nastran, and would you recommend it, or against it? 

[darrenlovesmusic]

You can use any modeling approach and verify with an experiment or research papers if they exist.