Welds?

[MJuric]

I wanted to do a simulation on a weld. The weld connect won't work because the items are not perpendicular. I then did a bond only with the edge. That just doesn't look right and I think ignores the strength/Size of the weld itself.

So my thought here is does it make sense to actually model up the weld "Merging" the two parts and do the simulation that way? Is there a better way?

[bnemec]

We have been modeling the weld bead for the simulation model. But we haven't been able to gain a consistent correlation between SW FEA results around welds to physical fatigue testing. We're typically not that interested in static loading of welded joints, so don't have anything on that. We have not figured out what to do with the sharp gradient "hot spots" in the FEA at the start/stop of weld beads. Have tried a few methods some help.

[MJuric]

We have been modeling the weld bead for the simulation model. But we haven't been able to gain a consistent correlation between SW FEA results around welds to physical fatigue testing. We're typically not that interested in static loading of welded joints, so don't have anything on that. We have not figured out what to do with the sharp gradient "hot spots" in the FEA at the start/stop of weld beads. Have tried a few methods some help.

Mine is static so probably easier and probably a better way to do it than I'm thinking of.

[SPerman]

There might be something of value here.



I think this does a good job of summarizing the problems you will/are facing.

[JSculley]

There might be something of value here.
...
...

Or here:

https://apolloedge.com/2018/08/30/how-t ... lysis-fea/

They are using ANSYS but the methodology would work with other FEA software. The bottom line is to not use FEA to check the welds, but rather use FEA to estimate the actual loads the welds will see and then use traditional weld sizing techniques to design the weld.

[Frederick_Law]

Theoretically, a perfect weld is as strong as a solid part.
So, no reason to FEA weld.
If the part did not fail without weld in FEA, it will not fail with perfect weld.

Of course the key word is "perfect" weld.
A real weld will have so many stress concentrate locations the FEA will just show red.
Hence all the testing, procedure, certification etc. to get a good weld.

We were making fracking pumps frame, doesn't matter how much FEA the customer did, the fracking thing still crack.

[MJuric]

Or here:

https://apolloedge.com/2018/08/30/how-t ... lysis-fea/

They are using ANSYS but the methodology would work with other FEA software. The bottom line is to not use FEA to check the welds, but rather use FEA to estimate the actual loads the welds will see and then use traditional weld sizing techniques to design the weld.

Pretty much what the above video stated.

[FEAnalyst]

Welds are rarely modeled in stress analysis (usually only when calculating fatigue) but it might be necessary to include the geometry of welds in the model. Here you can find an interesting comparison of different approaches to weld modeling: https://apolloedge.com/2018/07/19/model ... lysis-fea/