Bending Moment Applied to Fillet Welds

[Maturin]

Attached is a sketch that shows a small W-section connected to a larger W-section by fillet welds at the flange tips. The smaller beam is relatively short and transfers a tension load to the larger beam. The welds are the only connections between the beams (i.e. no bolts, end plates, etc.). The flange thickness of the two beams is the same.

I've always avoided using details like this as I feel the fillet welds would not be able to transfer the bending moment at the welds. (The bending capacity across fillet weld throats being very small.)

The only time I could see a detail like this working is if one of the flanges was much thicker than the other and the thinner flange allowed the thicker flange to rotate a small amount (similar to a deep girder and slender columns). In this case the load would predominately be transferred in shear.

Is there any other way of looking at this detail that would give it a higher capacity than I've assumed?

 

[berkshire]

What is the objection to through bolting?
B.E.

 

[Maturin]

This is an existing detail that I'm evaluating. If I was desiging it from scratch then bolting the flanges would be my preference.

 

[kingnero]

http://www.eng-tips.com/viewthread.cfm?qid=340910

this thread has a very similar question, you might want to browse it through for ideas...

 

[dhengr]

Maturin:
The problem with that detail is that the load causes the flanges to bend as if cantilevers from the webs and semi-fixed and loaded out at the flange tips. There is a prying action at the root of the weld, and it puts the weld roots in tension across the axis of the weld. The very worst possible condition at a location of poor weld quality; lack of fusion, poor/no penetration, they faying surface is a crack starter lined up across the stress field. Maybe put a few bolts through the flanges at high load points, just to clamp the flanges together.

 

[dhengr]

The original designer would have provided a better hanger detail by using a WT section, where its web was welded to the support beam flange at the center line. That’s a much more direct load path and treats the same welds better. Then I would still reinforce the 3/16" fillets for their first and last 3" of length by welding out the near the end (not to the end) of the WT web, and then welding right back over the first weld by about 3". These end welds would end up being 1/4 or 5/16 fillets. The ends of those kinds of welds are somewhat susceptible to unzipping. You don’t weld out to the end of the web because there is too much of a chance of notching the web, and that’s no good either.

 

[renderu]

The most complicated aspect of this arrangement for me would be to figure out how the load is transferred from below to the weld. As someone suggested in the thread referenced above, maybe assume the effective weld length coming up at a 30 or 45 degree from the load?

After the above is assumed, this is what I think you can do to figure out the flange bending effect:

1) To check the flange in bending, assume simply supported at the welds.

2) To check the additional stresses in the welds due to the flange bending, assume the flange is fixed at both ends and transfer that end-moment into your weld (using throat thickness for section).

And of course superimpose all other stresses you would normally calculate (shear stress across weld due to section bending.... etc).

 

[renderu]

Another note, when transferring your bending moment to your weld I would suggest applying an appropriate stress concentration factor due to the abrupt change in bending cross section. I'd also apply a more conservative FOS than usual to this kind of problem.

 

[Maturin]

Thanks for the comments. There appears to be strong agreement that this connection detail should never be used. I retract the second comment in my original post as I don't think there is any configuration where the weld would not be overstressed.

 

[renderu]

Maturin, do you mean that you will refuse to do calculations on this arrangement? Just curious.

 

[racookpe1978]

No - Its not that the weld will or will not be overstressed - it's that the weld condition and the load conditions as-is are very, very difficult to predict accurately, and so it is very EASY to make some design assumptions that may MISS show over-stressing the weld will occur.

Do you see the difference? That the weld has not failed in practice may simply be that it has never been loaded yet to a fatal condition. And, in truth, it may never be loaded that way.

But we can't show that. We can't reliably calculate that.

 

[renderu]

I agree with racookpe1978, partially. Use the correct reason: that you find it too complicated to calculate. Not that it will be overstressed no matter what.

 

[Maturin]

I did some approximate preliminary calculations and the weld is overstressed under the worst loading condition. It's not really that difficult to calculate as I assume that there is a fixed connection at the web and at the weld line and a point of inflection in between (similar to bolt prying action). I will recommend that the connection be reinforced by adding end plates to the shorter beam (stiffeners in the upper beam are likely not required). The reason for the post is that I thought there might be an alternate way of analysing the connection. I've discussed this detail with other engineers in the past and they usually don't see it as an issue.

 

[desertfox]

Might be late in the day now but have a look at this site for welding stress calculations.

http://www.roymech.co.uk/Useful_Tables/Form/Weld_strength.html

 

[Ron]

I would consider this to be more of a shear flow issue than bending. You have bi directional shear issues in the weld.

 

[chicopee]

My two cents would be to weld stiffeners on both sides of webs of the top and bottom beams in the event prying of the contacting flanges is determined to be a problem.

 

[ToadJones]

I think this type of weld/connection is specifically prohibited by AISC. Similar to the idea that one should never use a single sided fillet weld on a plate that is in tension.