Tee Connection Axial + Shear 1

[amengr]

I've got a case where we have a beam that has an end reaction of vertical shear and an axial load. The steel fabricator would like to use a WT connection that is welded to the supporting column web and bolted to the supported steel beam. I've run through the simple shear connection requirements in the aisc and am working on the axial and the combined loading cases and my question is: Are there any additional requirements for the weld from the WT flanges to the column web when the axial load is acting away from the connection - similar to a prying action check for a bolted hanger tee connection? Or is the WT flange acting as a plate bending about two-pinned ends and all I need to check is the plate bending?

 

[SandwichEngine]

Why not just use a knife plate? Then you eliminate the bending. Is the column a tube and they're trying to save it from local plastification or something?

 

[human909]

Where I'm from welded shear tabs are the standard go to connection for beams. They also work axial force with barely a second though. If the axial for is large then you might need to check plate thickness to account for the AXIAL eccentricity. (which is only ~10mm or so but it still matters.)

A welded T just seems to complicate things due to prying of the welds.

 

[KootK]

If you only weld the flange tips of the double tees, you'd design the welds for the vector summation of:

1) Vertical shear.

2) Tension due to axial load.

3) Maybe tension due to load eccentricity to the bolt line.

In this instance, there would be no prying.

If you weld the top and bottom of the tee stub, in addition, to the flange tips, that gets a bit more complex.

 

[JoshPlumSE]

There is a long discussion about WT's (or double angles) welded and subjected to both shear and axial load in a previous thread:

https://www.eng-tips.com/viewthread.cfm?qid=358106

It's kind of a long read, but it's very informative and comprehensive. A brief summary:
a) If you're using a WT connection with a weld at the edges of the Tee, then this is NOT allowed per Bill Thorton's Sept 2011 Modern Steel Construction discussion.

https://www.aisc.org/globalassets/modern-steel/archives/2011/09/2011v09_interchange.pdf

b) The program I worked on at the time (RISA) had a version of this connection that (when subject to axial loads) used a revised version of this weld configuration. This was a point of contention about whether it should be allowed at all when axial force was present.

c) As a result of that discussion, the RISAConnection program greatly revised the treatment of axial force for the connection. Adding a number of warnings if the axial force exceeded a certain percentage of the shear force or such. The intent being that this type of connection should only be used when the primary force is shear.

d) Unfortunately, it seems that since my leaving RISA in 2017, some of these safeguards and warnings have been removed. This is exactly what I feared when Nemetchek / Amber took over and cleaned house of many of the most experienced engineers. It's honestly a bit scary to me. I learned a lot in that previous discussion / thread. So, it's scary for me to see so much of that institutional knowledge disappear like this.... I feel especially bad for the RISA users who trust this program to warn them when they do something outside of normal, expected practice.

 

[human909]

In this instance, there would be no prying.

I'm thinking an axial tension load here and a relatively thin T-section would cause prying as the weld lines a significantly offset from the axial load.

I simply don't see the advantage of a T section unless the column web is understrength for a simple plate.

It's kind of a long read, but it's very informative and comprehensive. A brief summary:
a) If you're using a WT connection with a weld at the edges of the Tee, then this is NOT allowed per Bill Thorton's Sept 2011 Modern Steel Construction discussion.

Thanks for confirming my gut reaction. Personally I don't deal with clip angles / double clip angle and Ts. More about local practice and availability of Ts, so it is good to have another source to back up my gut reaction.

 

[KootK]

I'm thinking an axial tension load here and a relatively thin T-section would cause prying as the weld lines a significantly offset from the axial load.

This might be a regional terminology thing but, in my world, you really can't have prying without some manner of backspan on the thing exerting the prying force. Here, the welds will see a bit of root flexure as the WT flange bends but that's just straight flexure, not prying.

 

[KootK]

With regard to JP's stuff, I note that both the Thornton article and the pervious thread deal with double angles, not WT's. Certainly, I agree with the concept for double angles. Granted, two angles positioned back do tend to remind one of a tee so, perhaps, the situations are similar enough that the same logic applies.

 

[human909]

This might be a regional terminology thing but, in my world, you really can't have prying without some manner of backspan on the thing exerting the prying force. Here, the welds will see a bit of root flexure as the WT flange bends but that's just straight flexure, not prying.

Ok. I'll give you that. Prying is not the correct use of the word. What I meant to emphasise is what is better explained in JoshPlumSE's link.

 

[KootK]

@ JP / human909,

By extension of the same principles, do we then have concern for the kind of connection shown below? It seems as though we might. This case is much more common in my practice as a way to bypass potentially thin HSS walls that would otherwise fail in flexure. I've seen this as a shear + axial connection but, also, as a predominantly axial connection in the case of vertically braced frames. If it's bunk owing to the weld prying business, I'll need to put the brakes on that in the future.

 

[KootK]

I used to have access to the article snipped below which, I believe, did a good job of elucidating this subject. In particular, the box beam example.

 

[human909]

By extension of the same principles, do we then have concern for the kind of connection shown below? It seems as though we might. This case is much more common in my practice as a way to bypass potentially thin HSS walls that would otherwise fail in flexure. I've seen this as a shear + axial connection but, also, as a predominantly axial connection in the case of vertically braced frames. If it's bunk owing to the weld prying business, I'll need to put the brakes on that in the future.

Good question. Personally my gut feel on the connection that you have shown there doesn't give me confidence due to the weld bending under axial tension. But that is just my gut feel.

I normally get around this challenge of axial loads into HSS by using thicker wall section. Though in the case of a recently designed 22m tower I used a thinner wall section as my main columns and ensured that all cleats only transferred shear into the columns. Large single cleats often with 3 incoming HSS members mean that the only load being transferred to the column is in shear. Good controlling of detailing is important here.

On an slightly related note I saw big welds crack recently from the actions of a pneumatic hammer. Relevant because the pneumatic hammer very much would have been putting these welds in cyclical bending. It lasted about 3 months of use before the welds cracked... A few errors were made in installation which meant it premature failure of the welds. (I saw premature because there is a reason why these devices come with catch wires. They are known for cracking welds.)

 

[Settingsun]

It would be nice to read the Blodgtt article. At the bottom right, it shows a box girder with welds either outside the web only or both sides of the web. Is the both-side weld the solution or was it pointing out that it isn't? Seems to me that the weld needs to distort to the deflected shape of the flange in either case.

My approach would be to stiffen the flange (of the box beam or KootK's T-connection) to make the slope small compared to what the weld can tolerate. A weld pulled transverse to its length elongates about 1mm before snapping AFAIK. So say 0.1 to 0.2mm across the leg dimension = 1/60 to 1/30 slope as a starting point for discussion. Static loading only per Human909's caution.

More cautiously, keep well below the 0.1 to 0.2mm yield deformation under service loading.

 

[Ingenuity]

It would be nice to read the Blodgett article.

Link

 

[JoshPlumSE]

By extension of the same principles, do we then have concern for the kind of connection shown below?

Yes, I tend to think the issue applied for that case as well. However, if the Tee were welded fully across the tops and bottom on the Tee (instead of just the sided) then you move into a better weld configuration. But, it probably ceases to have the rotational ductility to be called a shear connection.

That's the real dilemma... getting a weld configuration that works without overly restraining the connection.

Also, to me, the connection is the same whether it's a Double Angle or a WT, provided the welding is the same.

 

[human909]

Coming back to KootK's question. It is all well and good to be armchair engineer and criticise his connection for being unideal. But is is really so unideal that it is cause for concern? I think Steve has best articulated it, though I'd expect most of us had already come to that conclusion, we need keep the plates stiff to limit the weld bending.

The simplest answer I come to is to use thicker walled sections or significant axial connection. There is an interesting article here.

https://www.mdpi.com/2076-3417/10/11/3793/htm

Though we are wringing out hands over a connection here that other engineers would just design badly and get away with it. The number of photos I could show you of poor design from a recent site visit would make you cry. Almost all members were HSS and the details were terrible.

 

[KootK]

unideal

Unideal...I like that. Probably describes 2/3 of what I do.

...we need to keep the plates stiff to limit the weld bending.

Agreed. Somewhere, there's a plate stiffness that's acceptable. I wish there was an official version of steveh49's calc. "A fillet weld of this size may rotate this many radians." Fat chance, who'd ever want to commit to that?

I'm sure that the failure noted in the Blodgett article had a great deal with the component being cyclically loaded in a way that would exacerbate fatigue issues.

If you think about it, even an uplift base plate on a wide flange column is likely to see some bending in the fillet weld roots at some locations. Something that I've long wondered about is the extent to which a two sided fillet weld alleviates this condition (or doesn't).

 

[WillisV]

Of some relation/interest is that the 1.5 "directionality" increase for fillet welds in tension has been eliminated for rectangular HSS in tension in the 2022 AISC Specification (that was just under public review) as an increased safety factor/hedge against rotation of the weld root (something which presents itself in many other situations as discussed in this thread but was not part of the specific tests so was not targeted in this Spec update). I agree that some sort of stiffness/rotation allowance would be best - but we aren't there yet.

 

[human909]

Unideal...I like that. Probably describes 2/3 of what I do.

Unideal is a phrase I use so often that I forget that I'm using it and that it may not be a ubiquitous engineering phrase. I use it at least as commonly as other engineering phrases descriptors like conservative and unconservative. As engineers we also have to deal with the real world challenges. I aim for ideal but sometimes it isn't achievable.

Generally whenever a connection or other configuration arises in my design that is "unideal" I'll hand wave it away using "engineer judgement" and knowledge that my structure is still comfortably exceeds ultimate limit state. When I do end up checking these "unideal" circumstances I spend copious amounts of time inspecting every last detail for a connection in this scenario here. Normally the result is negligible. That said, unless every now and again there is a time 'unideal' aspects can bite harder than you expect.

Eccentric cleat connections for struts is one such example and research has shown how severely unideal previous design guides in Australia and New Zealand were. (I'm not sure how prevalent this connection affected other localities, I still see poorly designed connection in new builds here.)