Load Path / Eccentricities 4

[slickdeals]

Folks,
This is not a design for a real project and merely a hypothetical question.

The sketch shows two options. Option B is what is probably always modeled in analysis (centroids are coincident). But more than likely, the detail provided in the drawing is Option A.

I believe AISC lets you ignore connection eccentricities in statically loaded members (J1.7). I am not sure if this qualifies for the same.

I am presuming that based on the gusset geometry, it is impossible (or very difficult) to get enough weld to transmit the moment due to eccentricity (weak-axis bending of gusset). In that case, the WT member will have to be sized up to pick up axial + moment at the end connection. Am I right in stating this?

How would you folks approach such a problem? I am not trying to make it a research problem (maybe it is), but get an idea on how one would go about designing such a connection.

Thanks in advance.

 

[slickdeals]

@dhengr:
Could you please elaborate on "clipping the lower corner of the gusset plate" information? If I understand you correctly, you are suggesting the intersection between the column, gusset plate and base plate to look like the clip similar to the one at a bearing stiffener?

 

[connectegr]

Typically the corners of gusset plates are clipped at the baseplate/column flange corner. (And similar at the column web/baseplate corner) In additional to avoiding welding stresses at the corner, it is usually necessary to clear the welds of the column flange to baseplate. This is also rarely a "perfect" corner in fabrication. The fit-up could create gaps in the vertical or horizontal welds.

We typically show a clip in our details. Or even a "cope" if access is required for a groove weld.

http://www.FerrellEngineering.com

 

[connectegr]

If you are interested. Here are some details showing clipped or coped gusset plates.

http://www.FerrellEngineering.com

 

[dhengr]

I think you understood me correctly, and I guess that horiz. line could be a base pl. as well as the top flg. of a beam, I’ll give you that. At the low 90° corner, clip the gusset pl. 1" x 1" x45°, size = col./base pl. fillet + 3/4", and then stay back from that with the start of your gusset welds, two reasons; you have to miss the weld btwn. the col. flg. and the base pl. during fit-up of the gusset, and you do not want someone welding the gusset/base pl. fillets right into the fillet btwn the col. and the base pl. That corner, with three converging welds on each side, is a bitch from the tri-axial stress, weld cracking, weld imperfections, stress raiser, residual stress, standpoint. Don’t do that kind of 3-D corner detail.

 

[dhengr]

If I understand your ‘bearing stiffener’ terminology correctly that would be the same situation. On a built-up member you’re missing the web/flg. fillet wed, on a WF shape you’re missing the web/flg. radius which is a very expensive fit-up to try to make properly, as connectegr suggests. In both cases you are preventing the welder from welding into that corner, a good thing from your and my standpoint. The clip is generally less expensive to fab, particularly on the gusset pl. where it’s just a straight shear operation, but the cope is possible too, but more costly.

 

[JAE]

OK - please read the AISC Commentary on D3.3. Specifically this paragraph:

[red]Significant eccentricity may exist within the connection if U is less than 0.6. For values of U less than 0.6 the connection may be used only if the provisions for members subject to combined bending and axial force are satisfied in the design of the member.[/red]

So this tells me that using D3.3 gives you a U value and if the U value is greater than 0.6, the eccentricity discussed here doesn't need to be included in the design.

 

[connectegr]

JAE
I disagree...
This applies to a concentrically loaded gusset connections. For example, a double angle brace connection, with an angle on each side of the gusset plate. The loading of the gusset is concentric. But shear lag still applies to the net capacity of the single leg connected to the gusset. If U<0.6 then the designer must consider the moment in the design of the member.

But, in slickdeals example the gusset is eccentrically loaded. Even if the flange of the WT is designed as a plate, there is an eccentricity of 1/2 tf + 1/2 tg. If the neutral axis is used the eccentricity is much greater. This eccentricity results in weak axis bending of the gusset plate.

Shear lag is a reduction in the net capacity, due to the load not being applied directly to all elements of the cross-section. For example, a wide flange brace connected with web plates only (or flange plates only). The connection is concentric, but shear lag still applies. Or a slotted HSS welded to the gusset plate at the centerline of the HSS. If the HSS is 12x2 with the 12" sides welded to the gusset (not realistic), U will be very small. And additional design considerations are necessary, even if only for the local stresses at the connection.

A U<0.6 identifies a large eccentricity from the connected element to the neutral axis. Therefore requiring additional consideration in the design of the member.

http://www.FerrellEngineering.com

 

[Lion06]

My take on shear lag is this. One of the symptoms of shear lag is that there is typically eccentricities involved, but that isn't what the shear lag is about. The shear lag is aobut how much of the section is available for the tension capacity right at the connection.

Think about this - If the WT were turned such that the web of the WT connected to the gusset and the bolts lined up with the neutral axis of the WT, there would still be a shear lag effect because the connection is to only one part of the section. There is now zero eccentricity, but there is still a shear lag effect, right?

 

[slickdeals]

Yes, there would be shear lag because at the connection the whole section is not effective. The entire section is effective only when all "lagging" shear is distributed over the entire cross section of the member. How long this takes depends on how many parts are connected or how eccentric the connection is.

 

[slickdeals]

A question regarding the bolts. In Option 1, it appears that the WT will be designed for a constant bending moment P*e. In that case, shouldn't the bolts also be designed for Direct shear + Moment? I am assuming the moment due to eccentricity will be small for the bolt group, but just want to make sure that the bolts will need to be designed for moment as well.

Similarly, if the flange of the WT was welded to the gusset plate, the weld would need to be sized for axial load and moment due to eccentricity. Right?

 

[Lion06]

I don't think it would need to be sized for the axial load and moment. Even if you go that route, the eccentricity is much smaller than for moment that the WT is seeing.

The way I see it is that the gussets are providing an eccentric axial load to the WT, nothing more nothing less. If you have a pin (in the strong axis of the WT) at the flange of the WT with a cable (this detail at each end) and put the cables in tension then the pin obviously takes no moment, the cable delivers an eccentric shear to the WT. There is nothing wrong with that. The gusset isn't a pin, but I think it is flexible enough to ignore any moment that might be there.

 

[BAretired]

Agree with SEIT. The bolts are in single shear. Moment is taken by the WT.

BA

 

[JAE]

The plate to WT flange connection interface doesn't "know" that there is a full WT beyond, or perhaps a gigantic W36, beyond the connection as long as the shear lag effect kicks in due to the connection length being long enough.

All it "knows" is that there is a plate lapping it with 100% of the tension coming through that plate (the web of the WT has 0 tension in it at its end due to shear lag).

connectegr - I think there is eccentricity in the connection (1/2 tf + 1/2 tg) as you suggest - agree with you there.

But it appears to me that AISC states you can ignore the relatively minor eccentricity (1/2 tf + 1/2 tg) if U is > 0.6.

We could check with AISC to verify.

 

[connectegr]

I have discussed a similar eccentricity condition with AISC. An axial loaded beam to a single plate connection. Without lateral stability the shear tab must consider weak axis bending.

We presently have engineering staff on the AISC Spec Committee and the Manual and Textbook Committee.

http://www.FerrellEngineering.com

 

[connectegr]

see attached

Weak-axis bending of the gusset plate is not a shear lab issue. Without lateral resistance the gusset thickness will probably be controlled by weak-axis bending. AISC does not ignore these localized connection stresses.

I agree that if U is greater than 0.6, the WT does not need to consider the eccentricity.

http://www.FerrellEngineering.com

 

[slickdeals]

Am I understanding correctly that if U > 0.6, then the WT can be designed only for axial loads and no moments?

 

[connectegr]

That is my understanding. A U less than 0.6 would imply a significant eccentricity in the member and would require consideration in the member design. If U is greater than 0.6, the it is not necessary to consider the moment in the member. Also, reinforcement or increased connection length may be necessary due to shear lag and the local connection stresses. These options can change U, by moving the neutral axis or increasing the length. NOTE: The type of connection can significantly impact whether additional design of the member is required.

Often these issues are not communicated to the fabricator or connection engineer. Without knowing what connection the fabricator may choose, the designer may need to design the brace conservatively. Or note that the connection engineer must verify that the connection selected provides a U > 0.6. (In your example, this can be as simple as increasing the bolt spacing or decreasing the bolt diameter. The net section of the WT or WT flange may require reinforcement due to shear lag. Therefore providing an increased cross-sectional area at the connection to allow the force to transmit to the entire shape.)


There are always extreme cases, but they are rarely practical considerations.

AND thanks for the thread. I have enjoyed contributing.

http://www.FerrellEngineering.com

 

[TXStructural]

Option 1. The connection is concentric. Any bending in the WT is induced by the WT shape and eccentricity of the gusset-WT connection, not the gusset-column-beam connection.
Using Option 2, you have introduced eccentricity of the connection, and still have a WT which will try to bend. There will be no moment at the ends of the WT (aside from those introduced by the same mechanisms that create shear lag), since the gusset is flexible.

The nature of the gusset is that it will bend to match the ends of the brace, which may induce buckling under compression.

To remove the eccentricity, modify the WT-to-gusset connection.

 

[slickdeals]

So this tells me that using D3.3 gives you a U value and if the U value is greater than 0.6, the eccentricity discussed here doesn't need to be included in the design.

JAE, it appears you are interpreting the code correctly, and I am overly complicating it.

 

[dhengr]

I think SEIT has it about right @ 1APR 13:08, and I’ll try saying it in slightly different words. As I said above, I think the moment in the WT is P*e at the region of the mid length of the brace member, and this moment and its amplifying effect is very important in the design of the tension/compression brace, so you must design for it plus P. The moment in the WT near the gusset is less than P*e and is a function of the flexibility of the gusset pl./bolt connection detail (relaxation, joint rotation, semi-rigid). Look at the way Slick drew his gusset pl. (col./base pl./gusset/WT geometry) and think yield line analysis or weak direction pl. flexure; this would be a good candidate for FEA if we could afford the time on each and every joint, or you could probably make a Ph.D. thesis out of this study.

This bolt stresses issue, due to bending, is just not the same as the bolt prying, etc. for the bolts though an end pl., above and below the tension flg. of a beam, on a moment connection; the gusset pl. and WT flgs. can’t act that strongly, wrt the bolts. I don’t mean to suggest some prying may not exist do to moment, but put it in perspective, and consider how the joint really acts. The WT is the stiffer element and the gusset pl. will flex a bit due to the relative stiffness of the WT, if the WT takes a curvature. The bigger question on this bolted joint is that some of the bolts will start to yield in single shear or bearing on the gusset pl. and WT flgs. before all of the bolts come into play, due to fab. and bolt hole tolerances. So, when do you finally achieve bolt shear = P/# of bolts, and do you ever really have that as an average bolt shear? The gusset and the WT will conform (compatibility) through the bolted connection, and in this detail I do not see that adding much to the bolt stresses, but this might be an interesting FEA problem too.

I sure don’t have all the answers, but all the U’s> or < .6 and load factors and material reduction factors, etc. won’t eliminate the need for us to have a good basic understanding of how the structure actually works. I don’t have the latest AISC, would someone **PLEASE POST** a few pertinent pages covering “U” and its usage? So, I understand what I think I’m talking about. This type of topic is a great discussion topic for our brown bag lunch sessions, in the office and they’re good questions here too. But, once we have a fundamental understanding we certainly can’t afford to devote this kind of rigor to every joint, or we’ll be out of business before we get the joints designed, let alone the whole bldg. designed and built.

Connectegr certainly seems to know what he’s talking about on this topic, a smart guy, well grounded, thanks for your contribution. “AISC does not ignore these localized connection stresses,” but nor can they have a separate code section for every imaginable condition. We’re the engineers and should see these. My only lament is that the codes and bldg. designs have gotten so complicated that we and the fabricators now need a separate engineering firm to design our steel joint and connection details. And, I’m not trying to put you out of business Connectegr. In another life, I worked for a steel fabricator, in a special div. other than the structural steel dept. I would get involved primarily when, for fab. or erection reasons, they wanted a redesign presentation, for some repetitive joints, a design more to their liking, of the details shown by EOR.

RE: SEIT’s 1APR 10:24 post: I would say shear lag is a means of (explanation for) getting a concentrated load input distributed into a member, over some length, eccentricities may or may not be involved. This concentrated load is primarily distributed by shear stress or shear strain to the whole member. And, the implication of this is that at the bolted connection a smaller net section must be considered, until the shear lag distribution has occurred to the whole member..

Happy Passover & Easter to all.