design for gusset plate not at a column

[jeg1976]

Can anyone point me to any references regarding how to design a gusset plate when not at a beam-column joint (i.e. for a knee brace away from a column or for chevron/x-bracing at the midspan of a beam)? I've read that the UFM (Uniform Force Method) or KISS method are not appropriate for this type of connection. Thanks.

 

[hokie66]

gwynn,
No reason, except that is "not the way it is done". Ignoring the eccentric part has the same effect as not welding it or removing it. You could make the gusset any size, but the answer is the same when one line of bolts delivers an axial force.

 

[WillisV]

I agree with Hokie. If the part of the gusset concentric with the single bolt line is adequate to develop forces then it the weld associated with the concentric portion can be designed solely for the axial force and shear components of the line of action.

 

[slickdeals]

See attached

 

[slickdeals]

And another one from the Seismic Design Manual.

 

[nutte]

In your case, the weld that attaches the gusset to the beam will have to be designed for a moment = horizontal component of brace force * half the depth of the beam.

This is not correct. The moment here, Ph*eb (where Ph is the horizontal component of the brace force, and eb is half the beam depth), will be countered by another moment, Pv*ev (where Pv is the vertical component of the brace force, and ev is the horizontal distance from the work point, where the brace work line intersects the beam centerline, to the center of the weld group). As Hokie mentions, if the weld is concentric about the brace work line, then these two components are equal, and of opposite signs, so the moment is zero.

I also agree with Hokie that if the concentric portion of the weld and gusset provide adequate capacity, that you haven't made the connection worse by adding some extra redundant gusset on the right-hand side of the connection. If you check this longer weld length, then you do have a moment on the weld, but you also have more weld. The result shouldn't be a weaker weld. But you have made the analysis more complicated, perhaps with no gain in connection efficiency.

 

[slickdeals]

Nutte/Hokie,
You are both correct. I hadn't drawn the sketch ......

 

[BAretired]

Well, hokie, WillisV, nutte and slickdeals, you are not entirely correct. From an elastic point of view, you have made the situation worse by adding material to the gusset. That is to say, the maximum elastic stress has been increased in an eccentric gusset from that of a concentric gusset.

I agree that, under static loading, from an ultimate strength point of view that the situation has not been made worse. But under cyclic loading, I am not so sure. I would suggest that the higher elastic stress in the elastic condition may produce undesired effects under cyclic loading.

BA

 

[jeg1976]

slickdeals, I agree with your sketch. The only time there won't be a moment is if the brace angle is 45 degrees and e_v = e_h.

Note: The below statement is only in regards to single angle bracing, where the bracing workline is not at the member centerline

Per standard detailing (per my experience - where the gusset is usually offset a fixed distance from the edges of the single angle brace), the case where the midpoint of the gusset at the beam interface coincides with the intersection of the bracing workline and the beam interface, rarely if ever occurs for single angles, since the bracing workline is not centered on the the brace. Hence, in general, for single angle bracing, a moment exists at the interface point (whether this moment can be ignored is left to engineering judgement). Attached are sketches for reference.

 

[nutte]

JEG, your statement about 45 degrees is not right. Sketch it out for a different angle and see. If the weld is centered on the brace work line, the moment is still zero. Your eccentricities are no longer equal to each other (as for the 45 degree case), but your load components Pv and Ph aren't, either. They're all related to each other, based on the sine and cosine of the brace angle.

To look at it a different way, the brace work line goes right through the weld CG. The eccentricity is zero. How can there be moment?

 

[jeg1976]

nutte: I was assuming that the weld length is equal to the gusset length at the interface. I'm not a detailer, so I can't say for sure if they would typically make the weld symmetric about the brace workpoint at the interface, or just weld the full length of the gusset edge. If it's the latter, then I stand by my original statement.

 

[hokie66]

BA,
Make the gusset infinitely long in one direction. Do you still think the redundant material makes the connection worse?

By the way, my assumption in all of this is that the brace is a double angle, i.e. symmetric about the gusset in the other plane. If a single angle, then out of plane bending of the gusset becomes an important consideration, as that type connection has been found wanting.

 

[BAretired]

hokie, I don't believe that standard beam theory holds when a member becomes infinitely deep, so the answer to your question is it depends. For a very thin gusset, the mode of failure is not clear. But we are not talking about plates with infinite width.

Consider a plate 1 by 4 units (doesn't matter whether it's inches, cm or mm). If the plate is loaded concentrically with load P, the maximum stress is uniform throughout and is equal to P/4 or 0.25P.

Now, let's consider a plate 1 by 6 units with load applied, as before, 2 units from one edge such that the eccentricity is 1 unit. Now, A = 1*6 = 6 and S = 1*6²/6 = 6. The maximum stress is P/A + M/S = P/6 + P*1/6 = P/3 or 0.3333P which is greater than 0.25P. It is not my opinion...it is a fact that the maximum stress is greater than it was for the concentric case. Of course, the minimum stress is P/6 - P/6 = 0. The average stress is P/6 which is only two thirds of the average stress in the concentric plate but if the loading is cyclic, I am suggesting this case may be more critical than the uniformly loaded plate.

If the load is static tension, then the concentric plate is considered to fail at A*Fy or 4Fy for the 1x4 plate. The 1x6 plate fails when the entire section is stressed to yield, mostly in tension but partially in compression. This occurs when P = 4.325Fy and M = 4.325Fy, slightly in excess of 4Fy for the smaller plate. So the eccentrically loaded 1x6 plate carries about 8% more load than the concentrically loaded 1x4 plate.

BA

 

[hokie66]

BA,
Eloquently argued. It would be interesting to see that tested in a lab. Does anyone know of such a test? We use a lot of simplifications in our modeling of structures, and oftentimes the structure is smarter than we are.

 

[BAretired]

slickdeals, I agree with your sketch. The only time there won't be a moment is if the brace angle is 45 degrees and e_v = e_h.

Not true. There will be no moment in the gusset if the force is applied through its centroid irrespective of the brace angle.

BA