Shear Lag in CSA S16

[pbc825]

Good day.

I'm designing a connection for which I believe shear lag may be relevant. In this case, the supported beam is subject to both shear and axial force. So I've got a web stiffener welded onto the supporting beam so as to transfer the axial load to the flanges of the supporting beam. There is a single line of bolts through this web stiffener plate and the beam is coped top and flange is block coped bottom. I've also added a bolted angle to the other side to help with shear resistance. I think it could be reasonably assumed that the welded web stiffener could be more rigid than the angle connection. The link shows a quick sketch. Although this is not the traditional consideration for shear lag, I think the shear lag principle is relevant here in that the force is not transferred simultaneously to all elements of the force resisting components. I have two questions as follow:

1. I've reviewed the shear lag literature in AISC, but I'm not able to locate it in CISC literature. Can anyone direct me to the CISC literature?
2. Has anyone designed a connection similar to this connection? Was there any consideration for shear lag?

Thank you in advance.

 

[canwesteng]

Shear lag is in clause 12.3.3. Why it isn't included with the rest of the bolt and weld resistance clauses is a mystery. Shear lag would apply for axial forcse on your detail, unless the critical net area is only the web. Given that you only have 3 bolts taking the shear and tension I'd probably be inclined to say that is the case.

 

[dhengr]

Pbc825:
Generally, when you consider shear lag, you are talking about a concentrated load or reaction being applied and then how that loading/force/stress is distributed into the member in some sort of a gradual way, and as a function of material properties, individual element stiffness and strength, etc. With the bolted connection you have, you have the concentrated loading, but at first, a fairly discrete distribution of those forces into the various elements until you get some distance away from each bolt (or the bolts). I don’t have your CISC lit., so I can’t see exactly what it says. It would be nice if you attached that to your post for our review. For the most part, the discrete way that bolts act, in a bolted joint, is pretty well baked into the design by the allowable shear and bearing stresses and the design procedures dictated by the codes/stds. Given fabricating tolerances etc., one or two bolts come into play first, and yield in shear and bearing a bit before the third bolt really even sees much load. So, where is the shear lag in that detail, it hasn’t even had a chance to start to develop. But, you sure do have a localized stress problem, again, pretty well baked into what the codes tell us to do. The concept of shear lag is helpful in understanding how stresses are distributed in each element, around a bolt hole, etc. The study of Theory of Elasticity says that stresses at a point (in a very confined area) can’t change from one value or direction to another value or direction over a very short distance except as the mechanical properties, design strengths, stiffness’, etc. allow this to happen.

Shear lag would be more appropriate as a discussion topic if you had a line of 8 bolts connecting an angle iron to a plate and the whole system was in tension parallel to the line of bolts. The question would then be, how (and how quickly) does the load transfer from the angle to the plate, and what is the tensile stress distribution in each member as you move along the length of that connection. And, that’s an internal shear deformation problem, causing that progressive change in the tensile stress. But, in a bolted connection you still have the hole/bolt dia. tolerancing issue which causes the load to be applied at only several discrete points until considerable local deformation has occured. This latter complexity is eliminated if the line of bolts is replaced by a couple lines of welds, because at least the discrete bolt loading issue goes away.

 

[TLHS]

I just realized that I don't think the question you seem to be asking is what you're actually asking. The traditional application of shear lag in this detail would involve getting the load from the bolts in the web of the wide flange into the whole body of the wide flange, and the appropriate reductions to account for that.

I *think* you might actually be asking what's going to happen on the other side of the connection where you have bolts that pass through a bolted clip angle on one side, the web of the wide flange, and then the welded stiffener behind. Is that right?

This isn't a shear lag question, but it's a question of load path and compatibility. While this isn't the traditional arrangement, I would say that you're combining bolts and welds in a connection, because the restraint conditions of your connection plates vary due to those two styles of connection. On top of that, you have added flexibility due to the bending in the leg of the clip angle on the bolted connection.

Personally, I wouldn't consider the bolted portion of this connection as contributing. Proving compatibility between the two different forms of connection is going to be impossible with this arrangement. You can't even make the bolted portion properly slip critical, because of the bending in the angle. You can read the S16 clauses on combining bolts and welds, but I don't see how you could meet the intent.

 

[pbc825]

Thank you all for you advice. I agree that I've mis-conceptualized the term shear lag. I reviewed some hand-written notes from a grad class that's added to my confusion. My apologies.

As TLHS describes, I'm interested in load path and compatibility of assuming vertical shear load is shared between the angle side and the welded web stiffener tab on the connection. Or rather, if that assumption is reasonable. The welded connection side is doing the lion's share of the work, and in some instances, the bolts are sized such that there is no need for the angle on the opposite side. I'd offer that we're typically recognizing a ultimate state that occurs just prior to failure. If failure were to occur then I would estimate the angle would pick up some load.

The big reason I ask is that the for this project in particular, some connections are by us, and others are done by the engineer of record. The engineer of record has included the same style of connection, so we're trying to copy it. But I was struggling to understand what the load path between the two sides could be or if load equalization could be assumed. That's what led me to review the grad class notes.