Dogbone Design 4

[Althalus]

I'm getting further and further into seismic design lately. Most of the methods described in AISC 341 are straight forward and easy to understand.

I became aware of the use of dogbones to combat oversized connections. The principle seemed simple enough. But when I brought it up to another consultant, he raised his eyebrows and warned that this was a very complex design methodology and you really need to know your stuff. This came from the manager of this company's structural group who has a guy whose master thesis was on seismic design.

So, I'm wondering if I'm missing something. Here's what I understand. Tell me what I'm missing.

1) One requirement of joint design requires that it be designed for the "maximum load that can be transmitted through the joint".
a) This means that theoretically, the full strength of the member (including the factor for material being stronger than the nominal strength) being loaded to full load capacity at the connection.
b) Assuming maximum moment controls. I'm still unclear on whether that allows for combinations of axial, bending, and shear. But it seems that (at the very least) it does not take into account variables like unbraced length, k value, etc. (correct me if I'm wrong.)

2) One way of combating that is to weaken the beam at a location where the load is rather low.
a) Since the maximum moment being transmitted through the weakened point is much lower than the maximum beam capacity, the resulting moment at the connection is also lower.

3) Placing a "dogbone" near the end of a beam is a common method.
a) The dogbone must be formed using smooth shapes such as circles and elipses, such that no load concentration occurs due to the shape of the hole.
b) It must be placed in a location that takes advantage of the low loads at a particular location, but still close enough to the ends such that the additional moment built up in the intervening distance
does not overload the connection.

So, what is the complex part of the design I'm missing?
And if anyone knows of a good online source (or possibly a publication) that goes over the steps, I'd really appreciate a reference.

 

[Agent666]

The dogbone is typically referred to as a structural system as a Reduced Beam Section (RBS).

By the sounds of what you've written you seem to have a reasonable grasp of the concepts involved. It really is no different to a normal ductile moment frame design, except that you're effectively limiting the moment capacity of your beam and hence the capacity demand at connections.

It can be a great way of producing an efficient moment frame design. Allowing you to tune the stiffness somewhat independent of the strength.

I believe AISC has design rules for these types of systems.

Locally to me in NZ, one of our industry bodies has created a few design examples and discussion on design of these systems that might be helpful. Basically they have taken the AISC guidance and localised it to the way we do things. Here, here & here.

https://engineervsheep.com

 

[JoshPlumSE]

You have to get a copy of AISC 341 and 358 to understand the RBS type of connection completely. My thoughts on it's "complexity":

1) The concept is relatively simple. You want to reduce the demand on the beam to column flange welds. You do this by forcing the hinging of the beam to be at a lower moment slightly away from this connection.

2) There are limits to what can be used based on the existing test data (beam and column sizes, clear span and such).

3) Lots of rules about the weld and connection geometry (rat holes, protected zones, weld inspection/testing, et cetera) that have to make their way onto the drawings.

Overall, it's not crazy hard to come up with a basic engineering design. However, you have to realize that it's not just engineering, it's material specs, detailing, fabrication and erection....

 

[KootK]

But it seems that (at the very least) it does not take into account variables like unbraced length, k value, etc. (correct me if I'm wrong.)

Your beam design will still take account of all of those things in addition to being designed for the over strength moment capacity at the dogbones/hinges and combination with other load effects..

One way of combating that is to weaken the beam at a location where the load is rather low.

No, what you're doing with the dogbone is actually weakening the beam at a location where the moment demand is near to its maximum value. The idea is to use the dogbone as a fuse of sorts to limit the amount of load experienced in other parts of the structure such as beam to column joints which often posses less ductility than is available at the dogbone.

So, what is the complex part of the design I'm missing?

It may be this:

However, you have to realize that it's not just engineering, it's material specs, detailing, fabrication and erection....

In high seismic regions where this strategy is commonly employed, it's not a big deal to execute. It can be a pain in the rear for all involved to try to execute the dogbones in markets that haven't dealt with them regularly before however. This may have been what your concerned colleague was alluding too. If your project is in Texas, it's probably governed by wind design rather than seismic. As such, local fabricators and contractors not have much experience with the dog bones.

 

[JoshPlumSE]

One thing I should mention is that if you were to design it as an OMF then a lot of the requirements of SMF / IMF could fall away. Allowing you to use an RBS connection without having to have all your ducks in a row on the drawings / details and such. IMO, it would be a good idea to give them a shot (especially the protected zone, and the rules on the geometry of the rat holes and backer plates). But, since you're not relying on the extra ductility for your design, you could probably get away with ignoring some of the portions of the concept that may be onerous for your area.

 

[Guest090822]

I did some research about 20 years ago on the use of viscounts-elastic dampers at connections for seismic reasons. That's about all I remember though LOL

 

[Althalus]

I would think that the same phenomenon would need to be addressed at the base of the column to the baseplate and foundation. But I've never seen a picture of a reduced column section. Why is that?

 

[KootK]

1) Foundations tend to be thick and stiff.

2) Columns tend to be squat.

1 + 2 = it's relatively easy to force the plastic hinge to form in the column rather than the foundation.

 

[JoshPlumSE]

Althalus -

Part of the general philosophy of seismic design is to encourage the inelasticity in the beams and discourage it in the columns. The idea is that when beams go inelastic, the structural integrity is NOT compromised. And, in a post event repair, it is much easier to replace a beam than a column. However, if you introduce a hinge in the column, then the gravity load carrying system may be compromised. The caveat to this is at the roof level where it's not uncommon to reverse the moment connection and force the inelasticity into the top of the column.

 

[Agent666]

The caveat to this is at the roof level where it's not uncommon to reverse the moment connection and force the inelasticity into the top of the column.

The same generally applies at the base of the columns as well. Foundation and baseplate stiffness relative to beams/column stiffness in the first storey often forces a hinge near the base of the structure. Provided their is no corresponding hinge at the underside of the first level you don't have a soft storey.

I don't think doing the same dogbone detail to the base of the columns is appropriate, nor warrented.

https://engineervsheep.com

 

[KootK]

I suspect that Althalus' point is that most conventional, steel moment frames require plastic hinge formation in the columns at the base in order to form a complete mechanism. Mechanically, I dig it as a means of forcing the hinge location to occur away from the base connection where behavior tends to be less predictable and more subject to non-ductile failure modes. It's not as though you'd be sacrificing a meaningful amount of axial capacity in most situations.

The dogboning has a real cost associated with it so, I agree, we'll not be seeing it at column baes unless the engineering community comes to feel that it is warranted for technical reasons. Frankly, I wouldn't be surprised to see that come to pass over time given how complex the joint behavior is likely to be and how little of this stuff has been tested and/or tested at anything resembling full scale.

 

[Althalus]

Thanks for the responses, y'all. But I found out the answer through analysis.

The whole reason we need the dogbone is to satisfy the "highest load capable of being transmitted..."

With the beams reduced, the maximum load transferred to the columns (and therefore, through the columns)is below the column capacity. If not, we just make the column stronger.

 

[KootK]

So "Strong Beam - Weak Column" was all that you really needed to hear on this?

 

[sandman21]

I suspect that Althalus' point is that most conventional, steel moment frames require plastic hinge formation in the columns at the base in order to form a complete mechanism. Mechanically, I dig it as a means of forcing the hinge location to occur away from the base connection where behavior tends to be less predictable and more subject to non-ductile failure modes. It's not as though you'd be sacrificing a meaningful amount of axial capacity in most situations.

The dogboning has a real cost associated with it so, I agree, we'll not be seeing it at column baes unless the engineering community comes to feel that it is warranted for technical reasons. Frankly, I wouldn't be surprised to see that come to pass over time given how complex the joint behavior is likely to be and how little of this stuff has been tested and/or tested at anything resembling full scale.

Are you suggesting that we would add a dogbone at the base? AISC already covers the base connection requiring a ductile limit state control or the connection is designed for capacity of column. The plastic hinge at the base occurs after the levels above have formed their hinges.

 

[KootK]

Are you suggesting that we would add a dogbone at the base?

Yes, in a hypothetical, futuristic kind of way.

AISC already covers the base connection requiring a ductile limit state control or the connection is designed for capacity of column.

I realize that but, based on what I've seen of these connections, I feel that it is often questionable how ductile they actually really are. That, particularly given the complexity of the connections that one sometimes encounters. Mixed materials, edge distances, assumptions about stress distributions... you know the drill.

A column dogbone would:

1) Make for a nicely predictable plastic hinge mechanism.

2) Allow one to shield the base connection from potential damage, as we do with beam-column connections.

3) Reduce the flexural demand at the base connection where it would be designed for capacity / overstrenght.

That's all I'm saying.

The plastic hinge at the base occurs after the levels above have formed their hinges.

I wasn't aware that we knew the order of operations with any particular certainty. What informs your opinion on that? I would have thought that it would depend on the proportions of the frames, the proportions of the frame members, and the relative base fixity expected.

 

[Althalus]

So "Strong Beam - Weak Column" was all that you really needed to hear on this?

Not exactly. Although, it "points in the general direction".

Remember that I started this thread because the dogbone is a new methodology for me. So, I'm not thinking in that mindset.

I just know that "the highest load that can be transferred" is often section dependent rather than calculated load dependent. Now, I'm allowed to actually calculate the load based on a reduced section (which is closer to the actual analyzed load).

The column then cannot be loaded higher than all the loads flowing into it. So, the column is also given some reprieve from the beams having a reduced section.

This was a new direction of thought for me.

 

[JoshPlumSE]

I don't tend to like the idea of a dog bone close to the base. Shielding the base connection from damage seems odd when you're forcing the damage into the column instead.

That being said, the idea of reducing the plastic demand at the base is kinda nice. How about alternate ways:
1) Something akin to the Dura Fuse connection.
2) Something like a cross between a flange plate and a dog bone where the only thing that needs to be replaced is something like an RBS plate.
3) I've seen one company use long chaired anchor bolts to force more ductility into the anchorage system. Not sure if it really reduced the demand at the base plate or not though.

 

[KootK]

Shielding the base connection from damage seems odd when you're forcing the damage into the column instead.

Are we not already do this when we take this path:

...or the connection is designed for capacity of column.

I get it, intentionally weakening a column has a cringey feel to it. But, philosophically, I'm not sure that's deserved.

 

[JoshPlumSE]

You make a good point. I don't really think along those lines we're usually designing the anchorage to be ductile. Right? So, we're hoping to get some inelastic elongation of the anchor rods and such.

I'm really just thinking about the post test RBS connections that I've seen and what they look like. Add a bunch of axial compression into that connection as well and I'm even less comfortable.

The overall question though, is a really good one. I know we have a ton of testing on various moment connections to ensure they can develop ductility. But, then we've got essentially the same thing happening at the base and we don't have nearly as much testing... at least not that I know of. Maybe in 20+ years we're going to see a boat load of Northridge type weld fractures at the column to base plate connections.

Note: This wouldn't be totally unprecedented... I know it's a different scenario (i.e. a braced frame), but the Oviatt Library at CSN had base plate fractures.

 

[KootK]

I don't really think along those lines we're usually designing the anchorage to be ductile. Right?

That is certainly a common approach. My preference is to actually embed the columns within grade beams if I can which definitely attempts to force hinging into the columns.

I'm really just thinking about the post test RBS connections that I've seen and what they look like. Add a bunch of axial compression into that connection as well and I'm even less comfortable.

I'll concede that point, that does sound terrible. I'd expect that much would depend on proportions though. A yielded, dog boned W18x35 is likely to look a lot different than a yielded, dog boned W14x90. This is definitely not something that I'd be willing to do without some testing in play to validated it.

Maybe in 20+ years we're going to see a boat load of Northridge type weld fractures at the column to base plate connections.

I share the very same thought. Structural engineering is very much a reactive enterprise in that we just keep fixing things until new things go wrong. We've got the beam-column joints sorted so what will it be in the next earthquake? Base connections?? Who knows? Steel to concrete anchorage has long been a sore spot for me in this respect. My perception is that, every time that we go and test a "first principles" anchorage design method, we find out the capacity is half of what we originally thought.