The Chevron Effect and Analysis of Chevron Beams—A Paradigm Shift 1

[MacGruber22]

Who has read this paper? We had a webinar a couple of months ago with Patrick Fortney and it was very enlightening. I have a new 10-story braced frame project starting as I speak and need to fully understand the consequences. After spending a day reading the paper and working through example problems, I am very bothered by what seems to be a simple issue at the heart of it. How did we miss this? I wonder when this will be in the code as well - if it as critical as Fortney and Thornton say, then you would think that it would be incorporated in a new printing. It is not as if this paper was written by some random engineers...

AISC Chevron Effect Paper

-Mac

 

[JAE]

I looked for follow-up comments in the Journal but didn't see anything.

Might show up in subsequent codes. If you design for Unity = 0.95-1.0 then maybe there's a problem.
If U < 0.85 then maybe not.

Check out Eng-Tips Forum's Policies here:
faq731-376

 

[WARose]

It is not as if this paper was written by some random engineers...

I'll pass on that one.

How did we miss this? I wonder when this will be in the code as well - if it as critical as Fortney and Thornton say, then you would think that it would be incorporated in a new printing.

What we do isn't exactly 100% settled science. Remember the quote:

Engineering is the art of modeling materials we do not wholly understand, into shapes we cannot precisely analyze, so as to withstand forces we cannot properly assess, in such a way that the public has no reason to suspect the extent of our ignorance.

 

[MacGruber22]

If you design for Unity = 0.95-1.0 then maybe there's a problem.
If U < 0.85 then maybe not.

I will try to summarize some key parts of the paper, albeit poorly I'm sure. Attached are the presentation slides if you don't feel like reading the original.

First, let's assume that the paper is reasonably correct in its assessment. Sure, having fat to the design may cover it without needing to specifically evaluate in many circumstances. I am not sure I am comfortable in using that logic for all cases based on what I have read. The example in the report that is potentially concerning is the condition where brace forces are near balanced in magnitude. Thus, the beam is not subject to much global load other than axial (assuming the beam is not taking much gravity load itself).

The report has suggested that this condition could result in the beam being significantly under capacity in shear and/or moment within the gusset length. The argument here is that the 'real' forces in the beam are dominated by horizontal gusset-to-beam local forces, rather than by global beam forces generated by significantly high unbalanced vertical brace forces. Traditional centerline frame analysis doesn't account for this local effect, naturally. The moments, shears, and beam end reactions are unaffected by this effect since global equilibrium of the frame analysis remains.

What we do isn't exactly settled science

Never claimed nor intended to imply. I am merely bring up an issue from two well-known AISC code writers, of which they contest is very important to consider. I would appreciate specific contribution to the topic.


-Mac

 

[Settingsun]

I've only skimmed the presentation, which I admit is risky since the discussion by the presenter is often critical to interpretation. I also don't know the procedure of delegated connection design as have always been responsible for both member and cxn design. I'm not inclined to buy the referenced article however since it appears that this is just a rebadging of St Venant's principle - is that correct? The real distribution of forces on the beam (via the gusset) is replaced in the global analysis by a statically-equivalent set of concentrated loads so the analysis loses accuracy over the extent of this substitution.

The presentation seems to imply that there isn't coordination between the beam designer and connection designer or that the coordination is limited, perhaps one-way: the beam designer gives parameters to the cxn designer and the cxn designer can't give feedback that the design may be more economical with a larger beam instead of complex connection. The beam designer therefore should do at least a concept design of the connection to understand its likely effects on the beam economy. Goes without saying really. Connection design is serious business - connections fail, not members (usually).

The bending moments attributed to the connection designer (eg +/- 3502 on page 119) look ridiculous. Where does the concentrated applied torque at each end of the connection come from? I really hope I've misunderstood this, or the author is being unfair on the cxn designers.

I haven't thought this through at all, but the examples had substantial gussets compared with beam size. Does that provide the necessary local increase in beam capacity by happy coincidence? We've all read test reports that said something along the lines of 'although the effect can be shown theoretically, the tests indicate traditional design is adequate'. Perhaps this is one of those fortuitous situations.

 

[JoshPlumSE]

That presentation is interesting. Thanks for sharing it.

This is something that (when presented clearly like this) is very easy for most engineers to understand. The forces imparted into the beam from the chevron bracing are not infinitely small point loads applied at the work points. Rather they are really spread out over the width of the gusset plate. I have to think most engineers understand this, they just consider that a connection design issue. Whereas this paper / presentation is arguing that it can affect the beam design as well.

I don't know that there really is an emergency concept to take away from this. Yeah, there could be some localized shear issues in the beam. I have to think that is probably pretty rare. For seismic cases, at least, we're usually over designing those beams by a decent amount. So, I don't know that shear is much of a worry for me.... Though I will think about it now. I suppose if we make the beam a lot deeper, but the web doesn't increase in thickness then it could be an issue.

 

[MacGruber22]

I also don't know the procedure of delegated connection design...however since it appears that this is just a rebadging of St Venant's principle - is that correct?

Absolutely on the later. I only ever design connections for small projects and not often are braced frames, and never chevron types. I have only ever used the uniform force method since all of my braced frame connections have been brace to column - very straightforward in most cases.

For delegated designs, I have always proceeded as if the connection engineer should take care of local effects and report back if there were issues that could be more economically resolved by increasing member size. A typical example would be moment frames. RAM Structural System will check WF column joints and report if column stiffeners, doublers, etc. will be required, but it will not provide a design of same. Thus, where required, my connection detail notes instruct the connection engineer to check all local effects on member (wall plastification, etc.). Until I came across this paper, I thought my instructions to the connection engineer would be sufficient. It seems that it may not be true in all cases, as the paper asserts that conventional local effect analysis of chevron connections by delegated designers do not always capture the effects properly, and sometimes by a great margin (see edit to this comment below). Maybe I am missing something, but that is my takeaway. If this is true, I appreciate the paper if only for me to keep this in mind when designing and reviewing connection calcs.

The bending moments attributed to the connection designer (eg +/- 3502 on page 119) look ridiculous.

After re-looking at the diagrams, I think I may have myself confused on the intent of the topic. The 'conventional' design of these connections do seem to have huge moment increases as you point out. The so-called symbiotic method has moment increases that are much less. So, does this mean that true issue is that connection designers are actually designing for unnecessarily high local effects, and that EOR's are often under-sizing beams, resulting in a lot of stiffeners/plating? If this is true, even if I used the CE method in member design (and communicate same to connection engineer), the issue isn't mitigated unless the connection engineer analyzes in the same manner. How likely is it that I can force the connection designer to do this?



-Mac