unbraced length 9seemingly very fundamental0 question

huangyhg

unbraced length (seemingly very fundamental) question
i want to explain a situation that recently happened in my office and i would appreciate some feedback.
i am working on a project for which i am the main contact for our firm.  i am doing the design and coordination with the architect.  i check in periodically with a senior engineer to go over what i am doing.  
i had an exchange this week that troubled me a bit.  
there is one situation where we have a four sided, self supporting "closet" (so to speak) to house some sliding doors.  there are some very small members that span wall to wall and bear on top of the wall.  they are supporting little load.  i decided to make them small hss so that i could attach a light gauge clip angle to both sides of the tube where it bears on the wall so that these two clips would prevent twist of the member at this location to provide a brace point.  i told him the reason i did this was that if i couldn't brace the end, i wouldn't be able to calculate an unbraced length because it wasn't restrained anywhere.  even if i provided one clip for a channel, it could still twist at the ends and just flop over.  
he proceeded to tell me that if you have a beam, set it on a wall (without restraining the top flange at a single location and without restraining the ends against torsion) that the unbraced length would be the actual length of the beam.  i expressed some concerns about this, and even stated that aisc would not allow this (pointed to the example of a seat angle having the requirement for the top angle to brace the ends).  he proceeded to just tell me i was wrong and that the unbraced length is never larger than the actual length (which i agree with as long as it is actually braced at its ends).  
he also talked about how he was talking about strength and not overall stability of the member.  isn't lateral-torsional buckling a stability failure and not a strength failure.  as far as i am concerned that is a stability issue.
anyway, i think he came away from that meeting with the impression that i am lacking some fundamental understanding of behavior, but i am not seeing it that way.  
can anyone here comment on my thought process?
this person has been with the company for about 10 years and has about 20 years of experience.  he only recently came to our office and this was one of the first interactions i've had with him.  he is well-respected in the office and i hate the idea of him thinking that i am lacking fundamental knowledge.
it ended up being a very akward situation for me, because i felt like he was missing the fundamental idea i was trying to get across (which i thought i did pretty clearly), but how does a guy with 1.5 years experience say that to a 20 year guy?
any thoughts?
check out our whitepaper library.
if you set a beam on a wall and have no connection other than direct bearing and the load applied to the beam is perfectlyfff"> concentric, then the beam will simply span from wall to wall and the unbraced length would theoretically be the span length because:  the compressive bearing stress at the end would inhibit one side of the tube lifting off the wall.
however, there is no such thing as a purely concentric loading or perfectly straight beam so there will always be some level of lateral torsion or twist in the member.
so with some level of torsion (due to beam sweep or load eccentricity) the beam would have to have some sort of load path to resist this torsion.  again - lifting of one flange tip or beam edge off the wall would be resisted to some extent by the compression bearing stress.  when the beam tips sideways, the bearing reaction moves to the other side and you have an effective resistance to the torsional moment in the other direction.
example:  take it to a crazy extreme....you have a 20" x 20" tube sitting on the wall.  with a l lb load how would it ever twist off the wall?  it just intuitively wouldn't.
but your main question was unbraced length.  
the question then is:  does the bearing compressive stress in the end of the beam serve to inhibit section rotation at that point?  
i don't think, in normal terms, that it does so in such a manner that can be checked, quantified or relied upon.  
even with my crazy extreme example above, as a structural engineer (with 29 years experience) would i count on it except for very small loads or   
jae-
i believe even if you bolted the flanges down or welded them that this would still not constitute a brace point.  it is certainly more braced than not welding or bolting, but aisc require stiffeners along with the bolts where beams continue over columns (or other beams) in addition to the bolts, correct?
for a seat angle where the beam is welded to the seat, you still are required to put an angle at the top flange (or near the top of the web), correct.
i can see the theory of a perfectly concentric load as you noted above, but that situation doesn't occur in real life so why go into discussion about it during a meeting?
as for a 20"x20" tube, that is inherently stable because it is a section with equivalent moi in both directions, correct?  if you have weak axis bending there are no lateral-torsional buckling considerations, correct?
i agree with the example you gave about the 20" x 20" tube with a 1 lb load, but as you point out that just isn't realistic.
thanks for the input.
but aisc require stiffeners along with the bolts where beams continue over columns (or other beams) in addition to the bolts, correct?
aisc section k1.5 requires a check against sidesway web buckling.  theoretically, sometimes you do not need stiffeners at continuous beams over columns but every collapse i've ever seen has involved the lack of these stiffeners so i always use them no matter what.
for a seat angle where the beam is welded to the seat, you still are required to put an angle at the top flange (or near the top of the web), correct.
yes but i think it is simply a good practice, not necessarily implying that in every case it would fail if you didn't.  
as for a 20"x20" tube, that is inherently stable because it is a section with equivalent moi in both directions, correct?  if you have weak axis bending there are no lateral-torsional buckling considerations, correct?
correct.  i agree.  i was just using the 20 x 20 example to help describe the force couple developed when the tube tries to overturn.  it would be applicable for a 20" tall by 18" wide tube.
keep in mind my post:  i agree that a connection is required...aisc requires it as i indicated in my last paragraph.
jae-
i appreciate your input.  the last thing i would like you to comment on is this - he talked about lateral-torsional buckling (and the associated unbraced length) and overall stability as if they were mutually exclusive.  
my understanding is that lateral torsional buckling is a stability failure.  when the moment strength of the beam is limited by lateral torsional buckling, then you have stability issues that require reducing the amount of load a beam can handle compared to its full strength.
i think the disconnect might be due to the type of   
i know that the ltb equations are based on the ends of a beam beaming restrained. but i also agree that it should be restrained even when ltb is not required to be checked just based on global instability coming from tolerances, etc.. and also since it is stated to be required aisc.
structuraleit:
i would entertain the idea of the required torsional restraint being provided at the ends of a w section, or hss for that matter, if the ends were bolted to the top of the wall through bearing plates welded to the beam end, and an additional end plate welded to the beam and bearing plate to transfer the torsional forces to the bearing plate.  the bolts would then take the torsional force to the wall structure.
this is essentially the scenario for a cantilever beam bolted to a concrete wall, but the torsion is resisted by the bolts in tension as opposed to shear for the cantilever.
mike mccann
mccann engineering
so what should i do now?  it seems like he has the impression that i don't know what i'm doing (maybe thinking) and i know his opinion will be valued when it comes time for my review.  i wouldn't be as concerned if this weren't my first interaction with him of any reasonable length (you know what they say about first impressions).
jmho:
you might just print off this thread and show him that intelligent, experienced, structural engineers disagree about seemingly basic questions. if everything in our field was really black and white, its doubtful this board would exist.
our office consists of several very expereinced engineers and 1 eit. the eit often shows the rest of us the errors of our ways. the experienced engineers are wise enough to realize that they don't know everything ( and really only dimmly grasp the outlines of what they don't know in a lot of cases), and can always learn something new.
not everyone will like you, or agree with you, or think of you as you think of yourself.  deal with it and move on with life.