roof deck lateral buckling

huangyhg

roof deck lateral buckling
just curious as to what other engineers do in this case:
steel beam roof framing (could be joists as well) where a metal roof deck spans perpendicular to the exterior wall.  when lateral forces are applied to the wall (say a stud wall) there is a lateral reaction at the top of the wall.  usually, we provide a continuous angle that runs down the length of the edge beam to facilitate connection to the wall studs.
now once the lateral force is transmitted into the angle, and then into the perimeter beam, we assume that it is then transferred into the metal deck diaphragm.  however, the question is:  can a light gage deck take the lateral force.
the deck is acting as a compression element, taking axial loads, with an unbraced length equal to the beam/joists spacing.  many times, especially in seismic areas, we provide additional steel angles, perpendicular to the wall, directly below the deck and connected to the deck.  these angles extend one, two or more roof member spaces into the diaphragm before terminating.  
is there a way to check the deck - alone - as taking this axial load verses always adding these struts to transfer the load into the deck more gradually?  especially so in low seismic areas as in high seismic i'd probably want the hard steel anyway.  
you have an unbraced length, an ix of the deck per foot, and a cross sectional area.  can you just get a kl/r and use aisc compressive capacity equations?  aisi?
what you describe is what causes the common practice of "hard steel" "wind beams", laying or aso along the roof, for example in hangar doors where obviously the dynamic solicitation, weight etc make that a 3d beam (maybe a truss) be better.
in any case the question about using the roof deck set orthogonally to the edge as bracing reverts to its behaviour as a web. the consideration of saya roof module as a column per aisi i think would be ok but of itself doesn't solves the bracing problem as long the load path is not completed. if the roof was entirely flat and you add diagonal "a la pratt" or so diagonal   
yet the analogy of mörsch approach is a risky one since it is obvious to anyone that only an undefined band width of roof deck near panel points can be taken...furthermore the panel deformation (one diagonal shortening, other -along the pratt diagonal-in elongation) shows the deck in the panel be in shear deformation, this reverting us to the question of the shear rigidity of deck panels.
so the fem folded plates model that captures what elasticity affords should be better. then upon seeing the forces one can thnk on how to tune the stiffnesses to show believable results that are at the same usable for safe design.
thanks for the reply.
i don't think that, for most structures, developing a 3d finite element model of a roof deck is the best way to approach the problem.
the deck is usually flat in this situation and as the load is applied to the "edge" of the deck, there is direct compression into the flutes and locally, the deck is acting like a column.  globally, it is taking a compressive edge force and resisting it as a flexible diaphragm.
aisi is such a headache to use that i was hoping aisc would be applicable for the deck compressive check.
have you considered creating a horizontal truss with angles between the first joists space? this is more positive than relying on the shear and compression properties of the deck.
the aisi specs for light gauge are a pain in the rear, but you should be able to analyze a unit section of deck, say one foot wide, for buckling. i would assume some fixity - say 0.75 x span for the actual buckling length.
redhead...yes, that's what i've thought about...just not sure if using aisc lrfd or asd compression   
some things...
i am not proposing to make a 3d model of decks for every case; yet it is recurrent the question of the shear rigidity of these decks, that the 3d model can capture and one take own appraisal of what elasticity plus pdelta at least says about. in any case, to make such model, the array features in autocad make it very simple, then import to the analysis program.
respect analysis, of course one for thin sections, whichever the code. aisi has mathcad sheets that portrait the specified procedures, contact them on how to get such.
also, the roof acting as a strut as part of the wind resisting system can and will coexist with some gravity forces in sonow areas. it is then not a pure axial force case, but one with some flexure.
and the truss to which redhead refers to is the "wind beam" or "wind bracing" at roof level, for small structures sometimes being even rods in x taking the entire slope, also used to deliver indeformability.
jae:
i don't think aisc is applicable to the buckling of the deck because it is light gauge and local buckling considerations are a major factor. also, ishvaaag is correct. the problem is complicated by the concurrent bending from snow load.
stay with the independent bracing is my advice.
teng22 (visitor)20 feb 02 21:03
    i think " steel deck  diaphragm design manual" by prof. larry d. luttrell would be a good reference. you can buy it from sdi. his research shows that compressive strength of deck is controlled by local buckling at each flute, and he develop some kind of eqivalent spring concept to check it.
teng, we have the book, thanks, i'll check it out.
jae,
another good book on steel deck diaphragms is "designing with steel joists, joist girders and steel decks" by fisher,west and van de pas.   it was distributed by nucor corporation, sometimes free of charge.
aef
jae, a few thoughts...
i think the column buckling analogy is applicable, but is severely range dependent, meaning that it has applicability over a short range of material behavior.  the difficulty is quantifying that range!
localized bucking has a significant effect; however, i think more important is the suddenness of the buckling potential in light gage materials.  has to do with "oil canning".  this makes the critical length consideration in column buckling fall into a very short range, probably quantifiable to a great degree with conventional buckling approach and slenderness.  these effects will vary with deck style, for instance type b would be greatly susceptible, type f less so, and type n even less.
using the "strip" concept is conservative as there are plate interactions and effective lateral bracing that occur without adequate treatment in the analysis.