cross bracing in compression

huangyhg

cross bracing in compression
a question for you folks.
let's suppose you have a structure supported by steel columns with cross bracing in an x-shaped configuration between columns.  the structure is loaded vertically and with a shear load at the top.  under what conditions is it permissible to neglect compressive stresses in the design of the cross braces?
if the cross braces are very thin (ie, rods with turnbuckles), i can see how they would never carry any compressive load, yet would not be damaged by a minor amount of movement in the compressive direction- they would just sag a little more.
or, if the member sizes were comparable to the columns themselves, then the braces would be expected to carry part of the vertical loads as well as tension or compression due to shear at the top.
but at what point do you transition from one case to the other?  is it the whim of the designer?  or do you pretend that any member that is inadequate to carry its calculated compressive load just doesn't have any load?  is there any particular l/r ratio that you could use as a cutoff?
i'd be interested to know your thoughts on the topic, and would be especially interested in any references that might clarify this.
design codes could vary, but aisc-asd would be typical.
check out our whitepaper library.
i usually consider tension members only (but not under cyclic/seismic loads!). i just check that   
you can check the   
you're speaking about the stiffness matrix.  if the design is by hand, then you do need to set some limitations as the previous posts imply.  but it's all about deflection/flexibility/stiffness, something we usually let a frame program figure out.  probably most of us have solved a stiffness matrix by hand for a frame in school, but i doubt there's many that can do it later in life.  i know i can't.
ignore the compression strut, and assume simple span of the diaphram. the compression   
thanks for the input, all.  jae, your approach to the problem seems very reasonable, and it appears the other responses could be considered as case #1 in your response.
the problem i ran into some years back was that treating the braces as tension only made them large enough that i didn't feel like i could neglect compression loading in them.  in a single x-braced frame, the tension and compression in the two struts from shear are about equal.  once you start designing the braces for compression, the larger l/r gives a low compressive stress, which makes the braces much bigger still.  and of course, when you make the braces bigger relative to the column size, even more of the vertical load is carried in them.
i've never seen this issue addressed anywhere, and it is nice to see that there is some variation in the approaches used here!  thanks again.