This subject is addressed in Merritt's "Standard Handbook for Civil Engineers, 3rd Edition". The discussion is in the chapter about Wood Design & Construction (Chapter 11) but should apply just as well for steel members.

Here is the relevant quote: "...general experience indicates that a brace is adequate if it supplies a restraining force equal to 2% of the applied load on a column or the force in the compression flange of a beam. It does not take much force to hold a member in line, but once it gets out of alignment, the force then necessary to hold it is substantial."

I have always found this to be good guidance.

I myself am these days reviewing the brace things and feel that not all is well settled as of now. In particular, and waiting for my entire revamped understanding of what out there, I have the vague feeling of that LRFD 99 approvers have felt something of the same and have gone in some way conservative.

In any case, clear definition on how to proceed you have there, at least on force and stiffness demand for orthogonal braces. This you may find in chapter C of LRFD 99. I think I have worksheets to estimate these demands according to such code except for torsional bracing, but I would love to have the better understanding I am looking for before making them available at the Collaboratory site.

Once you know the bracing force and stiffness to find a tube that meets this is quite easy, or maybe a rod if  tension only braces.

AISC - Section C3.4a. deals with lateral bracing of beams - you are right on with the idea that stiffness AND strength are the two requirements.  You didn't mention what country/spec/code you are designing under so AISC may not be applicable.

alanderbyalfiler's statement about LOCAL flange buckling has nothing to do with lateral torsional buckling in a beam.

Around 2% is what I have heard before also. I would keep the relative stiffness of the brace in axial, to the main member in bending that direction, pretty high.

Thank you all for your comments.

I have heard of the "2 percent" rule to calculate flange brace forces but am aware of at lease four different variations:

1) Use 2% of the force in the compression flange of the member being braced

2) Use .2% of the bending moment in the member being braced

3) Use 2% of the nominal strength [(Fy)(bf)(tf)] of the compression flange of the member being braced

4) Use 10% of the applied axial load in the member being braced.

All of the above are rules that I have heard from other more senior Engineers but have never actually seen in writing. Is anyone familiar with any of the above?

JAE, When you say "AISC- Section C3.4a." Do you mean the ninth edition of the ASD Manual? If not, could you please give the full reference for what you are describing. I am designing for the US using the ASD Steel Manual (9th Ed.).

Ishvaag, Can you please give me a clearer picture of what you are referring to in LRFD 99? From what I see, LRFD 99 does not seem to address the issue of determing loads in flange bracing.

Thanks.
JS.

Salmon & Johnson's textbook "Steel Structures" has detailed description, procedures and example on this topic.  I followed it before.  Check it out.

It's not in the ASD 9th edition, although it or something like it may be in subsequent addenda to the 9th edition (ASD Spec is really getting old).  I was referring to the LRFD manual - sorry.

This stuff has been developed over the years by Joe Yura at Univ. of Texas - Austin.  The 2% rule is rough, very rough, and probably usually conservative.  It would be based on the compressive force in the top flange.

In Australian practice, a restraining member to a compression member must resist, at a minimum, "0.025 times the maximum axial compression force in the member at the position of restraint" (AS4100-1998, Clause 6.6.2).

The commentary to AS4100 provides the following references:

   Mutton, B.R. and Trahair, N.S.
   "Stiffness Requirements for Lateral Bracing"
   Journal of the Structural Division
   ASCE
   Vol 99, No ST10, Oct 1973, pp 2167-2182

   Mutton, B.R. and Trahair, N.S.
   "Design Requirements for Column Braces"
   Civil Engineering Transactions
   Institution of Engineers, Australia
   Vol CE17, No 1, 1975, pp 30-36

In th Uk we use 2% of the squash load of the compression column flange.
Forced required in Brace = 0.02BTpy.
B=  Flange width
T=  Flange thickness
py= Steel strength.

ishvaaag mentioned some rule of thumb guide sheets and a collaboration site. I'm looking for some on this and a related topic: Lifting lug design. Any suggestions?
Thanks-
C.Fee
cfee@ev1.net