You need to address both local and global buckling of the twin girder system.
Note that the information in Appendix C does not adequately cover a two-beam "global" buckling mode. A closed form solution is provided by Yura and Helwig in this paper which I highly recommend reading for this type of system:
Global Lateral Buckling of I-Shaped Girder Systems
J. Struct. Engrg. Volume 134, Issue 9, pp. 1487-1494 (September 2008)
Obtainable here:
http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JSENDH000134000009001487000001&idtype=cvips&gifs=yesAs a quick and dirty guesstimate - if the moment of inertia of the twin girder system about the system's y-axis (using the parallel axis theorem - i.e. sum of Ad^2 about the centerline between the two beams) exceeds the sum of the moments of inertia of each individual beam about their strong axis, then you will not have a global buckling problem. This is not always the case though as indicated in the above referenced paper.
Assuming that the global buckling mode is satisfied, this then leaves the design of the individual cross-frames to serve as brace points to reduce the unbraced length for the buckling of the individual beams between each frame. The angle between the top (compression) flange will be serving as a nodal brace (Appendix 6.3.1b) and will need to supply the required brace strength from eqn. A-6-7 and required stiffness from eqn A-6-8. The supplied stiffness will be equal to 6EI/L of the brace member assuming it is bent in double curvature between the two beams. The other possibility since you have top and bottom bracing is to treat them together as a torsional nodal brace per A6.3.2a and break the required moment down into a axial couple for each member as well as satisfy the stiffness requirements.