I believe 3.7.4 does include double angles back to back when used in transmission towers because they are used all the time. The gold standard software is PLS-CADD's TOWER program which includes properties for double angles with both short and long legs back to back as well as equal leg angles. You have to be careful to connect the angles so that the individual L/r is lower than the composite L/r or they do not behave as one member. The new version of ASCE 10 is in the final stages of approval and is going to the publisher soon.

I would caution you that the compression formulas in ASCE 10 were done after years of full scale testing on tower structures, and your catwalk support may be a little out of line for the assumptions inherent in T-Line towers. Our structures are analyzed as trusses with the compression formulas derived from testing. In the very old days when I started designing them, all work was done as a 2D determinate truss using a graphical method of joints and no computers. Our towers use very low factors of safety (compared to bridges and buildings) because we have wide rights-of-way and if one falls down in an ice storm or hurricane, we go out and hang the wire back up in the air on temporary poles and rebuild the downed tower. Our industry is unique in that we will allow a few failures. We design for a 50 year return period storm so if we get a 200 year storm and we lose a few structures, we accept the failure and go out and rebuild the line.

The building and bridge engineers would never accept a failed structure or be sued into the poor house. If your catwalk is used by people and a failure could hurt someone, I would steer away from the compression allowables in ASCE 10. I don't use AISC much but why don't you use the compression equations for double angles in AISC?

_____________________________________
I have been called "A storehouse of worthless information" many times.

The tower supports a catwalk that holds a conveyor used to transport material. No people involved unless someone is up working on the conveyor and then no one is going to be up there when the wind is blowing 90+ mph. Everyone will be in the basement because the only time the wind blows like that around here is during a tornado. I am analyzing the tower as a cantilevered truss with pin connected bracing. I am using wind load from ASCE7-05. Towards the top of the tower I have single angle bracing members. I have been using a modified slenderness ratio for the single angles per AISC Section E5 and plugging it into the equation for critical buckling stress from Section E3 of AISC. The equation for the critical buckling stress in AISC is the euler buckling formula and is identical to Eqn. 3.6-2 in ASCE10 with the exception of a 0.877 factor which accounts for the effect of member out-of-straightness. The modified slenderness equation given in AISC for single angles looks a little different than that given in ASCE10 but I get nearly the same slenderness ratio (KL/r) from both equations. In fact, the commentary to AISC Section E5 states that the modified slenderness equation is "essentially equivalent to those employed for equal-leg angles as web members in latticed transmission towers in ASCE10-97". The commentary also goes on to state that the equation accounts for a decrease in effective length and a reduction in end eccentricity due to end restraint from the connection. I should mention that I am using at least two bolts in my connections.

Towards the bottom of the tower my forces get larger and I have to use double angles. I would assume you get the same reduction in effective length and reduction in end eccentricity due to end restraint as you do for a single angle, however, AISC doesn't say anything on double angles used in truss members. I could use K=1 and evaluate the interaction equation for axial load and bending with M=Pe but I think this would be way too conservative. So the question is what value of K do I use and how much does the end restraint reduce the moment due to connection eccentricity? I would be alot more comfortable using a slenderness ratio (KL/r) based on years of testing than one based on an educated guess. I am just not 100% sure that the equations in ASCE10 Section 3.7.4.2 apply to double angles connected back to back. The only thing ASCE10 Section 3.7 says is "the provisions of this section are applicable only for 90deg angles". Is this to mean single and double angles?

All this has led to another question or maybe a clarification. It appears that the tower industry models towers as 3D trusses with pin connected joints. However, if the member connection is made with welds or at least two bolts, the industry considers some end restraint in determining effective member lengths used in design. On the other hand, any forces resulting from this assumed end restraint are ignored in the design of the connection. The connection is designed simply for the compression or tension in the member. Is this correct?

The compression adjustment equations for KL/r do apply to double angles because they are used extensively in T-Line design for lacing members. The leg bracing equations also are applied to built up leg members and I believe they are called cruciform where 4 angles have their heels bolted together with spacers to form a cross. Take a look at 3.11 for stitch bolts that apply for double angles so the individual L/r is lower than the composite L/r. I know most of the committee members that wrote ASCE 10 (and are currently doing the revision) and I can assure you the double angles are covered.

You are correct in the 3D truss assumption. The tower is analyzed as a truss and the moments in the members and connections are usually neglected. We also do not use A-325 bolts torqued to prevent slip. We only care that there is a bolt in the hole and the load is transferred through bearing on the hole and shear in the bolt shaft. We use a special limited thread length A-394 tower bolt to try to keep the shear load out of the plane of the threads.

The legs were usually lap spliced when the members were not very thick. We had a full scale tower test many years ago where the legs failed at the lap splice at about 75% of the maximum load, so we now use butt splices for the thicker leg members. Our very large towers use a welded up 12x12x1.5 single angle leg made from welding 2 flat plates together.

_____________________________________
I have been called "A storehouse of worthless information" many times.