Double channel box beam vs. tube section

[jay156]

This is an interesting thing that happened when I was designing reinforcing for an existing C6x8.2.

I checked it in RAM Advanse for the torsion load of a new gangway they want to add to it, and it failed, spectacularly (d/c=12.6). So I thought to reinforce it with another C6x8.2 welded at the tips of the flanges to make a box. It still failed, though precisely half as bad (d/c=6.3). Then just to see what would happen, I changed the section to a TS6x4x.3125 and it passed with a ratio of 0.4!

The software is very obviously treating the channels separately. It says the torsion constant J of the box section is 0.13 in^4, though it really should be ~25.

My only question is, is there a reason that, if the flanges are welded together, that I shouldn't consider the channels to be acting together as a box?

 

[frv]

Hmm.. Obviously, it would certainly be better than considering two separate channels if you weld them together (although VQ/I doesn't apply to side-by-side "composite" sections- it certainly helps with torsion)

However, I think that ever since the Hyatt regency collapse in Kansas City, the idea of welding channels flange-to-flange makes many engineers recoil.

 

[csd72]

frv,

The problem at the hyatt was with the connection of the rods, not with the members as such.

jay156,

This is why you need to try and understand the assumptions behind the program. Just as every law has loopholes so too do every analysis program.

 

[frv]

csd 72,

That's true in the sense that the change in connection design primarily led to the failure. However, the assembly did fail (in part) at the welded (boxed) section. See the link to a Wikipedia photo:

http://en.wikipedia.org/wiki/File:Hyatt_Regency_collapse_support.PNG

 

[dhengr]

Jay156:
How would the computer program know it should treat the two channels, toe to toe, as a box section if you didn’t tell it to do so in some way? As a bending member the two channels act as the sum of two channels, except that their lateral stability is improved by being attached; but in torsion the box section is far superior, and we know that from our first Strength of Materials course. A box section might be made up of top and bottom plates and two side plates, but you can’t input them to the program as four separate plates and expect the program to just automatically interpret that as a box section, you have to connect them in some way so that the program understands how to treat them. Then design your welds accordingly.

The problem is that that toe to toe weld is a difficult weld to make when you put the toes together. How do you prepare the tips to get anything other than a cover weld on top of the flanges, with a little penetration and a poor root condition? And, then how is that weld stressed? You must understand how that member and joint are going to be loaded and stressed under usage so you know how to design and detail that weld joint. And, that’s what wasn’t done (went wrong with) the Hyatt Regency collapse. The Arch. or Engr. of record showed a detail that couldn’t practically be built; the detailer showed a detail which could be built, but none of the three paid any attention to how that member was going to be loaded or stressed; then nobody really checked the shop drawings of these members, and the rest is history. If they had taken that load into the channel webs with some stiffeners around the hanger rods, and to prevent the weld from ripping at the root, everyone would still be dancing. We really have to pay attention to everything we are doing as Structural Engineers, it’s all important, and the smallest detail can come back to bite you, even if you do run a FEA on it.

 

[a2mfk]

All good points. Like Dave said, the fabricator changed the continuous rod design since that would have been a bit$% to install. So the top hanger connection now supported the lower levels which the EOR had not designed it for, so a pull-thru failure at the bottom flange was inevitable since that single connection now supported the lower three levels of walkway.

Back to the OP, and going off what dhengr said, I'd probably add top and bottom plates and avoid the toe welds to create the box, if that is the route you want to go. Pretty simple fillet welding though the underside is overhead. Calc the shear flow for the weld size, and then you end up with a pretty nice box shape. I am not sure exactly what that does for you in torsion, I'd have to give it more thought...