Steel Beam Torsion 10

[damo74]

Hi folks.

I have a job on site in which I designed a steel beam to carry some precast slabs and blockwork wall panels. The loads from these elements are basically acting through the centre of the beam. Consequently I designed a beam strong about the x-axis and weak about the y-axis. The client now wishes to place another block wall adjacent to the beam. However I would require an angle fixed to the beam to support it. The beam which has been erected fails due to the additional torsion, when I check it. Has anyone any ideas how I could stiffen my beam to provide additional torsional resistance? Do you think welding stiffeners in between the flanges would be of much use?

 

[flamby]

You will have to carry the reaction right to the centre of beam ie web if you have an I beam. You can otherwise use top or bottom central points of your beam to transfer the load and avoid torsion.

Ciao.

 

[UcfSE]

You could weld a continuous plate between the flanges to create a box section. You could also weld stiffeners at diagonals to make something like x-bracing between the flanges for the length of the beam. That of course assumes you have a wide flange or other I-shaped beam. "Design of Welded Structures" by Blodgett describes the latter. Vertical web stiffeners like are normally used at bearing points will do nothing to help strengthen the beam against torsion. I'm sure there are other ways as well, these two take a lot of welding. It seems that no matter what you do if you want the beam to take additional torsion as well as the additional bending that occurs anyway you will have a lot of welding.

 

[JAE]

I'd chime in that UcfSE is correct that making a box section by adding plates will do the most good in terms of torsional stiffness and capacity.

 

[shepherd]

Assuming you have adjacent beams and deck supporting top flange, you could provide kickers at close intervals off the bottom flange (or stiffener) to an adjacent beam to resist the torsion. Or if you don't have a deck on the top flange you could place diaphragms between the beams.

 

[UcfSE]

What do you guys think about welding small angles to each flange with the legs out to make a small "lip" at the end of the flange and using metal deck instead of a continuous plate? It seems like that would be cheaper than a continuous steel plate but I don't know right off hand how you would get torsional properties for that set-up.

 

[jike]

Can the precast slabs carry the load from the new wall? If so, let it carry the load to the center of the beam.

Another idea, is to modify the detail between the beam and the precast slabs so that the torsion in the beam is resisted by the bending in the precast slab. You may have to consult the precast slab designer to coordinate the connections and evaluate the bending capacity of the precast slabs.

 

[J1D]

damo74

How is new block wall is adjacent to the beam? Along the beam but off the center? like two walls standing side by side closely? I imagine:

1.the beam is an edge beam, and the new wall will be supported by the beam directly. The best way is to provide another beam under the new block wall, if it is feasible for construction and architecture. Second way is to provide torsional restraint by connecting bottom flange to the precast slab (on the other side of the new wall) like jike suggested. Welding stiffener plates between flanges to form a tube seems complicated and expensive in this situation.

2.the beam is an interior beam, the off-set new wall will be sitting on the precast slab. No torsion then since the load of the new wall is transferred to the beam through the slab bending.

 

[GregLocock]

UcfSE

The easiest way to work out torsional constants for fabricated sections is FEA. I'm pretty sure Timoshenko describes several methods to calculate it, but I often see factors of 2 difference between different estimates.

The difficulty is that plane sections do not remain plane, and if the section is non symmetric then figuring out where the torsional axis is can be a nightmare.

So I use FEA.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[tincan]

If the proposed set up is a typical spandrel beam, then you will have both normal bending and torsional bending.
Normal bending stresses +- warping normal stresses combine to increase the total flange stress.

About the only way to account for the additional stresses is to increase the area of the flanges. Since the beam is inplace, you could weld strips to the underside of the top flange and a plate to the bottom flange. Check the stresses before and after, since the strips will carry no dead load (existing in place loading)

If there is any way you can jack up the beam you could weld between the jacks, then relocate and weld in the voids. There's an old oil field expression, "make it hell for stout and then double it"

Have done the above on a couple of last resort, no other way jobs. Fortunately, there were no codes in the area to govern/restrict us.

"Design of Steel Structures" Bresler, Lin, Scalzi, John Wiley & Sons has a section on spandrels. Also, have developed spreadsheets for the solution.

Best, Tincan

 

[pba]

Don't fix the angle to the beam. Fix it to the edge of the slab and use straps on the top surface. - If this can be done the problem goes away.

Closed box sections are, as you know, better for torsion. I have in the past welded a square hollow section on to an I beam web to support eccentric loads.

Do remember that the load path for any torsional load will act through the beam end connections and onto the columns - Both will need checking for the torsional effects.

 

[DRC1]

If you can stiffen the ends of the beam against torsional rotation, the resistance of the beam will increase. I ha a project where I had to land a bridge on a beam, but the beam was not loaded thru the center. The torsion capaity of the beam was increased by welding large plates to the end. AISC has a design guide for torsion, As mentioned before, Omar Blodgett's Design of Welded Structures is an excellent reference.