Torsion On Tube Girder

[jmggks]

Folks,

I'm a long time reader, first time poster looking for help with torsion. I have a HSS girder that connects 2 HSS columns in an industrial structure. The columns continue up past the girder. The girder supports WF beams on 1 side of it. There is grating clipped to the WF beams (no deck or diaphragm action). I was planning on shear tab connections for the WF beams to the girder.

In my analysis program, I put rigid end offsets on the ends of the WF beams equal to 1/2 half the tube girder width. This results in a torque at the ends of the tube girder of 2600 ft. lb. We want a bolt up connection between the girder and the columns, and we also want the ability to handle a small amount of variation in the spacing between the columns, so I was thinking of putting shear tabs on the columns, Ts on the ends of the girder, and using short slots in 1 of the column tabs to handle the variation in length.

The problem is that the torque is way too much for a shear tab unless the tab is very thick, and making it very thick violates the assumption of a simple shear connection between the girder and the column. My working plan is to use the shear tab on 1 end of the girder and do something like a shear endplate on the other end of the girder to handle the torque. I'm not thrilled with this plan, but it is where I am at.

My questions are (1) do people agree that torque needs to be considered here (note that if the girder was an I beam, I would have not have included any eccentricity in the beam to girder connections, and there would be no torque at the end of the girder) (2) assuming torque does need to be considered, are there better ideas for how to connect this girder to the columns in a way that can transmit torque but allow some axial adjustment (3) if I use my assumption of the shear tab one 1 end and the endplate on the other, is it appropriate to think that the endplate will take all of the torque because it is torsionally stiffer than the shear tab?

Thanks for your comments and suggestions!

 

[KootK]

1) I would try to find a way to convert torque in your girder into bending in your wide flange beams.

2) If you provide some sizes, I'll take a swing at it.

3) Depends. The non-end plate end will rotate however much it rotates. If that end doesn't permit that rotation relatively freely, it will pick up some torque.



I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[KootK]

Curious: what is driving the use of HSS here? Could you just use a wide flange girder?

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[jmggks]

This is for a plating operation and they prefer HSS beams for cleanliness (fewer ledges to collect dirt). I know that is contradicted by the WF beams connected to the girder. I would rather make it all WF but the client wants the tube girders.

I'm not sure that I understand the suggestion to convert torque in the girder to bending in the wide flanges. The torque is originating from the wide flange end reactions acting eccentrically from the tube centerline. Can you elaborate on this idea?

Thanks!

 

[KootK]

Yeah, clients just want what they want sometimes.

When your girder attempts to rotate under the applied torsion, there will be a couple of load paths available for resisting that torsion:

1) Torsional restraint at one or both girder ends. This is the mechanism that you've been describing.

2) If the girder ends can be envisioned as free to rotate torsionally, the torque can be resisted by end moments in the supported beams. This is the crux of my recommendation above.

If you attached your wide flange beams with shear tabs, single angles, or double angles welded to the HSS and bolted to the wide flanges w/o slotted holes, I feel that you would satisfy the requirements of path #2. It's tough to make a recommendation without a good sense for the member proportions however.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[KootK]

Here's a sketch illustrating the concept for relying on moment transfer to the W-beams. I've also thrown in a detailing option if you want to stick with your original concept. It would be sensitive to the member proportions, however, as it involves an angle leg in torsion which could get a little dicey from a stiffness standpoint.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[KootK]

Sketch.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[jmggks]

I think I understand. I am analyzing the shear tabs on the sides of the girder as pins, but in reality they have some moment capacity. This moment capacity is insignificant in terms of beam bending, but it is not insignificant when we are thinking about the torsion capacity the shear tabs at the ends of the girder. When load is applied to the beams, it wants to rotate the girder. This can be resisted either in the girder to column connections or in the beam to girder connections. We don't really know which is stiffer, but if one of these is strong enough, then ultimately it should work.

Here are some specifics. The beams are W8x24 with end reactions of about 5k each. There are 4 beams connected to the girder. The girder is HSS8x6x3/8". The girder spans 140" and supports 4 beams. 4 beams x 5k x 0.25 ft. eccentricity = 5000 ft. lb. to be resisted. When I analyzed the girder as torsionally pinned both ends, I got 2500 ft. lb. torque at each end of the girder.

I planned on the shear tabs on the sides of the girder being 6" tall x 3/8" thick w/ (2) 3/4" A325 bolts on 3" centers. Bolt shear is the lowest limit state, so this connection is good for 21.2k shear (much more than required). So if this connection is strong enough for 5k shear and 1250 ft. lb. moment, then I can use connections of the ends of the girder that are designed for the girder shear reactions and not worry about the torsion? If so, that makes the details much more simple than what I was thinking I had to do.

Please let me know if I am understanding this correctly. Thanks!

 

[KootK]

Nailed it. KootK approved.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[WARose]

If the column to girder connection is HSS to HSS (if understand this correctly)......why not make it welded? It will transfer the torque you have then.

 

[jmggks]

Two things are leading us toward bolting. First, we are not confident in the spacing of the columns, so if we can allow a small amount of adjustment in the girder end connections that helps a lot. Secondly, this is a plating operation and the fumes eat steel. The steel is shop powder coated for better corrosion resistance. Touch up painting on field welds does not match the powder coat (looks bad) and does not hold up to the environment nearly as well.

 

[BAretired]

It makes much more sense to take the torsion out of the HSS girder by designing the connection at each supported beam for the moment 5,000*3 = 15,000"# and considering the girder/column connection as torsionally hinged.

BA

 

[SAIL3]

HSS members require end pls internally or externally or an external equivalent to develop their torsional capacity....

 

[KootK]

HSS members require end pls internally or externally or an external equivalent to develop their torsional capacity....

Interesting. I see where you're coming from but I'm not sure that this is strictly true. In my detail above, for example, the St. Venant shear in the tube would get converted into a couple comprised of two vertical plates in strong axis bending at the connection. There would be some localized distortion of course. It's also worth remembering that the loads that create the torsion in the first place are almost never delivered to the resisting member via all four tube faces either.



I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[SAIL3]

without this tube-type diaphragm one can get excessive distortion(racking) of the HSS member.....

 

[KootK]

Agreed. In a perfect world, there would be a through plate at all locations where torque is applied or resisted.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[jmggks]

KootK, I think I am going to revise the connection to be like your sketch. When I look at the shear tabs on the WF to tube connections, I cannot treat these as slip critical connections because of the powder coating on the steel, and we don't want to mask and field touch up. With standard holes in the beam webs and shear tabs and only 2 bolts on 3" centers, the connection can rotate about 2.4 degrees before the bolts are bearing on both the web and the tab. I then look at the single plate that I was going to put at the girder to column connection. The tube girder is very stiff in torsion, so I am neglecting any twist along the length of the tube. If I try to get 2.4 degrees rotation out of the single plate, it would have a shear stress of about 135 ksi. So even though the WF to girder connections are strong enough, I am not confident that they would take any load until the single plates at the end are already overloaded.

Your design gives me just what I need - bolt up, ability to incorporate some short slots, and the ability to handle a small amount of torsion. Thanks!

Regarding SAIL3's comments, I would just note that the 8x6x3/8" tube has a torsion capacity of 496 ft. kips and I only need 2.5 ft. kips, so even though the connection that KookK sketched may not be able to develop the full torsional capacity of the tube due to distortion, it should work in my case.