Connections for torsion.

[3doorsdwn]

I’ve got a situation where I have torsion on an I-beam. At the end of the beams I have simple shear connections (i.e. angles and bolts). What I’m having a hard time deciding is: what to design the angles for (in terms of the load from torsion; throwing the (block) shear demands aside for a moment). Intuitively, you’d design for a bearing force (tension or compression) on the angle (and the bolts) for the couple created by the torque. BUT this type of connection is torsionally “pinned” is it not? So, theoretically, it shouldn’t see any load from the torque. (But it actuality (as we all know) there really isn’t any such thing as a connection that is perfectly “pinned” or “rigid”).

Thoughts?

 

[Lion06]

I will be interested to hear some responses to this. I have always wondered how you have a connection that is pinned against torsion. In my mind, I imagine the beam just continually twisting around since the pin (for torsion) offers no rotational restraint (which is what torsion is).

 

[UcfSE]

A pinned torsion connection differentiates whether the section is free to warp at the supports, not whether it has any twisting moment. The torsion is still max at the supports and has to go somewhere. It isn't quite analagous to the flexural "pinned" and "fixed" terms we use.

I would try sizing an angle at the top and bottom flange for the torsion moment.

 

[3doorsdwn]

UcfSE, when you say "I would try sizing an angle at the top and bottom flange for the torsion moment.", are you saying I should connect to the flanges? This is a beam to beam connection.

 

[csd72]

3doorsdwn,

virtually all the torsion resistance of an I beam is in the flanges, the web plays very little part. So any connection that does not connect rigidly to the flanges is not a torsion connection.

If you are saying that this is connected to a perpendicular beam then this is a difficult detail. You can either detail it so that there is negligible torsion transferred into the supporting beam, or you can brace the supporting beam (with fly braces e.t.c.)to take any torsion.

A flexible end plate is sometimes used that can transfer torsion (in plane) but is too flexible to transfer out of plane loads.

I would suggest you read the AISC design guide on this as this really is a complex topic.

csd

 

[3doorsdwn]

Actually, I have the AISC design guide on this topic and (like most publications) doesn't specifically address this topic (accept to address what constitutes torsionally fixed and trosionally pinned connections (i.e. on p.9 & 107)). I have not addressed tensional connections (in the past) without some type of restraint for the flange (and it appears that that may be necessary for this case; perhaps (since it’s beam to beam) a full pen weld on each flange may be required).

 

[3doorsdwn]

Repost, too many misspellings above:

Actually, I have the AISC design guide on this topic and (like most publications) doesn't specifically address this topic (accept to address what constitutes torsionally fixed and torsionally pinned connections (i.e. on p.9 & 107)). I have not addressed torsion connections (in the past) without some type of restraint for the flange (and it appears that that may be necessary for this case; perhaps (since it’s beam to beam) a full pen weld on each flange may be required).

 

[JedClampett]

A lifetime ago, when I worked on Nuclear Power Plant design, we did a lot of study and developed criteria and details for torsional connections.
If the connection is fully fixed, it's easy. If the member is OK, the connection is OK. But for a torsionally pinned connection, you can either check the angles or shear plates for the torsional shear forces or design a special connection. You'll find out pretty quickly that the angles or shear plates have to be very thick to carry any significant torsion. So we would put plates top and bottom with shear tabs on each side of the flange to carry the torsion. The top and bottom plates (only welded to the main member) can develop the couple. Not too expensive, but not free either.
You didn't ask, but be sure to get your AISC Torsional Design Guide to analyze the warping stresses on the member.
Once you design some of these, you'll see the advantage of avoiding the torsion all together with kickers or some either method.

 

[3doorsdwn]

jed, I have Enercalc for the analysis. When you say you designed for the "torsional shear forces" for simple shear connections (i.e. bolted angles) are you referring to resolving the torque into a couple (with the centroid at the center of the connection)?

 

[DaveAtkins]

I agree with JedClampett. And I think this type of connection can resist the torsion from the SUPPORTED beam, without putting torsion into the SUPPORTING girder (due to the vertical end reaction of the SUPPORTED beam). The girder will twist a little bit, but not enough to worry about.

I hope I am not muddying the waters.

DaveAtkins

 

[UcfSE]

When I mentioned angles, I was thinking beam to column connection. I suppose I would use plates for beam to beam.

Note that torsion doesn't transfer from the twisting beam into it's support as torsion, usually. Twisting in the beam usually becomes flexure in the supporting beam or column when it's orthogonal to the twisting beam. You get a concentrated moment into your support.

 

[JedClampett]

I'm away from my Enercalc machine at work. I don't think it analyzes warping stresses, however.
If you can treat the two angles as pure shear members acting as a couple and keep the shear through them under the allowable, you'll be OK. But the moment arm is so small, that's going to be tough.

 

[JedClampett]

I was wrong. Enercalc does check torsional warping stresses.

 

[DaveAtkins]

To clarify my earlier post, the torsion reaction at the end of the supported beam will induce a concentrated bending moment into the supporting girder, as UcfSE said. But if you connect the flanges of the supported beam to the supporting girder, the supported beam will twist the supporting girder, because the end of the supported beam is a pinned end (for vertical loads). It is this twist that can be ignored.

DaveAtkins

 

[3doorsdwn]

by Jed:

"I'm away from my Enercalc machine at work. I don't think it analyzes warping stresses, however.
If you can treat the two angles as pure shear members acting as a couple and keep the shear through them under the allowable, you'll be OK. But the moment arm is so small, that's going to be tough."
--------------------
My thinking was somewhat similar but I was planning on treating the end force like a couple that would put forces on the angle as if it was a seated connection. It's a 2-bolt connection so the "d" of the twist would be about 3". Since 4 bolts connect the sub beam to the beam it is framing into (the torque is on the sub) that gives me 4 bolts to resist the T/C couple. With that accounted for (and the force on the angles) I should be covered.

 

[SE008]

open section under torsion, check

http://www.webcivil.com/thintorsion.aspx

 

[JedClampett]

If you need the connection to carry the torsion, the bolts will not be your weak link. Your analysis is sound once you get to the bolts. The connection of the beam to the angles is usually torsionally good. But in between there is a section that consists of two plates (the angle legs) separated by the web. The torsion there is carried by a couple with arm length of tw + angle thickness. Since that distance is usually about 5/8 inch, the forces are pretty large on each side.
As I said before, in Nuclear Plant design when we tracked all the forces through their various paths, we learned a lot about the connection behavior. And that torsion was a bear to design for.

 

[MikeHalloran]

Have you evaluated the deflection at the end of whatever arm applies the torque, due to the beam's twist?

I ask because I've seen crane beam assemblies writhe under their own weight while being hoisted, and I'd have to describe the torsional and lateral deflections as 'scary', at least to a civilian.








Mike Halloran
Pembroke Pines, FL, USA

 

[3doorsdwn]

Jed: "If you need the connection to carry the torsion, the bolts will not be your weak link. Your analysis is sound once you get to the bolts. The connection of the beam to the angles is usually torsionally good. But in between there is a section that consists of two plates (the angle legs) separated by the web. The torsion there is carried by a couple with arm length of tw + angle thickness. Since that distance is usually about 5/8 inch, the forces are pretty large on each side."
---------------------------------
But that is the part I'm unsure of: Are you *sure* that is how the transfer would take place (i.e. with the T/C couple happening vertically)? It may seem intuitive that is how it would occur (considering the shear flow in the web is vertical). But you would think that the beam's rotation would carry the couple directly into the angles (i.e. making them act as a seat as I said above, thus making the force on the bolts horizontal).

 

[3doorsdwn]

Mike: "Have you evaluated the deflection at the end of whatever arm applies the torque, due to the beam's twist?"
------------------
Yes, rotation and deflections are minimal/tolerable.