When to consider I-beam torsion

[McT178]

I am working with an I-beam grid that can be approximated as a "U" that is fixed at the two ends of the U. A load is applied to the middle of the beam that would act as the bottom of the U. Is there any guidelines as to when torsion should be considered? For instance, is there a ratio of length to web height when exceeded torsion needs to be considered? I understand that torsion will exist to some degree where the beam at the bottom of the U meets the side beams, but this would be limited to amount that the bottom beam will deflect. I appreciate any advise.

 

[flash3780]

I say consider a torsion load if a torsion load exists. The one way to be sure that it's not a problem is to calculate the stresses.

 

[hokie66]

I don't understand your word picture, but I-beams are not much good in resisting torsion.

 

[Lion06]

Can you post a sketch?

I'm having a hard time following your description.... at least in any way that makes sense to me.

 

[BAretired]

In that situation, I would ignore torsion as it is not required for strength.

What you have is two cantilevered beams with a simple span framing between the two ends. Torsion could be calculated based on the simple beam slope, but it will be negligible.

BA

 

[prex]

Confirming what BAretired says, I would describe more precisely the rule to follow this way:
-consider torsion when it is due directly to the action of external loads (a so called primary effect)
-do not consider torsion when it is due to the internal constraint of the structure, in other words it is caused by the deformation of the structure under the external loads (a so called secondary effect)

prex

http://www.xcalcs.com

: Online engineering calculations

http://www.megamag.it

: Magnetic brakes and launchers for fun rides

http://www.levitans.com

: Air bearing pads

 

[McT178]

Thanks for the replies. I have attached a very simplified sketch of what I am working with. I have never seen torsion considered in case like this, but it was specifically asked for. The calculations I have are only for max torsion, which return very high numbers. I know that these values are very localized and are not a true representation. I have also calculated the max deflection at the center of the beam assuming the ends are fixed. From this I get very small rotation which should prove that torsion is negligible. Is this what you are implying BAretired? Or, as prex states, I can simply ignore because the torsion is produced by secondary forces.

 

[csd72]

jmcternan,

You should always 'consider' torsion when it exists. You may not need to put specific numbers to it but you should always take it into account even if it is just allowing 10% overcapacity to allow for its effects.

In the case that you describe the amount of torsion depends on the flexibility of the end beam, the more flexible that end beam, the more torsion you will get.

what I have done with very similar situations is to do hand calculations based on the corners being pinned (i.e. two cantilvers with an end load and one simply supported beam) and then use the resultant forces in an alnalysis to calculate final stresses and deflections. The analysis is also important to calculate weld stresses.

If you are not confident to make these sort of engineering judgements then you should calculate for all these things until such a stage that you are confident enough to discount them.

 

[SAIL3]

As BA indicated it would be negligible and in any case it is self-limiting..so no I would not normaly consider torsion in this case.
Now, maybe in an case where the loaded bm is very flexible and of a very long length and moment connected to the two cantilevered bms, then maybe I would look at it..even in this case it is self-limiting.

 

[ron9876]

I think that csd72 said it right. The only way to develop the engineering judgement to know when you can ignore torsion or any other force/stress you should calculate it and see the magnitude of the results.

It is my opinion that proper mentoring of young engineers will take them thru the trail of the various types of structures with detailed calculations for the elements.

 

[rb1957]

if the loaded member is pinned, then the torsion is "only" due to the shear reaction not being thru the SC of the other supporting members. if the loaded member continued under the supporting members then there'd probably be no torsion (at all).

if the loaded member is fixed, the fixed end moment creates torsion on the supporting member.

 

[flash3780]

I don't see any torsion load in your diagram. Roarks has an analytical solution to this problem (Table 8.2).

 

[rb1957]

flash,
what will the out-of-plane load do to the supporting members ?

is the out-of-plane load applied thru the shear center of the uspporting members ? no, but the torque created is probably roughly 1/2 of squat.

if you Really wanted to, i'd consider the torque as a couple acting along the supporting I-beam caps, so it would augment the bending streses.

 

[flash3780]

When I think of torque, I think of twisting - as in a shaft. The vertical members will be in bending (albeit statically indeterminate bending). Maybe I'm not understanding the problem correctly. I think Roarks table 8.2 has this an analytical solution for this situation.

 

[McT178]

The structure is on a horizontal plane and the load is applied vertically (sketch shows top view and side view). I have calculated the max deflection of the loaded beam which is what limits the the torsion. Using this deflection I have calculated the max angle of rotation for torsion on the side beams, and calculated the torsion from there. Is this method sound? Torsion appears to be just about 1/2 of squat as rb1957 suggests.

 

[BAretired]

The web of the loaded beam should be connected to the web of each cantilever beam using a standard shear connection, not a moment connection.

The loaded beam carries a concentrated load W at midspan. The simple span bending moment, M is WL/4 where L is the span. The end rotation is M*L/4EI = W*L^2/16EI which, in the worst case, is also the rotation of the end of each cantilever beam.

The rotation per unit of length of each cantilever beam is W*L^2/16EIC where C is the length of cantilever. Torsional stresses can be determined from that using torsional constants for the beams. They should be very small.

BA

 

[flash3780]

Try this.

 

[BAretired]

flash,

That is a different scenario. In this case, all beams are in the horizontal plane.

BA

 

[rb1957]

and the load is out-of-plane (not in-plane as the Roark examples consider

 

[rb1957]

a more interesting point to consider is that the shear reaction from the loaded beam is applied to only one flange of the supporting beams ... how quickly will it diffuse to the other flange ?