torsion in concrete beams.

I've seen many application of torsion to concrete beams; my experience is that if you have it, you'll know it.

Many of the applications I've seen are rectangular beams with edge blockouts (or seats) for supporting some other structural member. The concrete beam is usually supported at the centroid of the orginal rectangle. In many cases the support for the beam varies from large diameter concrete piles to smaller H-piles. With the latter application, the rotation just screams to be addressed with the concrete piles you really have to wonder. I guess I recomend that you assess the rotation with the slab or whatever edge stiffening member you have.

I know this doesn't answer the question, just more of what my experiences have been. [sig][/sig]

Well, to get right to the point, I check torsion whenever such as situation arises. I always start with the code and work my way through it. But when faced with your situation, I always check the rotation or twist with the slab (diaphragm) just as the slab can be flexible given the right conditions, thickness, supports, span lengths etc. [sig][/sig]

Two other common incidences of torsion occur in soffit beams and in beams where you have differential live loads, such as a parking garage, even if you consider the slabs. Longer spans also induce torsion, so as Qshake says, just keep on checkin'. [sig][/sig]

Out of interest have any papers been written concerning the torsional restraint offered by slabs? [sig][/sig]

Check the ACI Structural Journal and the ASCE Structural Journal. Very likely you'll find something on the subject. [sig][/sig]

For many years I designed concrete structures in Texas (as they are quite common there) and we usually didn't focus a lot of attention on torsion. Typically, we used either beam and slab or beam, joist and slab systems. With these, we would typically specify closed stirrups for perimeter beams where the structural slab was on one side only.

ACI dictates that if you design the adjoining slab (whether it be a slab, joist, or other system) based on a SIMPLE end condition (i.e. the joists or slab don't depend upon the torsional rigidity or strength of the perimeter edge beam) then you can reduce torsional effects to a small amount (see ACI 11.6.2.2) This we did, and calculated the "typical" torsion on various size members and determined what kind of stirrups and longitudinal steel this produced. Then we would add that in to our normal gravity analysis and design when scheduling rebar in the beams. [sig][/sig]

If you are considering that the slab/diaphragm offers resistance to torsion in beams, then you should check that wether that slab can resist that much amount of additional moment (that is being transferred to it because of torsion). The safest way is to design the beam for the torsion, rather than trying to assume that slab can take tat additional loads.

To answer the other part, encountering torsion, since almost all the structures here in India are deigned in RCC, all elements more or less monolithic, we come across torsion very frequently, like projections on lintels, projections not lying as the same level as the slab, free cantlever steps from a beam, brackets in beams supportin cross beams/slabs etc. [sig][/sig]

The torsion should always be checked.

You mention calculating the twist/deflection. These parameters are typically calculated for "service" conditions, and should not be confused with "limit states". At "limit states" or ultimate strength upon which most concrete design is based the cracking, twists, deflection etc. are significant and redistribution of loads will form, possibly negating your redundancy.

Always provide redundancy in your designs. Since torsional failure is sudden and non-ductile, you should always check it and confirm adequate strength in the primary member. [sig][/sig]

It seems that there are many ideas offered and most imply that it is best to check the torsion - even if the resulting reinforcement detail becomes somewhat excessive.

The failure mode in torsion of a beam and slab system is very interesting. I would have expected the beam to rotate under torsion and the slab to act as a simple strut or tie to stop the rotation (simple pencils and ruler theory)! The effect would be to impose an axial force in the slab rather than a moment. Eg a beam spinning about a longitudinal axis could be restrained by a stap perpendicular to the axis.

At the same time a monolithic connection could introduce a moment into the slab..................

I have designed many beams for torsion but it is good to question the rationale behind the codes from time to time.

Thanks nb [sig][/sig]

Can someone explain to me the mechanism of torsion in the simplest manner? The code requires additional longitutinal reinforcements at the mid section of the beam for torsion resistance. If the addtiional reinforcecement is located at the top or the bottom of the beam, what would happen? Is it necessary to place the additional bars at the mid section even if the beam isn't "deep" say, 12" overall depth beam?

One point to be considered is, where is the torsion coming
from? If you have a slab on one side, monolithically connected,
the torsion in the beam is caused by the slab, and the beam
has to be designed for the torsion. The effect would be slightly
less for slabs of unequal spans on either side.
However, if the torsion is caused by other elements (cantilever
projections etc), I feel the restraining effect of the slab could
be taken advantage of. Note that in this case we should
consider the most conservative (least) restraining force from
the slab, i.e. say only dead load effects. A separate load condition
of slab alone fully loaded may have a more critical torsional
load on the beam!

Torsion in beams cannot be neglected.

I think I can help you with some theoretical explanation:

Torsion in beams is indeed a special situation. It does not represent a regular problem, for Polar inertia of cross sections of beams is very low- you can compare their values to the inertia of a same cross section about bending axes. The stress distribution obtained manually or by automatic calculation is proportional to the stiffness of the members and therefore you don’t normally have stress due to torsion, because it distributes to other members and in more probable effects- like bending.

In special, typical and well known cases where there is no place else for the load to go, you will get torsion stress in members even with low polar rigidity. They will, probably be the only ones to be resisting!
There are two kinds of torsion stress in beams:
- Compatibility;
- Equilibrium.
The first one can be allowed and happens when you have, for example, two beams crossing at one point (without any support at that point) and torsion comes from the need to have compatible stress at that joint. The torsion is low in this case.
If the second case occurs this means that you have a bad conception and that no other member is resisting to the loads, than the beam in torsion or that you have no choice.
Beams in torsion don’t behave well and their resistance is very low and especially unreliable.