Torsion on a RC Beam

Hi

1. Is it the correct way to release moment Mx if I do not want to design the beam to take torsion?

Though it is unlikely that a RC beam is allowed to rotate about its own axis or cannot resist torsional moments but if you want to simulate the same in analysis, releasing MX is the correct method.

On the other hand, if your intention is to ignore torsional moment while designing, TORSION parameter exist in some design code for the same purpose. In Concrete Design Mode (RC Designer), you will find a check box in Design Briefs to design for torsion or not.

2. Will the release of Mx cause the moment to be transfer to the support?

Releasing MX would mean that the member is free to rotate about its own axis and this will definitely affect member forces which are calculated on the basis of displacement results. In short, member forces are going to be redistributed. 

Dear Sir/Madam,
I would like to add a few points that are, in my opinion, very important in this case...
1. There are two types of Torsion - Equilibrium Torsion and Compatibility Torsion. Equilibrium torsion is present inherently in special cases like cantilever resting on beam and beams curved in plan where the release of torsion would result in instability. Compatibility torsion occurs due to the ability of a beam to take a basic amount of torsion. Release of torsion in this case would not result in instability.
2. Most codes allow compatibility torsion to be neglected with certain provisions. Be sure to understand the codal provisions while designing. For example, Indian code IS 456 allows torsion to be neglected in design if it is neglected in analysis without instability issue s (ie, for compatibility torsion). Here, the important thing to notice is that before you neglect torsion in design, you should analyse the structure without torsion (using Start MX release) and ensure there is no instability.
3. In some cases, such as a floor grid with non-orthogonal beams, using Start MX can give warnings of instability. Usually in such cases, people use START MPX 0.99 to minimize torsion.
4. When we use START MPX 0.99, even though the torsion is nominal, it can significantly alter reinforcement depending on codal provisions. In such cases, if we check the analysis results and confirm that the torsion is indeed nominal, we can use TORSION parameter to neglect torsion in design.
5. If we use TORSION parameter in design without doing the torsion release first, it is not correct. For example, let us say a secondary beam of 3 m length and having load of 8 kN/m is sitting on primary beams on either side. The mid span moment will be 3 x 3 x 8 / 8 = 9 kN-m theoretically. This is what you would need to design the beam for when you neglect torsion effect. However, if you run the analysis without releasing torsion, you will probably have a small hogging moment at the start and end of the secondary beam such as 1 kN-m and the mid span moment would be less, say 7 kN-m. Now, in such a case, if you design by neglecting torsion, you would essentially design the secondary beam for the reduced bending moment 7 kN-m while the hogging moment which would have become torsion on the primary member would be neglected. Hence, such a structure design is not safe.

Hope this is helpful.
Arunkumar