when do we need to design for torsion in a structural member

huangyhg

when do we need to design for torsion in a structural member
i've got a question about when to design torsion in structural members, say steel, but maybe reinforced concrete as well.  a colleague of mine was designing structural steel members for bending and axial loads, but not for torsion and was put on the "hot seat" when his calcs were checked by the supervisor because torsion was not allowed for.
my question is under what circumstances is it compulsory to consider torsion? is it when the load is applied away from the shear center only? like loads on a cantilevered balcony or some other eccentric load?  when the member on which the load is being applied is unsymmetric?
are they other conditions where we have to considet torsion?is any particular shape (w shapes vs square tubes, etc.) more susceptible to torsion?
when we do encounter torsion, what is the best design approach to ensuring that torsion has been designed for and will not be a problem?  what do we have to check for, specifically?
thanks in advance.
i had a professor in grad school that said the best way to design for torsion is to eliminate it! i tend to agree!
check for torsion when:
1.  you have a cladding being attached back to the underside (or top) of a beam that will exert a horizontal load that is not through the centroid of the section.
2.  you have a vertical load that is being cantilevered from a beam and does not load the beam through its web.  
i am sure there are other blatantly obvious situations that i am missing and others here will mention.  
a wf (or any open section) performs much worse than an hss (closed section does) as far as torsion is concerned.
i agree with jike, the best way is to eliminate the problem.
torsion is never a good thing.
there may be unavoidable situations when torsion cannot be fully eliminated. for an example, built in perimeter beams, etc.
the design for torsion of the concrete beams is covered under chapter-11 of aci-318 building code.
no question there are many situations in commercial construction where torsion cannot be eliminated, for instance where hanging brick. you can minimize this torsion by kicking back to the structure at frequent intervals.
re: aisc steel design guide 9, torsional analysis of structural steel members, 1997.
i-beam shear connections resist little beam torsion. you must use other means to transfer the i-beam torsion.  there are many ways to do this.  you can reduce or eliminate i-beam torsion by bracing the flanges.  if you can not use bracing, then you must weld the flanges solid at it ends and design the i-beam with torsion included.  stiffeners will not help you with i-beam torsion.  however, they are good for local flange torsion though.
well, there are two issues with this.
one - serviceability - if you are concerned about the torsional distortion of a   
eliminate it, or check it always.  
what do you think the answer is to your own question?  how much research have you done?  
look for a text by salmon and johnson.  their text is commonly used as graduate class text and is very helpful.  since this is a basic question and you seem to have little understanding of it, your best bet is to study it thoroughly yourself.  in the end that will help you the most.