Torsion Constant 1

[bazflexure]

Hi all..
I have seen some etabs model in which torsion constant for beam and columns is set as 0.01 .. why is it so..
If you can refer any code reference or ready, I shall be greatful..

Regards,

 

[slickdeals]

This could be because interior beams are designed for compatibility torsion, and the program is not smart enough to recognize that compatibility torsion will get redistributed as positive moment. Hence, some engineers might choose to design interior beams with no torsional stiffness.

I am not sure about the columns though.

 

[bazflexure]

thanks man! one more thing..
If I am using bearing wall system and there are two or three columns also present.. so for lateral load analysis the correct way would be to take the columns in the analysis and put shear and moment modifiers to 0 rite?? or is there any other way..
I hope that i have explained the question clearly.. if not please tell me I'll repost it /./.

 

[asixth]

bazflexure

1. Like slickdeals has pointed out, by setting the torsional constant to a small value the member will not attract any torsion and the loads will be redistributed elsewhere. I use a general purpose frame package so when designing an rc slab I will generate a grillage and multiply the torsional constant that is calculated from mechanics by 10^-3, this is usually a large enough reduction for the member to redistribute any torsion that it wants to attract.

Many textbooks on concrete mechanics will say that the torsional stiffness is greatly reduced from the onset of cracking (90% reduction) so a reduced torsional constant is correct if you want a more realistic distribution of forces. There has been a good concurrent thread running about 3D modelling so that would be worth a read because some good points where raised.

2. I think you need to perform the analysis twice. Once with the shear wall taking all of the lateral forces and then a second run with the columns taking the lateral forces they will take from lateral displacement. It is a poor practice to design the columns without them contributing to the lateral force resisting system. The will not be as stiff as the shear wall but they will attract some force.

 

[bazflexure]

@asixth
Yes, I have also read that about torsional stiffness. But what about using it for edge beams.. that wont be correct?? rite?

As far as point 2 is concerned,
the reason why I am doing it so is that because the structural system is bearing wall system and applying stiffness modifiers values 0.01 to colums for shear and moment would justy R value.

Moreover, I am using the same model for designing shear wall so using columns with stiffness modifiers would give more shear in shear walls and a better approxmiation of results.

what do you say?

 

[bazflexure]

slickdeals what do u say?

 

[asixth]

I can't comment on R-values or seismic design to ASCE-7 because I am un-familiar with that code. I just don't think it is appropriate to design a column for gravity loads only when in reality it will be attracting some lateral loads when the occur.

For your second point, I don't believe in trying to model everything from one central model. I say this because if you make one design assumption for a particular condition which is conservative for what you are trying to design, that same assumption may be less conservative for another element you need to design for.

The column/bearing-wall is a good example for this. Saying that all the lateral force goes to the wall is a conservative assumption for the design of the wall but that same assumption will not be conservative when designing the column.

Another example I had recently was the design of precast prestressed beams and an assumption of Young's modulus. If a larger Young's modulus is assumed the prestress losses are less and the stresses of the section at release are greater, but if I follow this assumption through when designing for camber and use the higher Young's modulus, the camber will be less which may be detrimental to the design if I underestimate the camber.

 

[slickdeals]

Like asixth says, you probably need more than one model to design the building. I typically look at cracked columns with all shear going into my wall and a model with column stiffness so that you don't under-design your slab-column joint. Granted that when the slab column joint cracks, it won't see any more moment , but you will still be left with an ugly crack. Not to mention that the unbalanced moments at the column can affect your punching shear strength which is often the controlling parameter in flat slab design.

For spandrel beams, I have designed beams framing into the spandrel beam as pinned at the spandrel beam (provide more positive moment reinforcement) and design the spandrel beam for threshold torsion (to control torsional cracking). A spandrel beam is also a case of compatibility torsion if you have torsional moment redistribution. As asixth said, torsional stiffness degenerates rapidly with cracking.

There was a paper published by Collins regarding torsion and its redistribution back in the 80s. I would suggest you look for that paper.

 

[bazflexure]

Thank you guyz..
you are right.. I am using two models.. one for gravity and one for lateral....
I just wanted to know that is that way right or not..
asixth, even if we dont take lateral stiffness of columns into account we do provide seismic detailing for them.. I was reading NEHRP Commentry and it said that in 1994 Northridge earthquake majority of failure occured because of the fact that some columns that were throught not to be a part of lateral force resisting system did attract load and failed during an earthquake..