Concrete Frame Approximation

[haynewp]

According to ACI 318 section 8.8.3, we are permitted to model a gravity loaded frame as a frame with fixed columns at the floor directly above and beneath. This is the way I see it done in all the concrete texts that I have as well.

However, my structural analysis book points out that this is not exactly correct. This is because the exterior columns will actually be in double curvature, and a 50% increase in inertia is required to compensate for this (which will of course change the moment distributions).

However, I do not see where ACI addresses any exterior column inertia modification requirements for double curvature when using this simplified model. All I can find is in section 8.6.1. And I can't find this modification being taken into account in my concrete texts either.

I was wondering what those of you who use this approximation (instead of full multi story computer model) are doing? Do you leave the column inertias as they are or modify the exterior ones to account for the double curvature when this may actually occur?

 

[JAE]

Isn't every column...both interior and exterior, in double curvature in all cases? I guess I don't see any difference between the interior and exterior? What am I missing?

 

[UcfSE]

I agree with JAE. Are they not all in double curvature?

 

[haynewp]

They are in double curvature, but the rotation of the interior joints is small, therefore the ACI fixed end model assumption is reasonable. This is the validation my book gives at the interior columns and it makes sense to me. Sorry I made sound like the interior columns were not in double curvature also.

 

[haynewp]

I made models of the 3 cases: a full frame, simplified with fixed ends at the top and bottom, and simplified with fixed ends at the top and bottom with 1.5I at the exterior columns only. The moments I get in the exterior columns and end girders are closer to the full model when using the 1.5I modification. There are pretty small moments in the interior columns using either simplified way.

But both simplified ways in my opinion are still a decent amount off from the full model at the ends of the frame. I guess most people are using computers to make full models these days anyway.

 

[rapt]

haynewp,

Why do you think that modelling the full model will give more accurate results. That does not take into account conctruction sequence, load application sequence or live load patterning (between different floors), construction loads, restraint effects and a lot more variables.

More accurate compared to what?

 

[haynewp]

Rapt,

I agree, I also brought the point up about the potential live load differences between floors to the person that brought this whole thing up to me as well. I used a comparison with the final construction to see how it turned out. I know there are tons more variables and situations you can add into this especially with construction sequence.
I said the simplified way was a decent amount off from from the full model, I never used the word "accurate" or discounted construction sequence.

I put the question out to see if anybody is doing anything different than just using the simplified model. By the way, what do you do? Do other codes allow this same simplified approach?

 

[rapt]

haynewp,

It seemed that you were assuming that the full frame answers were the correct ones and everything else had to conform. I have had a hard last week with people thinking that concrete floors designed using FE analysis/design software are really accurate compared to strip methods. They do not understand concrete design, unfortunately, and are being misled with default design approaches pushed by some of the purveyors of this type of software, especially PT design. Not understanding what they are doing in design, they are assuming the programs are designing correctly and are accepting some really unconservative designs as "more accurate" because they think something more hightech like FE can produce lighter designs. Too bad about cracking, creep, shrinkage and everything else that FE is not modelling that completely erode the supposed accuracy of FE.

All codes I deal with, ACI, AS, BS and Euro, and derivatives of these, allow the simplified frame with columns fixed at the far end.

You should also use the Equivalent Column approach to determine the column stiffness (this will reduce the column stiffness.

No matter what, for the beam/slab member design, the difference in the column moments will result in a small redistribution of the moments and shears. The total moment will still be designed for so they are safe.

It is only for the column design that these moments might be a problem if they are too small. If you look at the actual moments going into columns due to secondary effects, the vertical load effects are relatively small. If the moments are larger than the full frame values then they are conservative anyway.