How do you model PT slab in your full structure model? 1

[Dstru]

As far as I know, most engineers will simply model PT transfer slab as plate in FEA package for full structure and only add tendons etc for slab design in isolated floor model.
My question is, how do you take the reduced deflection by PT into consideration in the full structure model? the reduced deflection will affect the load distribution and affect the resultant forces in elements including walls & columns. I realize that if I delete the PT slab the wall design is quite normal but not so if PT is included but simply modelled as plate without given consideration to the PT effect. The abnormal wall design results (such as lager bending moment 'dragged' by lower PT floor) are due to the deflection of slab I guess but with PT the deflection should be lesser and thus the wall design should be normal and near the one with PT floor deleted in my full model.
So what do you do? do you increase the stiffness of the PT slab or is there properties that you would revise?

 

[MrHershey]

We typically will include an extra 25% stiffness or so for post-tensioned slabs in our service analysis. That's on top of what ACI recommends and based on some language in ACI's SP-240, specifically the portion by Horvilleur et al. So if ACI stiffness is 0.35 for a slab at service level, we'd use more like 0.45 or even 0.5. For strength analysis we tend to not rely on the slabs too much and actually use less stiffness than ACI indicates for two way slabs to try and limit our reliance of any slab coupling we may get and force the load into our primary LFRS. We'll still include slabs as shells, we won't do a membrane and completely neglect flexural stiffness. But we'll turn the stiffness down to like 0.05-0.15 Ig, low enough that it's still present but likely to work with just the minimal code-mandated bonded reinforcement or prestressing.

We mainly do this in wind-governed and low to moderate seismic applications though, where being less stiff is conservative both for performance and load generation (more flexible building = higher gust factor = higher load). We'd need to be more careful and look at capturing true stiffness a little bit better for high seismic applications where lowering slab stiffness may actually lower our demand.

 

[captain_slow]

@OP,

This question has been on my mind lately and here are some thoughts. I see that @MrHershey has weighed in with some code-related feedback regarding lateral loads. I will chip in and give some opinion on the state of vertical loads...

- addition of PT tendons does not really alter stiffness of concrete floor plates too much; regular RC concrete is usually modelled at 0.6-0.7 Ig and when you add PT you typically work with 1 Ig for evaluation of the floor plate itself. Tendons add uplift and change our assumptions about the concrete section (cracked vs untracked) but they do not directly alter flexural stiffness of slabs.​

- this ultimately affects distribution of vertical load into vertical support elements; I have a feeling that the end effect of adding PT does not have a massive influence on distribution of vertical loads, however I have no data to back this up. Next time I do a concrete structure I may go in and compare PT vs non-PT reactions on walls and columns.​

- all of this becomes tricky when lateral loads are being considered; there is a wealth of opinion out there and there is some code guidance that may be relevant to your situation. @MrHershey's response is probably more useful than my own in this respect. My opinion on this matter is that lateral models should contain as many structural elements as possible and floor plates should be modelled with 1.0 Ig (if they are PT) - this produces the best possible approximation of the real-life structure and shows you where the lateral load goes. I know that people also like to supplement this with an additional (and separate) building model where only lateral resisting structural elements are included so as to have an upper bound approach assuming that all of the lateral load is dealt with by the primary system; any such model would exclude columns, non-lateral walls and etc... .​

At the end of the day the real question is whether you are happy with the degree of rigour when taking reactions from the overall building model. Reactions from a developed PT floorplate model are more rigorous and thus also are more accurate (for vertical loads) and so nobody will be able to fault you if you choose to use those. I will however say here that people do use reactions taken directly from building models and it is not an uncommon practice.

 

[Retrograde]

I have a feeling that the end effect of adding PT does not have a massive influence on distribution of vertical loads

I would agree with this statement. I think the difference would be negligible.

 

[r13]

Just wondering if you should treat the PT slab as weigh-less plate for the static load analysis (with dead load applied at end supports), and provide lateral effects only for seismic analysis.