Transfer Design Reactions - Static Analysis or Tributary Area? 2

[OzEng80]

Hi

Apologies for reposting this here, but I failed to get any bites in the FE forum.

We have building analysis software that allows reactions to be provided from: tributary area analysis (similar result as hand calcs); or by static analysis (utilizes superstructure stiffness). My current approach to building design is to use the tributary area analysis reactions to design transfer elements and to use the worst case of static analysis and tributary area results for columns/walls. I believe this is being overly conservative.

AS3600, Ch7 seems to indicate that static analysis could be used, but the simplification listed in 7.6.2(b) ‘each level thereof together with the column as they occur above and below analysed separately’ has me a little concerned.

What is the appropriate/industry standard method for design of transfer slabs regarding load run down? What are consultants using?

Are there any requirements outside of AS3610 (propping, stripping times etc..) that should be maintained to ensure assumed behavior?

Any input is much appreciated.
Thanks

 

[csd72]

OzEng80,

I dont recall ever seeing anything in the Australian Standards but this is definately a case that needs close attention.

Any deflection of the transfer beam will alter the moment distribution of the beam/slab over.

The only way to truly take this into account is to model the structure as a whole.

 

[rapt]

1 This is never mentioned in a code as it is structural mechanics, not something requiring a design limit. The code simply says to take into account all loads on a member.

2 The problem with using a full 3B building model is that it does not take into account conctruction sequence. Redistributions of loads will ofter occur due to deflections/shortening in some supports etc. But these will be different depending on conctruction sequence. For example, if the deflection of the transfer beam under load is 40mm with a central column, the analysus will show that a lot of load will actually transfer back to the columns through frame action rather than coming down the transfer column to the transfer beam.

And the deflection of the transfer beam when the next floor above is built is basically zero while the overall building analysis will think it wwas 40mm at that time. When the last floor is built, the beam will have already deflected 40mm and this floor will experience only the deflection caused by its weight but its weight case redistributions of the loads from the lower floors. Follow this up the building and you will see that the 3D model is very unconservative and the tributary area is possibly conservative. Add stage prestressing affecting the deflected state and who knows where you are.

But have you designed the rest of the structure for this to happen. Probably not.

So I think tributary areas are best. Maybe conservative in some cases but 3D models are definitely very unconservative.

 

[OzEng80]

Thanks RAPT

That’s the direction I was heading. I am aware of the effects of support stiffness being increased due to a ‘construction sequence’ analysis and the subsequent increases in stresses for the transfer. The software we are using has a construction sequence analysis option but the actual ‘stages’ with respect to backpropping are a bit confusing to me.

Say for a multistory building with 5 carpark levels a 6th floor transfer and an additional 10 stories above.

Stage 1: level 0 – level 5. Deflections in these levels do not affect the stiffness of supports for the level above (no transfer).
Stage 2: level 6 (transfer) - The transfer itself will deflect under SW unless the backprops and the subfloors are infinitely stiff.
Stage 3: level 7 – props below transfer are likely to be removed resulting in minor deflection of transfer.
Stage 4: level 8 – slab will be built on the deflected level 7 supports assigning more load to the transfer and causing more deflection.
Stages 5-12: levels 9-16 – similar to level 8

Assume this building is constructed with 3 floors of ‘undisturbed supports’. Is it an appropriate simplification to analyse the building in stages of 3 levels (levels 7-10, 10-13, 13-16)? Or is the fact that the 3 undisturbed floors ‘climb’ the building negate this reasoning?

Also, to throw another one in there: the in-plane stiffness of the supports themselves contribute significantly to this effect. Walls in the structure above transfer can offer stiffer load paths and will resist even the ‘staged deflections’. Then there is the modeling of the stiffness of these elements themselves Ief……

Given the complicated and unpredictable nature of this load distribution, the lack of guidance in the codes and the frequency with which these principles are applied shouldn’t there be some specific guidelines for this?

So….. regarding my original question, what is the industry standard for building analysis and do standard backpropping practices support this approach? Can the results from a static analysis be adopted if full backpropping is specified? Does AS3600 CH7 ‘each level thereof together with the column as they occur above and below analysed separately’ read ‘ignore differential support stiffness’?

It is clear that a ‘tributary area’, transfer analysis is conservative but at times it seems overly so (lots of walls). Perhaps an approach such as having: a tributary area strength design and static analysis for serviceability could yield appropriate designs?

Your assistance/thoughts are very much appreciated.

I love this site.

 

[rapt]

Ozeng80,

Agreed the correct answer is probably somewhere between the 2 methods.

But if you are going to use a multistorey 3D analysis to determine it, what stiffnesses do you use. You have to allow for cracking and possibly creep and shrinkage.

Assuming all floors above the transfer are the same, the average of the 2 methods would probably be closest, as the 1st floor above the transfer will experience the full frame deflection effect, reducing gradually to basically none for the top floor. As the frame ethod assumes all floors experience it the deflection and the trib area method assumes none experience it, the average of the effect on all of the floors would be the average between the tributary area approach and the full frame approach.

If there is PT in the transfer member, the PT uplift would need to be included in the frame approach along with the averaging to get the answer anywhere near correct.

csd72
The full analysis approach is definitely giving an incorrectly low estimate. It is not modelling it in the same manner in which it is built.

 

[hokie66]

I agree with rapt. In my experience, the load rundown is done, and those column/wall loads are used to design the transfer floor. The issue of what deflection does in the floors above is why we make transfer floors very stiff. No matter how much thought is put into this type of interaction, the structure will often prove smarter than us.

 

[OzEng80]

Thanks for your responses.

RAPT, the actual (early age) stiffness of the superstructure is obviously a nightmare to determine. Assuming fully cracked walls and slabs will result in the softest superstructure with minimal redistribution, larger transfer stresses & a conservative design. One approach I have adopted (for determining preliminary reactions from an externally designed PT transfer) was to model the superstructure as cracked and the transfer as uncracked. This provided a stiffer transfer and attracted more load (loadings provided to PT designer were still tributary area).

Hokie, the load done run is undoubtedly the most conservative approach. My experience with transfers is that the short term deflection is never as large as calculated (I’m too green to assess the long term!). I would be interested to hear your experiences.

It would be satisfying to design the transfer for the actual loads that it sees (well a better approximation of them anyway). In recent jobs it really feels like I am double dipping (validating superstructure for static analysis and designing transfer for tributary area) and ending up with an overly conservative structure. Given that my transfer is inevitably governed by the span/500 LT service requirement I was looking to sharpen the pencil a bit.

So…. In summary:

The appropriate methodology for transfer design is to use tributary area reactions not static analysis.

A construction based analysis could be used with a considered review of the results from the other methods.



Are there any other methods being used? Given the amount of computer horsepower and sophisticated software in circulation I would have thought an area analysis would have been outdated….

Finally, what are your thoughts on the following approaches:

- Static analysis (fully cracked sections) with increased transfer stiffness (Ig) or 2x as stiff
- Construction staged static analysis service design, tributary area strength design?
- A combination of the two above

Thanks again!

 

[hokie66]

OzEng80,

I always think of a transfer floor as taking the place of footings. Deflection in the transfer floor has the same effect on the structure above as does footing settlement, and you don't want that to be much. Remember that transfer deflection and footing settlement are additive. I think conservatism is warranted, and it is a good thing that deflections are not as much as predicted. Some of that is due to interaction with the superstructure, but as you say, quantifying that is not readily done.

Transfer floors should ideally be avoided, but architectural considerations often dictate, especially in residential buildings.

 

[rapt]

Ozeng80,

I will extend Hokie66's comment to when you have to design a transfer member/floor, the overall cost of that member/floor is very small compared to the cost of the overall building so being 20% over capacity will have very little cost impact but will allow me to sleep a lot better considering the importance of the transfer member/floor.

Re the computer horsepower (meaningless) and the "sophisticated software", the one you are looking for does not exist yet.