Hi, I'm an italian structural engineer.
A question about PT two way slab: is it possible to design reinforcement layout with ordinary reinforcement and unbonded tendons in "column strip" for each direction and just ordinary reinforcement in "middle strip"?
(banded tendons in two way direction just in column strips)
Thank you.  

If you are following the ACI code, then it is not permitted. You are required to have cables placed no further than 5 ft or 8 * Slab thickness. This is in section 18.12.4 of ACI 318-02. This forces you to do a banded-distributed pattern.

Thank you for reply.
About structural research or technical paper, is it possible to design that layout about unbonded tendons?

I think you could design it that way but as slickdeals said, the ACI code apparently doesn't allow it (not sure about the Italian code).  It would be interesting to know why ACI requires the banded approach.

By the way - I was in Italy two summers ago.  What a great country!



 

Thank you for reply.

There is no italian code about two way slab. IMHO it is important to ask about it where there is knowledge.

I think it is possible to band tendonds just inside the column strips, because you could provide satisfactory sectional behaviour.
But I need knowledge and experience, so I ask to forum.

A practical reason for bands/uniforms is placement. If bands in both directions there will be a conflict at columns.

I agree that the banded in one direct with unbanded in the other is a practical way of avoiding the congestion you would have with banded in both directions. the section typically are so small, any loss in cable height leads to a big change in vertical component of the cable force.

But I can't think of a structural reason why you couldn't have banded in both directions  

History

The banded/distributed system was basically created for unbonded sustems, because the column/middle strip system requires weaving of tendons. This is reasonably easy with bionded PT as the tendons are not normally pre-manufactured full length, they are made up on sode out of 6m lengths of duct. It is very hard to do with full length unbonded tendons. So the USA with unbonded tendons went this way and because of this they cannot do partial prestressed slabs.

As an aside, the ACI code is stupid on this as it was written by unbonded people for unbonded industries. Also, the PT software available from the US at the time was written to handle only single strips and the developers of those programs had a big influence on the direction the USA PT industry went and on the ACI design approach. Because of this it limits people from doing bonded patially prestressed slabs and achieving a better result in many cases.

In Australia and Asia with bonded tendons, column/middle strip arrangements are the norm as is partial prestressing for flat slabs.

Lisia
There is no reason why you cannot do it. It was done quite often in Netherlands in the 1970/80's (called support strip stresing I think).
As long as you put column strips of tendons in both directions, some top reinfroceemnt over the columns and  and check for strength, crack control and cracked deflections properly, it will work properly as a partially prestressed slab for the column strips and a reinforced slab for the middle strips. Eurocode does not preclude this approach.
 

Rapt has given a good synopsis of the situation of unbonded vs. bonded/partially prestressed.  As a US trained engineer working in Australia, I agree that the ACI supported method is "stupid", and would encourage US consultants to do something about it.

Thank you for reply. I agree with Rapt.
Is there a link to technical paper about this "partially prestressed" two way slab?
 

Lisia,
The second edition of TR43 by the British Concrete Society covers it reasonably well, though the explanation in some areas leaves things up to engineering logic and a little reading between the lines as it covers both methods. It is not available online. You would need to purchase a copy.

Otherwise, there is nothing different to it than RC flat slab design to get the moments for column and middle strips in each direction and then designing each strip in each direction as a partially prestressed member, same as you would for an RC slab design.

Hohie66,
I have been suggesting that for 20 odd years. I am sure you know the response someone from Australia crtiicising a US methodology would get from a US expert or even a US design engineer! That is part of the reason that most US engineers who have stayed at home have missed out on the pleasure of using RAPT software for RC and PT design!

@rapt,
Do you really have a personal flame war with Bijan Alami? Just kidding.

Although I practice in the US, I see the merits of using bonded cables, although I am yet to use one in my practice.

Slickdeals,

Why would you think that? There was more than 1 PT software developer in the US in the 1980/90's. I would not limit the cause of the problems to Bijan, though he is a major contributor, but at least one other was/is just as bad or possibly worse in my opinion and he is in an influential position at the moment.

The benifits of bonded cables are many but it is also the design logic that needs to be modified and this is the big problem. You cannot go to partial prestressing until you go away from banded/distributed tendons. It is not unbonded that is the basic problem, though yoiu will never do column/middle strip tendons with unbonded and you need to to do partial prestress.
About 10 years ago VSL (now basically deceased in US building practice) tried to introduce a bonded slab system but, for some reason partly to do with PTI requirements and partly beacuse they did not know better, it was extremely expensive to use (far more expensive than the systems used in the rest of the world) and died a very quick death.

 

Rapt
Help me out here.  Why can't you use unbonded prestress for partially prestressed slabs.  And why the need to do away with banded and unbanded?  

I would see the design procedure as follows:
1. Determine a practical layout for the unbonded tendons (the banded will then produce high upward loads at concentrated areas while the unbanded will give a spread load)
2. Calculate the nett forces from this tendon layout   
3. Model the slab plate with dead, live and upward prestress loads  
4. design the bonded normal reinforcing for the nett remaining moments?    

Thank you for reply.
It's very interesting to read your opinion.

Is there a link to an example (builded floor) about partially prestressed?

I tried by google, but I need key words.

Lisia,

There have been thousands of buildings designed this way in Australia over the last 40 years. Look at the prestress company websites for examples or at www.ptia.org.au there are links to the PT companies.

Also, the UK PT companies and many of the Singapore ones have been doing it for years.

In Italy, Arup in Milan have done some.

wildehond,

You can design partial prestress with unbonded tendons though it is not as efficient as bonded tendons. The problem is you cannot partial prestress flat slabs if you are not dividing the slab up into column/middle strip in each direction and allowing for the moment and stress concentrations where they occur.

The design method used to ACI/PTI bases the design on an average moment across the whole width of the slab. This is not real. There are moment and stress concentrations. You can justify ultimate strength the US way as long as you provide a load path with the reinforcement and the banded/distributed tendon pattern does this (that is why distributed tendons in both directions does not work, no load path). basically tyou get a one way yield line result. However, you cannot make th=ese assumptions at service. You have to look at the actual stresses and concentrations where they occur. You cannot take an average. This is where ACI318 is completely wrong (see hokie66's comment earlier).

You can even do a aprtial prestress design using banded/distributed tendons, but it is not efficient as the areas with no tendons have to be reinforced for the stresses in those areas, so you are doubling up on reinforceemntas the ultimate strength reinforcement is not in the right place compared to the moments/stresses so you need to put more reinforcement in.

The most efficient system is to lay out your tendons and reinforcement in the same pattern as the moments (as you do for RC flat slab design). This results in column/middle strip layout of tendons in each direction and extra reinforcement in the same pattern. Unfortunately, this is not easy to do with unbonded tendons. That is part of the reason why the PTI/ACI went the way they did. But unfortunately, that is not the way the slab wants to work so you pay a penalty.
Even for the current ACI design method, the service stress design is wrong and unconservative. Because the stress is not averaged over the full width, the real stresses are actually higher than those being calculated and the stress limit to decide on cracked/uncracked (a bad term but I will use it as ACI does) should be significantly less, especially at the support but also for positive moments.  

Rapt

Thanks for your reply.  About 10 yrs ago I contributed to a course on PT run by a local polytech here in South Africa.  In reading around the subject, I was astounded at the number of texts that were flawed in that they were prescribing calculations that made assumptions of section properties as if the sections were uncracked.  This is all "wishfull" in partially prestressed slabs where one acknowledges that the slab will crack.

Is the flaw in the method I suggest the ultimate economy?  Because if the moments and shears (and compression) in the slab are all going to be calculated at ultimate limit, then the PT load multiplier is up for discussion.  I've been conservative and used 0.9. => 1.2DL+1.6LL+0.9PT   

But I've got around the column/middle strip problem by modelling the slab (with the PT loads all correctly positioned upward and downward to suite the tendon profile) using a Finite Element Slab programme.  Is this a structurally flawed method, or just conservative?  

I'll defer to rapt in answering your questions, but the PT should not be considered a load, it is part of the resistance.

wildehond,

The easiest example is alwaays single span, no end restraints uncracked. So that is what they put in books.

You can model the PT as equivalent loads to try to work out the Mp. Then subtract P.e from this to get the Secondary prestress moments so Msec = Mp - P.e.

In your ultimate load case, the combination would then be
1.2DL + 1.6LL + 1.0 Msec.
 I hope you were not putting in Mp for this because as Hokie66 says, the main prestress moment is an internal action, not an external load. The only effect of the prestress that is considered to be an external load is the secondary moments.

Some designers agree that this should have a .9 factor (RAPT software allows this but it defaiults to the code value) but most codes use 1.0. There is a theory that it is actually zero at ultimate strength as after the plastic hinges form, there should be no secondary moments as there is no continuity!

As long as you consider the stress concentrations from the applied loads, then FEM is ok. It is not concervative, it is correct.

The ACI code method is unconservative. Many designers using FEM for PT design still take averages over the full slab width. This is useless. FEM is predicting the concentrations and the designer is ignoring them after going to the trouble of calculating them. Also, if the concentrations are ignored some funny things can start to happen, e.g.
1   when people average the moments in irregular buildings, often they are averaging across a zone that has both negative and positive moments. Theoretically, if the negatives equal the positives, the result is zero! Not very sensible as both moments exist.
2   If you are designing for a one way failure mechanism, as the US method does, you cannot analyse for it on FEM as it is considering 2way action. The torsions and effects of stiff elements from different areas of the slab affect the moments in the area you are designing and can lead to stupid results and gross underdesign (similer to 1 above).
3    You cannot average effects across a change in section (eg drop panels)
 

Rapt
Thanks for your detailed answer.  But I'm unsure that I'm getting the my point accross.  
If you consider the simplest of harped tendon profiles, then each change of angle produces a vertical load, some up some down.  Across the length of the profile these forces total zero.  So there is no nett vertical load on the structure.  But if you model the positions of all these loads at the positions of the changes of angle (say in continuous beam analysis software), then the resulting moments and shears will be the values that must be reinforced for.  In continuous spans, this will even account for the secondary moments.

The only effect that hasn't been modelled in the above is compression in the concrete.  This compression in slabs must in any case be treated with some judgement in the end spans, but it is simple to calculate the moment capacity of an even compression load and then adjust the moments from the software results before choosing the bonded reinforcing.

If you accept the above logic as correct, then it is a short step from there to use FEM software to analyze a plate with multiple spans in both directions

 

RAPT,

So doing a column-middle strip configuration is too hard for unbonded tendons? -I was never aware of that. Is it more of a construction issue rather than design?

I guess i'm asking because last year we had a PT slab contracted out to a PT company. And it was designed by one of the authors of TR43 himself. He actually used unbonded tendons with a column-middle strip arrangement.

Do you think he wasn't actually aware, or maybe he found a way around the problem?

Thanks...

wildehond,

I understand you perfectly.

You are not understanding me. Yes, you can model the effects of the PT as applied forces to determine a PT moment/shear diagram. This produces MP. This can then be represented as 2 components at any cross-section, P*e (e = eccentricity of P at that cross-section) and Msec = Mp - P*e.

But in design, we calculate a section capacity based on the steel in the section and the concrete. The capacity of the cross-section includes the P*e component of the prestress effect on the concrete.

In design we have an applied moment M' and a concrete capacity Mu. The P*e component of the prestress effect is placed in the Mu side of the equation as it is part of the ultimate strength calculation, increasing the strength. It cannot also be used on the applied moment side reducing the applied moment, so the ap[plied moment is calculated as

M' = 1.2DL + 1.5LL + 1.0 Msec

Mu must then be greater than M'.

As hokie66 said, the prestress effects are treated as an interanl action, not an applied load. They cannot be treated as both.

What you are suggesting by ntreating the PT as a series of applied loads is simply a way of calculating the prestress effects on a slab.

You then have to understand how to use them. If you have been including it on both sides then you are grossly under-designing.

iyota,

Yes, it is a construction issue. It is possible but it is difficult. Unbonded tendons are made up full length before being delivered to site. So for a 50m square building, they would be 50m long.

For a column/middle strip layout, the tendons do not sit in specific layers, the tendons in the 2 directions interweave
throughout the slab to achieve their profiles. This is very hard to do with 50m long tendons.

All members of the TR43 committee understand this (I am one) as far as I know. One of the members is an unbonded proponent so I can guess who it was, I just cannot believe that he has done a proper column/middle strip layout with unbonded tendons. Did he actually have column and middle strip layout of tendons in both directions? Or was it banded/distributed with some extra intermediate tendons between the bands?

Rapt, Thanks for your patient reply.  No, I wasn't suggesting that you could use the PT on "both sides of the equation".  But that by using it on the load side, it provided a simpler way of solving 2 way spanning slabs with odd support conditions.  And Mu is simply provided with bonded "normal" reinforcing        

Just out of curiosity:
I have never used the effects of the vertical component in shear strength Vp, as permitted under ACI 318.

Do you typically use the vertical component in shear calculations, at least on deep transfer beams? If so, what is the effect of the reversing of curvature on continuous PT beams? How far away from the face of the support are you allowed to account for Vp?
 

Rapt,

   I'm pretty sure he used column-middle strip arrangement. It was a building in London we recently finished and the column spacing was 7.5x7.5m square bays. That's why I'm sure  if I saw a banded-distributed arrangement, I would have commented on it.

   Also, his firm advocates unbonded tendons because they can apparently zig-zag around obstructions better than bonded tendons. I agree in theory, but now that you mentioned a fixed length on-site, is that even possible??

   However,, his firm maybe also fabricates its own tendons and does the layout on-site, just like a number of others in the UK.  I'm not sure about that...

 

iyota,
Yes, the only benefit of unbonded tendons in my opinion is that they can move horizontally relatively easily.

But there are a lot of negatives, many oif which are more important to me, so I would never touch the stuff (I did many years ago when I was young and stupid).

If he is a PT supplier then he is not the person I was thinking of. The person I was thinking of runs a design business and does design for serveral of the suppliers.

Wildehond,

OK, it is just that several people previously have basically suggested the possibility.

That would be conservative as there is more capacity available from the tendons than you would get from the longterm tendon forces only, even with unbonded tendons.

Except that the P*e effect should then be reduced by the capacity reduction factor, or for South Africa, the material factor which I think is 1.15, so your load factor should have been .87 for the PT moments!

slickdeals,

Yes, you can allow for it in the Vc calculation (not the principal shear, only the flexural). But you must take into account the reverse curve regions properly (as some US software does not!!!).  

Rapt,

   No, this guy has a design firm.  Whether or not his firm also supplies the strands and does the tensioning, I don't know. Retired now and apparently does PT design himself in projects whenever he feels like it.

   But anyway, thanks for thanks for the info. Nice to hear things I never knew before.  A number of people in my office  like unbonded systems. They think the risks are over-rated. So I just wanted to get an opinion from people who discourage unbonded PT.

Iyota,

Same person then. They are not involved in the construction directly, only design. He has been pushing unbonded PT with frame contractors in lieu of bonded PT from PT companies because they do not want the extra subcontractor on site. A backward view in my opinion, similar to PT operations in USA. I would much prefer to have specialists doing it! We agree to disagree on this quite often!

An senior engineer in Hawaii was once heard to ask "why would you put reinforcing steel (unbonded tendons) into concrete and not bond it to the concrete?" Seems to defeat the purpose doesn't it?

Rapt,

Just out of curiosity, you had mentioned that VSL had issues with their bonded pt system. I had worked for them for about two years. They still try to sell the idea of a bonded system, which I am a proponent of, but indeed they are typically faced, with the issue of cost being about 150% of the unbonded systems.  There were two different types of bonded systems; one was to be used in flat slabs, this had a flat duct with the ability to place up to (4) 0.6 strands. The other was a typical multi-strand system which was for transfer slabs, mats etc.  How do these systems differ from systems outside the US?

Also, might the other PT software developer you seem to have differences with be Allen Bommer? He is in a very influential position.  I have seen him post on this site once or twice.

Regards,

AUCE98
 

Is Rapt going to tell us who the mole is? [:)]

Rapt

I am not sure I am following the discussion. I am a US engineer and have designed several unbonded PT slabs using bands/distributed and they have performed quite well. And many 100's or maybe 1000's of slabs other than mine have performed well also.

And they are somewhat partial prestressed because we are allowed to see some tension during service.

If they (unbonded slabs) perform well and are less expensive shouldn't we advocate there use?

In Australia, we don't have the problem of deciding between bonded and unbonded, because unbonded is prohibited.  But I am interested in the cost difference.  A 50% premium seems a lot for an inferior system.  I would have thought grease was more expensive than cement.

Auce98,

No, I have no problems with Allan Bommer. I was thinking of someone from the 70/80's and I am not naming names, but he has helped to make a mess of the PT section of the ACI code in my opinion.

The unbonded system VSL were pushing in the 90's in US was a plastic duct 2 strand system. It was very expensive compared to the 4/5 strand flat metal duct systems used elsewhere.

The difference in the costing is not just the grouting. Outside the US, PT companies provide a full site service from installation, supervision, checking, stressing and grouting. That does not tend to be the case in US.

Also, In comparing the costing, you have to look at everything, not just the PT price.

But this thread was not about unbonded vs bonded, it was about, tendon layouts in 2 way slabs. The problem with unbonded came into it because it is very difficult to lay unbonded tendons in a column/middle strip layout which has basically resulted in the ACi code having a design method for flat slabs that defies all engineering logic and places stupid limits on how flat slabs can be designed.

Partial prestress is not allowed by the code simply because of this design methodology and unbonded "forced" the design methodology.

It is a comvoluted arguement as the dog keeds chasing its tail!

ash060,

See my previous reply.

I have other problems with unbonded and yes, there are problems with bonded too if it is not conctructed properly, but unbonded is much more susceptible to problems simply becasue the whole strength of the structure long term is relying on the anchorages. This is not the case in bonded if it ius built properly.

My main arguement is against the design logic in ACI code, the design of flat slabs based on the full moment over the full width, the stresses used to decide when the concrete is cracked, the logic of Class U up to a stress equal to the tensile strength of the concrete.
The whole thing is very misleading. The designer does not know what he is really getting. Yes, the buildings stand up, but there is a lot of cracking  and as I repeat, the designer does not know what is really happening in his slab. Maybe he deos not care, but I think he should. You are ending up with generations of designers who think that is how PT slabs actually work and think they understand what they are doing with design. Then you get people who start saying that, if the tendon distribution over the width does not matter, as the ACI code says, then you can use equally spaced tendons in both directions, or you can use the total moment on total width logic for slabs with drop panels or with concentrated loads. Because they do not understand how slabs really work. And this has been encouraged for years by those people I mentioned (No not AB).