Long Term Deflection & Compression Reinforcing 1

[OzEng80]

I recall a side comment by RAPT on a thread (that I haven’t been able to find) regarding the appropriate use of compressive reinforcing for reducing long term creep and shrinkage deflections. As AS3600, CL 8.5.3.2 now clearly states that compression steel needs to be located in the compression zone (who would have thought!) – I wanted to clarify the application of Asc in relation to slabs.

I had a quick look at Ku values for typical slabs that I have designed (deflection controlled) and Ku seems to generally be in the order of 0.1 (treating as singly reinforced). Adopting this value of Ku =0.1 and assuming 25mm cover and d = D – 25 – 10, the minimum slab thickness to just scrape any reinforcing into the compressive zone is D = 300mm (kud = 26.5mm). As this is based on a singly reinforced section – accounting for this ‘compressive reinforcing’ would actually lower Ku further (or does it – if it is not in compression?!). I have read that under sustained loading (ie appropriate to the consideration of long term deflections) the neutral axis lowers further as internal stresses are redistributed into the compression reinforcing – I have not found anything to quantify this however.

So…
How do you determine the long term neutral axis depth and is this appropriate for the assessment of ‘compressive reinforcing’.

How far into the compression zone does reinforcing have to be to be effective - Should the reinforcing be located within the ‘compressive stress block’?

Is there a limit to how much compression reinforcing can be effective (right up to 100%)?

Should the Kcs reference be removed from the slab chapter (is it really on appropriate to beams)?

What is industry practice? Every engineer I have encountered (including myself) has used 'compressive' reinforcing to reduce deflections in slabs much thinner than 300mm… Since the industry dictates predominantly slabs with complete top and bottom mats of reinforcing (for ease of construction), I am loathe to end up specifying thicker slabs than everyone else.

I understand that the formulas are significant simplifications of very complicated theory and are ‘deemed to comply’ – I am not disputing their appropriateness, I just want to know to apply them correctly.

Thanks in advance!

 

[asixth]

OzEng80

1. Cannot comment on your first three questions, however, I would use a triangular stress distribution when considering service loads and hence calculating deflections.

2. I believe Prof. Ian Gilbert who chairs the sub-committee is against kcs factors for calculating deflections. I find it to be a very helpful multiple when calculating additional deflections due to creep and shrinkage. When I say I find it a "helpful multiple", I mean using a multiple of 2 gives my spreadsheet deflections very similar answers to what RAPT (software) does.

3. I would never use a kcs of less than 2.0 for slab design because it is not industry practice to reinforce slabs in compression regions hence no help will be provided by the reinforcement to reduce the deflections.

I have attached my spreadsheet.

The process I go through basically follows:

1. Calculate service moments midspan and calculate Ieff based on these service moments.
2. Calculate instantaneous deflections based on short-term loading and Ieff (Normally G+0.7*Q).
3. Calculate additional deflections from creep and shrinkage based on Ieff and kcs=2.0 (2.0*G+2.0*0.4*Q).
4. Sum the two to give a total long-term deflection.

I have only ever used this approach to verify the outputs I have obtained from RAPT (software).

 

[rowingengineer]

I believe RAPT is in the UK at the moment on holidays I imagine, having fun sun banking on their beaches on doubt. So I think you may have to wait a while for him to chime in.

Ian Gilbert says “The use of the deflection multiplier kcs to calculate time-dependent deflections is simple and convenient and, provided the section is initially cracked under short term loads, it sometimes provides a ‘ball-park’ estimate of final deflection. However, to calculate the shrinkage induced deflection by multiplying the load induced short-term deflection by a long-term deflection multiplier is fundamentally wrong. Shrinkage can cause significant deflection even in unloaded members (where the short-term deflection is zero). The approach ignores the creep and shrinkage characteristics of the concrete, the environment, the age at first loading and so on. At best, it provides a very approximate estimate. At worst, it is not worth the time involved in making the calculation.”
(I doubt I can say it any better.)

From the paper:

http://www.ejse.org/Archives/Fulltext/200101/02/20010102.htm

I have also heard Ian say you should never use a KCS less than two if you must use the kcs method.

I believe you should treat the compression steel as design compression steel, such that you do you complete analysis as you would in a beam and then % of yield is % of effectiveness. So you would use the compressive block after analysis with the compression steel.

Now the real question is what about PT slabs??? That one I can’t answer I don’t know how you would ever get a reasonable answer from the KCS method for PT slabs.

Should KCS be removed, from the standards? YES, it should, we have the ability to better estimate the deflection these days, thus KCS should be put to death, I believe it has only survived this long because of Finite element programs, that needed the KCS to do the deflection calcs in an easy fashion.

As for doing what everyone else is doing, be careful here, I don’t use compression steel in slab generally and neither do any of the designers in my office, so you may be the only person using compression steel in slabs. And if you find anyone whom is using the compression steel shoot them down.


Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[OzEng80]

Thanks for the responses.

If Kcs = 2.0 then Acs = 0 so I guess it should not be referenced in the slab section. I will go through the paper and try and get my head around it better.

In regards to industry practice, it is common place in my region for entire top and bottom mats of reinforcing to be provided (with additional top reinforcing as required). It is apparently more cost effective to do this (less tying and individual bars). I have specified slabs with staggered/curtailed and discontinuous top mats only to arrive on site with the builder opting to put an entire top mat in (note that I tend to deal more with residential applications with short span, thin slabs). To this end it is commonplace to have reinforcing on the opposite face to the flexural reinforcing (I am too wary to call it compression reinforcing now!).

I have only worked in two offices but in both it was considered appropriate to use compression reinforcing with an upper limit of Asc of about 50-60% of flexural – Kcs of about 1.4. Through another project I was in contact with a senior engineer who insisted that an Asc of up to 100% could be adopted. I have a set of drawings in the office at the moment for another consultancy for a previous stage of a project where the slab thicknesses can only be justified by a Kcs less than 1. Bit of an epidemic really – unless I am just mixing with the wrong people!

Thanks for the spreadsheet asixth – I have attached mine (which needs a major rehash) since we are sharing.

 

[rowingengineer]

OzEng80,
What part of Oz are you in? Adelaide?

Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[OzEng80]

Cairns. What have got against Adelaide?

 

[rapt]

Who would go to the UK in winter for a holiday. There is no sun on the beaches (is there such a thing there, beaches or sun as I did not see either), or anywhere. It was freezing, and it was worse in Europe where we were snowed in a couple of times! Fortunately I was busy working all of the time spreading the RAPT gospel in training (mainly) and marketing sessions so I was in heated comfort, except when I had to walk to the train station (about 1000m) in a raging blizzard that grounded all flights out of Copenhagen that night, including mine).

RE the topic, RowingEngineer and Asixth have covered it pretty well.

The comment on the code about compression reinforceemnt having to be in compression was a compromise but was meant to emphasis that just being at the opposite face to the tension reinforcement does not make it effective compression reinforcement. Personally, I think that it has to be at least in the top half of the compression zone to be effective.

RE the compression zone to use, yes the cracked elastic stress zone places the neutral axis much higher than would be the case allowing for the effects of creep and tension stiffening. So it would have an effect at lower depths than the cracked section analysis would show but the effect is not great until it is fairly high in the compression zone anyway, so I would ignore it until is in the cracked section compression zone anyway in terms of kcs.

Even in heavily reinforced RC beams, I have never seen the effect of compression reinforcement be the equivalent of the Kcs factor of .8. Equal top and bottom reinforcement can have an effect from 0 (no effect) to maybe a maximum of 20% in a very heavily reinforced member deeper memebr, never 50% reduction in deflection that a Kcs factor of .8 might suggest.

RAPT actually calculates the strain and stress patterns for the section allowing for shrinkage, creep and tension stiffening (as long as the default tension stiffening method is used, not Bransons guess formula) so it allows for this effect for you and shows the real (relatively) effect of tension face and compression face reinforcement.

From what we have found,
- compression reinforcement has little effect in lightly loaded memebrs and shallow members (eg slabs). It is more useful in deeper members, rectangular beams without a flange (if you can ever find one) and more heavily reinforced members.
- extra tension reinforcement is more useful in lightly reinforced memebrs

RE PT members, no code suggests the use of Kcs for PT members. Some software developers use it because it is simple, and they do not understand what they are doing, but it is not correct. ACI and AS3600 only use it for RC members specifically. For instance, if the PT balanced load exactly equals the permanent load, there is no permanent deflection so ther is no creep/shrinkage deflection as kcs * 0 = 0! This is patently stupid as there are still creep and shrinkage effects causing deflection.

R.E. (not you RE) the other designers, yes it is a problem and the code is part of the problem as has been aluded to above. All I know is that people who have done suryeys of slabs after a long period (20 years of service) have all indicated that RAPT's results are in the ballpark of what they are measuring on site, while their code checks have been way out and very unconservative in estimating deflection effects for slabs. The general conclusion is that, unless you account for it properly (not using code simplifications) ignore compression steel in deflection calculations.

 

[rowingengineer]

RAPT good to see you survived.

Following on from your above comments, how about compression steel over a column. I would have thought compression steel would be very helpful at this point.


OzEng80,
Nothing against Adelaide, just had a few discussion with engineers from that area, and they seem to consider the codes to be books of untruths. Mind you this is only 3 engineer of many so I shouldn't be so harsh.

So are you a JCU man then? I'm an NQ man myself, and attended JCU a while back, I believe Pat Murray is the concrete design lecture these days, Might be worth giving him a call to see what the local NQ standard is, but I can tell you in my office Kcs = 2 for slabs is standard.





Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[rapt]

RE,

Midspan reinforcement seems to have the most effect on deflection, either tension or compression reinforcement depending on the situation as described above. Modifying the reinforcement at the supports does not seem to be nearly as useful.

Are there more than 3 engineers in Adelaide! Actually most of the designers I converse with in Adelaide would never use Kcs of .8, more normally 2 except for deeper members, but then I suppose they are RAPT devotees if I am talking to them so they believe us (and their slabs do not deflect too much!) and try to calculate it properly rather than wasting their time guessing re Ian Gilberts quoted comment.

I did not realise that NQ concrete performed differently to that in other areas of Australia!

 

[OzEng80]

RE - UQ actually, but I just designed the JCU Dental school if that count's for something. Ironically, the engineer you mentioned may have been the one that insisted that 100% Asc could be used (the engineer’s company was doing construction supervision of a project and were advocating the removal of a precamber that I had specified on the basis of the compression reinf)..... I trust the wording of this statement is suitable? Let me know if it's not?

We should catch up if you come up this way? I seem to recall you owe me a beer!

RAPT - concrete is the same in NQ but the higher wind speed helps to hold the floors up.

Thanks for all the help – much appreciated as always!

 

[rowingengineer]

RAPT,
Interesting about the support steel, I have always lapped my bottom steel over columns hoping to reduce deflection.

as for concrete in NQ, It doesn't shrink as much, due to the high humidity. Roof tops and walls can sweat if too skinny. other than that concrete is concrete, not that any of these are concrete properties just behavior.

I suggested he call Pat so that Pat can tell him that they use Kcs of 2, more because if Ozeng 80 hears it from Pat he may believe it than a guy on the net by the name of rowingengineer.

Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[hokie66]

Interesting discussion. I try to control deflection with adequate depth, but do admit to throwing in some "compression" bars from time to time. RE, I never figured the lapped bottom bars at columns assisted much in deflections, but they do a lot for overall robustness. Having said that, I can remember several occasions when I used a bunch of compression bars to limit the deflection of cantilevers, tried to compute how much there would be, and had much less than I expected. Calculating deflections in concrete structures is a very imprecise thing.

 

[rapt]

RE,

I agree with Hokie66 that the best reason for lapping bottom steel is for robustness. It does wonders for ductility of connections (punching shear failure capacity is increased immensly) plus helps a lot for high wind/earthquake effects and also in accidental damage situations (eg if a column fails) that some codes now require (BA8110, EC2).

Interesting comment about shrinkage in NQ. Don't they have airconditioning!

Hokie66
In Cantilevers, if they are beams, compression steel will be more effective in them than in a normal beam as the flange is not at the compression face, so the neutral axis depth is normally much deeper!

 

[rowingengineer]

OzEng80,
I'm always keen for a beer, you should comment on my FAQ. Forget I mentioned Pats name then if he is giving out that advice, He is considered to be the local expert and I would have thought he would have know better, BUT I guess the guy whom suggested it was named Andrew as Pat has been unwell.

I will give Pat a call and chew his ear when i get the chance.

Hokie,
I think if you add tension and compression reo your transformed modulus would increase, and shrinkage deflection would reduce, but Kcs doesn't tell us this. Yes I agree it dose do a lot more then help with deflections, But i would like to test my theory. But alas I doubt I will ever get the chance, maybe I can swing a uni to do it as a thesis.

Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[hokie66]

RE, I agree, except that at supports you often have a deeper section for some width (beam, band, column) which already acts to stiffen that area. Bearing walls would be the exception. In adding compression steel to help with deflections, I have always thought of it as making the concrete act like steel as much as possible, where the reinforcing forms the flanges. Deflection of steel members is predictable. The problem with that analogy is that the webs are not steel.

 

[asixth]

A few comments:

1. Nice spreadsheet OzEng. I am going through it and seeing if there is anything I can steal for my spreadsheet.
2. I'm keen for a beer.
3. How is the job market in Cairns and Adelaide. I'm ready to up and leave Brisbane.
4. The basic shrinkage strain for Cairns recently increased from 850E-6 to 1000E-6. What I am led to believe is that local aggregates can greatly affect properties of the concrete such as shrinkage and elastic modulus. So concrete's aren't concrete's.
5. The kcs approach doesn't account for the creep and shrinkage characteristics of the concrete, the environment, the age at first loading etc. But does anyone go into RAPT (software) and change the relative humidity from 50% or the time of loading from 10 days? I am unfamiliar with Inducta or the percentage of the Australian market that is using Inducta at the moment.
6. I have come across two buildings lately constructed in the 1980's that in my opinion would not satisfy today's standards or design methodologies. One of the buildings is our office building (designed by others) which has very noticeable end span deflections. I haven't surveyed it but I would expect the deflections to be in the order of 40-50mm for an 8400 span. Tomorrow afternoon, I may survey the slab to measure the deflections, and compare to software (and my spreadsheet) to see how accurate the calculations are. See the attachment, span between Grid 4 and 5 on Grid C, 8400 span, 250mm thick slab. Of course, this will all depend on how many beers I have.

 

[OzEng80]

asixth

1. Help yourself.
2. Me too
3. Can't speak for Adelaide but Cairns is still in a hole. Apparently one of the worst unemployment rates in Australia. I thought things were starting to pick up a bit – but today we had a SE with 5 years experience apply for a drafting role... Not a good sign.
4. I had no idea. Where did you get that info?
5. We have inducta (and RAPT) which contains humidity and hypothetical thicknesses etc. as well as the ability to analyse to other standards. I haven’t played enough with it yet, but the Australian method of analysis results in significantly larger deflections then the other codes. The software still relies on you to enter your own load combinations (and therefore interpret Kcs).
6. At least you have drop panels. I am working on a new classroom block which is basically a carbon copy of 3 previous stages since 94. The classrooms have 8200 clear spans 220 thick with a short tapering (to 170mm) balcony cantilever. Classing this as an ‘end span’ (which is pushing it) and using a Kcs = 0.8 (Asc was less than Ast even if it was effective) results in a 40-50mm deflection. I went out there today expecting to have to duck my head into the classrooms or to roll an ankle as I negotiated the steeply sloping floors – but alas, there was no discernible deflection. Given that two of the blocks are now 15 years old would make it pretty hard to justify the 25% increase in slab thickness (though the architect hasn’t noticed – yet).

 

[IDS]

Some comments from the heavier end of structural engineering:

It is standard practice in non-building structures to reinforce every face, in fact it's a requirement of the Bridge Code (AS 5100), but I agree with earlier comments that for thin slabs the "compression" reinforcement is likely to be at about the level of the neutral axis, and will make little or no difference to deflections.

In recent years I have been involved in a number of projects where deflections were much greater than predicted in statically determinate structures, where prediction of deflections would appear to be simple. The main factors that I have found to be neglected (or underestimated) in a simplistic analysis are:

1) Tension stiffening is much less than predicted by the "Branson" equation for lightly reinforced members with maximum moments just over the cracking moment.

2) The reduction in cracking moment due to shrinkage stresses has a huge effect on deflections, and must be properly taken into account.

3) Differential thermal stresses may also result in a significant reduction in cracking moment, especially if the tension face is in full sun. Even in the UK!

4) After cracking the section is asymmetric, and shrinkage will have a significant effect on deflections, even if the section is symmetrically reinforced.

5) It is important to take account of temporary load sitiuations during construction or due to live loads, if these are higher than the long term loads. Cracked sections stay cracked!

Taking account of these factors I have found a good match between measured and predicted deflections, with the deflections being about three times greater than predicted by a simplistic analysis, complying with code requirements.

More details are in a paper I presented at the recent Concrete Conference in Sydney:

http://interactiveds.com.au/Publications/Deflections09.doc

Finally it should be emphasised that in many situations the best we can do is come up with an upper bound prediction of deflections. In real structures deflections vary widely, even in nominally identical members. Actual deflections will be much less than a predicted upper bound if for any reason a section that has been predicted to crack remains uncracked. The best we can do is to consider the consequences of the upper bound deflections occurring, and if these are unacceptable take measures to reduce them.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[rowingengineer]

RAPT,
I think you know what PT3D is, maybe you should ask Jeff what he thinks of this statement taken from there home page (maybe it is a typo) But i knew at least one PT company had stated that they were using Kcs for deflections.

http://www.inducta.com.au/pt3d.htm

"Evaluating of the deflections using Ieff

Based on the adopted reinforcement, PT3D will evaluate Ieff in each point, in both orthogonal directions and will calculate the short-term deflections based on the reduced section properties. The long term deflections are evaluate using Kcs factor, and increased load factors."



Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it

 

[rowingengineer]

Thankyou Doug for your input, nice paper.

Following on two points, I would like to discuss shrinkage and thermal stresses; these are all related to restraint, either by reinforcement or supports. The reinforcement is an easy restraint to account for as there are papers dealing with this issue, however there are very few references dealing with support restraint forces, and associated deflections. The only method I know of that attempts to account for support restraint forces is the PCA method. Then I use a bit of bluff and mirrors and convert this from forces in the member, to a reduction in flexural strength.

This is one area that the AS codes do not treat at all well, with one clause that is a bit puzzling giving extra reo based crack control and if is restrained or not, with a partially restrained clause giving you judgement choice. But no discussion about restraint associated deflection.

1. I noticed Asixth you have the ACI method in your spreadsheet, you may wish to down load Alex’s metric version, for the other parts (shear & torision).
2. Beers is good
3. Job markets all over the place are slow, but if you’re good you will find a position. I am/have changed jobs twice in the last 8 months, no problems. Once forced by conflict with my employer and then got an offer I couldn’t refuse. In Brisbane/Townsville at the moment a few companies looking for Good people.
4. Where did you get that info asixth? Is it just cairns or Townsville too?
5. I have access to both RAPT, Inducta and Strand7. I change all the inputs in everything be it RAPT, Inducta or Strand7. From a marketing point of very RAPT hasn’t spent much time in the NQ, so Inducta is more established, but is developed by a software guru not a structural engineer as RAPT. Inducta have just made available the AAEM and EC2 methods for deflection, I would recommend any engineer use these over the Kcs method. Inducta also has a note in there manual and will give out advice that Mxy can be ignored for slabs due to what the standards have told them. This is incorrect and RAPT will post a message soon showing the standards stance on this issue.
6. With older building you have to remember that often the construction pace was slower thus suspended slabs were on forms for longer and the concrete agg was better generally. Also with deflections, I am less conservative in my modelling then with strength design. Make sure you don’t be to conservative int eh deflection modelling, with regards to column stiffness ect.


Arguing with an engineer is like wrestling with a pig in mud. After a while you realize that they like it