long-term deflection due to creep 1

[SCss]

Hi all,

I am doing some FEA and FEM and was trying to fully grasp the governing design concepts for ultimate as well as serviceability limit state design. From my understanding long term deflection due to creep (serviceability) usually governs slab thicknesses, while ultimate loads usually govern reinforcement design. Do you think this is an accurate assumption? And if so why?

 

[KootK]

In general, I'd say that's pretty accurate. For sensible slab and beam proportions, it is a relatively simple thing to add rebar in response to increasing load as moment capacity scales up more or less linearly with reinforcing quantity. Adding rebar is an effective way to improve bending capacity. Adding rebar will improve deflection, but it won't be anywhere near as effective as adding depth. Hence serviceability concerns having a large impact on thickness choices. I have limited faith in FEM deflection results. Too much depends on construction methods rather than design. I'm still mostly a span/depth ratio guy.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[CELinOttawa]

I agree with Kootk. I have a building I'm dealing with right now that has 3/4" less slab thickness than the prescriptive thickness (or else "do a specific check of deflection") table requires. It is reasonably reinforced, but long term creep is an obvious issue with the building. The extra rebar has not been effective, and an extra inch in the 1970s would have saved everyone a good deal of trouble.

Do not skimp on the required thickness, and do not truss the FEM deflections. Just meet the minimum prescriptive thicknesses to avoid specific checks; they embody a great deal of real-world trial and error, or at least error avoidance.

 

[rapt]

It is actually long term shrinkage as well as creep that causes the increased long term deflection, both about the same effect. Plus FEM software also often ignores cracking!

Also, we have found that the "deemed to comply" L/D ratios in design codes can be significantly un-conservative, especially for flat slabs still resulting in slabs with severe deflection problems.

Comparisons consultants have done with real world building deflections compared to RAPT's long term deflection estimates have shown that deflation calculations done properly fully considering cracking, tension stiffening, shrinkage and creep can provide reasonable estimates of long term slab deflections.

 

[CELinOttawa]

Hmm... Interesting RAPT. I've always found that an extra inch saves a lot of trouble. I typically round up to, or if close anyway, add an extra inch to the results of the span over check.

The problem site I'm dealing with now checks as needed a minimum of 8 3/4"; I would have used 10" slab if doing the design - They built an 8" flat PLATE. Things would be better with a flat slab, and much better with some beams... But I cannot change what was done forty years ago.

I'll keep your comment in mind about the span to depth checks not always being conservative.

 

[IDS]

Further to the comment from rapt I would add:

- Shrinkage is a significant issue even if the reinforcement is symmetrical (contrary to what some codes suggest). Once the concrete has cracked the section is no longer symmetrical.
- Shrinkage will significantly reduce the cracking moment.
- Differential temperature also significantly reduces the cracking moment, especially for sections with a tensile face exposed to direct sun.
- The ACI (and AS 3600) formulas for deflection significantly over-estimate the stiffness of lightly reinforced sections at just over the cracking moment.
- For structures with a long design life look up recent work from Bazant on creep deflections.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[hokie66]

I am in the boat with KootK and CEL. Although I respect the work done by many to try to predict deflections, in my opinion it is still just guesswork because of so many variables. I have never known a structure to have deflection issues which did not also fail traditional and conservative span/depth rules.

This discussion is "deja vu all over again", as I am sure that rapt remembers.

 

[IDS]

It is perfectly possible, and not that difficult, to get an upper bound estimate of deflections that is much better than guesswork. Sometimes actual deflections will be significantly less, but they should not be significantly more (given materials and construction as specified).

Is it worth doing it?
For standard size buildings with a long history of similar structures; probably not.
For anything unusual, or without a long history, or where greater than expected deflections will be a big problem; definitely.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[rapt]

Hokie,

you obviously have not had anything to do with the 9m square grid flat plates in the 1970's office buildings in Melbourne. all were designed to the span/depth rules and according to them worked properly.

In general the deflections are in the order of 90-100mm, so L/deflection of about 140 or worse, and the deflections are very noticeable.

If I am after L/250 deflection, I would like to know I was somewhere in the ballpark with my calculations, rather than ending up with a building that my client cannot sell 20 years later and wanting to know why!

L/D rations and simplified calculation methods where "designers" use reduced multipliers for compression face reinforcement etc just cannot give you that.

Doug - agree with all of your comments

 

[asixth]

I have increasingly seen consultants reinforce the top face of there slabs midspan, either for transfer stresses or to reduce the long-term deflections.

I work on a long-term deflection criteria of L/360. It seems to come up in many client briefs these days. And it seems spans keep getting longer.

The big problem I face and I guess many other consultants are in the same boat is that many builders and developers only want to pay for the absolute minimum material quantity that will get the job done. I have used concrete design packages mentioned in this thread and haven't had a slab design even close to having questionable deflections after construction. Granted that most of them have been designed using loads higher than what the real-life load would be. Most people tend to break down services, ceiling finish, floor finish and partitions as a dead load and will allow up to 200kg/sq.m for them. And this is on top of a constant long-term live load of 40-60% of the design live load.

Saying that I am an advocate for going to the effort and actually determining the slab thickness by refined calculations to keep the concrete quantities to a minimum.

 

[hokie66]

No, rapt, I have never had anything to do with a 9m span flat plate. Do you know the thickness?

One of the big variables which can bring even an appropriate design undone is lack of inspection, e.g. using the wrong size chairs for the top steel in a flat plate makes a big difference, and without capable inspection, it happens.

 

[rapt]

Hokie,

Sorry, I do not have the numbers any more. This comment was from discussions I had with consultants in the late 1980's when we first started selling RAPT in Melbourne.

But the designs were in accordance with the code span/depth ratios at the time (later 1960's and early 1970's) and they were the same as the current BS8110 limits (not sure about the British Eurocode ones).

Consultants were refusing to design flat plates of those spans because of the trouble they had with buildings built in the 1970's to those rules. They did start to relent when we were able to show them that RAPT was actually predicting the deflections they were getting in those old buildings, not the expected values from code span/depth ratios.

I still remember consultants in the 1970's when I first started in prestressing who said they did not need prestressing to reduce deflections because they could justify L/D ratios of 35 for RC flat plates on spans like this while prestress could only reduce it to L/D of 40. They were from Melbourne also!

Asixth,
Compression reinforcement in slabs can give some benefit, but nowhere near to the extend that kcs in AS3600 and the deflection multiplier in ACI predict. I checked this for a 280 flat slab recently with equal top and bottom reinforcement and RAPT's TLT deflection prediction was 2.6 times Short term deflection, reduced from the codes 3 times without compression reinforcement, but still significantly higher than the code kcs of 1.8 for that case.

 

[KootK]

Just stumbled upon this paper from Bentley on how to use FEM for deflection estimates: Link

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[Brad805]

@kootk, here are a couple of reviews from the University of Toronto regarding the current (well 2002) state the non-linear FEM analysis methods.

http://www.civ.utoronto.ca/vector/journal_publications/jp40.pdf

http://www.civ.utoronto.ca/vector/journal_publications/jp42.pdf

 

[rowingengineer]

When writing my finite element article for Engineers Australia, I used the works of Peter Taylor for one of my case studies. Peter did the forensic investigation on a slab with drop panels and spanned about 10m and 8m. He reported the results in a few articles around the place, cheapest of which is in the below link. He is retired now, but back in the day this was a big deal in the NWS structural group.

Thickness of slab was 220mm which give span on depth ratios of 43 and 36 (fairly high but close to the rules of thumb that we would use these days) and drop panel thickness of 400mm. a 15mm pre-camber was specified for the middle of the spans. I think at the time the british code allow a span on depth ratio of 36 for 5 kPa and with pre-camber you could justify your design???

http://www.tlbengineers.com/index.php?option=com_content&view=article&id=79&Itemid=104

I don't know if there are many other publications like this, but it does highlight that good predictions are better than bad guesses.

http://www.nceng.com.au/

"Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning."

 

[racookpe1978]

I'm not a structural PE, so take this with a full salt shaker of salt, but it sounds like the 'thumb rules" for long-time creep of horizontal concrete slabs need to be "beefed up" to require thicker sections.

Or accept that long-term horizontal concrete is going to creep towards failure over a fifty-seventy year building life.

 

[KootK]

@Brad/Rowing: thanks for the docs. I'm well on my way to having a little "FEM slab deflection" library.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.