Column shortening in tall buildings

[Lion06]

Does anyone know of a good reference that talks about column shortening in tall buildings generally and differential shortening specifically? I'm getting total axial shortening at the upper floorof 2.5" for gravity columns, but only 3/8" for lateral columns. I'm concerned about what this is going to do to the levelness of the floors and how much extra concrete the contractor will pour.

 

[Ron]

SEIT...2.5" seems high for upper floor condition. Check your calcs.

 

[abusementpark]

How tall is the building? Steel or concrete columns?

 

[Lion06]

The calcs are definitely right. It's a 30+ story building, steel construction. The individual story shortening is less than 0.1", but when you accumulate that over the entire height of the building it comes out to 2.5".

 

[abusementpark]

I have some webinar notes at my office that discuss this issue in high-rise buildings. I'll check on Monday to see if there is any thing useful.

Shouldn't there be people in your office who have dealt with this before? For a job of that magnitude, I would sure hope so.

 

[Lion06]

This is the first building of this height we've done, so this is kind of a joint effort.

 

[bookowski]

That sounds right. Usually the problem occurs between lateral elements (like the core) and gravity elements. Under gravity loading your lateral elements typically have very low levels of stress and thus very little shortening compared to the gravity elements. Since it's a steel building at least you don't have to worry about long term creep.

If you look at articles about burj dubai you'll see that they sized all of the elements to have a similar stress ratio even though this was overkill to deal with this problem. In your case this obviously doesn't make sense.

Most of this can be built out. The steel fabricator and erector can take a portion of this differential into the columns so that under your dead load (or a portion of it) you level out, this is common. You can work with them to provide the info that they need. Also, your slabs will be poured level so the only differential affecting floor level will be from superimposed loads which should be quite a bit less.

 

[Lion06]

bookowski-

If you pour say the 15th floor level, then it shortens another 1" compared to the adjacent lateral column (due to the additional 20 stories on top of it) won't that result in a floor that is substantially out of level?

 

[bookowski]

Sorry, I didn't mean to imply that eliminated the problem but that you should only be looking at shortening from part of the load, not all of it. First off I wouldn't include live load or if you do only a very small portion of it, you'll never see all of your design live on all floors. Above that I meant that you can assume for any floor being built that it can be built level or sloped up as required, so the shortening is only from subsequent loading - like you said in your case 1" for the additional 20 stories. The 1" differential over 20 equates to 0.05/floor which means about a 10ksi differential in stress in the column (assuming about 12' floors) for dead loads - that sounds reasonable.

If this differential from the subsequent loading is too much the fabricator can build this out by adjusting the location of the connections, i.e. move each connection up 1/8" every 3 floors or whatever you require and adjust the slab as well - similar to cambering. You should first see if it's even a problem though - if you look in the code of standard practice they talk about column shortening and tolerances (do a search in the pdf if you have it), you may not even have an issue.

There's a document titled "Column Shortening in Tall Structures" by Mark Fintel, S.K. Gosh and Hal Lyengar. It's mostly geared towards concrete since this is a bigger problem with concrete construction but it still has some useful info.

 

[msquared48]

That's not unreasonable for high rises Ron. We had to add more steel, ie, make the columns larger than required as we went down to account for this.

SEIT: You can also try to shift more vertical load to the lateral columns to balance the system.

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[msquared48]

Additionally, SEIT, you could, every few floors, just make the columns, well, say 1/4" longer as an example to gradually take out the vertical deflection as you procedd up the structure, depending on the vertical deflection pattern.

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[rscassar]

What we try to do achieve uniform axial strain throughout all vertical elements. Normally this is done by assuring that all vertical elements are subject to a similar stress but on one particular building, a high E-modulus concrete mix was used to lessen the strain for the higher stressed elements.

When there is a differential strain profile across the building there is also problems associated with stiff outrigger systems transferring load from the highly strain elements to the lightly strained elements.

 

[Ron]

Mike/SEIT....I interpreted the original post incorrectly, apparently. I thought that SEIT was saying that there was 2.5" in the upper floors...made no sense to me. 2.5" in 30 floors is not surprising.

SEIT.....I think you might be overthinking this one. Yes it's a consideration. No, it is not a big deal on any individual floor....just the total.

 

[hokie66]

Ron,
It is a big deal on the upper floors if the external columns have shortened 2.5" and the core hasn't. Or the other way around. Gets tricky, with lots of ways to compensate. Have to also consider affects on facade.

 

[msquared48]

But, unless compensated for Ron, the total vertical deflection will be read at the top floor, and gradually less as one proceeds down from the top.

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[hokie66]

If all the floors are cast level, the problem will actually be worse at about half height than at either the bottom or top.

 

[Lion06]

My biggest concern is the pouring of the slabs. Even if the shortening is compensated for with slight movement of the connections it almost seems like some of the slabs need to be poured out of level.

Looking at say the 20th floor of a 35 story building. Let's say there is 1.5" of differential shortening between adjacent columns at the 16th floor due to total DL only. Let's also say that this 1.5" is accounted for with elongated columns and moving the connection up at the problem column incrementally such that at the 16th floor the top of steel is higher at the problem column (say 1" higher since some of the shortening has already happened) . This means that the contractor needs to provide a sloped slab of constant 6" thickness so that when the final 1" of shortening takes place the slab will be level. That seems very finicky. I also imagine this would be extremely problematic for F(F) and F(L) numbers.

Hokie, this makes me definitely see your point about the problem being worse at half-height than at either top or bottom. The top has little shortening left to go by the time the slab is cast and the bottom never sees much accumulated shortening to begin with. The real problem (in terms of compensating for it) is where there is a fair amount of shortening that happens AFTER the slab at a given floor takes place.

 

[Ron]

I understand the issue and agree it's tricky, including the actual predictability/accuracy of loading, distribution and path. Did a load test some years back using exterior columns as reaction and loading the interior columns. Building was only 8 stories. We measured compression strain in the columns simultaneous with load application and found the elastic shortening was much less than predicted, though we were getting load transfer.

 

[Lion06]

Ron-

That's very interesting. I wonder if E comes in higher than the 29000 ksi used in design. I know we often see Fy come in higher.

 

[Ron]

SEIT...the building was a reinforced concrete/post-tensioned structure, so you would expect some discrepancy in predicted vs. actual. Steel should be a bit more predictable; however, actual load path and distribution are much more diverse than we are willing to assume. I've seen this in doing strain measurements on static and dynamic structures. In some cases, it has been more difficult to achieve a load at the expected locations than would be prodicted. The more diverse the connections and the more shared the expected load paths become, the more disparity between predicted and actual.

You might consider some in-place strain monitoring during the construction of the building. For a few thousand bucks you can get a lot of data that will allow you to modify some things as necessary as the building goes up. Cheap compared to some of the problems you can have. Hokie66 noted a particular concern of facades. Strain compatibility is sometimes difficult to control, thus more "slop" is put in the design/connection of facades, thus the more difficult the waterproofing....a "vicious cycle" to say the least.