SAFE v12 Deflections with Creep 1

[Corey7320]

What is the best way to determine deflections in Version 12?

In Version 8 my firms procedure was to create a load combination called LONG and scale load cases to match the time period being investigated. For example the combo would have 2.0xDL, 2.0xSDL, and 0.5xLL for final deflection (5years+) Also part of the procedure involved setting the long term deflection multiplier to 1.0 for all load cases and selecting Normal and cracked deflections under analysis options.

The new version is quite different in setup. You now have to specify if a load case is linear or nonlinear, what type of nonlinear, starting conditions of the nonlinear, you can create load cases that include other load cases at scaled factors, Load combinations can include linear and nonlinear load cases, you can select Service initial, normal or longterm for load combinations. I have tried to match our old method but there are too many variables with the new version. I have not been able to duplicate the results from the old version 8.

I have an inquiry in to CSI if they respond with some valuable insight i will share it.

Any help is appreciated.

 

[Corey7320]

This is the reply from CSI

The cracked deflection analysis as implemented in SAFE v12 is documented in Example 16 of Analysis Verification problem along with hand calculations to verify SAFE v12 results. We also provided the S16.FDB file in the Verification folder so you can run this model and look at the details documented in Example 16 verification document.

There are two common situation for investigating cracked deflection i.e., immediate cracked deflection and long-term cracked deflection. This can be set up as follows:

1) Immediate Cracked Deflection:
a. Add Dead Load case using Nonlinear (Cracked) started from Zero Initial Condition.
b. Add SDL Load case using Nonlinear (Cracked) started from using Continue from State at End of Nonlinear case “Dead”.
c. Add LL Load case using Nonlinear (Cracked) started from using Continue from State at End of Nonlinear case “SDL”.

2) Long Term Cracked Deflection:
a. Add Dead Load case using Nonlinear (Long Term Cracked) started from Zero Initial Condition.
b. Add SDL Load case using Nonlinear (Long Term) started from using Continue from State at End of Nonlinear case “Dead”.
c. Add LL Load case using Nonlinear (Cracked) started from using Continue from State at End of Nonlinear case “SDL”.

The reason to start from a previous case ensures that the change in stiffness due to cracking developed in previous load case is considered when additional loading is applied.

I hope this clarifies this issue.

After looking at the file they mention i am not sure how the creep coefficient and shrinkage strain are determined. For 5years or more the time-dependent factor is 2.0 in ACI 318-05. If anyone has comments on these factors i would like to hear them.

Thanks

 

[seny]

i got exactly the same answer, probably, typical respond on this matter.
in addition - you do not need to use modifiers as per 10.11, when load assigned as nonlinear cracked cracking conditions internally calculated by safe on element by element base w/ determination of Ie. when load assigned as nonlinear cracked long term, safe calculates long term deflection as 3 times immediate, no compressed reinforcing is taken into consideration.

 

[Corey7320]

I have done some research and the new version of safe uses shrinkage strain and creep coefficient for long term deflection analysis. MacGregors text and probably any other good concrete text covers how to calculate these parameters. I have taken field measurements of deflection in a building that was poured in Nov/Dec of 2007 and compared to the results that SAFE 12 gives. I came up with a shrinkage strain of 0.000404 and a creep coefficient of 3.088. I used some basic assumptions on how the floor was loaded during construction. I had 4 sequential nonlinear cases and used the same shrinkage strain and creep coefficient for all case. (not sure if that makes sense, I originally tried calculating different shrinkage and creep for each load step but did not acheive good result) I have obtained good results that are in the realm of what i measured in the field. One large factor was how supports were modeled in SAFE. The floor export from ETABS just placed point supports at the columns which retrained rotation about X and Y. This modeling did not have nearly the deflections measured. I then added columns that extended a half story above and below the slab and pinned the top and bottom of the column stubs. This acheived better results. I also considered using full-height columns above and below and fixing the ends but never tested that modeling case. If anyone has comments on my procedure i would be glad to hear them.

 

[seny]

great job - nothing can be more real than field measurement.
for safe 8 it was very simply immediate x 3 (default, branson recomende 1+3 ) = long term.
in safe 12 it seems more accurate.
i did by hand comparison for different kind of assigning for load case - linear, non-, non- long term.
it seems more accurate than i said before - it seems they implemented 435r procedure.

 

[zeemas]

Well done guys.

 

[Corey7320]

Some final comments on the subject ... I spent some time studying the support conditions and after examining many possibilities, i think that using full height columns below the slab only with either pinned or fixed supports is most realistic. It results in good estimate of deflections and seems conservative for bottom steel design. The top steel design varied more and no case was the clear conservative solution. One other thing i convinced myself of is that during cast-in-place concrete construction any given floor, except the top two stories, will experience about 1.5 x Self Weight max due to the shoring and curing sequence. This is based on 10 days between floor pours and shoring until 30 days resulting in 3 floors supporting 4 floors of weight worst case. I also roughly estimated the relative stiffness of the the 3 supporting floors with the lowest floor being more cured than the floor above and so on. Hope these comments are helpful.