Interpreting CSI SAFE results - PT Concrete 1

[SoCal_Structural]

Hello everyone,

I have run into multiple issues in designing Post-Tensioned Parking Structure, and hoping to get some feedback from you guys. A quick summary of the parking structure building:

- 8" PT SLAB (Slab is thicker than usual, Live load requirements (100 psf) for this building are higher than parking structures
- 14"x38" PT Beams spaced at 18' O.C.
- Some 24" wide PT girders
- 24"x24" Concrete columns

I have attached a pdf with snapshots of output from software as well as a quick hand check for prestressing force requirement.

Observations:

1) - Moments in beams are much smaller than anticipated - There is 35-40% difference between ETABS/Handcheck vs SAFE (SAFE MOMENTS SMALLER)

2)- All beams are failing - I am having to require insane amounts of Pre-stressing loads (and still failing) eventhough the moments are smaller than actual moments, this is going to get worse once I fix the moment issue.

3)- Flexure Reinforcing (non-prestressed) requirements are higher at mid span (for positive moments) than for end moments (negative), even though the end moments are larger.

Below are links to PDF results and SAFE model:

https://files.engineering.com/getfi...-4d00-b66e-d0bb8f99e664&file=SAFE_Results.pdf

https://files.engineering.com/getfi...2b-036b56e44548&file=Level_4_Without_Ramp.FDB

I tried messing the beam section properties in the model, by making it a T-beam and assigning it an effective width, but no effect on results. Tried without effective width (beam stem only) still no luck. Any feedback would be appreciated.

Thanks,

 

[SoCal_Structural]

Responding to my OP since I initially posted this under a wrong category.

Thanks,

 

[Ingenuity]

Kudos for recognizing that there is an error!

I do not use SAFE for PT design, but here is what I would do:

1. Forget about doing this as full 3D framing model until you have a grip on what is happening with SAFE on a 2D run.

2. Select a single grid line and model a frame: columns and beams over a single story.

3. Do some hald calcs of this single frame model and compare to SAFE’s output.
​

Based upon your loads and spans, you should be able to make this work with around 15 to 20 tendons, with balancing say 75% of DL. 30 to 50 tendons is not correct.

I would even suggest that you first model a simply supported 64’ span beam with pin supports (no columns) and check that output to your hand calcs. Then venture to a column/beam 2D model.

Start simple, understand what SAFE is doing, then maybe extend that to 3D if truly warranted.

Based upon 17 tendons, I get 360 psi of flexural tension at midspan under service conditions, about 5.5 in² of bottom midspan rebar (for ultimate), and long-term deflection of less than 1”. This was modeled on 100% column fixity with an assumed column height of 10’ so there is a negative moment demand at ultimate requiring about 7 in² of bonded rebar.

As you alluded to, for a 18' span slab with a 8" thickness (even with LL of 100 psf) that is pretty thick: L/D = 27, and results possibly in 'unnecessary' additional load to take up on the PT beams.

 

[SoCal_Structural]

Thanks for the insight, Igenuity.

After playing with many variables, I decided to re-import my model from ETABS. However, this time for every small change made I ran the analysis, so I could narrow down when/if something went wrong.

My results are much better than the previous model. I noticed that the moments dropped in the beams after adding tendons in the beams, while the load case was dead only, I noticed that the program considers post tensioning regardless of the load case selected. Subsequently ran my model with 20 tendons (quick check), the beams appeared to all pass.

I hit my milestone for the evening, will be back at it in the morning. You are right, I thinking it will end up being somewhere between 17-20 tendons.

If I run into issues on this model, I will follow what you said and start with just a member design.

All said, I think the culprit may have been the column/support stiffnesses (i may have messed with the supports in my first model).

Thanks again!

 

[rapt]

Never used it, but you should be able to get individual load case values or total ultimate, unaffected by load balancing effects. They should be similar to Etabs.