Sudden jump in Pier Shear Forces 1

[erginc]

Hi all,
I am new to eng-tips, found most of my answers in old topics but this time I need to write and ask about it.

I have a 37 story building including 3 basement floors. Wall Pier P4 extends from BASE to 29th story. Its length at lower stories (first 3 story to be exact) is 4.7m, and upper levels it is 2.72. There is a sudden change in length. I will refer to 4.7m wall as P4L, and for 2.72m it will be P4U. (L stands for lower and U for upper)

As a result I got quite a big change in shear force at the third floor where pier length drops suddenly. I have no idea how ETABS is calculating Pier forces from parts of Wall Meshes and Mesh Internal Forces, therefore I can only speculate about why I am getting this type of sudden change.
My guess is the offset between P4L and P4U which is about 1m will creates this sudden increase. Axial load transferred from P4U to P4L with that offset will create a huge moment. But this should only be applicable if plane section remains plane assumption is true. But in this case P4L is with 4.7m cannot stay plane as it is not continues along height.

To illustrate it I have added a ZIP file with a little screen shot of ShearForce for total DeadLoad, and an XLS file with pier forces of P4 on lower levels where you can observe the jump for total dead, total live and SpectrumY loadings. I hope this helps. (Unit: ton force-meter)

So I have two questions.
1) How ETABS is calculating Pier force from wall internal forces/deformations?
2) How can we describe above case?
3) Will it be correct to name 2 different pier names for 4.7m pier, one name for 2.72m same as P4U and rest with a different name so ETABS calculates pier forces better.

This actually a general FE analysis question may be, but it might be simply related to ETABS way of calculating Pier forces, therefore I posted it here.
Thanks in advance,
Ergin

 

[irizarry]

I really can't answer with certainty, because I haven't inspected your model, but I do know some versions of ETABS had some problems when assigning shear forces, especially from earthquake loads (although I'm not sure if this problem was also happening for Gravity loads). I think you should double check your model for connectivity issues and also check if some of the walls don't have supports assigned at some point.
About how does ETABS calculate pier forces, I think it does use internal forces calculated from an FE analysis (if you mesh the wall), but frame elements are not divided into a mesh, so only its calculated rigidities will be included. I think you should try running the analysis without automeshing the piers (go to assign -> auto mesh options -> no automehing(use as structural)) and check results.
Also, if both piers are connected and continuous, no mattter the difference in length, both should be named equally (although you can still assign different section properties).

 

[zeemas]

What you have explained is usually common in tall buildings at lower floors where other walls are introduced and at the basement level, it is called as Back stay effect.
See the research paper at following link, i hope it will help.

http://ztwocents.wordpress.com/2010/08/10/back-stay-effect-in-tall-buildings/

 

[erginc]

Just came back from vacation and saw the replies.

Thanks irizarry, I have checked my model for support and connectivity it looks ok. For pier forces I also got the following response from CSI "Program integrates the wall forces at the extreme top and bottom wall section using nodal forces of wall which has unique pier label. You can replicate the Pier forces by looking at joint forces of wall and integrating them at the wall centroid."

Actually I was hoping to get a little more detail from CSI like some article or some link to a paper but this is all I got from them. And for simple pier I have tried to integrate nodal forces, with no luck, only axial is close, rest is completely different between hand calculated and program calculated pier forces. I am missing something but don't know what yet...

After reading zeemas response and CSI answer, now I need to read the paper for back-stay-effect to understand it. Thanks for the paper zeemas.