Solid and subdivided columns

[sanya greek]

Hello!

I created 2 same constructions:
1)Building with solid columns
2)Building with subdivided columns

After push-over analysis i have two different curves with "Target Displacement".

Solid


Subdivided


Can you explain me please, why i have this difference?
Is it possible to have the same results and how?

Thank you so much!!!

[huffte]

What type of element are you using? Note the descriptions and limitations of each element type before deciding which one is most appropriate for your situation. Displacement-based or force-based? Concentrated plasticity PH elements or fiber elements? I would suggest a through study of each element type in the SeismoStruct manual and Help system. Such differences are not unusual and, if all elements gave the same answer, there would really be no need for multiple element types. It is important to know your element capabilities. Best of luck in your work.

[seismosoft]

The second curve (subdivided model) shows some convergence instabilities in the highly inelastic range. This is to be expected if for instance you have 4-5 infrmFB element per member with 4-5 integration section each (leading to 20 integration sections per member which i way too much). If in the subdivided columns you employ the infrmDB element type, the results would probably be very close.
Seismosoft Support

[sanya greek]

In my situation i created 1 building and after subdivided columns in Pre-Processor of SeismoStruct. So, i analysed first structure without subdividing of columns, and second structure i analysed with subdividing of columns. Thats why i belive that have the same conditions.

Type of Convergence Check Criterion: Displacement/Rotation based
In Frame Element Modelling i used : Inelastic plastic-hinge force-based frame element

" Concentrated plasticity PH elements or fiber elements " - i can not understand, where i can find in programm about it.
I upload 2 files of my example, can you help me please?

https://drive.google.com/open?id=1ZMfvd ... _vu0svAgkv - without subdividing of columns
https://drive.google.com/open?id=1qQW2H ... PrZyQ5CcJD - with subdividing of columns

Thank you so much!

[huffte]

In looking over your model, it appears that you have used 'infrmFBPH' (force-based) elements for the columns and 'infrmDB' (displacement-based) elements for all beams. The 'infrmFBPH' element is an inelastic element with inelasticity distributed over a fixed length and the 'infrmDB' element is a fiber-based element. As noted in the Help system, accurate results may often be obtained with force-based elements without subdivision of the element. Also from the Help system, I provide the following quote regarding displacement-based elements (for which you have divided all beams into 3 subdivisions, it appears):

'Consequently, infrmDB should be employed with members of small length, leading to the need for a mesh refinement, in order to achieve good accuracy in the case of higher order distributions of deformations. Typically, 4 to 6 elements per structural member need to be defined, hence users need to post-process nodal displacements/rotation in order to estimate the members chord-rotations.'

Perhaps a more realistic comparison might be (a) the model with no subdivision using force-based elements for the columns and (b) the model with 4 to 6 subdivisions for columns defined using displacement-based elements. Just a thought.

[sanya greek]

Thank you Tim Huff , and thank you Seismosoft! You helped me so much!