A quick question about force- and disp-based elements

[linguan118]

Dear administrators,
I have been discussing with many colleagues using Opensees. When a whole structure (e.g. a whole bridge, framed building, even high-rise building) is modeled, usually it is hard to get converged results using force-based elements. Generally, disp-based elements are used instead also for lower computational cost (? haven't tested whether five disp-based elements or one force-based element with 5 ips runs faster). The much more complicated iteration processes in force-based elements are basically the reasons for that.
I wonder if force-based elements have been used in modeling of a whole structure of a real project? What can we do generally to increase the possibility of converged results using force-based elements?

Thank you very much in advance!

[fmk]

yes they are used in real projects. the force-based elements fails because of what happens at the sections, if sections nicely behaved the force beam column works great. if not, not so great. To improve the chance of convergence you need to ensure the behaviour of the sections. of course sections fail and maybe you can do nothing (except through some arficial small elastic stiffness in there so sectio flexibility matrix does not become singular)

however, i have also noticed that for these problems and a discretiztation of the disp beam column into 5 elements the disp beam model typically experiences global convergence issues at around the same time

[wuhaoshrek]

Hi Dr fmk and lin guan, I have similar concerns about modeling a real (or close to real) high-rise building using force-based beam column element. For modeling shear walls or core walls in a high-rise building, is appropriate to modeling the walls using a equivalent frame model? I mean using FBE for the wall at center, and using rigid element to connect the wall node to the adjacent coupling beams. thanks.

[fmk]

Prof. Lowes from W.Washington uses FBE to model the walls. Prof. Filippou uses continuum elements. For tall walls with bending the dominant mode both are fine, though for short walls the continuum approach should be used. HAve a look at presentations they gave awhile back.

http://opensees.berkeley.edu/wiki/index ... esDays2012