Flexible integration scheme

[wuhaoshrek]

Hi, all,
I wanna implement a flexible integration scheme for force-based beamColumn element
in OpenSees: changing integration point and weights in every time step, aiming to model
the plasticity spreading during the analysis of RC structure member.

First I wanna know if there is a way to implement this in the current version of OpenSees.exe?
Or is there any workaround can be used?

Otherwise, do I have to modify the source code of FBE in OpenSees? and how can I make
it?

Thanks a lot.

[fmk]

it is not there now .. code must be written, either a new integration class of change the element to accept a new integration class during the analysis in a setResponse or update parameter call. you will just need to keep the same number of integration pt's

[wuhaoshrek]

fmk wrote:
> it is not there now .. code must be written, either a new integration class
> of change the element to accept a new integration class during the analysis
> in a setResponse or update parameter call. you will just need to keep the
> same number of integration pt's

Hi, fmk,

I'll try to write code for FBE later.

By the way I got another problem. I model with one FBE for a cantilever column, the sectional
Action-deformation relationship is Bi-linear. My problem is how to make it when the bi-linear
is softening, which means post-yielding stiffness is negative and so does the element stiffness
matrix. I used displ. control, giving the node the displ. incr I wanna apply.

Thanks a lot, fmk.

Hao

[fmk]

use a non postive definite solver, i.e. bandGeneral or Umfpcak. every solver will fail if you get a 0 on the diagonal.

[wuhaoshrek]

fmk wrote:
> use a non postive definite solver, i.e. bandGeneral or Umfpcak. every
> solver will fail if you get a 0 on the diagonal.

Thanks fmk, you are right.

[mhscott]

You can try using the UserDefined beam integration option for the forceBeamColumn.

http://opensees.berkeley.edu/wiki/index ... nTypes.pdf

Then update parameters in order to change the locations and weights of integration points.

parameter 1 element 1 pt 2 ;# location of IP #2 of element 1
parameter 2 element 1 wt 2 ;# weight of IP #2 of element 1

updateParameter 1 $newLoc
updateParameter 2 $newWt

From a theoretical point of view, changing the location of integration points during a material nonlinear analysis is not a good idea due to path-dependency. Beware!