number of Integration points of nonlinear beamcolumn element

[kwaklab]

I modeled SAC building with nonlinear beamcolumn element.

When I used 5 integration points for nonlinear beamcolumne element, opensees gave good result, which was consitent with established study.

But after reducing integration point to 2, it gave wrong result ( structure was more flexible; first period was changed from 2.2s to 3.5s).

What is the reason for this? Is is possible?

Could you recommnand of number of integration points for nonliear beamcolumn element to get good result?

[silvia]

the number of integration points goes along with the integration scheme.
if you use too few integration points you don't have a realistic distribution of deformations. typically 4-5 integrations points give good results.
if you have a strengthening system. if your system is softening, you should use the beamwithhinges element.

[Fahimeh]

silvia wrote:
> the number of integration points goes along with the integration scheme.
> if you use too few integration points you don't have a realistic
> distribution of deformations. typically 4-5 integrations points give good
> results.
> if you have a strengthening system. if your system is softening, you should
> use the beamwithhinges element.

Does the number of integration points imply the plastic hinge length? I was told that the post elastic stiffness factor we use for material properties depends on the hinge length. But we don't define any plastic hinge length in fiber modelling. How does it work?
Thanks,

[IsmBaha]

Integration point weight wIP defines the integration length of each integration point, as LIP = wIP*Lcol
where LIP integration point length, Lcol is the height of the column element.

For example, as I remember for NIP=5 with gauss-lobatto, 1st integration point weight was -> wIP1~= 0.05.
So with that, LIP1=0.05*Lcol. This LIP1 value recommended to be close to Lp value you have in your column. If not close you can change the NIP=3,4,5,6,7... etc to check which one is closer to Lp.

If your column section is hardening I'd just use Nip=5 as recommended by others.

[Fahimeh]

IsmBaha wrote:
> Integration point weight wIP defines the integration length of each integration
> point, as LIP = wIP*Lcol
> where LIP integration point length, Lcol is the height of the column element.
>
> For example, as I remember for NIP=5 with gauss-lobatto, 1st integration point weight
> was -> wIP1~= 0.05.
> So with that, LIP1=0.05*Lcol. This LIP1 value recommended to be close to Lp value you
> have in your column. If not close you can change the NIP=3,4,5,6,7... etc to check
> which one is closer to Lp.
>
> If your column section is hardening I'd just use Nip=5 as recommended by others.


Thanks so much for the advice.
Where can I find the weights of integration points? The value 0.05 seems very small compared to plastic hinge length in reality.
I appreciate you give me some reference on that.

[nimatehran]

Hello ;
Download this file

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

[IsmBaha]

Fahimeh,

You can check the Dr. Scott's paper on the different integration methods as "nimatehran" pasted above.

Integration weigths change with respect to Nip. So you can check the cases when Nip=3,4,5,6,7,8,9. Each will have different weight.
Or you can read through Dr. Scott's paper and manually change your weight yourself (which is tricky).

For example, to see what the weight of Nip=5 (of Gauss-Lobatto integration), just go to wikipedia and see the value. http://en.wikipedia.org/wiki/Gaussian_quadrature
You will see at the bottom for N=5, weight of 1st and 5th IP is 1/10=0.1. Since our elements have boundaries from 0 to 1 (not -1 to 1), this weight in your FB element will be 1/10/2=0.05.