This is right for softening behavior, since they are the same in accuracy. So why did you say that applying modified G-R to FBE is more reasonable? I think BWH element is more reasonable for softening behavior because of simplicity and clear concept.
For hardening, when applying modified G-R to FBE, all the IPs have opportunity to undergo inelasticity, so this will lead to inaccuracy since there are not enough IPs in the middle region. FBE with other normal integration scheme will be preferred for hardening.
the difference between the dispbeamcolumn and nonlinearbeamc
Moderators: silvia, selimgunay, Moderators
-
linguan118
- Posts: 140
- Joined: Sun Oct 03, 2010 11:36 pm
- Location: Hong Kong
Re: the difference between the dispbeamcolumn and nonlinearb
Research Assistant Professor, The Hong Kong Polytechnic University
guanlin@polyu.edu.hk
guanlin@polyu.edu.hk
-
wuhaoshrek
- Posts: 122
- Joined: Tue Oct 28, 2008 4:01 am
Re: the difference between the dispbeamcolumn and nonlinearb
linguan118 wrote:
> This is right for softening behavior, since they are the same in accuracy.
> So why did you say that applying modified G-R to FBE is more reasonable? I
> think BWH element is more reasonable for softening behavior because of
> simplicity and clear concept.
> For hardening, when applying modified G-R to FBE, all the IPs have
> opportunity to undergo inelasticity, so this will lead to inaccuracy since
> there are not enough IPs in the middle region. FBE with other normal
> integration scheme will be preferred for hardening.
What I mean is that I think use modified Radau can also take the hardening case into
consideration. For this case, I would like use 3 elements, 2 end elements with Modified
G-R method, and with several Lobatto points in the middle element. For BWH, it is straight
and efficient with softening case, but bring inaccuracy with hardening. What do you think?
> This is right for softening behavior, since they are the same in accuracy.
> So why did you say that applying modified G-R to FBE is more reasonable? I
> think BWH element is more reasonable for softening behavior because of
> simplicity and clear concept.
> For hardening, when applying modified G-R to FBE, all the IPs have
> opportunity to undergo inelasticity, so this will lead to inaccuracy since
> there are not enough IPs in the middle region. FBE with other normal
> integration scheme will be preferred for hardening.
What I mean is that I think use modified Radau can also take the hardening case into
consideration. For this case, I would like use 3 elements, 2 end elements with Modified
G-R method, and with several Lobatto points in the middle element. For BWH, it is straight
and efficient with softening case, but bring inaccuracy with hardening. What do you think?
-
linguan118
- Posts: 140
- Joined: Sun Oct 03, 2010 11:36 pm
- Location: Hong Kong
Re: the difference between the dispbeamcolumn and nonlinearb
Yes, I think so. It is not suitable for hardening behavior.
Research Assistant Professor, The Hong Kong Polytechnic University
guanlin@polyu.edu.hk
guanlin@polyu.edu.hk
-
wuhaoshrek
- Posts: 122
- Joined: Tue Oct 28, 2008 4:01 am
Re: the difference between the dispbeamcolumn and nonlinearb
linguan118 wrote:
> Yes, I think so. It is not suitable for hardening behavior.
Hi, linguan118,
I intended to use 1 element to model plastic hinge, with element length equal
to the actual PH length. 2 int. points is assigned to the end nodes, with weights
1.0 and a very small No. like 1E-10; I used to think it should be appropriate to
model the PH, compared with Modified G-R it is more straight forward, physically.
I used it in a real column simulation, and the outcome is good. BUT recently I tested
it with a simple bilinear model, with hardening the result make sense, but with
EPP or softening it the element can not be converged.
Could you make any comment on this, what's the problem with integration in this
way.
Thanks a lot.
> Yes, I think so. It is not suitable for hardening behavior.
Hi, linguan118,
I intended to use 1 element to model plastic hinge, with element length equal
to the actual PH length. 2 int. points is assigned to the end nodes, with weights
1.0 and a very small No. like 1E-10; I used to think it should be appropriate to
model the PH, compared with Modified G-R it is more straight forward, physically.
I used it in a real column simulation, and the outcome is good. BUT recently I tested
it with a simple bilinear model, with hardening the result make sense, but with
EPP or softening it the element can not be converged.
Could you make any comment on this, what's the problem with integration in this
way.
Thanks a lot.