Error in RC Frame Analysis (ZeroLength)

Post Reply
ankurjain1992
Posts: 38
Joined: Mon Aug 03, 2020 5:38 pm

Error in RC Frame Analysis (ZeroLength)

Post by ankurjain1992 » Mon Aug 31, 2020 6:20 pm

What is the use of equalDOF command in STKO ?
I have seen one of your webinar for lumped plasticity of an RC frame.........you guys have not used equalDOF in that.....I have seen in one analysis that they have used equalDOF command.

I am attaching one of the file I downloaded from your Webinar mostly 7th one.....

Firstly, It is showing error when I run it...
Secondly, I tried to find Eigen value...but the resultant Eigen are coming wrong from this file..
Thirdly, Is it possible to do Pushover in this figure using ZeroLength element ?

Can you pls correct this file for me ?


Thanks in advance
building_lumped_zero_length.zip
(179.12 KiB) Downloaded 11 times

STKO Team
Posts: 321
Joined: Tue Oct 29, 2019 8:45 am

Re: Error in RC Frame Analysis (ZeroLength)

Post by STKO Team » Tue Sep 01, 2020 7:57 am

Dear user,

That file was built and run with a previous version of OpenSees. Now we use OpenSees 3.2 so probably something has changed in the new OpenSees version. But I think it's only some convergence setting inside the force beam column, because just changing some settings in the pushover analysis (algorithm -> Krylov, time step -> adaptive) it now runs.

Regarding your question about the equalDOF command: The equalDOF command is used to constrain two nodes to have the same value in one or multiple DOFs. You often see it in opensees examples when they want to attach a node to another in some DOFs. A lumped plasticity column is one example:
Immagine you have node 1 and node 2, and a zeroLength (plastic hinge) connecting them. Now this zeroLength has a nonlinear spring only about the Rx axis for example. In this case you use the equalDOF to constrain all the other DOFs.
Why didn't we use it in our example?
Because we are using the HingedBeam, which is an STKO automatism that automatically creates an assembly of 1 column + 2 zeroLength. Now the column can be arbitrarily oriented in space, and the zeroLength will have the same orientation.
This is a problem for the equalDOF command, because it only works in global coordinates...
So when we use the HingedBeam, and we would linke to equalDOF some DOFs, we simply use a stiff material in the zeroLength for that DOF. Of course the stiffness of that spring should be not too large, otherwise your matrix will be numerically ill-posed. what I typically do is to evaluate the internal beam stiffness for that DOF, and then I use a spring whose stiffness is 2 or 3 order of magniture larger..

Here is the modified file:
building_lumped_zero_length.zip
(182.76 KiB) Downloaded 12 times

ankurjain1992
Posts: 38
Joined: Mon Aug 03, 2020 5:38 pm

Re: Error in RC Frame Analysis (ZeroLength)

Post by ankurjain1992 » Tue Sep 01, 2020 6:20 pm

Thanks for the reply STKO Team....You guys are doing great work...

In this example I got some more doubts which I wanna ask..

1. Why you have assigned elasticBeamColumn element for all 4 Columns and ForceBeamColumn element for all 4 Beams ?

2. Can I use only ForceBeamColumn element for the whole Frame ? What difference will it make ?

3. How you have define the Moment of Elasticity as 12000 for the elasticBeamColumn element ?

4. How you have define Modulus of Elasticity of a Uniaxial material with material tag 1 ? The value of E for Elastic material you defined is 1E+13. Can I keep it 1E08 ? What difference will it make if I do so ?

5. For zeroLength element, how to decide which DOF we have to release by assigning a spric material and which one to assign stiff material. ?


Thank you in advance.

STKO Team
Posts: 321
Joined: Tue Oct 29, 2019 8:45 am

Re: Error in RC Frame Analysis (ZeroLength)

Post by STKO Team » Wed Sep 02, 2020 8:29 am

1. Why you have assigned elasticBeamColumn element for all 4 Columns and ForceBeamColumn element for all 4 Beams ?
Simply because in that webinar we wanted to show that you can use any kind of element as the inner element of the HingedBeam, both a fiber or an elastic beam. Of course in the practice you will use the elastic element because the nonlinearity is lumped in the hinges.
2. Can I use only ForceBeamColumn element for the whole Frame ? What difference will it make ?
Of course you can. But since you know that the inner element is elastic, it's simply much faster to use an elastic element rather then a force-based beam element. The force-based element is a general purpose (linear and/or nonlinear) finite element that is very effective when you are using nonlinear materials, because in that case the sectional properties will change (due to nonlinearity) during the analysis. With the force-based element, since it is based on equilibrium, you can use just 1 element per member, while with a displacement based element you would need more elements.
3. How you have define the Moment of Elasticity as 12000 for the elasticBeamColumn element ?
Since the internal element remains elastic, it cannot capture stiffness loss due to concrete cracking, so assigning the full Young's modulus of concrete would over-estimate the initial stiffness of the element. In that example I had to reduce the flexural stiffness (E*Iyy and E*Izz) at about 35% percent of the elastic one. So I decided to reduce the Young's moduls... However, reducing the Young's modulus would reduce also the axial stiffness (E*A) and this is not correct since in compression the cracking of concrete does not reduce stiffness... Now in this new version of STKO we introduced the stiffness modifiers, so you can use them to reduce only the flexural inertiea:
stiffness_mod.png
stiffness_mod.png (49.44 KiB) Viewed 289 times
4. How you have define Modulus of Elasticity of a Uniaxial material with material tag 1 ? The value of E for Elastic material you defined is 1E+13. Can I keep it 1E08 ? What difference will it make if I do so ?
E of conrete has order of magnitude = 4 (3.5E4)
A has order of magnitude = 5 (see the section properties)
Iyy and Izz have order of magnitude = 9 (see the section properties)
so E*I has order of magnitude = 4+9 = 13.
That material is used to simulate a rigid material to impose continuity in the DOFs where you don't have plastic hinges, so it had to be at least as stiff, or stiffer, to impose continuity. Make sure you do not make too stiffer otherwise it may give numerical problems... In this example you can use a couple of order or magnitudes more (say 1.0E15) but not more than that.
5. For zeroLength element, how to decide which DOF we have to release by assigning a spric material and which one to assign stiff material. ?
It depends on you. In this example I knew my columns will bend only about their local Z axis, that's why I applied the nonlinear spring only at the (local) 6th DOF of the zero length
hinge direction.png
hinge direction.png (290.26 KiB) Viewed 289 times

ankurjain1992
Posts: 38
Joined: Mon Aug 03, 2020 5:38 pm

Re: Error in RC Frame Analysis (ZeroLength)

Post by ankurjain1992 » Thu Sep 03, 2020 5:16 pm

Awesome guys ... You people are doing great work... Thank you so much for this beautiful explanation ...
and sorry for the delayed reply.

ankurjain1992
Posts: 38
Joined: Mon Aug 03, 2020 5:38 pm

Re: Error in RC Frame Analysis (ZeroLength)

Post by ankurjain1992 » Fri Sep 04, 2020 12:16 pm

Sir If i used Lumped plasticity approach for my model and I use the ElasticBeamColumn element then can I apply UDL load over the Beam or Lateral UDL load over the column....??

Because I read some where that for applying UDL load and all, distributed plasticity approach is only good and not Lumped plasticity approach ?


Thanks in advance....

STKO Team
Posts: 321
Joined: Tue Oct 29, 2019 8:45 am

Re: Error in RC Frame Analysis (ZeroLength)

Post by STKO Team » Mon Sep 07, 2020 8:26 am

Of course you can apply uniform distiributed loads on elastic beam columns using the lumped approach. Probably what they meant (wherever you read that sentence) is that, if your distributed load is so high that can lead to non-linearity in the middle of the element, of course you need distributed plasticity to capture that.
lumped plasticity is good when you expect the plastic hinges to appear only at the beam-ends

Clarabella
Posts: 31
Joined: Wed Jul 08, 2020 3:50 pm

Re: Error in RC Frame Analysis (ZeroLength)

Post by Clarabella » Mon Sep 07, 2020 8:31 am

Sir If i used Lumped plasticity approach for my model and I use the ElasticBeamColumn element then can I apply UDL load over the Beam or Lateral UDL load over the column....??
yes you can use UDL

Post Reply