rigidLink Beam

Post Reply
hr_park
Posts: 20
Joined: Tue Mar 31, 2020 7:44 am

rigidLink Beam

Post by hr_park » Tue May 26, 2020 3:10 am

Dear STKO Team

Sorry for asking too many questions :cry:

I want to make rigiLink beam on modeling.

To connect two nodes with rigid link beam, do I create "node to node link" in the interaction, and then generate rigidLink in the mp of the contraint? Do I need anything else?

After modeling like this, I opened the mpco file, and the rigidlink beam part looked a little strange, so I post a question.
3.png
3.png (86.61 KiB) Viewed 754 times
4.png
4.png (46.5 KiB) Viewed 754 times

STKO Team
Posts: 434
Joined: Tue Oct 29, 2019 8:45 am

Re: rigidLink Beam

Post by STKO Team » Tue May 26, 2020 4:36 pm

Dear user,
don't you worry, we are here to help.
You created rigid link in the right way, but deformed shape is a little bit strange.
If you share your model with us, we will check it.
Thank you

STKO Team
Posts: 434
Joined: Tue Oct 29, 2019 8:45 am

Re: rigidLink Beam

Post by STKO Team » Wed May 27, 2020 11:12 am

Dear user,

We had a look at your file and we did some changes to improve the convergence of the pushover analysis and now we managed to push it further up to a displacement of 0.6 units.

However the main reason for non-convergence is that you are using many materials that can suddenly fail and their strength can drop down to 0. This is very challenging for convergence. For example both concrete models you are using in the columns have a 0 residual stress! same for many MinMax materials...
With those models, the best choice is to either reduce the time step to very tiny values, or to provide some residual strength.

Here is a list of changes that we made in the attached file.
  1. Fiber cross section must have the -GJ option. In version 3.2.0 it is mandatory... Since you already provided a Torsion in the aggregator, I simply put the section -GJ to 1.0e-10.
  2. Some of your ElasticPPGap have the -damage option, but not the -eta option (default eta = 0). If you use the damage option, then the eta option is mandatory. So I checked them and left them to 0. (I know, this is a wrong documentation in OpenSees)
  3. In the analyses, I changed the test from Energy to DispIncr. Why? Since you are using Penalty, the Unbalance will never go to 0. But if you use Energy, keep in mind that the Energy comes from the inner product of the incremental displacement and the unbalance. So I suggest to always use DispIncr when using Penalty for constraints
  4. I changed the defaul algorithm to KrylovNewton, setting the -maxDim to 200, so that the tangent is computed only once per step (-maxDim, means how many iteration between two next updates of the tangent matrix...), this is much more stable for this kind of problems.
  5. Finally I used many more time steps... typical when you have these sudden load drops!
After all these changes, I managed to push the analysis further up to 0.6 units of target displacement. However, after the first load drop at about 0.15 units, there is a strange increase of the load multiplier. I think this is due to the way you modeled all those interaction. Now it's up to you to assess the correctness of the model in this regard.

hr_park_pushover_curve.png
hr_park_pushover_curve.png (32.08 KiB) Viewed 736 times
Attachments
FPS_fold_push.zip
(235.89 KiB) Downloaded 43 times

hr_park
Posts: 20
Joined: Tue Mar 31, 2020 7:44 am

Re: rigidLink Beam

Post by hr_park » Tue Jun 02, 2020 1:32 am

Dear STKO Team

Thank you for your kind and detailed reply!
Considering all the suggestions, I'll try modeling again
Thank you very much!

Post Reply